Pettit (2013) On The People's Terms - A Republican Theory and Model of Democracy

more information – www.cambridge.
org/9781107005112
On the People’s Terms
According to republican political theory, people’s freedom as persons

requires that they be publicly protected against subjection or
domination in the exercise of basic liberties. But there is no public
protection without a coercive state and that raises a problem since, by all
accounts, coercion takes away from the freedom of the coerced. In
addressing this problem, Philip Pettit argues that state coercion does not
involve subjection or domination if people share equally in democratic
control of the direction it takes. He proposes a normative theory of
democracy under which the goal is to ensure that political coercion is
non-dominating and, linking philosophy with policy, he supplements
the theory with a realistic model of institutions that might promote that
goal. On the People’s Terms is an original account of the rationale and
organization of democracy, offering a new direction for democratic
thought. It fully lives up to the high ideals of the Seeley Lectures.
Philip Pettit is L. S. Rockefeller University Professor of Politics and

Human Values at Princeton University and also Distinguished
Professor of Philosophy at the Australian National University. His
books include The Common Mind; Republicanism; Rules, Reasons and
Norms; and Made with Words: Hobbes on Language, Mind and Politics.
Amongst his recent co-authored books are The Economy of Esteem, with
Geoffrey Brennan; A Political Philosophy in Public Life, with Jose Marti;
and Group Agency, with Christian List. A collection of papers on his
work, Common Minds: Themes from the Philosophy of Philip Pettit,
appeared in 2007.
the seeley lectures
The John Robert Seeley Lectures have been established by the University
of Cambridge as a biennial lecture series in social and political studies,
sponsored jointly by the Faculty of History and the University Press. The
Seeley Lectures provide a unique forum for distinguished scholars of
international reputation to address, in an accessible manner, themes of
broad and topical interest in social and political studies. Subsequent to their
public delivery in Cambridge the University Press publishes suitably
modified versions of each set of lectures. Professor James Tully delivered the
inaugural series of Seeley Lectures in 1994 on the theme of Constitutionalism
in an Age of Diversity.
The Seeley Lectures include
(1) Strange Multiplicity: Constitutionalism in an Age of Diversity

j a m e s t u ll y
i s b n 978 0 521 47694 2 (paperback)
Published 1995
(2) The Dignity of Legislation
jeremy waldron
i s b n 978 0 521 65092 2 (hardback) 978 0 521 65883 6 (paperback)
Published 1999
(3) Woman and Human Development: The Capabilities Approach
m a r t h a n u ss b a u m
Published 2000
(4) Value, Respect and Attachment
joseph raz
Published 2001
(5) The Rights of Others: Aliens, Residents and Citizens
s e y la b e n h a b i b
Published 2004
(6) Laws of Fear: Beyond the Precautionary Principle
cass r. sunstein
Published 2005
(7) Counter Democracy: Politics in an Age of Distrust
p i e r r e r o s a n v a l lo n
Published 2008
(8) On the People’s Terms: A Republican Theory and Model of Democracy
philip pettit
Published 2012
ON THE PEOPLE’S TERMS
A Republican Theory and Model of Democracy
PHILIP PETTIT
cambridge university press
Cambridge, New York, Melbourne, Madrid, Cape Town,
Singapore, São Paulo, Delhi, Mexico City
Cambridge University Press

The Edinburgh Building, Cambridge cb2 8ru, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9781107005112
© Cambridge University Press 2012
This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2012
Printed and bound in Great Britain by the MPG Books Group
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data

Pettit, Philip, 1945–
On the people’s terms : a republican theory and model of democracy / Philip Pettit.
p. cm. – (The Seeley lectures)
isbn 978-1-107-00511-2 (hardback)
1. Republicanism. 2. Political science – Philosophy. 3. Democracy. 4. State, The. I. Title.
jc423.p432 2012
321.80 6–dc23
2012020514
isbn 978-1-107-00511-2 Hardback

isbn 978-0-521-18212-6 Paperback
Cambridge University Press has no responsibility for the persistence or

accuracy of URLs for external or third-party internet websites referred to
in this publication, and does not guarantee that any content on such
websites is, or will remain, accurate or appropriate.
For Rory and Owen, in another season.
Contents
List of tables page x

Acknowledgements xi
Introduction. The republic, old and new 1

1 Freedom as non-domination 26
2 Social justice 75
3 Political legitimacy 130
4 Democratic influence 187
5 Democratic control 239
Conclusion. The argument, in summary 293
References 311
Name index 329
Subject index 333
ix
Tables
1.1 Variations in hindrance page 35

1.2 Modes of interference 56
3.1 The demands of justice and legitimacy 140
4.1 Illustrating the discursive dilemma 192
4.2 Mechanisms of representation 199
x
Acknowledgements
This book has been long in the making. I decided to give more attention to
the relationship between republican political theory and democratic theory
soon after publishing Republicanism: A Theory of Freedom and Government
in 1997 (Pettit 1997c) and I wrote some papers on the topic in the years
immediately following. I worked at developing an overall view of repub-
licanism and democracy in the Pufendorf Lectures in Philosophy at the
University of Lund in 2005, in the Albertus Magnus Lectures in Philosophy
at the University of Cologne in 2009 and, finally, in the Seeley Lectures in
Political Theory at the University of Cambridge in 2010. I wrote up a final
draft of the book while on a year’s research leave at the Center for Advanced
Study in the Behavioral Sciences at Stanford in 2010–11 and I reworked the
text in the course of a graduate seminar in Philosophy and Politics at
Princeton University in Fall Semester 2011. Late in the piece I presented
related material in the Frankfurt Lectures in Political Theory at the Goethe
University in Frankfurt, and in the Quain Lecture in Jurisprudence at
University College, London, and the intense discussions on each occasion
led to some final alterations. I am grateful to all those institutions for their
support, as well as to the John Guggenheim Foundation for the award of a
fellowship in 2010–11.
In the course of the years working on the text I built up an enormous
number of personal, as well as institutional, debts. Since I cannot hope to
acknowledge all my creditors, I hope just a few words of thanks will suffice. I
am grateful for the valuable contributions of my audiences and hosts at the
institutions where I lectured or gave seminars on the topic, for the contri-
butions of the graduate students with whom I have been fortunate to work
on related topics, and for the wholly invaluable commentary and criticism
that was provided week after week by the students who participated in the
Princeton seminar in 2011. I must also thank those who took the trouble to
read the full text and sent often extremely insightful and helpful comments:
David Plunkett, who used it in a course at UCLA; my colleague, Annie
xi
xii Acknowledgements
Stilz, who was a selfless, illuminating critic; and especially Niko Kolodny,
who wrote as a temporarily anonymous reviewer for Cambridge University
Press and forced me to reconsider a long list of issues.1
Sustained interaction with those Princeton colleagues associated with the
University Center for Human Values has been absolutely crucial in shaping
my views; the relevant players, on any count, have to include Anthony
Appiah, Chuck Beitz, Marc Fleurbaey, Dan Garber, Liz Harman, Nan
Keohane, Melissa Lane, Stephen Macedo, Victoria McGeer, Jan Werner
Mueller, Alan Patten, Gideon Rosen, Kim Scheppele, Peter Singer, Michael
Smith and Jeff Stout. There are many others with whom I have had
exchanges, some of them quite extended, on the topics covered. The list
includes John Braithwaite, Geoffrey Brennan, Annabel Brett, Ian Carter,
John Ferejohn, Rainer Forst, John Gardner, Moira Gatens, Bob Goodin,
Alan Hayek, Kinch Hoekstra, Istvan Hont, Duncan Ivison, Frank Jackson,
Susan James, Matthew Kramer, Martin Krygier, Chandran Kukathas,
Cecile Laborde, Niki Lacey, George Letsas, Christian List, Frank Lovett,
Jenny Mansbridge, Jose Marti, Josh Ober, Michael Otsuka, Paul Patton,
Joseph Raz, Magnus Ryan, Wojciech Sadurski, Tim Scanlon, Amartya Sen,
Ian Shapiro, Quentin Skinner, David Soskice, Nic Southwood, John
Tasioulas, Larry Temkin, Richard Tuck, Laura Valentini, Miguel Vatter,
Jeremy Waldron, Albert Weale, Barry Weingast, Jo Wolff and the late Iris
Marion Young.
The various people listed have each played an important role in pushing
and pulling me around the arena of debates covered in the book. My thanks
to all for their interest and their engagement and my apologies, of course, for
what many of them will take – or, as I would naturally like to think,
mistake – for sheer obstinacy. I owe a special debt to Quentin Skinner,
who is responsible for having introduced me to the potential of republican
ideas. And I owe an extra-special debt to Victoria McGeer, with whom
conversation and life never flag. Finally, I should express my gratitude to
Richard Fisher of Cambridge University Press for the enormous encourage-
ment and support that he offered from the time I was first invited to present
the Seeley Lectures.
1
N. Kolodny, in his unpublished paper, ‘Rule Over None: Social Equality and the Value of Democracy’,
offers an alternative way of developing some of the ideas that are basic to the approach taken in this book
and this was a source of constant challenges as I prepared the final version of my own text.
Introduction. The republic, old and new
the project
Every philosophy of the good society starts with an account of the canonical
complaint that the state should help to put right: the evil that the society
should drive out by means of political organization and initiative. The
complaints targeted for political rectification come in two broad families.
On the one side, personal afflictions like misery or poverty or inequality; on
the other, social failures like division or disorder or perhaps an excess of
customary restriction.
The more personal complaints generate a powerfully motivating agenda,
since most of us would rejoice in a state that silenced them. But these
complaints are liable to seem politically over-demanding. While it would be
good to be rid of misery or poverty or inequality, not everyone will agree
that the state could, or should, be given the job of dealing with them. The
removal of the less personal evils is not politically over-demanding in the
same way, for most people will think that the state is able to remedy such
failures. But these complaints may fail to motivate appropriately: their
rectification falls short of what many of us feel that we in a politically
organized society can and should collectively provide for our members.
Republican philosophy identifies a complaint that is meant to be at once
personally motivating and politically feasible. It indicts the evil of subjection
to another’s will – particularly in important areas of personal choice – as an ill
that we all recognize and recoil from and at the same time as an ill that the
state is well placed to deal with. I shall be arguing in the course of this book
that such subjection can be effectively corralled and reduced, though certainly
not wholly eliminated, by means of political initiative. And yet it takes only a
little imagination to realize just how repellent this subjection can be.
Think, by way of exercising such imagination, of how you would feel as a
student if you depended for not failing a course on the whim of an
instructor. Or as a wife if you had to rely on the mood of your husband
1
2 Introduction. The republic, old and new
for whether you could enjoy an unmolested day. Or as a worker if you hung
on the favour of a manager for whether you retained your job. Or as a debtor
if you were dependent on the goodwill of a creditor for whether you had to
face public ignominy. Or as someone destitute if you had to cast yourself on
the mercy of others just to survive or maintain your family. Or think about
how you would feel as the member of a cultural minority if you had to rely
on the humour of majority groups for whether you escaped humiliation; or
as an elderly person if you depended on escaping the notice of youth gangs
for walking safely home; or as a citizen if you were dependent on winning
the favour of some insider group for whether you or your kind ever caught
the eye of government.
It is a commonplace in most cultures that such involuntary exposure to the
will of others is inherently troubling and objectionable. Even when those
others do not exercise their power in actual interference, the very dependency
involved is something from which we naturally recoil. The possible modes of
subjection are many and diverse, as these examples already testify, but it
should be clear that the state is capable of curtailing them in various ways.
Without assuming the cast of a Leviathan in their lives, it can assure its people
of a level of protection, support and status that frees them from at least the
more egregious forms that such dependency can take.
Already in classical, republican Rome, the evil of subjection to the will of
others, whether or not such subjection led to actual interference, was identi-
fied and indicted as the iconic ill from which political organization should
liberate people, in particular those in the fortunate position of citizens. It was
described as the evil of being subject to a master, or dominus – suffering
dominatio – and was contrasted with the good of libertas, or ‘liberty’. The
accepted wisdom was that people could enjoy liberty, both in relation to one
another and to the collectivity, only by being invested with the power and
status of the civis, or ‘citizen’. Being a free person became synonymous with
being sufficiently empowered to stand on equal terms with others, as a citizen
amongst citizens (Wirszubski 1968: Chapter 1).
The idea that citizens could enjoy this equal standing in their society, and
not have to hang on the benevolence of their betters, became the signature
theme in the long and powerful tradition of republican thought. Familiar
from its instantiation in classical Rome, the idea was reignited in medieval
and Renaissance Italy; spread throughout Europe in the modern era,
sparking the English Civil War and the French Revolution; and inflamed
the passions of England’s American colonists in the late eighteenth century,
leading to the foundation of the world’s first modern democracy. With
citizenship becoming more and more inclusive as a category, the idea was
The project 3
that the state could provide for all citizens in such a measure that they would
each be able to walk tall, live without shame or indignity, and look one
another in the eye without any reason for fear or deference.
The recent revival of republican thought is built on this idea that there is
an ideal for the state to promote – freedom understood as non-domination –
that is both personally motivating and politically implementable. Freedom
in this sense is not meant to be the only value in life, or the only value that
ultimately matters. The claim is merely that it is a gateway good, suited to
guide the governments that people form and sustain. Let government look
after the freedom of citizens in this sense, so the line goes, and it will also
have to look after a plausible range of other goods and do so at a plausible
level of provision. It will have to guard against division and disorder and
intrusive regulation and it will have to provide in a decent measure against
misery and poverty, unfairness and inequality.
This book joins a growing body of contributions in political theory that
are guided by the republican ideal and more generally by the republican
tradition of thought.1 While the volume offers an outline history of the
tradition, an analysis of freedom as non-domination, and an account of
what the ideal requires by way of social justice in people’s relationships with
one another, the main focus is elsewhere. It is on what the ideal demands by
way of political legitimacy in the relationships between citizens and their
state. The book argues that while the state has to guard people against
private domination – that is, the requirement of social justice – it also needs
to guard against itself practising a form of public domination. The require-
ment of guarding against public domination, thereby delivering political
legitimacy, turns out to demand a rich array of popular controls over
government: in effect, a distinctive form of democracy. It enables us to
explain why and how government should be forced, in the title of the
volume, to operate on the people’s terms.
1
The recent movement, as I think of it, began from the historical work of Quentin Skinner (1978) on
the medieval foundations of modern political thought, and from his subsequent articles in the 1980s on
figures like Machiavelli, who wrote within the republican tradition identified by John Pocock (1975).
An up-to-date list of English works in contemporary republican thinking should include these books:
Pettit (1997c); Skinner (1998); Brugger (1999); Honohan (2002); Viroli (2002); Maynor (2003); Lovett
(2010); Marti and Pettit (2010); McGilvray (2011); these collections of papers: Van Gelderen and
Skinner (2002); Weinstock and Nadeau (2004); Honohan and Jennings (2006); Laborde and Maynor
(2007); Besson and Marti (2008); Niederberger and Schink (2012); and a number of studies that
deploy the conception of freedom as non-domination, broadly understood: Braithwaite and Pettit
(1990); Richardson (2002); Slaughter (2005); Bellamy (2007); Bohman (2007); Laborde (2008); White
and Leighton (2008).
This theory of democracy, which takes final shape in Chapter 3, will not
be very persuasive unless we can offer at least a rough model of how it might
be institutionally realized. That is what the final two chapters provide.
While the model developed there may be rejected or amended by many
who still want to stick with the basic republican theory, it should at least
help to show that the theory is not institutionally infeasible.
The model developed in chapters 4 and 5 suggests that democracy operates
at its best in a dual process involving, in the short haul, the exercise of popular
influence over government and, in the long haul, the imposition of a popular
direction on government. As a result of the short-term electoral and contest-
atory influence that democracy can give them, so the idea goes, the people
gain the power to force government over the longer term to conform to widely
accepted norms of policy-making. The combination of these two processes,
each with its own temporal register, can ensure that the demos, or ‘people’,
enjoy a significant degree of kratos, or ‘power’, over the laws that govern and
shape their lives, thereby avoiding public domination. The model offers a
picture of how public institutions might serve to implement the republican
version of the democratic ideal, giving people channels of influence that
conjoin to form a river of popular control. Readers who are interested in
this model, rather than in the republican theory on which it is based, might go
directly to the final two chapters, using the propositional summary offered in
the Conclusion to orientate their reading.
In the remainder of this Introduction, I outline the main ideas in the
historical tradition of republican thought, distinguishing them from liberal
and communitarian ideas, and explaining how I make use of them in the
philosophical argument that follows. That argument begins in Chapter 1 with
an account of the republican ideal of freedom as non-domination, and
continues in the following two chapters with the theory of social justice and
political legitimacy that republicanism would support. As already suggested,
social justice constrains the relations that the members of a society should
have with one another, and political legitimacy the relations that they should
have with their government and, more broadly, their state. The theory of
republican legitimacy turns out to offer a theory of democracy, since it
requires a very specific form of equally shared, popular control over govern-
ment. Having developed that theory in Chapter 3, I then go on in chapters 4
and 5 to outline the dual-aspect model of the sort of democracy required.
Even where it covers ground that I have traversed elsewhere, my pre-
sentation of republican history and theory has shifted somewhat as a result
of the many recent discussions of these topics. The historical outline in this
Introduction coheres with the story of republican development that I have
Three core ideas 5
presented in earlier writings, building on the work of Quentin Skinner, but
it sharpens the contrast between the Italian–Atlantic republicanism with
which I identify and the form of republican thought introduced by
Rousseau. The argument in the first three chapters is broadly faithful to
lines of thinking I have defended in other works but is novel on a number of
counts. It builds the discussion around the distinction between freedom of
choice and freedom of the person. It uses an analysis of the fundamental
liberties, and of their grounding in public norm and law, to articulate the
ideal of freedom as a person and the requirements of republican justice. And
it develops a theory of republican legitimacy at proper length, marking it off
from the theory of republican justice on the one side and non-republican
theories of legitimacy on the other.
three core ideas

Three ideas stand out as landmarks on the terrain of traditional republican
thought. While the ideas received different interpretations and emphases
in different periods and amongst different authors, they constitute points
of reference that were recognized and authorized by almost everyone
down to the late eighteenth century who has a claim to belong to the
tradition.
The first idea, unsurprisingly, is that the equal freedom of its citizens, in
particular their freedom as non-domination – the freedom that goes with
not having to live under the potentially harmful power of another – is the
primary concern of the state or republic. The second is that if the republic is
to secure the freedom of its citizens then it must satisfy a range of constitu-
tional constraints associated broadly with the mixed constitution. And the
third idea is that if the citizens are to keep the republic to its proper business
then they had better have the collective and individual virtue to track and
contest public policies and initiatives: the price of liberty, in the old
republican adage, is eternal vigilance.
The mixed constitution was meant to guarantee a rule of law –
a constitutional order – under which each citizen would be equal with
others and a separation and sharing of powers – a mixed order – that would
deny control over the law to any one individual or body. The contestatory
citizenry was the civic complement to this constitutional ideal: it was to be a
citizenry committed to interrogating all the elements of government and
imposing itself in the determination of law and policy. These institutional
measures were taken to be essential for organizing a government that would
promote the equal freedom of citizens without itself becoming a master in
their lives – in other words, that would protect against private forms of
domination without perpetrating public forms.2
Freedom as non-domination, the mixed constitution and the contest-
atory citizenry were all represented in Roman republican thought and
practice, and they were articulated in different ways amongst the many
writers who identified with Roman institutions. These authors included the
Greek-born historian, Polybius, the orator and lawyer, Marcus Tullius
Cicero, and the native Roman historian, Titus Livius or, as we know him,
Livy. While they drew freely on earlier Greek sources, including Plato and
Aristotle, they were united in the belief that it was Rome that first gave life
and recognition to the key republican ideas.3
Leading thinkers in medieval and Renaissance Italy drew heavily on
Polybius, Cicero and Livy when, more than a thousand years later, they
reworked the republican ideas in seeking a political philosophy that would
reflect the organization and experience of independent city-states like
Florence and Venice (Skinner 1978). The neo-Roman framework of
thought that they crafted in the course of this exercise – in particular the
framework outlined in Nicolo Machiavelli’s Discourses on Livy – served in
turn to provide terms of political self-understanding for northern European
countries that resisted or overthrew absolute monarchs.4 These included the
Polish republic of the nobles in the sixteenth and seventeenth centuries, the
seventeenth- and eighteenth-century Dutch republic and the English
republic of the 1640s and 1650s.
While the English republic was the shortest lived of these regimes, it
had the widest influence and the deepest impact. The republican ideas
2
There are three ways, according to most contemporary normative theories, in which a government and
state might fail to be satisfactory (Fukuyama 2011). It might fail to operate impartially by systematically
favouring members of a particular grouping, like a family or tribe. It might operate impartially but fail
to operate according to established, stable rules in decision-making; that is, it might be ad hoc or
capricious, rather than constitutional. Or it might operate impartially and constitutionally but fail to
be accountable to its subjects. The first danger introduces partial, as distinct from impartial, rule; the
second particularistic rule, as distinct from constitutional rule – the rule of law; and the third
paternalistic rule, rather than accountable rule. We might say that in the republican tradition the
mixture of the mixed constitution is meant to ensure impartial rule, the constitutionalism of the mixed
constitution to ensure constitutional rule, and the contestatory character of the citizenry to ensure
accountable rule.
3
Eric Nelson (2004) has identified a Greek tradition in later republican thought that coexisted with the
neo-Roman tradition in which I am interested. I do not give attention to this tradition here.
4
For a vigorous and impressive argument that Machiavelli gave the contestatory element such emphasis
that he should be seen as a distinctive figure in the tradition – a radical democrat rather than an
aristocratic republican – see McCormick (2011). Chapter 6 of that book takes me to task for not being
more Machiavellian in that sense and I hope that the current work may help to counter its depiction of
the republicanism I espouse as being aristocratic in character.
Three core ideas 7
that emerged in the thought of defenders such as James Harrington, John
Milton and Algernon Sidney became a staple of political thought in
eighteenth-century Britain and America, albeit often adapted to make
room for a constitutional monarchy (Raab 1965). And they were incorpo-
rated deeply, if not always overtly, into the enormously influential work of
the Baron de Montesquieu (1989) on The Spirit of the Laws. However
differently interpreted or applied, the ideas were more or less common
property to the Whig establishment in eighteenth-century Britain; to their
Tory opposition, at least as that was formulated by the 1st Viscount
Bolingbroke (Skinner 1974); to radical Whigs who were a constant sting
in the side of every establishment (Robbins 1959); and, of course, to the
American colonists, and their British apologists, who came to feel that the
Westminster Parliament ruled its colonies in a manner that betrayed
the ‘commonwealthman’ or republican heritage (Bailyn 1967; Reid 1988;
Sellers 1995). Republican ideas provided the framework for the arguments
made in support of the cause of American independence over the 1760s
and 1770s – including arguments made by contemporary English sup-
porters such as Richard Price (1991) and Joseph Priestley (1993) – and for
the arguments put forward in the constitutional debates of the 1780s
between federalists and anti-federalists (Madison, Hamilton and Jay
1987; Ketcham 2003).
Amongst the three ideas associated with the republican tradition, the
conception of freedom as non-domination is the most distinctive. If you are
to enjoy freedom as non-domination in certain choices, so the idea went,
then you must not be subject to the will of others in how you make those
choices; you must not suffer dominatio, in the word established in Roman
republican usage (Lovett 2010: Appendix i). That means that you must not
be exposed to a power of interference on the part of any others, even if they
happen to like you and do not exercise that power against you. The mere
fact that I can interfere at little cost in your choices – the mere fact that I can
track those choices and intervene when I like – means that you depend for
your ability to choose as you wish on my will remaining a goodwill. You are
not sui juris – or not ‘your own person’ – in the expression from Roman law.
You are unfree, as the eighteenth-century republican Richard Price (1991:
26) explained, because your access to the options will depend on an
‘indulgence’ or an ‘accidental mildness’ on my part. To quote from a
seventeenth-century republican, Algernon Sidney (1990: 17, 304), freedom
in this tradition requires ‘independency upon the will of another’ – an
‘exemption from dominion’ in relations with others. In an equivalent slogan
from a popular eighteenth-century tract, ‘Liberty is, to live upon one’s own
terms; slavery is, to live at the mere mercy of another’ (Trenchard and
Gordon 1971: ii, 249–50).
In arguing that the state should be concerned in the first place with the
equal freedom of its citizens, republicans held that citizens should each be
assured of enjoying non-domination in a sphere of choice that came to be
described as that of the fundamental or basic liberties (Libourne 1646; Pettit
2008a). This might be identified, in contemporary terms, with the sphere of
choice required for being able to function in the local society (see Sen 1985;
Nussbaum 2006). They thought that a state organized under a mixed
constitution, and disciplined by a contestatory citizenry, was the best
hope of promoting this ideal.
The citizenry was traditionally restricted to mainstream, usually pro-
pertied, males and, under the republican vision, a citizen would be a liber,
or a ‘free-man’, insofar as he enjoyed sufficient power and protection in
the sphere of the basic liberties to be able to walk tall amongst others and
look any in the eye without reason for fear or deference. John Milton
(1953–82: viii, 424–5) captured the idea nicely in arguing that, in a ‘free
Commonwealth’, ‘they who are greatest . . . are not elevated above their
brethren; live soberly in their families, walk the streets as other men, may
be spoken to freely, familiarly, friendly, without adoration’. In the vision
of contemporary republicans, this ideal ought to be extended to an
inclusive citizenry; freedom as non-domination ought to be secured for
all more or less permanent residents, independently of gender or property
or religion.
the liberal opposition

These remarks constitute the broadest of brush strokes but the pattern that
they project on to the intellectual and institutional swirl of political history is
not a capricious imposition; it is not like the figures that we may think we see
in the snow, or the clouds, or the stars. The Italian–Atlantic tradition that we
have been describing constitutes a firm reality that endured across classical,
medieval and modern times (Pocock 1975). The best sign of its independent
importance is that the set of ideas described constituted a vivid and salient
target of attack for those who espoused a rival way of thinking about liberty – a
way of thinking that eventually gave rise to classical liberalism – in the later
eighteenth century. The main figures here were utilitarian thinkers like Jeremy
Bentham and William Paley (Pettit 1997c: Chapter 1).5
5
For a somewhat divergent reading see Kalyvas and Katznelson (2008).
The liberal opposition 9
Hobbes had already set himself against the republican way of thinking
about freedom in the 1640s, offering a somewhat complex alternative –
though not one that had a lasting influence – in its place (Pettit 2008c:
Chapter 8; Skinner 2008b). Without explicitly drawing on that earlier
precedent, Bentham reported in the 1770s ‘a kind of discovery I had made,
that the idea of liberty, imported nothing in it that was positive: that it was
merely a negative one: and . . . accordingly I defined it “the absence of
restraint”’ (Long 1977: 54). On this definition you are free in a given
choice just insofar as others do not restrain the selection of any option: not
the option you actually prefer, for sure, but also – at least on what came to
be the standard reading (Berlin 1969: xxxix; Pettit 2011b) – not any option
you might have preferred but didn’t. This conception makes the absence
of actual interference on the part of others enough for the freedom of a
choice; it does not require the absence of a power of interference on their
part. Even though you avoid interference only because of my being good-
willed and indulgent, then – even though you can choose as you wish only
because I permit it – still, on this new approach, that is enough to make
you free.
Bentham and Paley and their ilk were reformers, committed to having
the state cater for the freedom – and more generally for the utility or
happiness – of the whole population, not just the freedom of mainstream,
propertied males that government had traditionally protected. So why
would they have weakened the ideal of freedom so that it is not compro-
mised by having to live under the power of another, only by active
interference? My own hunch is that it was more realistic to argue for
universal freedom if freedom was something that a wife could enjoy at the
hands of a kind husband, a worker under the rule of a tolerant employer – in
other words, if it was an ideal that, unlike universal freedom as non-
domination, did not require redressing the power imbalances allowed
under contemporary family and master–servant law (Pettit 1997b:
Chapter 1). It may be for this reason that Paley (2002: 315) described
freedom in broadly the republican sense – an idea that ‘places liberty in
security’, in accord with ‘common discourse’ (313) – as one of those versions
of ‘civil freedom’ that are ‘unattainable in experience, inflame expectations
that can never be gratified, and disturb the public content with complaints,
which no wisdom or benevolence of government can remove’.
The rejection of freedom as non-domination raised a question about the
linked ideas of a mixed constitution and a contestatory citizenry. Those
devices were required on the traditional, republican way of viewing things
because they were supposed to ensure that when the republic makes laws
that protect its citizens against private domination, it does not impose those
laws in a publicly dominating manner. The idea was that if the interference
imposed by the state is not under the control of any single agency, as the
mixed constitution more or less guarantees, and if it is itself subject to the
control of those on whom it is imposed, as a contestatory citizenry would
ideally ensure, then it will not be dominating. It will not involve subjecting
people to the unchecked will of a distinct, independent agent. It will be a
non-dominating – or, as it was often called, a non-arbitrary – form of
interference.
Once freedom came to be construed as non-interference, however, it
was no longer clear why such constraints were necessary. Every system of
law coerces and penalizes its subjects – and every system of law presup-
poses taxation – so that there is no law without interference. If freedom
means non-interference, therefore, then there is no freedom-based
requirement to make the interference non-dominating, as the mixed
constitution and the contestatory citizenry promised to do. The best
system will be that in which there is the least overall interference, public
or private. And it may just be that the best system is one in which a
benevolent despot coerces people so that they don’t interfere with one
another, yet keeps the coercion it perpetrates to a minimum. William
Paley (2002: 314), surely Bentham’s most clear-headed associate,
embraced the point when he noted as early as 1785 that the cause of liberty
as non-interference might be as well served, in some circumstances, by
‘the edicts of a despotic prince, as by the resolutions of a popular assem-
bly’. In such conditions, he said, ‘would an absolute form of government
be no less free than the purest democracy’ – and, by his lights, no less free
than the most classical republic.
While Bentham and Paley were mainly interested in advancing the
utilitarian programme, they shaped the way in which early nineteenth-
century liberals thought about freedom and the requirements of freedom.
We might define liberalism – somewhat tendentiously, in view of the many
meanings given to the term – as any approach to government that makes
freedom as non-interference paramount or central. And in that sense it
contrasts quite sharply with the republican approach in which freedom as
non-domination plays the central role. Liberalism in this sense may be
right-of-centre, as classical liberals or libertarians generally were, making
freedom as non-interference into the only concern of government. Or it
may be left-of-centre, making freedom as non-interference into just one of
government’s goals: perhaps a goal derived from the broader concern with
happiness, as in the case of utilitarians; perhaps a goal that is paired with a
The communitarian opposition 11
separate concern like equality, as in the case of John Rawls (1971, 1993,
2001), Ronald Dworkin (1978, 1986) and other egalitarians.6
No matter which form liberalism takes, it contrasts with republicanism on
how to understand freedom, a value to which each approach gives a prominent
place. On the republican construal, the real enemy of freedom is the power
that some people may have over others, whereas on the liberal understanding,
asymmetries in interpersonal power are not in themselves objectionable.
Right-of-centre liberalism is happy to tolerate such imbalances, so long as
active interference is avoided. And left-of-centre liberalism rejects them only
insofar as they have unwelcome effects on equality or welfare or whatever other
value is invoked to supplement the ideal of freedom.
Although there is a deep difference between the guiding ideals of the
approaches, the conflict between republicanism and liberalism should not
be overdrawn. For while Paley and others had seen that freedom as non-
interference does not strictly require a mixed constitution or a contestatory
citizenry, almost every form of liberalism has endorsed the main elements in
the idea of the mixed constitution and given some recognition to the
contestatory role of citizens.7 Liberal ideals like the rule of law, the separa-
tion of powers, and the liberties of speech and expression are reflections of
such earlier institutional ideals. Republicanism differs from liberalism in
espousing a more radical ideal of freedom, in arguing for a distinctive
connection between freedom in that sense and its twin institutional ideals,
and in giving a distinctive interpretation of those ideals, particularly that of a
contestatory citizenry. But nonetheless there are definite, discernible con-
tinuities between the traditions.8
the communitarian opposition

In the late eighteenth century the republican tradition was opposed, not just
by the emerging liberal approach to politics, but also by a tradition that
6
A further distinction within liberal doctrines in this sense is that between those in which freedom is
taken as a goal – I think, as indicated at the end of Chapter 1, that freedom as non-domination should
be taken in this way (Pettit 1997c: Chapter 3) – and those in which the rights associated with freedom
are taken in non-consequentialist form as side-constraints; the best example of the latter approach is
Nozick (1974). For a discussion of consequentialist and non-consequentialist approaches in political
philosophy, see Pettit (2001d, 2012a).
7
Indeed Paley (2002: 331) himself embraced the ‘combination of the three regular species of govern-
ment’ that he claimed to find in England in the 1780s.
8
There might be some merit, for this reason, in adopting Richard Dagger’s (1997) usage and speaking of
the approach defended here – which resembles Dagger’s own – as republican liberalism or liberal
republicanism.
originates with Jean Jacques Rousseau. I describe this approach as commu-
nitarian, for reasons that will become clear shortly, though it is often described
as republican; it represents a broadly Continental form of republicanism, as
distinct from the Italian–Atlantic tradition that I have been characterizing.9
It is easy to mistake the two versions of republicanism, for as liberalism
came to displace traditional republicanism as the main ideology of the
English-speaking world, the name ‘republicanism’ came to designate the
new Rousseauvian doctrine. It is primarily with the Continental, commu-
nitarian version of the doctrine, rather than the Italian–Atlantic tradition,
that critics of liberalism like Hannah Arendt (1958, 1973) and Michael
Sandel (1996) seem to identify, for example.10 It is this version of repub-
licanism that is rejected, along with liberalism, in the work of Juergen
Habermas (1994, 1995).11 And it is this doctrine that self-described liberals
often focus on in arguing for the merits of their own approach (Brennan and
Lomasky 2006).
Rousseau espoused a version of freedom as non-domination or non-
dependency, giving it pride of place in the concerns of the state. He says that
if people are to be free, each must be ‘perfectly independent of all the others’
(Rousseau 1997: ii.2.2). And he maintains that that ‘which ought to be the
end of every system of legislation is . . . freedom and equality’, where
freedom is understood in the sense of non-dependency and equality is
valued ‘because freedom cannot subsist without it’ (Rousseau 1997:
ii.11.1). But while he remained faithful to Italian–Atlantic republicanism
in this respect, he broke dramatically with it in arguing against the mixed
constitution and the contestatory image of the citizenry.
Rousseau rejects the idea of the mixed constitution under the lead of
absolutists like Jean Bodin (1967) in the sixteenth century, and Thomas
Hobbes (1994a, 1994b, 1998) in the seventeenth.12 He accepted their argu-
ment that every state has to have a single, absolute sovereign, individual or
corporate, and that no such agency can operate, as envisaged in the mixed
9
For a fuller discussion of this communitarian form of republicanism, with reference to Kant as well as
Rousseau, see Pettit (2012b).
10
In discussing this conception of the republic in earlier work, in particular the conception as it appears
in Sandel, I have sometimes described it as neo-Athenian (Pettit 1998). I regret that usage now, since
as a matter of history – if not in later representations, which were deeply influenced by Polybius –
Athens had many of the characteristics of a mixed constitution; it was not a city ruled by an assembly
with Rousseauvian powers. See Hansen (1991), and for a summary account that emphasizes this
mixture see Chapter 2 of Dowlen (2008).
11
For the record, I think that Habermas’s own views come close to republican views, as I conceptualize
and defend them.
12
For a wonderful overview of the French tradition in the background of Rousseau’s thinking see
Keohane (1980).
constitution, on the basis of coordination between different, mutually
checking centres of power. Hobbes had argued that the mixed constitution
supports ‘not one independent commonwealth, but three independent
factions; nor one representative person, but three’ (Hobbes 1994b: 29.16.
See too 1998: 7.4, 1994a: 20.15). Rousseau strikes a similar note of derision in
attacking the champions of this centrepiece of republican thought: ‘they
turn the Sovereign into a being that is fantastical and formed of disparate
parts; it is as if they were putting together man out of several bodies one of
which had eyes, another arms, another feet, and nothing else’ (1997: ii.2.2).
Bodin and Hobbes had argued that the sovereign could be a monarch,
which was their own favourite candidate, an aristocratic committee, or a
committee-of-the-whole. But they allowed that such a sovereign might del-
egate much of the work of government to another agency. Rousseau held that
only a unanimously endorsed committee-of-the-whole could serve in the
sovereign role, on the grounds that an aristocracy or a monarchy would
impose an alien will on people (1997: i.4.4). And he argued that while the
members of the sovereign assembly have to rely on majority voting in order to
make their decisions (iv.2.7), they should deliberate and vote, at least in the
ideal, on the basis of the common interest. Only if the members act in that
way, he thought, can the assembly claim to be enacting a general will in which
all figure as equal objects of concern. The members of Rousseau’s assembly are
to think as citizens, focused impartially on their common interest, and they
are to vote for any measure they support, not out of personal or factional
motives, but on the impersonal, deliberative basis that ‘it is advantageous to
the State . . . that this or that opinion pass’ (iv.1.6).13
How is this impartiality likely to be achieved? Rousseau thinks that
insofar as the people in assembly have to decide only on matters of general
law – and not on administrative or adjudicative measures involving indi-
viduals (1997: iii.17.5) – they will operate under a veil of abstraction that
keeps personal or factional concerns out of the picture. They will consider
‘the subjects in a body and their actions in the abstract, never any man as an
individual or a particular action’ (Rousseau 1997: ii.6.6). And so they will
tend to be moved, he hopes, not by particularistic motives, but by an
impartial, egalitarian concern ‘with their common preservation, and the
general welfare’ (Rousseau 1997: iv.1.1). They will form their minds and cast
13
On this reading of Rousseau, the common good is not defined procedurally as whatever good is
supported by suitably motivated majority voting; rather it is the good that suitably motivated majority
voting might be expected to track. See Cohen (2010).
their votes on the basis of open deliberation and debate about the require-
ments of the common good.14
Rousseau betrayed the earlier tradition of republicanism in this concep-
tion of popular sovereignty. In his idealized republic, individuals are con-
fronted by the single powerful presence of ‘the public person’, which
‘formerly assumed the name City and now assumes that of Republic or
body politic’ (1997: i.6.10). While he envisages an ideal under which people
are independent of one another as private persons, in line with the repub-
lican conception of freedom, he thinks that this mutual independence is
attainable only at the cost of a form of submission to the public person –
specifically, to the general or corporate will of the public person – that
would have been wholly at odds with Italian–Atlantic sentiments. While
every citizen should ‘be perfectly independent of all the others’, he says, this
is only going to be possible insofar as each is ‘excessively dependent on the
City’ (Rousseau 1997: ii.12.3). The totally novel, consciously outrageous
assumption Rousseau introduces is that ‘each, by giving himself to all, gives
himself to no one’ (1997: i.6.8).
The rejection of the mixed constitution in favour of a popular, majoritarian
sovereign led Rousseau to reject also the contestatory role that was given to
citizens under traditional republicanism. Following Bodin and Hobbes he
emphasizes that the sovereign assembly of the people has to have absolute
power and not be subject to interrogation by citizens, at least when they speak
outside the assembly on the basis of their rights as individuals. Hobbes (1994b:
18.4) had said if subjects could ‘pretend a breach of the covenant made by the
sovereign . . . there is in this case no judge to decide the controversy’. And in
the same vein, Rousseau claims that ‘if individuals were left some rights . . .
there would be no common power who might adjudicate between them and
the public’ (1997: i.6.7). The idea is that there has to be one, final, spokes-
person on what the law is and that if the people could individually contest the
legislature’s decisions outside the assembly, there would have to be another
body to rule between them. Once the legislative assembly has spoken,
according to this picture, it falls to individuals to comply, not complain.15
Citizens are no longer invigilators of government, alert to any possible
14
Rousseau makes some remarks that seem to cast doubt on the value of public deliberation, but in
general, as Joshua Cohen (2010: 75–7, 171–2) argues, he clearly expects deliberation to have a major
role in the proceedings of the assembly.
15
But how, Rousseau asks, ‘are the opponents both free and subject to laws to which they have not
consented’? His response is ‘that the question is badly framed. The citizen consents to all the laws,
even to those passed in spite of him, and even to those that punish him when he dares to violate any of
them’ (iv.2.8).
misdoing and ready to challenge and contest the legislative, executive and
judicial authorities. They are law-makers, not law-checkers, generators of law,
not testers of law. They serve in the production of public decisions, not in
controlling for the quality of decisions proposed or made.
This viewpoint is deeply opposed to accepted republican doctrine and in
adopting it Rousseau is moved by a thought first floated by that great enemy
of the doctrine, Thomas Hobbes (1994b: 18.6). This is the idea that no law
supported by the general impartial will ‘can be unjust, since no man can be
unjust towards himself’ (Rousseau 1997: ii.6.7). Far from every law being a
fair target for civic critique and challenge, each comes draped in an authority
and majesty that brooks no individual opposition. Having been party to the
creation of the popular sovereign no one as an individual retains the right of
contesting the decisions of the collectivity, even if those decisions are ones
that the person argued against in assembly. In an extraordinary reversal of
received ideas, the cause of freedom as non-domination, from which
Rousseau starts, is now linked with a new, communitarian form of the
very absolutism that republican doctrine had always challenged. The people
or community gets to be sacralized, as it assumes the role of the popular,
incontestable sovereign, incapable of doing wrong to its own members.
While Rousseau’s wider theory is deeply at odds with Italian–Atlantic
republicanism, however – and while, as I believe, it is normatively unattrac-
tive – it has one powerfully appealing element. This is his ideal of a deliber-
ative assembly in which everyone is able to speak and vote on an equal basis,
and yet is required to speak and vote on the basis of a concern with the
common good.16 Elaborated influentially in the work of Juergen Habermas
(1984–9, 1994, 1995), the constraint requires participants in a deliberative
assembly to argue for their different proposals on the basis of ‘reasons that
are persuasive to all who are committed to acting on the results of a free and
reasoned assessment of alternatives by equals’ (Cohen 1989: 23). Such consid-
erations will include general reasons to do with what promises more peace or
order or prosperity for the society as a whole. But they can also include reasons
16
While Rousseau does not address the issue of how precisely the parties should deliberate with one
another, his insistence on their abstracting from personal and factional considerations can be taken to
mandate something like this constraint. In any case, the constraint ought to be congenial from a
Rousseauvian point of view. Setting it up as a formal requirement would serve as a defence against the
intrusion of personal and factional concerns in the decisions of the assembly. If the participants are
explicitly required to seek out mutually acceptable reasons in support of their proposals, that in itself
may serve to elicit an attachment to the viewpoint of the public (Elster 1986). And once it is salient to
everyone that everyone is expected to do this, it may become shameful for participants to display
concerns of a more personal or factional character (Brennan and Pettit 2004). The topic is central to
the discussion in Chapter 5.
put forward on behalf of a particular subgroup or individual. The consider-
ation in this case will not be just that the subgroup or individual is unsatisfied,
as if all others should be moved by that thought, but that the group or
individual suffers a disadvantage to which – purportedly by general criteria
that are endorsed across the society – no one should be exposed.
This deliberative constraint will play a role in our later discussions and, to
that extent, the position taken in this book has Rousseauvian linkages. But
the point I want to emphasize here is that the communitarian vision he
sponsored is nonetheless hostile to the central tenets of Italian–Atlantic
republicanism. Rousseau himself remained faithful to that tradition in his
continuing to embrace something like the ideal of equal freedom as non-
domination. But even this connection was broken amongst those who
identified with his vision in later generations. As his ideas washed around
political circles, this last vestige of republican thought was swept away in the
flood. It became natural to think, in a vulgarization of Rousseau’s own
analysis (Spitz 1995), that freedom is nothing more or less than the pos-
session or exercise of the right to participate in popular decision-making.
Benjamin Constant was probably the major figure in accomplishing this
final step, although he did so as a critic, not a defender. Himself attracted to
what was seen as a brand of liberalism – although one that in many respects
kept close to the older republicanism – he gave a famous lecture in 1819 that
described the supposedly ancient way of thinking about politics and free-
dom with which the liberal view has to compete (Constant 1988). According
to this ancient way of thinking, he says, the people in a commonwealth
constitute the sovereign, the role of citizens is to participate as officials or
electors in sovereign decision-making and – this is the alteration from the
strict Rousseauvian picture – freedom consists in nothing more or less than
the right to participate in such communal self-determination: the right to
live under a regime of law that you have a certain participatory or electoral
role in creating.17
With this twist, Rousseau became associated with a set of ideas that
contrasts in every dimension with the Italian–Atlantic tradition. This new
ideology replaced freedom as non-domination with freedom as participa-
tion. It replaced the institutional ideal of the mixed constitution with that of
a popular, absolutely sovereign assembly. And it replaced the ideal of a
contestatory people with that of a participatory legislature against which
17
Yiftah Elazar has persuaded me that while the eighteenth-century thinker, Richard Price (1991),
preserves other aspects of the Italian–Atlantic tradition, he had also begun to emphasize this self-
legislative theme.
individuals had no rights in their own name. This broadly communitarian
family of ideas assumed many different forms, for example in Kant’s
political philosophy (Pettit 2012b), as it became associated with the nation-
state rather than a city-state like Rousseau’s Geneva, as the body of citizens
proper was taken to be less inclusive than Rousseau imagined and, in a final
departure, as participation was allowed to become electoral rather than
legislative in character. But no matter what form it assumed, the approach
remained decidedly at odds with the Italian–Atlantic tradition from which
it had sprung.
The contrast between the approaches shows up nicely in their different
images of the free person or citizen. The free-man in the Italian–Atlantic
tradition, the liber of Roman thought, was someone who lived in his own
domain – the masculine form fits with the habits of the time – on terms that
he himself set. Within that domain he lived sui juris, as it was put in Roman
law, ‘under his own jurisdiction’. He did not operate in potestate domini, ‘in
the power of a master’, and he did not have to make his choices cum
permissu, ‘with permission’. He could act without fear or deference, being
protected and empowered in relation to others, and even in relation to the
very law that helped establish his position.18
The tradition emphasized, of course, that no one can expect to enjoy this
status outside a community in which citizens are powerful and vigilant
enough to keep tabs and impose checks on government. It is because they
‘attain unto liberty’ on the basis of such political engagement, as James
Harrington (1992: 75) puts it in the seventeenth century, that free citizens
are ‘able to live of themselves’. But while the status of the free person can exist
only in the presence of such public virtue, it consists, according to this way of
thinking, in the enjoyment of equal freedom as non-domination across a
range of significant, personal choices: the fundamental or basic liberties. These
choices have to be available in common to all citizens, and available on the
common basis of a protective and empowering law and culture.
18
Although self-described liberals do not ordinarily emphasize the need for contestatory virtue, they do
often endorse this ideal of the free person, preserving a deep commonality with Italian–Atlantic
republicanism. Isaiah Berlin (1969: lx) embraces the ideal when he says that freedom requires ‘an
area . . . in which one is one’s own master’; a domain where one ‘is not obliged to account for his
activities to any man so far as this is compatible with the existence of organized society’. This domain
is to constitute ‘a certain minimum area of personal freedom which must on no account be violated’
(Berlin 1969: 123). It is not clear how you can be your own master without enjoying non-domination;
to enjoy non-interference by grace of just the goodwill of another would seem to leave you well short
of the independence required (Pettit 2011b). But putting that problem aside, it is clear that on this
issue republicanism and liberalism may often converge.
The image of the free person assumes a different cast in the Rousseauvian
tradition. According to the new, communitarian, way of thinking, the free
person is the active political figure whose highest fulfilment consists in
participating with others, at whatever level of community, in activities of
shared deliberation and decision-making. In Hannah Arendt’s (1958) favourite
phrase, it is the citizen who embraces the vita activa rather than the vita
contemplativa – the active rather than the contemplative life – and in particular
an active life in the realm of public affairs. This rather romantic picture of the
tirelessly engaged public figure stands in stark opposition to the image of the
free-man in older republican thought, though it may echo the emphasis on
the need for a contestatory citizenry. Whereas Italian–Atlantic thought hails
the enjoyment of a publicly protected freedom in the domain of private life –
a freedom, in the republican picture, that enables you to stand equal with
others, not depending on anyone’s grace or favour – this new vision tends to
downplay private life in favour of public engagement.
building philosophically on republican history

As already mentioned, this book joins a growing body of literature in
seeking to build a contemporary political philosophy with ideas drawn
from the Italian–Atlantic tradition (for an overview see Lovett and Pettit
2009). The main idea employed in the construction is that of freedom as
non-domination, which is more fully explicated in the first chapter. But the
ideas of a mixed constitution and a contestatory citizenry also play a role, as
we shall see, at various stages in the unfolding argument.
Two of the three main topics addressed in normative political theory have
already been introduced: social justice in the relations amongst the citizens
of a state; and political legitimacy in the relations between the state and its
citizens. The third topic, which is not addressed in this book, is the
sovereignty that ought to be available to each of the different peoples that
states ideally represent.19 Republican political theory, understood as an
approach built around the ideas reviewed here, offers a distinctive approach
on all three fronts (Pettit 2013).
The republican theory of social justice would argue that the state should
establish equal non-domination for its citizens in relation to one another.
The republican theory of political legitimacy would argue that in the course
of providing for them in this way the state ought not to dominate its
19
This description of the ideals may not appeal to all. Other accounts would cast them as ideals of social
justice, political justice and international justice.
Building philosophically on republican history 19
citizens. And the republican theory of global sovereignty would argue that
the state ought to help establish an international order under which it and
its people are not dominated from without, whether by other states or by
multinational organizations (see Pettit 2010a, 2010c). In a slogan, the state
ought to be an internationally undominated, domestically undominating
defender of its citizens’ freedom as non-domination. In this book I shall be
looking in outline at the republican theory of social justice and in greater
detail at the republican theory of political legitimacy or, as it turns out to be,
the republican theory of democracy. But I shall have little or nothing to say
on the republican theory of international sovereignty.
Since the theory presented in the book is primarily philosophical in
character, it may seem surprising that I should link it closely with a historical
tradition, commenting at various points on the positions maintained by the
champions and the critics of that tradition. After all, the fashion in most
contemporary philosophical work is to emulate the sciences and cut free
from the baggage of history. So why emphasize the traditional provenance
of the ideas developed within the theory?
I try to sustain the historical linkage for a number of reasons. Construing
the ideal of freedom in terms of non-domination is likely to be more widely
acceptable if the construal can be given historical credentials, particularly
credentials in the work of writers admired on all sides of politics. Building a
political philosophy out of the demands of such an ideal in the context of a
contemporary, inclusive, society is likely to be more persuasive if the
enterprise can be linked with similar enterprises by other thinkers in other
times. And, more generally, the existence of a historical pedigree for an
approach taken in political philosophy is bound to give the approach more
intellectual plausibility. How likely is it, after all, that any one of us would
discover afresh a wholly novel ideal for political life?
While the tradition within which I pursue political philosophy is impor-
tant to me, however, the philosophical use of historical ideas should not be
mistaken for the proper historical investigation of those ideas and of the
context in which they arose. I think of what I am doing as interacting with
intellectual history by drawing on the work of experts in the area and
sometimes challenging them with philosophically motivated hypotheses as
to what was in the minds of their subjects. I do not think of the exercise as
history proper, or even history improper.20
20
The closest I have come to doing intellectual history is in my work on Hobbes’s theory of language,
thought and politics (Pettit 2008c). But even there I think of what I do as proposing a reading or
reflective equilibrium
The normative claims defended in the book should not be judged by
historical criteria, therefore, but rather by philosophical. The philosophical
methodology I follow is that which John Rawls (1971) described as reflective
equilibrium. The idea is to set out general principles for the domain
investigated, whether it be justice or legitimacy or sovereignty; to use
empirical assumptions to derive their implications for specific cases; to see
how those implications fit with what we find credible on reflection; and to
go back and forth in the search for adjustments at either end that can
promote overall coherence.
The methodology is even relevant to the interpretation of freedom as
non-domination. I do not suppose that this is the only conception of
freedom in common usage, as Chapter 1 should make clear (on different
conceptions, see Schmidtz and Brennan 2010). My claim rather is that this
way of thinking about freedom has serious payoffs in normative thought. It
enables us to develop theories of social justice, political legitimacy and
international sovereignty that stand up well to the tests that reflective
equilibrium would support.21
This point is worth emphasizing because ordinary talk of freedom offers us a
network of flexible and, it may often seem, wayward idioms. We say that if you
can’t enact a certain option then it’s not a matter of your free choice. And yet
we also say that you may be free to do something – say, go and vote – even
when you are confined to bed and cannot get to the polling station. We say
hypothesis – that Hobbes was the first to think of language as an invention that transformed the nature of
its inventors – that others are better qualified to test. That also is how I think of the idea, as it was
formulated for example in Pettit (1996a, 1997c), that the main figures in the republican tradition
conceptualize freedom as non-domination; that idea was stimulated, of course, by the work of
Quentin Skinner, though it gave a somewhat different twist to what he had been saying (Pettit 2002b).
21
When we follow the method of reflective equilibrium, I assume that we share general ideas about the
nature of values like freedom, justice and legitimacy – the ideas come to us as connotations of the
terms – and that a normative theory should try to respect these in the interpretation of what the values
demand. I also assume that an overall normative theory will force us to weight these values against
each other and to try to reconcile their rival demands. And I assume, finally, that the demands of
values in this or that context – the demands of freedom-in-the-face-of-power, for example, or the
demands of justice-in-the-face-of-reasonable-disagreement – do not have to be derived from the pure
or abstract theory of freedom or justice, whatever that might be. I hope that in good part these
assumptions will be justified, arguendo, by the role they play in my argument. The first two
assumptions put me at loggerheads with the ‘more ambitious and hazardous’ holism that Ronald
Dworkin (2011: 263) supports, under which the demands of all values whatsoever are to be construed,
so that there are ‘no genuine conflicts’ (119) between them. For critical remarks on Dworkin’s
methodology see Pettit (2011a). The third assumption puts me at a distance from G. A. Cohen’s
(2008) foundationalist claim that the demands of justice under a contingent condition like reasonable
disagreement are to be derived from their demands in abstraction from any such contingent fact. For a
critique see Larmore (2012).
The emerging view of democracy 21
that even if something is not permitted, you may still have the freedom to
choose it. And yet we also say that you do not have the freedom to act in a
certain way – say, steal someone’s property – because it is forbidden. We say
that you may do something freely even if other options, unbeknownst to you,
are all closed. And yet we also say that you do not enjoy freedom in a certain
choice if there is only one alternative available: if, in effect, you are forced to
take that particular path. We say that you have a free choice between two
alternatives in a certain domain if those in charge there leave the decision up to
you. And yet we also say that you do not enjoy freedom in a choice if you can
make the decision only because others give you permission.
This variety of usage should not suggest that freedom-related words and
idioms are unconstrained and that like Humpty Dumpty we can make them
mean what we like. Describing a choice or action as free is always meant to
mark a contrast with some other choice or action. But one and the same act
may count as free in relation to one contrast and unfree in relation to another.
And so, as different contexts put different contrasts in play, we may find
ourselves ascribing freedom in the one case, unfreedom in the other. Take the
example where you are confined to bed on election day. By contrast with
another citizen who is not confined in that manner, you are not free to vote.
But by contrast with a non-citizen it remains perfectly sensible to say that you
are indeed free to vote. The predication of freedom is a useful way of marking
a contrast in each case, yet the contrast it marks differs from one case to the
other. And so we find ourselves saying what appear to be contradictory things
when, as a matter of fact, they are perfectly consistent.
These observations about the idioms of freedom mean that if we are to
build a political theory around the idea of freedom, then we have to regiment
things a little and rule on exactly how we are using the term and its cognates.
That is precisely what I do in Chapter 1, setting out the regimented, republican
understanding of freedom with which I shall be working. While the regimen-
tation builds on connotations of freedom that are recognized in ordinary talk –
otherwise it would not count as a theory of freedom – the final case for
preferring it to alternatives shows up in its capacity to underpin an independ-
ently attractive theory of justice, democracy and indeed sovereignty.
the emerging view of democracy

As already advertised, I offer a characterization of freedom as non-domination
in the first chapter, introducing a range of clarifications prompted by various
critics. I sketch a republican theory of social justice in the second chapter,
showing what would be broadly required for the members of a society to
enjoy equal freedom as non-domination in relation to one another. And that
then sets the scene for asking after what might ensure that the state which
supports its people’s equal enjoyment of freedom as non-domination is not
itself dominating – in other words, is a politically legitimate, freedom-
respecting entity.
My analysis of the requirements for such legitimacy gives us a theory of
democracy in the sense of a job specification for democratic institutions.
According to that theory, a state will be legitimate just insofar as it gives each
citizen an equal share in a system of popular control over government – that
is, a democratic system in which the demos, or ‘people’, enjoy kratos, or
‘power’. Chapter 3 outlines that theory, arguing that a system of equally
shared control over government requires that people should enjoy a suitable
form of equally accessible influence and that this influence should move
the state in a direction that people find equally acceptable. And then the
following two chapters develop a model of how this theory might
be satisfied, giving a sketch of the sorts of institutions that would meet
the job specification. Chapter 4 looks at the system of popular influence that
might serve in the required role and Chapter 5 at how it might impose a
suitable direction on the state and constitute a system of popular control.
The book ends with a conclusion in which I offer a summary of the main
propositions defended; this is meant to make it easier to use a volume that is
longer and sometimes more complex than I would have wished.
The view of democracy that emerges from this argument is marked by
three sharp contrasts with positions maintained in the current literature. It
conflicts, first of all, with the approach of those like Paley who think that
democracy has little or nothing to do with freedom. Amongst contemporary
theorists, that viewpoint is well represented by Isaiah Berlin (1969: 7) when
he says that ‘there is no necessary connection between individual liberty and
democratic rule’ (see too Berlin 1969: 130–1). Against this point of view, I
argue that if we start from the republican conception of freedom as non-
domination, then we can derive the need for democracy, under a suitable
characterization, from the requirements of freedom. That conception of
freedom makes it clear that when people live under a government they do
not control, then they live in unfreedom.
The view defended in the book contrasts, secondly, with the view of
democracy that has become more or less standard in political science, under
the impact of the Austrian–American economist and social theorist, Joseph
Schumpeter. In his discussion of democracy in Capitalism, Socialism and
Democracy, first published in 1942, Schumpeter (1984: 272) argued that
democracy does not enable the people to ‘control their political leaders’ and
The emerging view of democracy 23
that it merely serves to give them a potentially wayward form of influence that
need not impose any controlling direction. I argue by contrast that while a
system of non-directive influence might do a little to protect citizens – it would
guard against dynastic regimes – it would not enable them to enjoy freedom as
non-domination in their relationship with government. Democracy has to
promote popular control, and it has to put in place the institutions that such
control requires, if it is to serve the cause of freedom as non-domination.
The thesis defended in the fourth and fifth chapters, as I mentioned
earlier, is that democracy can provide for such control by mobilizing a dual
process. It has to provide for people’s electoral and contestatory influence
over who is in power, what processes of decision-making they follow, and
what actual decisions they make. And it has to shape that short-haul
influence so that over the longer haul it forces those in power to conform
to widely shared norms in selecting the processes they follow and in
determining the policies they adopt. This popular control may leave a
good deal of discretion to the authorities within the state but it will only
do so when such discretion does not entail domination.
The third big contrast that marks out the view of democracy defended here
is with the very common assumption that democratic institutions are more or
less exclusively majoritarian and electoral and that they operate in essential
conflict with institutions that give constitutional protection to certain rights,
authorizing the judiciary – or some other unelected body – to override
legislation. While this distinction between democracy and constitutionalism
is widely maintained, it is often linked with William Riker (1982), who argues
that democratic populism has to be tempered with a constitutional liberalism.
I maintain against this sort of view that the institutions required to give people
control over government are likely to include many of those associated with
constitutional protection. There is no need for the dichotomy that many
current thinkers accept between the two institutional forms.22
Any plausible model of the form that democracy ought to take is bound
to support many of the institutions that are actually in place in existing
democracies. But no model will have normative bite unless it can be used as
a base for arguing for reform. I hope it will be clear by the end of the book
that the republican theory on offer here has a critical, reformist edge and is
not designed merely to vindicate democracy as we know it. While it
22
For other views on which democracy requires a degree of constitutional protection see Cohen (1989);
Habermas (1995); Holmes (1995). For concerns about the extent to which any degree of constitutional
protection can hamper democratic initiative, see Schwartzberg (2007) and Tully (2009). And for an
outright defence of the adequacy of electoral, majoritarian democracy see Waldron (1999a, 1999b).
supports constitutional as well as electoral measures, then there is no reason
why it should support the precise sorts of institutions that characterize
extant regimes. Which precise measures it should support in any area will
be determined by empirical considerations, as well as the more abstract,
philosophical arguments advanced here; that is in the nature of institutional
design. But even the broad-brush strokes offered by way of sketching
possible institutional designs should make it absolutely clear that the agenda
supported by the republican argument is reformist and progressive in its
implications.
justice and democracy

In the course of the argument developed in the book I suggest, without
offering any precise guidelines on how to conduct the exercise, that it is
possible to use the ideals of republican justice and republican democracy to
make judgements on how existing, imperfect, regimes compare in their
realization. But one issue I do not address is the relative importance of those
ideals. Suppose we had a choice between two regimes, one scoring better on
democracy, the other on justice. The question is whether there is any reason
in the abstract to be predisposed towards one or the other or whether,
depending on context, our answer might go either way.
For the record, my own view is that in general we should prefer the more
democratic alternative in any such choice. Freedom as non-domination,
whether in the social or political arena, requires not just the absence of
interference, as we have seen, but its robust absence: its absence over a range
of scenarios in which there are variations in what we ourselves want to do
and, crucially, in what others would want us to do. A failure in political
legitimacy would compromise the robustness of freedom more deeply than
a failure only in social justice. Where a lack of social justice alone would
make us vulnerable only to our fellow citizens, a lack of political legitimacy
would make us vulnerable on two fronts.
The reason is easy to see. If we are subject to a government that can
dominate us, as in an illegitimate regime, then we are going to lack control
over changes in that government’s will towards us and towards those of our
kind. But this lack of political control means that any social controls we
enjoy over changes in the will of our fellow citizens towards us are also likely
to be somewhat precarious. While the law may put social controls in place,
guarding us against private domination, those safeguards will only be as
reliable as the will of the government that establishes and maintains them.
Let legitimacy fail, therefore – let the government be a law unto itself – and
Justice and democracy 25
we will be vulnerable both in relation to the state and in relation to our
fellow citizens. Public domination has costs that run deeper than those of
private domination alone.
I conclude that the republican ideal of freedom as non-domination is
bound to put a certain premium on the value of legitimacy and on the
democratic control that it requires. Whatever policies the government
supports, and whatever policies any one of us wills on government, none
should be put in place unless it is implemented under a form of popular
control in which we all equally share. That is the only guarantee against the
doubly disabling effect of public domination.
Much contemporary political philosophy focuses on the requirements of
justice where, bar a few references to the importance of certain political
rights, this is more or less tantamount to social justice. From the republican
perspective of this book, that is an unfortunate priority. The most impor-
tant demand we can make on our state is that it should deliver policy under
a system of control to which we each have equal access. Recognizing our
equal claims as citizens, and our divergent views on political ideals, we
should accept that whatever the proposals any one of us makes on public
matters, they should only be implemented or maintained under the proviso
that, however democracy is institutionally realized, they command demo-
cratic support.
This position entails a commitment to democracy that, paradoxically, is
not itself subject to that democratic proviso. A society that makes decisions on
a democratic basis, with everyone sharing equally in control of government,
could in principle decide on a permanent renunciation of democratic rule,
denying future generations the right to shape their institutions on an equally
shared basis. The population might vote unanimously, to take an extreme
example, to deny votes to women. Since such a possibility would be anathema
from a republican viewpoint, it follows that we ought to exempt the recom-
mendation to do things democratically, and that recommendation alone,
from the democratic proviso. We ought not to recommend that our society
should give people an equal share in control of government, provided this
proposal is itself democratically endorsed. We ought to recommend that our
society should give people an equal share of control, period. This, in
Wittgenstein’s image, is where the spade turns. This is bedrock.23
23
Michael Walzer’s (1981) complaint about the way in which philosophers ignore democracy, speaking
as if they had the authority of the sage, is fully consistent, I believe, with my viewpoint here. Like him,
I allow that most recommendations in normative political theory – all but the commitment to a fall-
back democracy – are democratically negotiable.
chapter 1
Freedom as non-domination
In order to gain a good understanding of the concept of freedom as non-

domination it will be useful to focus first on what is required for freedom in
one or another choice. As we saw, republicans traditionally concentrated on
the freedom of the person, period – the free status of the liber, or ‘free-man’
or citizen – rather than on the freedom of a person’s particular choices
(Pettit 2007e; Skinner 2008a). But once we know what freedom of choice
requires, we can represent people’s status freedom as a function of their
freedom over a common range of choices, secured on the basis of common
norms and laws. We will return to that issue in the next chapter.
A choice is identified by a set of mutually exclusive, jointly exhaustive
options, as in the choice you may have between doing X, doing Y and doing
Z. The options are available insofar as two conditions hold, one objective, the
other cognitive. Objectively, it is true that you can do X, or you can do Y, or
you can do Z, and that’s it: there is nothing else you can do instead. And this
truth registers cognitively: it holds according to your own perceptions of
the scenario.1 Thus if you have a choice between X, Y and Z, then you
must understand each of those options under its relevant aspect, as a case of
X-ing or Y-ing or Z-ing; realize that you can choose any one of them, taking it
under that aspect; and recognize that there is no further alternative. Whether
you do X or Y or Z is up to you and you only; you can think truly ‘I can do X’,
‘I can do Y’ and ‘I can do Z’. It may not be the case that any option chosen is
logically guaranteed to materialize as a result of your decision – the letter you
1
The ability or capacity to perform any option may be represented as coming in degrees, depending on
how easy the choice of the option is. Or it may be taken to have an on–off character, with the degree of
ease being registered at least implicitly in the characterization of the option. On the first approach we
would speak of the ability at a certain level of ease, L, to do X; on the second we would speak of the
ability in an on–off sense to do X-at-level-of-ease-L. While I avoid addressing this issue explicitly in the
current text, I follow broadly the second approach. Thus I argue that imposing an ease-reducing
penalty on doing X involves replacing the original option, X, by a different option, penalized-X. Such a
replacement may be more or less invasive, as I shall put it, depending on how large the penalty is. I am
grateful to Raffi Krut-Landau for a discussion of this point.
26
Freedom as non-domination 27
mail may not reach its destination because of problems at the post office – but
it is enough that as a matter of contingent fact there are no obstacles in the
way. Sending a letter to your correspondent is an option for you just to the
extent that the world happens to be compliant – there are no postal problems –
and success is in your hands.
There is a metaphysical issue about what has to be true of you as an agent in
order for it to be true that a choice is up to you: that you can make it the case
that the world in a moment from now is an X-world or a Y-world or a Z-
world – a world where a letter is on its way to your friend, or a world where it
isn’t. The question in particular is whether you can play this choice-resolving
role and yet be constituted, like every other natural entity, out of physical
constituents that conform to physical laws. This is the issue of whether you
have metaphysical freedom of will and it falls well outside our concerns here.
I shall just assume that you have the power of free will, however that is to be
understood, and ask about what social and other factors can impact on your
exercise of the power in such a way that, intuitively, they reduce your freedom
of choice.2
Almost all schools of thought agree that any factor that can reduce your
freedom of choice in that sense constitutes a hindrance that makes one or
another option unavailable: it removes some of the objective or cognitive
resources required for accessing the option, or it restricts your use of those
resources; it impairs or impedes the capacity that those resources establish.
But there is disagreement on two broad fronts: first, on the issue of what
has to be hindered in order for the freedom of a choice to be reduced; and
second, on what sorts of hindrances can impose a reduction of freedom.
I address the first issue in Section 1 and discuss different aspects of the
second issue in the three sections following.
In Section 1 I argue that your freedom of choice will be reduced by a
hindrance to any one of the options that characterize it, not just by a hindrance
to the option you prefer. In Section 2 I argue that your free choice of an option
may be hindered in either of two significantly different ways: one involves
vitiating factors, as I shall call them, the other factors of an invasive kind. In
the following two sections I then go on to explore the ways in which you may
2
For the record, my own view is that we should identify free will in the metaphysical sense with
conversability: a capacity to register and respond to reasons, as they are presented in interpersonal
conversation (Pettit and Smith 1996; Pettit 2001b, 2005b, 2007a). This, broadly, is the commonality
shared by those who are fit to be negotiated with from within what Peter Strawson (1962) calls the
participant as distinct from the objective stance. ‘Orthonomy’, which I mention later, is the virtue
displayed by someone to the extent that they successfully exercise conversability across the normal
human range. It constitutes free will in the psychological, as distinct from the metaphysical, sense.
28 Freedom as non-domination
suffer a loss of freedom through invasion, as distinct from vitiation. I argue in
Section 3 that domination – exposure to another’s power of uncontrolled
interference – is necessary for invasion: interference in the absence of domi-
nating power is not enough. And I argue in Section 4 that domination is
sufficient for invasion: you can suffer invasion without actually suffering
interference. Section 5 looks in the light of this discussion at what we others
can do to ensure your freedom in any choice, guarding you against vitiation
and invasion.
Assuming that vitiation is not a problem, the lesson of the chapter is that
freedom in a choice requires just the absence of domination; it is equivalent
to the freedom that was hailed as an ideal in the long tradition of republican
thought. While this equation of freedom with non-domination allows us to
say that domination with interference is worse in various ways than dom-
ination without interference, it marks the threshold where freedom begins
to fail at the point where domination, even domination without interfer-
ence, kicks in. With the argument for this understanding of freedom in
place, we can go on in the next chapter to ask after what it would mean to
establish equality for people in the enjoyment of freedom: that is, to ensure
an equal status for them as free citizens. This ideal amounts, as we shall see,
to a republican ideal of social justice.
1. what has to be hindered to reduce

freedom of choice?
We saw that any choice is characterized by a set of options that are available
in virtue of the objective and cognitive resources that you can access and
use. What does a hindrance have to affect if you are not to enjoy freedom
in that choice? There are two answers in the literature. The first is that the
hindrance need only affect your preferred option. The second is that a
hindrance to any option, preferred or unpreferred, will reduce your free-
dom. I shall argue in favour of the second answer.3
Hobbes’s position
The best-known defender of the first answer is Thomas Hobbes. His view
of freedom – corporal freedom, to be exact (Pettit 2008c: Chapter 8) – is
3
I ignore a third possible answer, that in order to reduce your freedom every option other than at least
one has to be hindered – that you have to be more or less forced to perform the remaining option. This
answer hardly makes an appearance in the literature.
What has to be hindered to reduce freedom of choice? 29
summed up in his famous, consciously anti-republican definition of a free-
man (Skinner 2008b): ‘a free-man is he that in those things which by his
strength and wit he is able to do is not hindered to do what he has a will to’
(Hobbes 1994b: 21.2). According to this definition, a hindrance takes away
from your freedom in a choice only if it hinders an option that you have ‘a
will to’; only if you prefer the option that is subject to hindrance.
This claim is not just an implication of his definition that may have
escaped Hobbes’s attention. It is a thesis that he explicitly defends in a debate
with Bishop Bramhall about the preconditions for having a free choice
between playing tennis or not. Bramhall suggests that if you are considering
whether or not to play tennis – we assume a willing partner – and you decide
against doing so, then you may still have been wrong to think that you had a
free choice. After all, unbeknownst to you, someone may have shut the door
of the (‘real’) tennis court against you. Hobbes is undaunted by the claim,
asserting that for anyone in your position ‘it is no impediment to him that the
door is shut till he have a will to play’ (Hobbes and Bramhall 1999: 91).
The line that Hobbes takes here is that your freedom in a choice is reduced
only when there is a hindrance that frustrates you in the attempt to satisfy
your final preference between the options. There is no frustration if a
hindrance blocks you taking an option that, as it happens, you do not want
to take anyhow; and so in that case there is no loss of freedom either. We may
all agree that you freely decided against playing tennis and that you might
therefore be held responsible for this decision; after all, you thought you could
act on your preference, whatever that turned out to be, and as a matter of fact
you did act on the preference you formed (Frankfurt 1969).4 But is managing
to choose in accordance with the preference you form, even managing to do so
in the manner illustrated by the tennis example, all that freedom in a choice
requires? I argue not.5
There is a clear contrast between a case like that of the tennis example
and the case where all of the options, and not just the option preferred, are
unhindered. In the tennis-type case we can say that you made a choice
freely, but not that you had a choice between the option preferred and the
alternative, a choice between playing tennis and not playing tennis. In the
contrast case, both things are true: whatever you choose, you choose freely
and, moreover, you have a choice between the given options. The question
to consider is whether in fostering freedom of choice we ought to try to
4
In such a case the action you take is voluntary, in ordinary usage, being taken for reasons other than
that you have no (acceptable) alternative; on this conception of voluntariness see Olsaretti (2004).
5
For a different point of view, in which ‘voluntary’ is equated with ‘willing’, see Otsuka (2003: Chapter 5).
ensure that both things are true or just the one. Ought we to try to give
agents choice over which of the options to realize? Or ought we to try just to
ensure that their preference between the different options is satisfied?
These are distinct ideals and they have different ramifications for policy.
If we seek the richer goal, trying to keep all your options open, we will want
to invest resources in making each available, independently of what you are
likely to prefer. If we seek the more austere goal, trying merely to avoid
preference-frustration, we will want to invest resources in making an option
available only to the extent that it is likely that you will choose that option.
The richer ideal is more expensive in the resources required, the poorer
more economical. Where we do not have the resources to support the first
ideal, of course, it will make sense to focus on the second. But this does not
yet tell us which ideal is the more appealing, nor which answers better to the
ordinary notion of freedom of choice.
Berlin’s critique
Other things being equal, the richer ideal is certainly the more appealing,
since its satisfaction entails the satisfaction of the more austere counterpart:
if you can do whatever you might prefer, you will be able to do what you
happen actually to prefer. More importantly, however, that ideal also
answers better to our shared ideas about freedom of choice. It sustains the
natural assumption that if you enjoy freedom in a choice between certain
options, then it must be up to you whether one or another option is realized.
And it enables us to avoid some counter-intuitive results that plague the
Hobbesian view.
If we say that preference-satisfaction is enough for freedom of choice, as
Hobbes implies, then you may count as free in a choice, even when you are
forced to choose as you do; the force won’t matter so long as the action you
are forced to take is the one you prefer. We might choose to live with that
counter-intuitive result, arguing that other virtues in the Hobbesian approach
outweigh the cost of doing so. But there is still worse in store for us if we
stick with that approach, as Isaiah Berlin (1969: xxxix) has pointed out (Pettit
2011b).6
6
Berlin clearly focuses on this point only in the 1969 Introduction to the collection in which his 1958
lecture on ‘Two Concepts of Liberty’ was published and acknowledges doing so as a result of criticism
by an anonymous reviewer of the 1958 lecture in the Times Literary Supplement. That reviewer, it
appears, was Richard Wollheim. I am grateful to Albert Weale and Jonathan Wolff for throwing light
on this for me.
If we say that preference-satisfaction is enough for freedom of choice, then
we have to acknowledge that there are two distinct ways in which you can
make yourself free in a choice between two options, X and Y, where you want
X and it happens to be obstructed. You may remove the hindrance so that
you can get what you want, i.e. X. Or you may work at changing your
preference, so that you come to want what you can get, i.e. Y. But, as Berlin
(1969: xxxix) argues, it is quite absurd to think that you might make yourself
free in this second manner: ‘to teach a man that, if he cannot get what he
wants, he must learn to want only what he can get may contribute to his
happiness or his security; but it will not increase his civil or political freedom’.
Berlin’s argument is worth setting out in detail.
1. Suppose with Hobbes that you enjoy freedom in a choice between X and
Y just in case you avoid hindrance in the option that you actually choose;
suppose that non-frustration is enough for freedom.
2. By that supposition, you do not enjoy freedom of choice in the case
where I hinder X, not Y, and you choose X.
3. But, by supposition, you would enjoy freedom of choice in that case, if
you were to choose Y.
4. If you know the situation, therefore, it appears that you can give yourself
freedom of choice, without constraining my hindrance, just by adapting
your preferences and choosing Y.
5. But this is absurd. You cannot make yourself free in the choice just by
accommodating yourself to my disposition to hinder your choice.
6. Thus the original supposition that non-frustration is enough for free-
dom of choice must be false.
In order to illustrate the problem, imagine that you are a prisoner who,
being forcibly imprisoned, does not have freedom of choice between staying
behind bars and living in the outside world. Do you lack freedom just because
the option you happen to prefer is living outside prison, as the Hobbesian
approach implies? If so, then you can make yourself free – you can give
yourself freedom in the choice between living in prison or outside – just by
adapting your preferences and coming to want to stay in prison. As Berlin
(1969: 139) expresses the thought, ‘I need only contract or extinguish my
wishes and I am made free.’ But this is patently absurd and argues powerfully
for rejecting the Hobbesian approach. Adaptation to a constraint cannot count
on any plausible way of thinking as a form of liberation from that constraint.7
7
While adaptation cannot plausibly make you free in the sense required, there is an independent sense
in which it may serve freedom: it can expand the range of choices available to you. Suppose you would
like to be able to choose between spending or not spending the odd weekend with me, where you can
Think of each option in a choice as a door, where an option is available
just when the door is open (Berlin 1969: xlviii). The lesson of this argument
is that it is not enough for freedom of choice in the ordinary sense that the
door you actually choose to push on happens to be open. It must also be the
case that any other option in the choice, any door that you might have
pushed on instead, is also open. Suppose we are considering whether you are
free in a choice between two options, X and Y. If you are truly to count as
having a free choice in the actual world, then you must not be hindered in
the actual world where you prefer X but equally you must not be hindered
in the possible world where you prefer Y.8 You must get what you actually
want but it must also be the case that you would have got the other option
had you wanted it instead.9 All doors must be open.
As against this point of view, a partisan of the Hobbesian position might
argue that it would be irrational to care about what happens should you
choose a particular option, say Y, when you think it vanishingly unlikely that
you might choose Y. And, building on that observation, the opponent might
urge that you should care about the hindrance to an option only in a way that
reflects the probability that you assign to choosing it.
This objection is misdirected. When you perform as a deliberative agent,
you think of yourself as someone able to choose between the available options,
someone on whom the choice depends. You conceive of yourself, in other
words, as the arbiter and author of whether the world will be one in which this
option materializes or one in which some alternative materializes. This self-
conception is inconsistent with indifference to the prospect of any option
only spend a weekend with me if you are prepared to go hiking, a prospect that is currently
unattractive. You might provide yourself with such a choice by working on your preferences and
making hiking into a tolerable prospect.
8
Is it enough that you enjoy the absence of hindrance in the actual world where you prefer X and in the
nearest possible world in which you choose the other option, Y? Or is it required that you can get X in
any of a wider range of worlds, including the actual one, in which you prefer X; and can get Y in any of
a wider range of possible worlds, including the nearest possible one, in which you prefer Y? My own
view is that you must escape hindrance for each option in a wider range of worlds; that this range of
worlds is discernible only on an intuitive, context-sensitive basis; and that it does not include all
possible worlds. These issues about the relationship between non-hindrance and freedom – about how
robust non-hindrance must be (Pettit 2001a; List 2004, 2006b) – are parallel to issues in epistemology
about the relation between true belief and knowledge; see for example Williamson (2000). I abstract
from them in the current discussion.
9
Strictly, there is a problem in saying that to be free in the choice of X, it must be the case that you could
have chosen the alternative, Y, had you wanted to – had you preferred that option. This condition
might be incapable of fulfilment because you are the sort of person who would only want to do Y if it
was not an available option; the possibility will be salient from Groucho Marx’s quip that he would
only want to join a club that would not accept him as a member. The problem can be overcome if what
is required is that you could have chosen Y had you tried to do so, where it is not required in that
eventuality that you actually prefer Y. For expressive convenience, I shall ignore this complication in
the text. I am grateful to Lara Buchak for alerting me to the problem.
being hindered, even one you are unlikely to take. Any such hindrance will
compromise your status as an agent, pre-empting the capacity that you take
yourself – we assume, correctly – to possess. And so the prospect of any such
hindrance is bound to be unattractive.
This discussion began with the question as to what has to be hindered in
a given choice in order for the hindrance to reduce the agent’s freedom.
Following Berlin, I propose to understand freedom in such a way that in a
choice between given options, say X, Y and Z, or just X and Y, your freedom
can be reduced by the hindrance of any one of those options, regardless of
which you happen to prefer. Such a hindrance will make one or more of the
options unavailable, undermining your objective or cognitive resources. It
will mean that you are no longer in a position to say or think truly ‘I can
do X, or I can do Y, or I can do Z’. One or another of those disjuncts will be
false as a matter of fact or false as a matter of your perceptions.
Freedom, robustness and probability

Given this account of the hindrances that can reduce the freedom of a choice,
we might ask about what is required to promote the prospect of your freedom
in a particular choice between X and Y: roughly, to maximize expected
freedom in this sense. The most plausible answer is that we should minimize
two probabilities: the probability of your being hindered in the event of
choosing X and the probability of your being hindered in the event of
choosing Y. If X is to be an open door, then the probability of hindrance
in the event of your taking X should be low, and if Y is to be an open door,
then the probability of hindrance in the event of your taking Y should also be
low. Thus, we should minimize a function that reflects the two probabilities
in some way: say, to take an over-simple proposal, we should minimize the
sum that we get by adding them together.10 Adopting this additive proposal
10
If we go with the adding proposal, we ought to stipulate that neither probability should go to 1 or 0;
this is reasonable since under standard probability axioms only logical necessities attract 1, only logical
impossibilities 0. The reason for imposing the stipulation is, as Emily Chapman pointed out to me,
that the probability of freedom will be intuitively much smaller in the first than in the second of the
following cases, though the sum is the same in each. First case: the probability of interference with one
option is 1, with the other 0. Second case: the probability of interference with each option is 0.5. This
observation may suggest that instead of adding the probabilities to find the number to be minimized
we should conjoin them in a way that is sensitive to the variance between the probabilities of
interference associated with the different options; for example, we might do this by taking the
number to be minimized to be the sum of the square of P(H if X) and the square of P(H if Y). The
idea would be that the greater the difference between the probabilities of interference associated with
options, the more dangerous that is for the probability of freedom. This is an issue in the measure-
ment of freedom that I ignore, as I ignore many other measurement issues.
for illustrative purposes, we should minimize P(H if X) + P(H if Y), where ‘P’
stands for probability and ‘H’ for hindrance.11
We should seek to minimize this sum, at any rate, if the resources available
for reducing hindrances are not a problem. If resources are scarce, then we
should obviously invest them with a view to minimizing the probability of
hindrance with the option that you are more likely to choose. Since preference-
satisfaction is desirable, other things being equal, you will endure a worse
failure if you suffer hindrance with frustration than if you suffer hindrance
without frustration; you will suffer, not just a loss of freedom, but also a loss of
preference-satisfaction. And of course you will be worse off in the event of
suffering interference without frustration, the more probable the frustration is.
It would be irrational to invest scarce defensive resources where they are likely
to do less good and so it would be irrational to seek to defend an option you are
less likely to prefer rather than an option you are more likely to prefer.12
However plausible the claim that freedom requires us to minimize P(H if
X) + P(H if Y), some writers take a different approach. They say that if we are
concerned with minimizing hindrance then we should focus on minimizing
the probability of hindrance for a given option only to the extent that that
option is likely to be chosen (Goodin and Jackson 2007). We should try to
minimize, not the sum of the probabilities given above, but the sum of those
probabilities after they have each been discounted by the probability of the
relevant option being actually chosen. In other words, we should seek to
minimize the following sum: P(X)P(H if X) + P(Y)P(H if Y).
To take this line, however, would be to endorse the Hobbesian position
described earlier, rejecting Berlin’s open-doors view (Pettit 2008b). To see
why the approach is Hobbesian, consider how it might lead you to think
about what freedom demands of you in the choice between X and Y. If
freedom requires minimizing the sum of the discounted probabilities, not
the original sum, then it is likely to require that you should choose X or
choose Y, depending on which is the less likely to attract hindrance: depend-
ing on which of the probabilities, P(H if X) and P(H if Y), is lower. If P(H if
X) is higher, for example, then by choosing Y you will set P(X) at zero, so that
11
For reasons familiar from the debate between evidential and causal decision theory, the hypothetical
probabilities P(H if X) and P(H if Y) should be understood, not as conditional probabilities, but in
one of the alternative modes consistent with causal decision theory. See Joyce (1999). Take the
conditional probability P(H/X), the probability of H given X. This might be low, not because X is an
open door, but because those who are capable of hindering you in the choice want you to choose X.
12
Waldron (2007) uses the fact that it may be sensible to invest resources on the basis of probability of
choice to argue, in effect, for linking freedom with the absence of frustration. For a response see Pettit
(2007a). Waldron adopts a position similar to that taken by Goodin and Jackson (2007), which I go
on to discuss in the text.
Vitiating and invasive hindrances 35
Table 1.1 Variations in hindrance
Hindrance is more probable If you choose X If you choose Y
You actually 1. choose X 2. choose X

You actually 3. choose Y 4. choose Y
P(X)(H if X) + P(Y)(H if Y) will reduce to P(Y)P(H if Y). That probability

may be extremely low, say because those with a power of hindrance want you
to choose Y. And so by adjusting so as to prefer and choose Y you will have
lowered the probability of hindrance and, on this Hobbesian account, will
have raised the probability of enjoying freedom of choice.
It may be helpful, using this probabilistic language, to set out schemati-
cally the contrast between the positions that the two sides to this debate
adopt. I do this in Table 1.1.
The Hobbesian view entails that if scenario 2 or 3 obtains, then you are
likely to escape frustration and so be free in the choice between X and
Y. The open-doors view entails that the probability of your being free in
the choice is independent of what you actually choose; it is determined,
regardless of what you choose, by the probabilities, P(H if X) and P(H if Y).
While this contrast between the two positions is sharp and clear, it is
consistent with allowing, as mentioned earlier, that a hindrance to a preferred
option is worse than a hindrance to an unpreferred option. You will be unfree
in either event but you will suffer a greater harm in the first. Should we register
this greater harm in the measurement of freedom? That depends on whether
we decide that frustration compounds the loss of freedom inherent in the
hindrance of any option or whether it counts as a distinct sort of harm. I incline
in the latter direction, but the issue is one of book-keeping and measurement
and need not concern us greatly here. I mention some problems in the
measurement of freedom in Section 2, but I generally skirt them in this book.13
2. vitiating and invasive hindrances

Having seen that the freedom of a choice can be reduced by a hindrance to
any of the options by which it is characterized, we ask in this section about
13
If we do decide that it is in some measure a freedom-relevant loss, then there will be a corresponding
reason for weighting the harm associated with P(H if X) or P(H if Y) – say, on a scale between 0 and
1 – in a way that reflects the degree of preference and so the probability, P(X) or P(Y). So long as the
weighting is suitably low, the proposal will be distinct from the proposal supported by Goodin and
Jackson (2007).
the sorts of hindrances by which your freedom of choice may be reduced. In
this section I argue that it is important to distinguish between hindrances
that invade free choice, as I shall put it, and hindrances that vitiate free
choice, and that there is a sense in which invasive hindrances are more
serious. And then in the following two sections I look at the different ways
in which freedom of choice can be invaded.
The resources required for free choice

All hindrances to free choice affect the resources, objective and cognitive,
that freedom presupposes. And so it will be useful to begin with an overview
of the resources that are required if you are to be free in a choice between
any set of options such as X, Y and Z. These come in three broad varieties,
personal, natural and social.
At the personal level you must have the mental and bodily wherewithal
or know-how required for making the choice, and you must be aware of
having such a capacity. You must be consciously able to raise your hand or
not to raise your hand if you are to have a choice about which to do. You
must be consciously able to send an email or post a letter if you are to have a
choice between sending a message by one or the other medium.14
The natural resources required for being able to enact a certain option are
those conditions in your environment that are required to make the action
possible. You will not be able to raise your hand if it is so cold that you are
frozen stiff; you will not be able to send an email if an electrical storm has
knocked out the Internet. Thus you must avoid such problems in the natural
environment, if you are to have freedom in the choice between raising or not
raising your hand, sending or not sending an email. And you must also be in a
position, of course, to know that you do not face such problems.
There are no particular social resources needed for being able to raise a
hand, considered just as a physical movement. But such resources are going to
be necessary for using the raised hand to signal something. You will be able to
raise your hand to greet someone, or to call a foul in umpiring a game, only if
your society establishes conventions that make such an act of communication
possible and only if you are aware of the fact. And, even more obviously, you
will be able to send an email or post a letter only if your society provides you
14
In particular, you must be consciously able to do these things in the presence of a desire to raise or not
raise your hand, to send an email or post a letter; you must have an ability that can serve your
preference or will, whatever it should turn out to be, over the options. You would not have adequate
personal resources for choosing an option if, perversely, you were able to enact it only in the event of
not actually wanting to do so!
with the technology and infrastructure presupposed to such actions and,
again, you are aware of the fact. You will depend on there being enough
others in the society willing to supply such a service, and enough others willing
to demand and pay for it, that you have the choice between emailing or
posting a letter – or relying on some more pedestrian form of communication.
It may not be clear how far social resources extend. Consider a case where
you are able to raise your hand only with the help of a physiotherapist who
is coaching you back to health after an accident or where you are able to
send an email only with the assistance of someone who understands com-
puters. Do you have the ability in such a case to take the option mentioned?
Does the voluntary assistance of another person count as a relevant social
resource?
My response is to say that if you depend on the goodwill of a given
individual or set of individuals for being able to take the relevant option,
then you do not have the ability in the sense required for freedom. You will
have the ability required for free choices if others are independently con-
strained or motivated to provide the assistance needed, or if there is a queue of
people willing to provide it so that, should any assistants fail, there will be
others willing to take their place. But you will not have that ability, on the line
suggested, just because someone happens to be willing to help you out. The
response is intuitive insofar as there is a clear distinction between the case
where you depend on the goodwill of others for assistance and the case in
which you can command such assistance, as we might say – the case where
you can lay claim to it, as if it were yours to own. That distinction will be of
relevance later when we consider the conditions under which domination
occurs; the first case allows domination, as we shall see, the second does not.
Two kinds of hindrance

These observations show that if you are able to make a choice between
any set of options, then you must have objective and cognitive access to the
personal, natural and social resources required for selecting each. To make a
free choice between options, X, Y and Z, is to use such resources in order
to satisfy your preference or will over those options, whatever that will
turns out to be. It is to exercise the capacity that the resources put in your
possession in order to satisfy your will or preference over the options.
But if freedom of choice requires being able to use your resources in order
to satisfy your will, then we can distinguish between two possible sorts of
hindrances. On the one side, there will be hindrances that affect the use
of your resources for any purpose and so, in particular, for the purpose of
satisfying your will. On the other there will be hindrances that affect the use
of your resources for the specific purpose of satisfying your will, and not in a
generic way.
Suppose that you have a car that can in principle serve you in various
ways, including for driving into the local city centre. You may be hindered
from using your car to drive into the city by either of two saliently distinct
sorts of hindrances, generic or specific. You may be hindered by generic
obstacles that prevent you from using the car for any purpose and, a fortiori,
for driving into the city; fuel may be short, the engine may fail, or the car
may be damaged. Or you may be hindered by obstacles that specifically
affect your driving it into the city centre, being triggered by your using it or
trying to use it for that purpose. There may be a law against driving into the
city centre; others may be tempted by your action to emulate you, thereby
causing traffic congestion; or an environmental group may be prompted to
respond by disabling your vehicle.
As there may be generic and specific obstacles to your using your car for a
certain purpose, so there are generic and specific hindrances that may affect
your using available resources, objective or cognitive, for the purpose of
satisfying your will in a given choice. I describe specific hindrances as invaders
of free choice and generic hindrances as vitiators.15 Invaders are inherently
inimical to freedom of choice in the sense that they are only triggered by your
attempting – or by the prospect of your attempting – to satisfy your will by
using the resources at your disposal. Vitiators are incidentally rather than
inherently inimical, because they do not have to be triggered in this way;
they materialize for reasons that are not connected to your attempting, or
being able to attempt, to satisfy your will. Where vitiators affect the capacity
presupposed to your being able to satisfy your will, invaders target the capacity
to satisfy your will just as such: they are designed, as we might put it, to thwart
your will.
Whenever another person or body imposes their will on you, allowing
you to choose only within limits that they dictate or only on conditions that
they decide, their hindrance certainly targets your ability to satisfy your will
and constitutes an inherently inimical assault – an invasion of your choice.
Just by virtue of imposing their will on you, they affect the extent to which
you can act so as to satisfy your own will. The imposition of their will entails
the displacement of your will, as we might say; it competes with your will for
15
In earlier work I relied, not on the distinction between invading and vitiating hindrances, but on the
closely connected distinction between hindrances that compromise, and hindrances that condition,
freedom (Pettit 1997c: Chapter 2).
control of what you do. It may amount to usurping your control over the
choice, taking charge of what you do. Or it may amount to claiming a share
in that control, establishing limits on the extent to which you can select an
option according to your own preference.
This observation shows that subjection to the will of another, be it total
or partial, represents one way in which your choice may be invaded, being
subject to a specific rather than a generic hindrance. But such subjection to
another agent’s or agency’s will also looks to be the only way in which a choice
may be invaded. A hindrance that invades your choice between options has to
be triggered by your seeking to satisfy your will in that choice, rather than
materializing for independent reasons. And while that triggering condition
can be fulfilled in the presence of a will that competes with your own will for
control of what you do, it is hard to see how it could be met otherwise. Your
seeking to satisfy your will in a choice is hardly likely, for example, to trigger
the appearance of a natural obstacle to your getting your way. For these
reasons we may identify invasive hindrances to choice with hindrances that
reflect the will of another as to what you should do.
When we discuss the variety of invasive hindrances in the next section,
we shall see that others may impose their will on you without deliberately
hindering you – without, in that sense, interfering with you. But many
thinkers treat the interference that another inflicts on you as the paradigm of
an invasion of choice. They equate the invasion of choice with an inten-
tional attempt to usurp or claim a share in your control, or perhaps with a
quasi-intentional counterpart of such an attempt: say, the negligent failure
to take an expected level of care not to undermine your control (Miller
1984). Isaiah Berlin (1969: 122) is one of this group, holding that the primary
offence against freedom of choice is ‘the deliberate interference of other
human beings’.
What sorts of factors will count as vitiators rather than invaders of a free
choice? The answer should now be obvious. Any factors that deprive you of
resources required for freedom in that choice, or that limit the use to which
you can put those resources, without imposing the will of another as to what
you should do. Any factors, in other words, that impair or impede your
capacity to use your resources for satisfying your will but without deriving
from the intrusive will of another agent or agency. Thus any lack of personal,
natural or indeed social resources will vitiate your freedom in a choice,
provided it is not a lack deriving from the will of another agent or agency as
to how you in particular, or those in your particular category, should make
that choice. Such failures of resources may derive at any time from your own
illness or disability, from the limits of your natural environment, from the
continuing, damaging results on you or your environment of the invasion of
earlier choices, from the aggregate consequences of independently motivated
actions by others, or from the actions of another agent that are necessitated in
some way and not a matter of voluntary choice (Olsaretti 2004).
The line drawn between invasive and vitiating hindrances runs bright and
clear in most cases, but it is less salient with the hindrances that derive from
the actions that others take. There is a big difference between the case where
you deny me the option of reading the newspaper, as in invasive interference,
and the case where that option ceases to be available because most other
people come to rely on electronic sources of news, put the newspaper out
of business, and thereby vitiate my choice of reading the newspaper. But
what, for example, of the case where there is only one copy of the newspaper
available, you and I both wish to read it and, recognizing the competition
between us, you grab it first, thereby frustrating me? Or what about the case
where the union of workers in the newspaper office calls a strike in order to
punish the management but fully recognize that this will frustrate readers like
me? Does such an act count as invasion or vitiation?
I say in each case that it counts as invasion. You may not wish to impose
your will on me as such, lamenting the fact that we cannot both read the
newspaper, but you impose your will on me – you actively and deliberately
interfere with me – for the instrumental reason that that is the only way in
which you can satisfy your own will. Equally, the union in the newspaper
office may not wish to impose their will on readers like me just for its own sake;
they may only impose their will on us for the instrumental reason that that is
the best way in which they can pressurize management to improve wages or
conditions. We may not take such a contingently motivated invasion to be as
bad as an invasion pursued for its own sake. And, to anticipate the argument of
the next chapter, we may not think that invasions of that type ought always to
be prohibited in a free society. But on the line taken here we clearly have to say
that nonetheless it counts as a straightforward case of invasion.
In invasions of these kinds, whether inherently or contingently hostile, you
exercise a power of interfering or not interfering with me, depending on your
will. Both forms of interference contrast, however, with the vitiating sort of
constraint you impose on me when you have no choice, by whatever criteria,
but to impose a certain constraint: you intervene in my choice, but out of
necessity, not out of a will – a free will, as we might say – to intervene rather
than not intervene. Thus, if we think that you have no choice but to seek your
own preservation in a certain context, and if your doing so impacts negatively
on me, then the constraint you impose constitutes a vitiating, not an invasive,
hindrance to my choice. You impose more or less involuntarily on me.
The relevance of vitiating hindrances
Hobbes acknowledges the importance of vitiating hindrances when he says,
as we saw, that a person makes a free choice in virtue of not being hindered
‘in those things which by his strength and wit he is able to do’. He implies
thereby that you can be free to choose an option only if you have the ‘strength
and wit’ – in our terms, the objective and cognitive resources – to be able to
enact it; only if it lies, as we say, within your capacity. And he contrasts
the way in which you can be deprived of your freedom by a vitiating failure
on this front – as, for example, when you are like ‘a man fastened to his bed
by sickness’ – with the invasive way you may be deprived of freedom by
‘external impediments’ (Hobbes 1994b: 21.1). In the latter case you are made
properly unfree, as we might put it; in the former you are made non-free, not
being eligible as a candidate for the enjoyment of freedom (Pettit 2008c:
Chapter 8).
Our distinction between invaders and vitiators of choice diverges from
Hobbes’s in two connected ways. First, it recognizes resources beyond the
internal resources of strength and wit on which he focuses. And second,
it breaks with his approach in associating invading resources only with
will-imposing hindrances and not, as he does, with any external impedi-
ments.16 Still, abstracting from those differences, our claim that there are
two sharply divided kinds of hindrances to freedom is broadly faithful to
the Hobbesian view.
Where the standpoint adopted in the last section put us on Berlin’s side
and against Hobbes, it turns out that the standpoint adopted in this not only
puts us on Hobbes’s side; it also puts us against Berlin. For Berlin (1969: 122)
maintains, quite assertively, that freedom in a choice does not require the
absence of vitiators, only the absence of invaders. ‘Mere incapacity to attain a
goal is not lack of political liberty,’ he says, where he clearly means the sort of
incapacity produced by vitiation. The observation behind his claim is that we
can say that you are free to vote or not to vote, for example, just in case the
government does not deny you the vote and regardless of whether you have
the capacity to get to the polling station: regardless, for example, of whether
you are fastened to your bed by sickness. Berlin generalizes this conception of
liberty, arguing that the freedom to make a choice, even in the ordinary sense
16
Despite these differences, it is not misleading to think of Hobbes as looking for a distinction, similar
to ours, that will mark off inherently inimical from incidentally inimical deprivations of freedom.
Given that he thinks of all resources as internal, and of failures that deprive an agent of those resources
as incidental, he may think of external impediments as targeting the use of those resources – their use,
necessarily, in the external forum – in an attempt to satisfy the agent’s will.
in which we predicate such freedom, does not ever entail the capacity to make
that choice: it does not require the presence of the resources that would make
the choice feasible.
We do often speak of liberty in the sense in which it is enough for freedom
of choice that the choice is not blocked by government or by any other
agency. But, going back to a theme in the Introduction, I think that this usage
is quite special, being useful only in certain contexts. Thus in the context of
marking the person’s citizenship it might serve a useful purpose to ascribe
the freedom to vote to the person who is sick in bed. It would be highly
misleading to say more generally, however, that you can enjoy the freedom to
choose between certain options, even when you lack the capacity to exercise
the choice (Van Parijs 1995). In most contexts to say that you are free to choose
between the options – that in that sense you have freedom of choice – is to
imply that the option to be realized is up to you. And it can be up to you only
if you have the resources and capacity to make such a choice: only if you are
free of vitiation, not just free of invasion.
The connection between freedom and responsibility reinforces this intui-
tive claim (Pettit 2001e). Suppose you believe that you have freedom of choice
between options, X, Y and Z. Under standard assumptions about the con-
nection between freedom and responsibility, you will have to think that you
are fit to be held responsible for whatever you choose: you can be blamed or
praised, as appropriate, depending on the merits of the option selected. But if
you believe that you lacked the objective or cognitive resources needed for
choosing one or another of the options, then you could not think that you are
fit to be held responsible for choosing that option. Hence your belief that you
have freedom of choice between the three options requires you to believe that
you have all the resources required for choosing any of them.17
The difference between the view adopted here and that which Berlin
espouses is of importance to how we keep the books on freedom but it need
not mark a substantial divergence in ultimate commitments. The reason is
that while Berlin thinks that strictly speaking the freedom to choose between
certain options does not require the capacity to do so, he insists that freedom
will be of little value in the absence of such a capacity. ‘What is freedom to
those who cannot make use of it?’ he asks. ‘Without adequate conditions for
the use of freedom, what is the value of freedom?’ (Berlin 1969: 124; see too lii).
Treating freedom as a bundle of legal, not necessarily effective, liberties,
John Rawls (1971) also distinguishes between what freedom requires, strictly
17
I am grateful to Daniel Berntson for discussion on this point.
speaking, and what it requires if it is to be valuable. We shall see in the next
chapter that this is an important feature of Rawls’s view and that it is relevant
to the comparison between his theory of justice and the theory that a
republican approach would support.
The significance of invasive hindrances

We have been arguing that your freedom of choice may be reduced either by
vitiating factors that are not inherently inimical to your freedom or by
invasive factors that are. But invasive hindrances are naturally cast, in any
accounting of freedom, as particularly objectionable and significant. To
suffer the vitiation of choice is to be denied a precondition for enjoying
freedom of choice: to lack required resources. To suffer invasion is to be
denied the very condition by which freedom is identified: to be thwarted in
making the choice according to your will.
The significance we ascribe to invasion, as distinct from vitiation, shows
up in the difference between the reactions that we have to obstruction of the
two kinds. You may be frustrated and exasperated by hindrances that do
not impose another’s will: say, by your lack of certain skills or knowledge, by
the impact of the weather on your plans, by the growing level of traffic on
the roads, or just by the way the dog soiled the carpet. But if you suffer
will-imposed hindrances – hindrances that reflect subjection to the will of
another – then you will feel all of that and something more. Unless you
are possessed of a special degree of self-discipline, you will also burn with
resentment and indignation (Strawson 1962).
Suppose that as you park your car in an unknown part of town, someone
warns you that if you leave your car there, it is likely to be damaged by a
coming hailstorm. Imagine the worry about the hailstorm – and the gratitude
for the information provided – that the warning would trigger. And now
think by contrast about the reaction that you would have in the parallel case
where the person makes a threat, if you leave your car there, to impose a
similar level of damage. You will be anxious about the damage of which you
are warned in the first scenario but, unless you are not as other people, you
will be outraged about the damage with which you are threatened in the
second. The warning alerts you to a cost that it is certainly important to take
into account in deciding whether to park your car. But while the threat
reveals your exposure only to a similar cost, it also does more besides. It shows
that you are not acting under your own will in adjusting to your environ-
ment; you are in a situation where another will presumes to rule over your
actions. And that, intuitively, is much more objectionable.
The distinction between vitiating and invading obstacles to free choice, as
we have drawn it here, maps onto this psychologically resonant distinction
between restrictions that exasperate and restrictions that may also outrage. In
view of their connection with such outrage, these are the hindrances that
assume prominence in social life and that call out most urgently for rectifi-
cation. Drawing on a tradition that long preceded him, Kant (2005: 11)
emphasized the point in a comment prompted by reading Rousseau’s Social
Contract. ‘Find himself in what condition he will, the human being is
dependent on many external things . . . But what is harder and more unnatu-
ral than this yoke of necessity is the subjection of one human being under the
will of another. No misfortune can be more terrifying to one who is accus-
tomed to freedom.’18
This argues that the invasion of free choice is worse on the whole than its
mere vitiation. But it is important not to downplay the impact of vitiating
factors. Such factors affect the range of choices in which you can hope to
enjoy the absence of invasion. And, more than that, they may put such
limits on your range of choice that you are subject, as a result, to a greater
degree of invasion on the part of others. Thus the way things are organized
in a society may not be the work of will in a relevant sense and may not
invade people’s choices as such – it may be the unintended, aggregate
consequence of how people are independently motivated to act – but it
can impact on free choice in a way that is closely connected to invasion. It
may constitute a structure or pattern that facilitates the invasion by some
people of the choices available to others. It may amount to an indirect,
structural form of invasion, we might even say, as distinct from the direct,
personal form of invasion that it occasions.19
The measurement of freedom

The upshot of this discussion is that in charting hindrances to freedom we
should distinguish between the vitiation that denies you a choice between
18
The primacy of will-imposed hindrances – usually identified with acts of interference – is denied by
Van Parijs (1995). Others are sympathetic with this viewpoint but concede that will-imposed
hindrances should be given priority on the superficial grounds that they are more salient candidates
for political remedy than other sorts of hindrances (Steiner 1994; Carter 1999; Kramer 2003). On the
relevance of political salience see Carter (2008: 62). Most approaches, like that adopted here, take will-
imposed hindrances to have priority over other hindrances on deeper, less contingent grounds.
19
This comment explains why the theory of freedom developed here is capable of allowing for the
systemic dangers to freedom that Clarissa Hayward (2011) rightly emphasizes. It provides a reason for
recognizing what I later describe as structural domination. My thanks to Maeve Cook for a helpful
exchange on this point.
certain options and the subjection to another’s will, the invasion by another,
that denies you the chance to make the choice according to your will. The
emerging picture shows that there are two dimensions to freedom of choice:
on the one side, the freedom that goes with the unvitiated range of choices
available; and on the other the freedom that goes with not being invaded by
others in the exercise of those choices. The unvitiated resources at your
disposal define a range of effective opportunities, and to the extent to which
that range is wider, you have greater latitude for choice, greater freedom of
opportunity. Being in a position to use those resources without invasion –
not being subject to the will of another, however partially, on the matter of
how you exercise the capacity they give you – will enable you to exploit
those opportunities more effectively.20 Where the unvitiated resources
ensure your freedom of opportunity, the absence of subjection and invasion
ensures your freedom of exercise or control.21
This observation teaches an important lesson about measuring freedom of
choice. Suppose we want to determine how two people fare in their freedom
of choice. The most obvious way to do this will be to take a range of option-
sets or choices as reference points – the choice between X and Y, the choice
between V and W, and so on – and then to ask how the agents compare in
opportunity-freedom and exercise-freedom, how they compare in latitude
and control, with respect to those choices: how far they avoid vitiation, on the
one side, and invasion on the other.22
Even if we think that comparisons of freedom of choice ought to be made
by reference to the same range of option-sets, however, we may take any of
a number of routes. One approach would be to assume equal unvitiated
capacity across those option-sets – that is, to assume equal freedom of
20
Here I make use of the notion of opportunity and exercise popularized by Charles Taylor (1985a) in
his distinction between opportunity and exercise concepts of freedom.
21
Eric McGilvray (2011) gives an illuminating account of how the traditional republican way of thinking
about liberty came to be ousted by the new commercial or market conception of liberty in the
nineteenth century; he links the latter with the earlier notion of natural liberty that figured prominently
in the juristic tradition. Among the two conceptions he describes we might say that the republican
tended to focus on the absence of servitude, the commercial on the presence of latitude.
22
The most salient alternative to this approach would be to follow Hobbes and argue that how much
freedom two people enjoy depends on how far they each enjoy exercise-freedom in respect of the
different options within their personal, idiosyncratic capacity: how far each is unhindered in ‘the
things which by his wit and power he is able to do’. But this alternative would have the absurd
consequence that the prince who can do lots of things – the prince who has enormous freedom of
opportunity – may count as no more free than the pauper who has few if any opportunities for choice.
The measurement of freedom of choice ought to be option-relative, as Hobbes would agree, but
comparisons of freedom will make best sense if two people are measured against the same range of
option-sets rather than each being measured against a different range: those that happen to lie within
their own capacity and to constitute effective opportunities for choice.
opportunity – and to try to chart variations in the degree to which people are
subject to invasion from others: that is, variations in their freedom of exercise
or control. Another would be to assume equal freedom of exercise and to chart
variations in their freedom of opportunity: variations in their unvitiated
capacity to exercise choice across the different option-sets. And yet another
would be to assume neither sort of freedom and to try to track variations in
both dimensions across the same option-set. This project would require us to
assign relative weights to vitiating and invasive hindrances.
Whichever of these paths we take, the measurement of freedom raises
a daunting challenge. It is deeply unclear how exactly we should weight
vitiating and invasive hindrances against one another, though my earlier
comments argue for giving invasive hindrances some priority. And equally it
is unclear how we should measure and compare different levels of freedom
within the category of exercise on the one side or opportunity on the other.
We shall see in the next two sections that there are different ways in which
your freedom of exercise or control may be hindered: that is, different ways in
which your choices may be invaded. Your freedom to choose between X, Y
and Z, to anticipate that discussion, may be reduced by suffering uncontrolled
interference on the part of another – interference can involve the removal,
replacement or misrepresentation of one or more options – or just by suffer-
ing domination: that is, just by being exposed to a power of uncontrolled
interference on the part of the other. And to add to the complexity that this
distinction introduces, there are different degrees in which a given option may
be removed, replaced or misrepresented; there are different degrees in which
such interference may be uncontrolled by you; and there are different degrees
in which you may be exposed to a power of uncontrolled interference. Since
it is unclear how exactly to measure or commensurate such variations in
degree it is clearly going to be extremely difficult to measure the extent to
which a hindrance may reduce your exercise-freedom in the choice.
There are similar difficulties in store on the issue of how to measure and
compare degrees of opportunity-freedom, despite the fact that the issue has
received a good deal of attention in recent literature (Sugden 1998; Carter
1999; Kramer 2003). Just to illustrate the sort of problems that arise here, take
a set of options X, Y and Z in which you and another have the same freedom
of exercise: there is no one in a position to subject either of you to their will.
And imagine now that you only have the opportunity to choose X or Y and
the other has the opportunity to choose any of the three options. Presumably
that means that you have lesser opportunity-freedom. But what are we to say
about the degree by which your freedom falls short? Do we think that the
shortfall is the same, regardless of how far the new option, Z, is intuitively
different from X and Y, for example, or regardless of how far it represents an
option that you value or ought to value? If X and Y involve drinking one or
another beer, does it matter that Z is a beer rather than a wine option, or a
beer or wine option that you would particularly like to have?
Should these observations lead us to despair about the prospect of measur-
ing freedom of choice? I think that there is reason not to expect a plausible,
more or less mechanical algorithm for ordering all possible variations in
people’s fortunes on a common scale of freedom. But we need not despair
about being able to establish some accepted standards whereby, for a given
range of choices, we can compare and order the more salient differences found
amongst the citizens of a given society or, perhaps, across two or more distinct
societies.
This issue will be central to the next chapter, when we consider how we
might pursue the goal of securing equal freedom as non-domination for the
members of a society. We shall see in that context that it makes good sense
to identify a range of significant choices or option-sets – the basic liberties,
as I think of them – and to look for a system that safeguards against
shortfalls from a culturally established, common yardstick in the extent to
which people have the opportunity to make those choices and in the extent
to which they avoid subjection to the will of others in how they exploit that
opportunity. I take the relevant yardstick to be set by what I call the eyeball
test. At the level set by this test, the safeguards should enable people, by local
standards, to look one another in the eye without reason for fear or
deference. The achievement of that discernible and applicable ideal would
make, intuitively, for the equality of people in their status as free persons or
citizens: that is, in the free status that has long been an ideal in republican
thinking.
Revealed will and real will

At this point, we are positioned to discuss the different modes in which your
freedom of choice may be invaded, looking at the notions of interference
and domination just mentioned. But before turning to that discussion,
which will be developed over the next two sections, I should mention an
important issue that deserves some attention. Assume that you have the
unvitiated capacity to enact a choice between X, Y and Z, having all the
internal and external resources required, and that you are not subjected to
anyone else’s will in how you use those resources and exercise that capacity.
Does it follow that whatever you choose to do, you count as enjoying
freedom of choice in every possible sense of that phrase?
Not according to a certain psychological ideal of freedom or, as it is often
put, autonomy. You may be subject to a malaise – a sort of heteronomy –
that consists in forming a will over the relevant options that is not, as it is
often put, your reflective or stable will, your true or real will. You may be the
sort of person whose will-formation is distorted in such a way that the will
that forms within you takes the form of an alien force and does not carry
your authority. It may not reflect your highest aspiration for yourself but
rather a disposition that you regard as a weakness of will or intellect. In
giving into that disposition, then, you will tend to think of yourself as a slave
of the passions, in the traditional image, and of the various pressures that the
passions have traditionally served to exemplify (James 1997). You will not be
able to think of yourself as fully present in the action. Rather you will
probably think that you have been betrayed by pathologies like obsession,
compulsion and low impulse-control, on the side of desire, or idées fixes,
paranoia and credulity on the side of belief.
The topic of autonomy, understood in this sense, has generated a large
literature. Isaiah Berlin (1969) had autonomy in mind when he spoke of the
positive, psychological liberty that consists in being your own master, if not
in the social world, at least in the citadel of the soul.23 Harry Frankfurt
(1988) argues that such autonomy requires that the desires you act on are
desires that you identify with; at the least, unlike addictive yearnings, they
are desires that are controlled by a higher-order desire that you should be
moved by them (see too Dworkin 1988). Michael Bratman (2007) associates
such autonomy or self-governance, not with the control of a higher-order
desire, but rather with a form of control exercised by a longer-term plan. My
own preference, following Pettit and Smith (1996), is to cast it as an ideal of
orthonomy – a rule of the ‘orthos’ or the right rather than a rule of the
‘autos’ or self – arguing that it requires you to form and act on desires that
answer to your evaluations, in particular to evaluations that reflect whatever
count as the relevant facts (see too Watson 2003: Introduction, 2005).24
23
Here I follow MacCallum (1967) in assuming, for simplicity, that even the notion of positive liberty
can be cast as requiring the absence of an obstacle. But the notion of positive liberty is often associated
with late nineteenth-century liberalism in Britain – modern liberalism, as it was then called (Gaus
1983) – and in that tradition positive liberty consisted, not in escaping internal obstacles, but in
making positive use of their absence by achieving a form of self-realization (Baldwin 1984). Note that
while I go along with MacCallum’s schema here I break with it in the next chapter when I argue that
the freedom of a person does not just require the absence of certain obstacles but, more specifically, an
absence that is secured by a rule of accepted norm and law.
24
The notion of autonomy or self-governance is employed by another set of thinkers for purposes of
developing, not an ideal of personal life, but an ideal also for politics. See for example Christman (2009).
No invasion without domination 49
There is a perfectly good sense in which freedom of choice requires some
version of autonomy or orthonomy – some kind of psychological freedom,
as it might be put (Pettit 2001e) – but I will be ignoring it in the present
context. Our concern is with your freedom of revealed will, not with
whether the will that you reveal counts as your true or real will. This is
appropriate in a book on political philosophy. All of us face a challenge as to
whether the will we form in one or another domain of choice answers to
standards of autonomy or orthonomy. But that is a psychological or ethical
challenge, not one that can plausibly be laid to the charge of a collective,
coercive agency like the state.
I mentioned early in the chapter that we are not concerned in this book
with what gives you metaphysical free will: that is, with what makes you
into an agent with the capacity, however that is understood, to take one or
another option in a given choice. What we have just seen is that equally we
are not concerned with what gives you psychological free will, enabling
you to form your will autonomously, however autonomy is understood.
Our concern is solely with social free will or, in effect, political freedom: that
is, with what is required for it to be the case that however imperfectly
formed your will may be, you are in a position to make your choice, without
vitiation or invasion, according to that will.
3. no invasion without domination

The upshot of the argument in the last section is that given the unvitiated
capacity to make any choice – given any set of options between which you
have the effective opportunity to choose – the issue of how far you are free
to make the choice turns on whether you are subjected to the will of another
in exercising that capacity. But what are the ways in which you can be
subjected to another will, be that the will of an individual agent or the will of
an organized agency or group (List and Pettit 2011)?
The canonical term for the act whereby one agent or agency intervenes in
the choice of another, with a view to restricting that choice – or at least with
the negligent effect of restricting the choice (Miller 1984) – is ‘interference’.
Interference is taken to involve the intentional restriction of the choice – or,
as in the negligence case, its quasi-intentional restriction – so that someone
interferes in a choice only if they themselves have a choice between inter-
fering and not interfering. The main issue that arises in discussing what
subjection to the will of another involves is whether such subjection, such
invasion of your free choice, is equivalent to enduring the active interference
of that other.
The republican theory of freedom as non-domination, to anticipate the
position I defend, maintains that the restriction of choice constituted by
active interference is not sufficient or even necessary for such subjection
to the will of another and that it is only domination that fits the bill.
Domination is defined by reference to interference but is distinct from it.
Someone, A, will be dominated in a certain choice by another agent or
agency, B, to the extent that B has a power of interfering in the choice that is
not itself controlled by A. When I say that B has a power of interference
I mean that B has the unvitiated and uninvaded capacity to interfere or not
to interfere. And when I say that that power of interfering is not controlled
by A, I mean that it is not exercised on terms imposed by A: it is not
exercised in a direction or according to a pattern that A has the influence to
determine. In that respect, it is unlike the interference that someone invites
when they hire an agent, for example, to make certain decisions for them.
In this section I deal first of all with the issue of what interference should be
taken to comprise, assuming it is a way of actively restricting your choice and
so a possible way of subjecting you to my will. This discussion is important,
since the notion of interference is presupposed in the definition of domina-
tion that I have just given. While the discussion involves some detailed
analysis, the bottom line is fairly straightforward, and is summed up in a
simple matrix. With that discussion in place, I then argue that interference
without domination does not constitute subjection. In effect, as the title of
the section phrases it, there is no invasion without domination: domination is
necessary for the invasive reduction of freedom. In the next section I shall
argue for the parallel thesis that there is no domination without invasion, so
that domination is sufficient as well as necessary for the invasive reduction of
freedom.
The nature of interference

If interference is to restrict your choice between X, Y and Z, then it has to
affect your use of the otherwise accessible objective or cognitive resources in
virtue of which we say that you have the unvitiated capacity to do X or Y or
Z. It has to ensure that as a matter of objective fact, or as a matter of your
cognitive perception, you no longer have access to those options.
The most obvious way in which I can restrict the use of your objective
resources is by removing one or more of the options, so that it is no longer
available to you. If you face a choice between X, Y and Z, having the
unvitiated capacity to take one or another option, I can impose on your
choice by objectively removing one or more of the options. I can transform
the set, X, Y and Z, for example, into the set, X and Y. I may do this without
signalling or communicating the fact, but if I am anxious to get you to choose
one of the remaining options, I will inform you of the unavailability of the
removed option and not have you waste your time on trying to get it.25
If I can impose my will on you by removing one of your options, I can
presumably have the same effect by replacing one of the options in the
choice by a different option. Suppose that I change the options so that you
have to choose, not between X, Y and Z, but between X, Y and Z*. This
intervention, too, will mean that I have changed the profile of options you
face and thereby restricted your choice. As it was not possible in the earlier
case for you to choose the removed option, so it is not possible in this case to
choose the option replaced: Z is no longer available but only the option that
replaces it, Z*. As in the other case I may not communicate the fact that
I have replaced Z in this way, though I will do so if I am anxious to direct
your choice towards one of the other options rather than have you waste
your time on trying to get Z.
The standard way in which I might replace one or more of the options in
your choice between X, Y and Z would be to attach a penalty to the option;
this will change that option, say Z, into penalized-Z or, if you like, Z-minus.
I may do this without telling you, though if I want you to choose X or Y, as
distinct from just wanting to have a negative impact on your life, it will be a
good idea to communicate what I have done, thereby threatening or coercing
you. Penalizing an option in this way may not actually stop you from choosing
the penalized option – that is, stop you from choosing Z-with-the-penalty.
And if you do choose it, we may say that you chose it freely: chose it in a way
that makes you responsible for what you did (Frankfurt 1969). But still, my
intervention in penalizing and replacing the option, however slight it may be,
denies you the unrestricted choice between X, Y and Z. It replaces it with a
25
Perhaps because of wanting to make it easier to measure and compare the degree of freedom that people
enjoy, some recent thinkers have argued for limiting the interference that can reduce freedom of choice
to the removal of an option in this sense – to what is often described as prevention (Steiner 1994; Carter
1999; Kramer 2003). Representatives of this view of freedom, sometimes described as the pure negative
theory (Carter 2008), have elaborated the most systematic, freedom-based alternative to republican
theory. While I believe that the two theories have to be judged ultimately on the basis of a methodology
of reflective equilibrium, as indicated in the Introduction, I think that this rival theory has to live with
some strikingly counter-intuitive judgements on how freedom may be reduced in a given choice. On
the one side, it would force us to ignore activities whereby I do not remove an option in the choice but I
do replace it; for debate on the claim that still I reduce your ‘overall freedom’ in such a case see Kramer
(2008) and Pettit (2008d). On the other, it would force us to ignore activities in which I do not interfere
with how things are as a matter of actual fact but I do interfere with how they are according to your
perceptions; I impose a cognitive rather than an objective restriction on you. And, on the face of it, those
interventions have a capacity to impose my will on you, as argued in the text.
choice between X, Y and Z-minus. It prevents you from choosing Z, albeit not
in the sense in which the actual removal of that option would do this.
Why say that penalizing an option means replacing it? The question
raises some tricky issues about the individuation of options but, for the
record, my own approach is as follows. I assume that for any agent there are
features or properties such that their presence reliably supports the attrac-
tion or aversion felt for a given prospect or option (Pettit 1991, 2005c). I hold
that a given option ceases to be the same option – it shifts identity – in the
case of any alteration in such a desiderative feature. To impose a penalty or
even the chance of a penalty on one of your options, then, is presumably to
change its desiderative features and to replace it by a different option. But to
impose a change that does not alter the desiderative features is not to replace
the original option; it changes the option but does not replace it.26
Thus, to explain how this stipulation should be understood, I do not
replace the option of your reading, as distinct from not reading, a book just
by moving the book from your left side to your right. This changes the
option but not in a desiderative and relevant fashion. Nor do I replace that
option just by offering you a reward for not reading it. This makes for a
desiderative change but not for a change in the option itself.27 While the
offer raises the opportunity cost of that option, as economists say, this just
means that there is now a more attractive alternative than previously.28
Assuming that desiderative features are not subject to much variation across
people, or at least across people in a given culture, we can take the replace-
ment of an option always to involve a change in any socially recognized
26
On the approach sketched, I can change any of your options by associating it with a property that
matters to you, whether it matters in a personal, self-regarding way or matters in a wholly impartial
manner: it reflects your moral vision of the universe. We might vary the picture by restricting the
relevant properties to ones in the self-regarding category. This would guard against the consequence,
which I am prepared to live with, that I can restrict your freedom by ensuring that should you take a
certain option then some bad consequence will follow for people elsewhere for whom you have a
moral, though not a personal, concern.
27
I rely here on an intuitive sense of the distinction between changes that alter a thing itself and
Cambridge changes, as they are sometimes called: changes such as the change I undergo when
someone emigrates from the country and I become thereby the nth oldest resident. An option
changes when its character or consequences change, as we may put it, but it does not change just
because there is a change somewhere else in the world. In our example it does not change – or, to
anticipate the next footnote, does not necessarily change – because of a change in the alternatives.
28
Although increasing the opportunity cost of an option does not necessarily change the identity of the
option, the nature of alternatives can sometimes affect the identity of an option. Taking an apple and
leaving an orange for your companion is a different option from taking the apple where the alternative
is a smaller apple; in the former case it is a perfectly polite option, but in the latter it is not, and
politeness is a presumptively desiderative feature. See Pettit (1991).
feature of that kind: in any established desirability characterization, to use a
term of Elisabeth Anscombe’s (1957).29
Although the replacement of an option, Z, makes it strictly impossible
to choose Z, as distinct from Z-minus, it does so in a sense that allows us
to distinguish between many different grades of interference. There are
degrees in how far an option may be removed or replaced, since it may be
removed or replaced across a smaller or larger range of possible scenarios.
But there are also other, more salient differences in the degree to which an
option may be replaced. The penalty attached to Z may be lesser or
greater, to go back to our example, and Z-minus may vary in a lesser or
greater degree from Z. It may even be so small a penalty, as we say, that it
does little or nothing to stop you from ignoring the penalty and still
choosing Z-minus. The removal of the option, Z, would prevent you
across relevant variations of scenario from choosing any Z-like option; the
replacement of the option prevents you from choosing only the narrowly
individuated Z – in effect, Z unpenalized. Your freedom in the choice
will be reduced by any penalty and any replacement – you will strictly
be denied a choice between X, Y and Z – but it may not be reduced,
intuitively, by much. This observation points us to a particularly striking
respect in which the measurement of freedom is bound to be a complex
project.
If a penalty or threat of a penalty can invade your freedom of choice, do we
have to say that a reward or offer of a reward – an ordinary, non-mesmerizing
reward – can have the same effect? Do we have to say that I restrict your
choice by offering you a regular, non-mesmerizing reward for choosing X, for
example, introducing the enhanced option of X-rewarded or X-plus? That
depends on whether the reward is refusable or not. If it is not refusable, so
that your options are now X-plus, Y and Z, then we have to say that I have
replaced one option and restricted your choice between X, Y and Z. But if the
reward is refusable, as rewards generally are, then I will not have had any
restrictive effect. It will still be the case with each of the three original options
that you can choose that option; for all I have done, you will still have all the
objective resources required and, we may assume, all the cognitive resources
too. I will have added a further option, X-plus, to that set of options but I will
29
The assumption entails that we cannot easily change our attachment to one or another feature. And that
is independently important in my perspective. Otherwise, you might be able to make yourself free in
certain choices by adapting so that the options are individuated in a way that removes obstacles:
adaptation might promise liberation. I am grateful to Hrishikesh Joshi for pressing the point.
have left each of the original options, X, Y and Z, in place.30 Thus the offer of
a reward does not serve to restrict your choice.31
We have seen that I may restrict a choice of yours by removing or replacing
one or more of the options you face, changing the set of alternatives that is
objectively available. But I can also restrict your choice by changing how the
alternatives present themselves according to your perceptions: by putting
apparent but merely pretend options on the table, thereby misrepresenting
the actual options in place. An option will be available for your choice just
insofar as you can make use, first, of the objective resources required for taking
it rather than any alternative and, second, the cognitive resources needed to
identify and understand the options on offer, to see that you are in a position
to choose any one of them, and so to decide on what you prefer. Apart from
affecting the use of your objective resources in order to change the choice you
face, I can act to similar effect on the use of your cognitive resources. And that
type of action should count equally as a mode of interfering with you: a
potential way of restricting your exercise of choice.
There are two distinct strategies I might adopt with a view to misrepre-
senting your options, each of which comes, like removal and replacement, in
degrees. One is deception; the other manipulation. If I deceive you about the
options available in a choice, or about the consequences they are likely to
trigger, I can obviously lead you to believe that the options are other than,
as a matter of fact, they are. I can act on you so that the set of options you
confront, according to your perceptions, is not X, Y and Z, but a set in which
one or another option is removed or replaced or a further option is added.
I will do this for example if I misinform you about the options on offer, or
issue a bluff threat to penalize one of them, or even make an insincere offer to
reward its choice. These are all means whereby I can impose cognitive, if not
objective, limits on your exercise of choice.
This observation raises a natural question. It should be clear why I can
restrict your choice by removing or replacing an option according to your
perceptions, since removing or replacing an option as a matter of actual
fact has the same effect; in either case I will deny you access to the option
affected. But I do not restrict your choice by the objective addition of a
30
Adding X-plus will change the opportunity costs of taking any of the other options, in particular,
X. But I am assuming, as mentioned earlier in the text, that a change in the opportunity cost of an
option does not entail a change of identity.
31
What if the offer is exploitative, taking advantage of your lack of acceptable alternatives and offering
an X-plus that is manifestly unfair? Strictly, the offer will not reduce your freedom in that particular
choice, though you may count as taking the offered option involuntarily (Olsaretti 2004). But the
offer may establish a relationship between us, as exploitation is generally taken to do, in which I
dominate you. Similar remarks apply to the offer of the blackmailer.
further option, as when I make an offer and add X-plus to the set. So why
do I do so when I lead you to believe falsely that another option like X-plus
is available?
When I add an option objectively to your choice, allowing you to see the
actual options available, then it remains up to you whether you choose X or Y
or Z; it’s just that you have the extra possibility of choosing X-plus instead.
Thus I leave you in a position where you can act on your will or preference,
whatever that should turn out to be. When I deceive you about the availability
of an extra option like X-plus, however, the apparent availability of X-plus
affects the presentation of X. It makes the option of choosing X, which is all
that is actually available to you, look like an option that can have the reward
attached. In effect, it replaces the option X, as it is cognitively registered, by a
pair of options: X-and-refuse-the-reward, or X-and-accept-the-reward.
I will not succeed in deceiving you, and so not succeed in restricting your
freedom, if you realize that I am deceptive and ignore what I say. But what of
the case where I do successfully deceive you but do not stop you choosing as
you would have done in my absence? What of the case where I issue a bluff
threat to penalize an option, say Y, and this is not enough to make you change
your mind: you still opt for Y? Can I be said to restrict your choice in this sort
of case? I will have forced you to choose from amongst a set of apparent options
that did not conform to the set actually available. Where the actual choice was
between X, Y and Z, I forced you to choose as if it were a choice between X,
Y-minus and Z. It may be a happy result that having formed a will to take
Y-minus, you actually get something better, plain Y. But nonetheless, I forced
you to form your will in the dark, not on the basis of how things actually were.
Apart from deceiving you, I may deny you the use of your cognitive
resources in making a certain choice by recourse to what I call manipula-
tion. Manipulation denies you the possibility of making a choice on the
basis of a proper understanding of the options on offer. It applies pressures
that affect your exercise of the capacity to think straight about the consid-
erations for and against different choices and so about the nature of the
options on offer. Manipulation may involve hypnotizing you, mesmerizing
you with the prospect of extraordinary rewards, making you feel guilty
about not doing what I wish, snowing you with so much information that
you are putty in my hands, or exposing you, if I am capable of it, to the
undermining power of my rhetoric.
Like successful deception, successful manipulation will affect the exercise
of your cognitive capacity to choose between certain options, even if it
leaves your objective capacity in place. By means of manipulation, I may
succeed in getting you to choose as I wish, but even if I fail in that, I will still
Table 1.2 Modes of interference
Do I interfere in your choice a. objectively (whether signalling b. cognitively (but not

between X, Y, Z the fact or not) objectively)
1. by removing an option? Yes. Real obstacle. Yes. Pretend obstacle.

2. by replacing an option? Yes. Real penalty. Yes. Pretend penalty.
3. by adding an option? No. Real reward. Yes. Pretend reward.
have restricted your choice by getting you to face what seem like different
options from those that were objectively available. I rig things so that your
will between X, Y and Z – the will you would form if you saw those options
for what they truly are – does not dictate the action you take; rather it is
dictated by your will over the distorted versions of those options that
I conjure into existence.32
In summary of what we have argued, then, there are three ways in which
I may interfere with you and restrict your choice: by removing options, by
replacing options or by misrepresenting options. And misrepresentation
can involve removing, replacing or adding an option in your cognitive
perception of things. The different modes of interference are salient in the
questions and answers mapped in the accompanying Table 1.2.
Interference is not sufficient for subjection

Interference in the sense characterized always restricts your choice, affecting
the options that are actually or apparently available. But does that mean that
it always subjects you to the will of the interferer? Surprisingly, it does not.
32
Given that manipulation can affect free choice only insofar as it distorts the presentation of the options,
it is important to recognize that not all attempts to change the way in which an agent sees certain options
count as manipulative and hostile to freedom. Consider the nudge, as it has recently been described –
and proposed as an acceptable resource of government (Thaler and Sunstein 2008). For example,
consider the nudge involved in presenting people with an opt-out versus an opt-in choice when they
sign up for driving licences and are asked about whether they are willing, in the event of a fatal accident,
to donate their organs to others. The difference between these presentations has a great impact on
whether people sign up for donation or not: most sign up if they have to opt out of donation, most don’t
sign up if they have to opt in. So does nudging in a case like this amount to manipulation? While we
may not be able to provide a general algorithm for answering the question, I think it is clear in this sort of
example that the answer should be negative. The opt-out system probably evokes more positive choices
because it communicates the presumptively correct and non-deceptive message that there are no horrific
consequences attached to the choice; if there were, the opt-in system would scarcely have survived. The
nudging involved does not replace any of the options on offer with a distorted counterpart and it does
not have the potential to affect a person’s freedom in making the choice between signing up and not
signing up for organ donation. It serves to provide further information on the options, at least on the
presumption about the correctness of the message conveyed.
Suppose you wish to restrict your alcohol consumption and hand over
the key of your alcohol cupboard to me, making me promise to return the
key only at twenty-four hours’ notice and not in response to a request for its
immediate return. When I refuse a request for immediate return of the key,
I interfere with your choice, removing the option of having a drink now.
I deny you the possibility of choosing according to your current will. But
do I subject you to my will? Do I impose my will on you, for example, in a
way that might reasonably trigger resentment? Surely not.33
In refusing the key I act under your instructions, not on my own wishes
or impulses: not at my own will or discretion. And, so we may assume,
I refuse the key only because your instructions require me to do so. There is
no real possibility that I might try to keep it against your wishes, refusing to
hand it over at twenty-four hours’ notice; such an action would jeopardize a
relationship that matters to me. In these conditions it should be clear that
you control what I do in refusing your request that I give you the key now;
I interfere with you, but only on your terms.
The arrangement in place with the key is a means, we might say, whereby
you impose your own longer-term will on yourself, not a means whereby
I impose my will on you. You are the one who set up that arrangement and
you are the one who decided the conditions it imposes. You use me to give
effect to your own will, not relying on yourself to be able to do so. When
I shut off the option that you now want to take, interfering with you and
even frustrating you, I channel that will and enable it to have an impact on
your behaviour. I perform like a robot that is programmed to satisfy
your instructions. I act as your servant, not your master. The lesson of the
example is that the interference that I or any others practise in a choice of
yours will not impose an alien will, and not therefore invade your freedom
of choice, to the extent that my discretion in exercising interference is
subject to your control: it is shaped by your influence so as to assume a
form that appeals to you; it materializes on terms that you dictate.
Interference is defined broadly enough to cover any plausible form of
activity whereby others might deliberately restrict your choices. In effect it is
defined broadly enough to represent the only candidate for a form of active
invasion; in particular, a form of active invasion that need not require the
presence of an uncontrolled power in another. And so the lesson of the
33
I assume in this example, and I make analogous assumptions in the other cases where I apply the
lesson of the example, that the will that you display in giving instructions can be identified as suitably
stable or authoritative in comparison with the will that the instructions require me to frustrate. There
are a number of ways in which this assumption might be vindicated but I do not pursue them here.
example is that there can be no active invasion of choice unless there is
domination. The active, intentional restriction of your choice by any other
agent or agency will be invasive only to the extent that it reflects a will that
you do not control.
The republican lesson about uncontrolled interference has often been
formulated in the claim that, provided it is non-arbitrary, interference does
not subject you to an alien will. Arbitrary interference, on this interpretation, is
interference practised in accordance with the arbitrium, or ‘will’, of another. It
is precisely what I describe here as uncontrolled interference: that is, interfer-
ence that is exercised at the will or discretion of the interferer; interference that
is uncontrolled by the person on the receiving end (Skinner 2008a).
In what follows I shall make little or no use of the term ‘arbitrary’,
preferring to speak of uncontrolled interference. The reason is that while I
believe that in earlier republican usage the word had something close to the
meaning I ascribe, it has other, misleading connotations today. In one usage,
arbitrary interference is interference that is not subject to established rules.
But interference that conforms to rules, and is non-arbitrary in that sense, may
still be uncontrolled by you and can count as arbitrary in our sense. In another
usage, arbitrary interference is interference that is wrong or objectionable,
so that what is arbitrary from one evaluative standpoint may not be arbitrary
from another. But while uncontrolled interference is going to count as
uniformly objectionable in most moral views, even those who adopt a contrary
view – even someone, for example, like Hobbes – ought to be able to agree on
whether, and to what extent, an act of interference counts as uncontrolled
and arbitrary in our usage. On that usage the term has a perfectly descriptive,
determinable meaning and people can agree on when it applies and when it
does not apply, independently of differences in the values they espouse; it is
not a value-dependent or moralized term.34
It may be useful to add one further comment about treating licensed or
controlled interference as non-invasive of free choice. This is that the con-
cession does not allow paternalistic interference. When I am being paternal-
istic I interfere in your choice according to your interests, though not
necessarily according to your wishes. But interests are always open to inter-
pretation, since they consist roughly in the wishes that you would have for
yourself, if you were able to view your predicament accurately and rationally
34
For an understanding of ‘non-arbitrary’ that preserves the rule-conformity connotation, though
without creating a great gulf between us, see Lovett (2010). For an evaluative understanding of the
term, but again one that does not create any substantive divergence of views, see Richardson (2002).
For a critique of my position that turns on related issues see McMahon (2005) – and for related
discussion, Costa (2007) – and for my response to that critique see Pettit (2006a).
No domination without invasion 59
(Geuss 1981; Smith 1994). And to the extent that I impose my own inter-
pretation on your interests, discounting yours as inferior, I act paternalistically
(Shiffrin 2000). Such paternalistic intervention, in the nature of the case,
involves interfering according to my own arbitrium, or ‘will’, not yours, and is
an exemplar of domination. The most that might be allowed on the repub-
lican view is interference according to interests that you are disposed or ready
to avow, where that readiness is easily tested and established; only this could
give you the control required to avoid domination (Pettit 2001e).
4. no domination without invasion

The considerations presented show that I do not reduce your freedom in a
choice to the extent that I actively interfere without domination: that is, to
the extent that you control my interference. But the paired issue we have to
consider is whether I can invade your freedom in the exercise of a choice,
not by means of any activity – not by interfering with any of the options –
but just by having an uncontrolled power of interfering with the choice of
any option: just by dominating you.
By the argument provided so far, you are unfree in a choice to the extent
that you are subject to my will – or, of course, the will of any other agent – in
making that choice. I can impose my will on you in a choice between X, Y
and Z by taking steps, uncontrolled by you, that change the cognitive or
objective profile of the options. But it turns out that you can be subject to my
will in a choice, and this subjection can change the profile of options you face,
even if I do not actively impose my will in this way. You will suffer that sort of
subjection to the extent that I have the uncontrolled power of interfering in
the choice – the unvitiated, uninvaded capacity to interfere. You will suffer
that sort of subjection, in other words, to the extent that I dominate you.
Suppose that I have such a power of uncontrolled interference in your
choice between X, Y and Z. And suppose that I do not actually exercise that
power, because I am favourably disposed towards you. While I am not
constrained to let you choose as you will, I view you with affection or
indulgence and am happy to let you choose according to your own wishes.
Even in this benign case, it turns out that there is a sense in which you are
subject to my will and a sense in which your options are not what they
would have been in the absence of such a power on my part.
You will be subject to my will, in the sense that you depend on my will
remaining a good will, for the ability to choose as you wish. Let me remain
benevolent and you will be able to choose X or Y or Z, as you please. Let
me become malevolent, however, and this will cease to be so: I will exercise
my power of interference so as to change the profile of the options before
you. This dependency on the state of my will constitutes a form of
subjection to my will, even when I do not actively impose my will on
you. The subjection shows up in an alteration of the options you face. You
will no longer confront a free choice between X, Y and Z. Rather, the
objective or cognitive profile of the options will be changed so that what
you confront is a choice between X-if-it-please-me, Y-if-it-please-me and
Z-if-it-please-me. The options will alter in a manner that parallels the
alteration imposed by the removal, replacement or misrepresentation
associated with unlicensed and uncontrolled interference.
But why should you care about subjection to me, if I am a wholly benign
master? Why should you care about the possibility of my interfering with
you if, as it happens, you think my benevolence runs so deep that it is
vanishingly unlikely that I would ever interfere? Shouldn’t your concern
about the possibility of interference on my part, or on anyone else’s, be
scaled to the probability of such interference?
The answer is that you could think of me in this purely probabilistic way
only if you adopted the sort of objective stance in which I cease to deserve
reactive attitudes like gratitude or resentment and become reduced to the
status of a force of nature: in this case, a wholly benign force of nature
(Strawson 1962; Pettit and Smith 1996). If you treat me as a regular human
being operating in regular circumstances you must be ready to feel gratitude
or resentment, approval or indignation, as my choices merit: you must be
ready to hold me responsible for the choice. And so you must think of me as
having the capacity with any of the options I face to take one or to take
another; that is a precondition of being poised to treat me reactively, holding
me responsible for what I choose. From within the perspective that you are
more or less bound to adopt, then, you will see me as a centre of choice for
whom interference is equally available with non-interference: as someone for
whom the question of whether to interfere or not is an open issue and as
someone, therefore, on whose goodwill you depend. Thus my access to
interference is going to be challenging, even if you think it unlikely that I
will interfere. The accessibility of the interference will signal your dependence
on my will quite independently of the probability of my actually interfering.
Invigilation and intimidation

There are a number of ways in which I may dominate you without any
interference, as there are many ways of practising interference itself. They
divide into two broad types, invigilation and intimidation. While these
normally go together they can also come apart: there can be invigilation
without intimidation and intimidation without invigilation.
In invigilation I stand guard over what you do, ready to interfere should my
will tend that way. As things are, I may not be disposed to interfere in any way
but even so I remain in a position where I can and will interfere in the event of
a change in my disposition. As a result of this invigilation, and whether or not
you are aware of the fact, you will be subject to my will; you will be dependent
on my goodwill for retaining the capacity to exercise choice. The options you
face, then, will not be X, Y and Z, just as such, but rather X-if-it-please-me,
Y-if-it-please-me and Z-if-it-please-me. In other words, the objective profile
of the options will be changed by my presence so that, in the old phrase, you
can choose as you wish but only cum permissu: only ‘by my leave’, or ‘with my
permission’. Being able to choose as you wish you may be said to be at liberty
to choose X, Y or Z, but to be at liberty in that sense is not to be free in the
sense of escaping subjection to the will of another. Your choice may reflect
your will, but it does so only because your choosing according to your will is
something that accords with my will. I am in the position of a master.
If I invigilate your choice and you become aware of the fact, then the
options will change their cognitive, as well as their objective, profile. In that
case you are likely to suffer intimidation as well as invigilation, recognizing
your dependency and vulnerability. The intimidation will boost the effect of
the invigilation, giving you an incentive to be cautious and deferential, in the
same way that recognizing that I will penalize a particular option will boost the
effect of the penalty, giving you an incentive to avoid it. Indeed, it may lead
you to behave as I would obviously prefer that you behave, even when I am
willing to let you have your own way; it may ensure that my presumptive
preference guides your choice without my having to lift a hand or utter a word.
Intimidation may also have this effect, however, in the absence of invig-
ilation. In the standard case of invigilation-cum-intimidation, the resources of
interference on which I rely are independent of your belief that I have them
and of the intimidating effect of that belief. But even if I have no independent
resources, and even if I cannot strictly practise invigilation, the fact that you
believe that I have the resources, and the fact that this intimidates you,
will still give me power over you. It will enable me to deceive you, say by
making a credible threat – though, of course, a bluff threat – to penalize a
certain option, and so it will enable me to subject you in a certain measure to
my will. It will mean that the options you take to be present are not X, Y and
Z but rather X-if-it-please-me, Y-if-it-please-me and Z-if-it-please-me. This
observation illustrates the truth of Hobbes’s (1994b: 10.5) remark that whether
or not it is well founded, the ‘reputation of power is power’. It gives us reason
to recognize that intimidation counts, side by side with invigilation, as a way
in which I may subject you to my will, even when I do not actively impose
that will in interference.35
These observations are enough to show that I can reduce your freedom in
a choice to the extent that I can expose you to the possibility of interference:
to the extent that I have a power, uncontrolled by you, of interfering in your
choices. One thing worth noting in particular is that I may do this even
without wishing to have or exercise this power. Suppose that I have the
power on such a basis that I can neither renounce it nor do anything to
reduce it. I cannot help knowing how you act, so that I have the cognitive
resources required for interference. And I have the objective resources on an
inalienable basis – say, because of having certain legal advantages, more
physical prowess or greater social clout. I may wish not to dominate you in
such a case – my will may be that you are not dependent on my will – but
whether I like it or not you are dependent and I exercise domination. You
are dependent on my will, and subject to my will, even though it is not my
will that you should suffer such domination.
It may seem quixotic to hold that I can subject you to my will without it
being my will that you should be subject to me in that way. But there is
nothing impossible about such unwilled domination. It is likely to be
exemplified in any relationships of inalienable asymmetrical power, as with
the husband over the wife in a sexist culture, the employer over employees in
an unregulated economy, the professor over students in a traditional college.
If I dominate you in this way, you will depend on my will in the sense that
how you fare in relevant choices – whether or not, for example, you can
choose between X, Y and Z – is going to vary as I vary in my will towards you;
you will retain access to those options so long as I remain goodwilled, but you
will lose access as soon as that goodwill decreases or disappears. But I will not
relish this subjection that you suffer and, could I do something to rectify it,
I would. Our relationship means that I subject you to my will, and the fact
that I shrink from the domination thereby imposed means that I subject you
to my will non-intentionally.
35
Note that if we include the non-standard resources activated under your belief that I have a power of
interference, intimidation will count as a form of invigilation; the distinction depends on requiring
invigilation to depend on standard, presupposed resources of interference. Note also that while
intimidation without invigilation depends on your falsely believing that I have a power of interfer-
ence, it is distinct from deception, considered as a particular, intentional act of deceiving you about
the options available in the relevant choice. This is to clarify how I keep the books on these questions,
but not to suggest that it is a uniquely appropriate way of doing so. I am grateful to Dorothea Gadeke
for discussion on this point.
Can I alienate my power of interference, and suspend the domination
I practise, by persuading myself that it is wrong to interfere in your choice?
Will the moral constraints that I thereby recognize act as checks on my will
and serve to liberate you, whether wholly or in part? I do not think so. Even
if I form the view that it is wrong to interfere with you, it remains the case
that I may prefer to practise interference, whether out of weakness of will,
out of malice, or out of a will for evil; the option continues to lie within my
capacity. My power of interference in your choice, and my domination over
you, can only be contained by external checks that remove or replace the
interference option or put it cognitively off the menu. My believing in the
wrongness of interfering will only introduce an external check to the extent
that I am thereby exposed, as I see things, to the penalty of moral disap-
probation for behaving in that way.
For those who continue to baulk at the idea that domination without
interference constitutes an invasion of your freedom, subjecting you to
my will, there is one last consideration worth introducing. This is that such
subjection is enough in itself to trigger the resentment that, as we saw,
typically characterizes the invasion, as distinct from the vitiation, of choice.
You may not resent me for the power that I am given by social convention or
natural prowess over you, since I am not responsible for having such power.
But it would only be human of you to resent me for acquiescing in the power
and, in particular, to resent the failure of the society not to rectify that
imbalance and affirm your equal standing. You may burn with such resent-
ment, as documented in oral and literary tradition, even as you recognize that
I do not want to have a power of interference and even think it wrong to
exercise that power against you.
These observations are interesting in themselves, but they also suggest that
we should recognize an indirect or structural form of domination as well as the
direct or personal kind, willed or unwilled, that we have been describing
(Hayward 2011). It is usually because of the ways a society is organized,
culturally, economically or legally, that some people have such power in
relation to others that they dominate them directly, and dominate them
without necessarily wishing for domination or even approving of it. Thus it
is usually because of the way that marriage law or workplace law is structured
that husbands or employers have a dominating power over their wives or
workers. These modes of organization may vitiate, but not invade, choice, as
when they emerge for example from customary practice, but they can indi-
rectly facilitate the worst forms of invasion and domination in a society. The
republican theory of justice, which we discuss in the next chapter, is designed
in great part to target such structural domination, looking for measures
whereby people can be assured on a public basis of not being dominated by
others in the broad range of the basic liberties.
Three theories of freedom

Let us assume, as argued earlier, that any plausible theory of freedom should
accept that free choice requires the absence of vitiating factors. With that
assumption in place, we can distinguish three broadly different theories of
freedom, each giving a different account of what suffices for freedom on
the invasive front. One theory, Hobbesian in inspiration, would say that
freedom of choice consists in the absence of invasive obstruction to your
actually preferred option, thereby equating freedom with non-frustration. A
second, in the style of Berlin, would hold that freedom of choice consists in
the absence of invasive obstruction to any option, preferred or unpreferred,
thereby equating freedom with non-interference – non-frustration, actual
or counterfactual. And a third, republican in character, would hold that
freedom of choice consists in the absence of a power of interference on
the part of any other, thereby equating freedom with non-domination. In
bringing the discussion to a close it may be helpful to see that Berlin’s
argument against Hobbes – in effect, the argument for freedom as non-
interference and against freedom as non-frustration – suggests a parallel
argument for freedom as non-domination and against freedom as non-
interference (Pettit 2011b).
The argument against freedom as non-frustration is that it leads to an
absurdity; it entails, contrary to powerful intuitions, that adapting your
preferences can give you freedom of choice. The argument against freedom
as non-interference is that it leads, by a parallel argument, to a parallel
absurdity; it entails, contrary to equally powerful intuitions, that ingratia-
tion – toadying, kowtowing and cosying up to the powerful – can give you
freedom of choice. I understand ingratiation in this context as an interven-
tion that wins the indulgence of the powerful without exposing them to any
cost or penalty, not even the cost of disesteem.
I set out the argument in steps that highlight the parallel with Berlin’s
argument against Hobbes.
1. Suppose with Berlin that you enjoy freedom in a choice between X and Y
just to the extent that you avoid interference; suppose that non-interference,
with or without domination, is enough for freedom.
2. By that supposition, you do not enjoy freedom of choice in the case
where I have a power of interference and, being ill-willed, am disposed to
interfere with one or the other option.
3. But, by supposition, you would enjoy freedom of choice in that case if I
were disposed, notwithstanding my power of interference, to interfere
with neither.
4. If you know the situation, then, it appears that you can give yourself
freedom of choice, without reducing my power of interference, just by
ingratiating yourself with me and getting me to let you have your way.
5. But this is absurd. You cannot give yourself freedom in the choice just by
accommodating yourself to my power of interference.
6. Thus the original supposition that non-interference is enough for free-
dom of choice must be false.
The view that non-interference is enough for freedom of choice entails that
ingratiation is a possible means of liberation in the same way that Hobbes’s
view that non-frustration is enough entails that adaptation is a possible means
of liberation. And this entailment argues against the theory of freedom as
non-interference as the corresponding entailment argues against the theory of
freedom as non-frustration. Let the anti-adaptation assumption be granted
and the non-frustration theory must fail; let the anti-ingratiation assumption
be granted and the non-interference theory must fail.
The problem is not that adaptation or ingratiation in the relevant contexts is
intuitively objectionable, as it surely is, or even that it will occur very often,
which it may not do. The problem, rather, is that neither adaptation nor
ingratiation counts intuitively as a means of liberation in a given choice and any
theory that entails that it can serve such a liberating role has to be inadequate.
Adapting your preference so as to want what I let you get takes my interference
as a given and tries to secure a decidedly second-best result: the avoidance of
frustration. And ingratiating yourself with me in order to win my permission
for you to do what you want takes my power of interference, and so your own
subjection, as a given, and tries to secure a somewhat better but still second-best
result: the avoidance of actual interference. In each case we want to say: this is
not freedom in the most attractive sense – the sense in which it requires the
availability of each option and independence from the will of any other as to
what you do. There is a Pickwickian flavour to claiming that you have won
your freedom in a choice by dint of adapting to my interference or in virtue of
softening me up. Each is a way of sugaring a bitter pill, not a means of rejecting
the need to swallow it.36
36
The point is more easily lost in the ingratiation case, it should be noticed, than in the case of
adaptation. We can readily imagine cases in which one single episode of ingratiation secures access for
you to a range of undominated choices. And in such cases it is easy to forget about the question of
whether the choice in which ingratiation enabled you to get your way was itself free – I say it was not –
and to focus on the fact that in view of the new choices that become available, the ingratiation was
Adaptation may be a rational response to the frustration imposed by
another, since it does indeed give you a second-best: if not freedom to satisfy
your preference, regardless of what you prefer to do, at least the formation of
a preference you can satisfy. And ingratiation may equally be a rational
response to another’s power of interference, since it also gives you a second-
best: if not freedom to satisfy your preference, regardless of what others
prefer that you do, at least enough breathing space to let you act as you will.
It may not give you the benefits of a secure peace in your relations with the
powerful, but it will give you the consolations of a fragile cease-fire.
The gain secured in each case allows us to say, in ordinary usage, that by
contrast with the situation where you do not make such responses you can
win a degree of freedom for yourself by adaptation or ingratiation. This is
because in ordinary usage, as we have seen, talk of freedom is often used to
mark context-sensitive contrasts, so that one and the same choice may be
described as free or unfree, depending on the contexts under comparison.
But no matter what the context of usage, the benefit that you win by
adaptation or ingratiation is not freedom in the regimented sense that we
have been developing; in this sense, freedom of choice requires that every
option be accessible and that your access to any option be independent of
the preference of another as to what you should do.
We can give nice expression to this sense of freedom by elaborating on
Berlin’s open-doors metaphor. Where the anti-adaptation assumption
means that all the options in a free choice have to be open doors, the
anti-ingratiation claim implies that not any mode of openness will do. Are
you free just insofar as both doors are open in the choice between X and Y?
Not necessarily. What freedom ideally requires is not just that the doors
be open but that there be no door-keeper who has the power of closing a
door; there is no door-keeper on whose goodwill you depend for one or
another of the doors remaining open. If I am in the position of such a
door-keeper, therefore, your access to the X and Y options is not sup-
ported in the manner that freedom of choice strictly requires.
As the plausibility of the anti-adaptation assumption argues that all the
doors in a free choice must be open, then, so the plausibility of the anti-
ingratiation assumption argues that there must be no dependence on the
liberating in a distinct sense. Suppose, to go back to our old example, that you ingratiate yourself
sufficiently with the warden or parole board to have a choice between remaining in prison or living
outside. The ingratiation will mean that you did not have a free choice in that instance but it will be
liberating in the distinct sense that it enables you now to enjoy all the undominated choices that
become available in the world outside.
good graces of a door-keeper. When you ingratiate yourself with me and I
let you go by a door that I would otherwise have closed, you do not cease to
be subject to my will. You have not escaped the constraint that made you
unfree in the first place, nor done anything to reduce the effectiveness of the
constraint. While continuing to operate under the yoke of my will, you have
merely adjusted so as to make your life more comfortable.
Freedom, robustness and probability, again

In holding that all the options in a choice should be open doors, the view of
freedom as non-interference implies that ideally you must be unfrustrated
in the actual world where you prefer X and also in the possible world where
you prefer Y. It requires that you should enjoy non-interference, not just
actually, but robustly: that is, in a range of possible worlds associated with
the available options, however unlikely some of those worlds may be.
In requiring that all options should be open, and in addition that there
should be no door-keepers, the theory of freedom as non-domination
implies that non-interference should be modally robust in a greater measure
still (Pettit 2001a; List 2004).37 It is not enough to enjoy non-interference
across a range of worlds in which your preference or will changes, now
seeking X, now seeking Y. For full or ideal freedom of choice, you also have
to enjoy non-interference across a range of worlds in which there are also
changes in the preference or will of other agents as to what you should do.
Thus you have to enjoy non-interference in X-worlds where others are
friendly, but also in X-worlds where others are unfriendly. And you have to
enjoy non-interference in Y-worlds where others are friendly and Y-worlds
where others are unfriendly. You have to enjoy non-interference in a way
that is invariant across relevant alterations both in your own will and in the
will of others as to what you should do. We saw in Section 1 that in order to
increase the probability of your freedom in an X–Y choice, on the theory of
freedom as non-interference, we have to minimize the following sum,
assuming that addition is appropriate: P(H if X) + P(H if Y). Things are
going to be a bit more complex if we want to increase the probability of your
freedom as non-domination, for we will have to allow for worlds where
others are unfriendly (U) as well as friendly (F). Sticking with the additive
37
For a somewhat variant treatment see List (2006b) and the discussion of freedom, which follows this
treatment, in List and Pettit (2011: Chapter 6).
version of the approach, we will want to minimize the sum: P(H if X & F) +
P(H if X & U) + P(H if Y & F) + P(H if Y & U).38
Minimizing this sum will mean maximizing expected non-domination.
It will involve keeping or putting barriers in place against the possible
interference of those who would otherwise have a power of interference in
your relevant choices. And it will mean keeping or putting obstacles in place
against the formation of agents or agencies that could have such power, if
they did come into existence. Thus it might mean taking steps to guard
against the possibility of a loose collection of individuals – say, a belligerent
minority or a potentially oppressive majority – incorporating to form a
dominating presence in the lives of others.
I add one final comment on the relationship between domination,
interference and frustration. We noted earlier that hindrance with frustra-
tion is more harmful than hindrance without and that hindrance without
frustration gets worse – it holds out the prospect of greater harm – as the
probability of frustration increases. Equally, domination with interference is
intuitively more harmful than domination without, so that we can associate
progressively greater levels of harm with the following three conditions:
domination without interference or frustration; domination with interfer-
ence but without frustration; and domination with both interference and
frustration. And correspondingly, we can say that domination without
interference gets worse as interference becomes more probable, and that
interference without frustration gets worse as the probability of frustration
rises. My inclination is to cast the extra harms imposed by interference and
frustration as distinct harms from the loss of freedom but, as I mentioned in
38
In an earlier footnote on freedom as non-interference I suggested that in an X–Y choice it is not
enough for such freedom that you enjoy the absence of hindrance in the actual world where you prefer
X and in the nearest possible world in which you choose the other option, Y; it is required, plausibly,
that you can get X in any of a wider range of worlds, including the actual one, in which you prefer X;
and can get Y in any of a wider range of possible worlds, including the nearest possible one, in which
you prefer Y. I take a similar view of the range of possible worlds in which you must avoid interference
if you are to enjoy non-domination. And here as in the other case, I think that this range of worlds is
discernible only on an intuitive, context-sensitive basis and, of course, that it does not include all
possible worlds. The worlds where non-domination requires that you choose X without interference,
for example, may vary in the precise physical movements whereby your X-ing is realized and in the
degrees of friendliness or hostility that others display. But they do not extend to all possible worlds in
which you choose X. Thus you are not unfree to make the choice between X and Y just because you
would suffer interference with X, should those in whose presence you act not only turn hostile, but
also develop super-human powers. The need to rely on contextual assumptions to determine the
range of relevant possible worlds is regrettable, but from the point of view of formalizing the theory of
freedom it seems inevitable. As mentioned in the earlier footnote, the indeterminacy is akin to that
which arises in epistemology when we ask about the range of possible worlds in which true belief has
to be present – when we ask about the extent to which true belief has to be robust – if there is to be
something deserving of being called knowledge.
How can we ensure your freedom of choice? 69
discussing hindrance and frustration, there is no objection in principle to
seeing them as freedom-related harms.39 We might well think that while
domination is enough for the loss of freedom, interference and frustration
make the loss more severe still.40
5. how can we ensure your freedom of choice?
Resourcing and protecting

We have concentrated in this chapter on how you can fail to be free in a
choice between options like X, Y and Z. We argued that you can fail through
having any of the options hindered and that hindrance may take the form
either of vitiation by impersonal factors or of invasion by another agent or
agency. And we saw that invasion – subjection to the will of another – may or
may not involve the interference of the other, but requires in either case that
the other has a certain power of uncontrolled interference and counts in that
sense as dominating the choice.
How might we ensure, then, that you are free in the exercise of a choice?
Two things are required. First, we would have to resource or facilitate the
choice in the sense of ensuring that any of the required resources you happen
to lack – these may be personal, natural or social – are made available to you;
we would have to compensate, in other words, for any vitiation of the choice.
And second, we would have to protect you in the exercise of that choice;
assuming that protection is needed, we would have to guard against your
being subject to the will of another in how you exercise it: that is, against
invasion.
Resourcing or facilitating the choice could involve many different sorts of
things. Were it a choice like driving or cycling to a certain destination, for
example, it might involve making sure you have a car and bicycle, that you
possess the skills that driving and cycling require, that there is both a network
of roads and a set of conventions governing their use, and that you know the
rules of the road and the route to the destination. It might even involve
helping you to make up for certain unusual physical or psychological defi-
ciencies, providing suitable assistance or therapy. Most of the resources
39
If we do take this view then, by analogy with a point made in discussing the interference-frustration
case, we will have to give some weight to the probability of interference and to the probability of
frustration. See the earlier footnote on the interference-frustration case.
40
This observation shows how we may take on board the lesson emphasized by Sen (2002) – in part,
against the republican view that he finds in my work – that there are many dimensions relevant to the
measurement of freedom.
required for exercising the choice will be available to you independently of
any help from us and these steps will usually be unnecessary. But they
illustrate what resourcing the choice might come to.
I mentioned earlier that you will not be suitably resourced – you will not
be resourced in a way that gives you the capacity presupposed to freedom –
if you have to depend on the goodwill of another for getting the resources.
The rationale for that stipulation should now be clear: if you depended on
me for the resources needed to choose an option, then I would have a
power of uncontrolled interference in the choice and would dominate
you. This stipulation means that any resourcing we provide for you will
be suitable only if we can provide it in a way that does not introduce
dependence on any particular agent or agency. We would avoid introduc-
ing dependence by resourcing you on the same grounds that we resource
each one of us, as in the case of providing for shared roads and shared rules
of the road. And with resources that you in particular need, we might avoid
introducing dependence by giving you claims under general rules to the
resources needed.
Protecting, as distinct from resourcing, the choice involves guarding you,
where that is necessary, against subjection to the will of any other person or
group in how you exercise the choice. Such protection will be focused on
the potential interference of any other and will put blocks or burdens in
place which ensure, so far as possible, that no one has a power of interfering
in the choice – or at least that no one has a power of interfering that is not
controlled by you.
The most important point to register about protecting against such
interference – strictly, against such a power of interference – is that it is
quite distinct from promoting non-interference: that is, taking steps that
make non-interference as likely as possible. We might make interference
very unlikely, even more unlikely than under a protective regime, by buying
off potential interferers: that is, by offering them rewards for not interfering
with you. The fact that someone was persuaded on that basis not to interfere
in your choice, however, would not release you from subjection to their will.
They would still have access to the option of interfering, for all that our
rewards ensured; the rewards would just mean that apart from not interfer-
ing, period, they would have the further option of not interfering and
accepting our reward. The point in protection is not to maximize expected
non-interference but to maximize expected non-domination. It means
taking steps to ensure that you enjoy safeguards against interference that
remain in place across variations both in what you prefer to choose and in
what others prefer that you choose.
The requirements of protection
How are we to protect you against someone’s interference in one or another
choice? It might seem that we should just reduce the person’s freedom of
choice between interfering and not interfering. But that is not right, for two
important reasons.
The first is that the person may be denied the freedom to interfere or not
interfere because of facing a check against non-interference rather than
against interference. We will not protect you if we check someone’s option
of not interfering, thereby reducing their freedom to interfere or not to
interfere. If the check is serious enough – if it involves removing the option
of non-interference, for example – it will mean that strictly the person does
not interfere with your choice, not intentionally restricting it, but in that
case we will ourselves interfere insofar as we use them to restrict the choice.
And if the check is not so serious, it will allow the person to continue to
interfere, intentionally restricting the choice, and give us a role or a share in
that interference. In order to protect you against another person or party, we
must try to reduce their capacity to interfere, not their freedom to choose
between interference and non-interference.
The second reason why protecting you against someone’s interference
must do more than reduce the person’s freedom of choice between interfer-
ing and not interfering is that we can replace the interference option in
that choice, and reduce the person’s freedom, just by imposing an actual
or apparent obstacle or penalty – or an actual or apparent probability of
obstacle or penalty – that has little defensive or deterrent value. To protect
against interference is not just to interfere in any old way with the agent’s
interference option. In the limit case it is to prevent interference: to impose
an indubitable, insurmountable obstacle. And short of that limit it is to
impose such a probability of such an obstructive obstacle or such a deterrent
penalty – in either case actual or apparent – that the person’s power of
interference in your choices gets to be reduced to a point where, by local
standards, you have no good reason to be anxious. Interference may not be
impossible for the other but it will incur such difficulty or such danger for
the interferer that it is relatively unacceptable or ineligible.
We said earlier that someone will dominate you and compromise your
freedom in a choice to the extent that they have a power of uncontrolled
interference with one or another option. While preventive protection may
be preventive under a smaller or larger range of variations in the attitudes of
others – and while in that sense it can come in degrees – it removes that
power in a radical way. Other forms of protection do not have the same sort
of impact but they can reduce the person’s power of interference to the
point where interference is effectively ineligible and not interfering is a
forced or involuntary choice (Olsaretti 2004).
We need not worry about the subjectivity of this benchmark for deter-
mining whether protection is adequate, if in the local culture there is an
accepted sense of where the benchmark lies. In the next chapter, as already
advertised, I will be introducing the eyeball test to determine what level of
freedom you must enjoy in the exercise of the basic liberties, as I call them,
in order to count as a free person or citizen. The test suggests that you
should have access to a level of resources and protections for those choices
that enables you, by local standards, to look others in the eye without a
reason for fear or deference. It will provide the required benchmark for
adequacy in protection.
We know that some forms of interference are going to be less serious than
others, say because of involving the replacement of an option, X, by an
option that is just barely different from it: the option of doing X with a little
discomfort, or at a little extra cost. The effect of the eyeball test will be to
require a degree of protection against others that makes the interference
option ineligible, if not actually impossible, and that does this more and
more effectively with progressively more serious forms of interference.41
While protection against interference comes in degrees for these reasons, it
is worth stressing that this in no way softens the contrast between enjoying a
low prospect of interference that is induced by protective measures and
enjoying a low probability of interference that is just based on the unlike-
lihood of potential interferers turning nasty. It is only when you are pro-
tected, however probabilistic the protection may be, that you can achieve a
degree of independence from the goodwill of others for being able to choose
as you wish. The decreasing probability of (uncontrolled) interference is a
good, as we saw earlier, even if it comes of an increasing goodwill, perhaps
induced by your ingratiation, on the part of the powerful. The decreasing
probability of (uncontrolled) interference that is induced by protection is
good in the same way and also good in a second: not only does it make
41
This line of thought is connected with the old slogan, according to which you are unfree in a choice
when you are exposed to forms of interference, presumptively serious, that others can impose at will
and with impunity. Others will be able to interfere at will insofar as they are not prevented or not
likely to be prevented from interfering. And others will be able to interfere with impunity insofar as
they are not subject to a penalty that inhibits and deters interference, making it effectively ineligible.
We can make you free, so the idea goes, to the extent that we can prevent or inhibit interferers in a
suitable degree: we can deny them interference-at-an-acceptable-cost.
(uncontrolled) interference less likely, it also reduces the power of interfer-
ence on the part of the powerful.42
Actions and omissions

When we are in a position to resource and protect your freedom, there is
going to be a question about whether the failure to provide resourcing and
protection counts as inimical to freedom in the same way as the invasion of
that choice. Is the failure or omission that materializes here as serious in
freedom terms as the more active harm? And is the power of omission as
serious therefore as the power of interference that constitutes domination?
In some circumstances omission or failure of the kind envisaged may be
as morally, politically or even legally culpable as invasion. But a failure to
resource or protect is not itself a way of invading that capacity, nor is having
the power to fail a mode of domination. If such failure is an offence, as it will
often surely be, it is a distinct offence from invasion.
That said, however, there is a very fuzzy line between certain failures of
resourcing and protection and actual invasion. Suppose that I provide for
your resourcing and protection in a certain range of choice and that, as time
goes on, this comes to represent the default expectation, shared between us,
as to what is going to transpire. In the context of these expectations, and the
pattern of one-sided reliance it sets up, the negative action of refusing
further help can be indistinguishable from an invasion of your free choice.
Indeed, my assumption of such a routine resourcing or protective role may
place me in a position where I dominate you. You will depend on my
goodwill for my continuing to provide for you, and, as it will appear in the
new context, for not invading your relevant choices. Or at least this will be
the case insofar as I am the only one able to help – there is not a queue of
people lined up to take my place – and there are no pressures that more or
less constrain me to do so.
Our discussion of the nature of resourcing and protection leads us
naturally into the project pursued in this book. I shall argue in the coming
chapters that the freedom as non-domination of its citizens is of paramount
importance for the state insofar as it is interested in justice or legitimacy and
that its first duty in both justice and legitimacy is to try to make its people
42
The point of this paragraph might be nicely illustrated with the help of a graph in which the vertical
axis represents increasing opportunity costs on the interference of others in your affairs and the
horizontal represents increasing protective costs. The indifference curves that represent equally
valuable bundles from your point of view will be skewed in a way that reveals the greater value at
any level of interference of being shielded by protective rather than opportunity costs.
free. Subject to the constraint of treating its citizens as equals, the state
ought to be guided by the idea of promoting such freedom, putting in place
the resources and protections that can guard people against domination.
In arguing that the state ought to promote freedom in this manner, I
endorse a broadly teleological or consequentialist view of the state’s task
(Pettit 1997a, 2001d, 2012a). The grounds for taking the consequentialist
approach are that it delivers satisfactory theories of justice and legitimacy:
that is, theories that fit well with our considered judgements and satisfy
reflective equilibrium. We shall see how the promotion of freedom as non-
domination is to be interpreted, under the guidance of republican tradition,
in our discussion of justice and legitimacy. And we shall see that a state that
promotes freedom under that interpretation of the project is going to
provide us with theories of legitimacy and justice that have a powerful,
independent appeal.43
43
As we shall see, I think that the state should pursue people’s enjoyment of freedom as non-
domination under the normative constraint that it treats all its citizens as equal. But this does not
involve giving up on consequentialism. It merely implies that the state’s treating people as equals is a
good in its own right and has to be built into the strategy that the state adopts. Consequentialism does
not put any restriction on the goods that may be given countenance; it merely stipulates that the
strategies it is right for any agent or agency to adopt are a function of the goods that it can thereby
expect to promote.
chapter 2
Social justice
Assuming for the moment that state and government are required in order
to organize life in a society on a coercive basis, there are two large questions
that any political philosophy must address in the context of that society.
First question: what decisions or policies should the state impose in order to
establish social justice in the relationships between its citizens? Second
question: what processes of decision-making should it follow, if it is to
count as a politically legitimate decision-maker for its citizens on questions
of justice, and indeed on related matters too? Both questions might be
treated as questions of justice in an encompassing sense of the term, but
since I think they are importantly distinct, I shall cast the first as a question
of social justice, the second as a question of political legitimacy.
I take citizens in this discussion to comprise, not just citizens in the
official sense, but all the more or less settled residents of a state who, being
adult and able-minded, can play an informed role at any time in conceptu-
alizing shared concerns and in shaping how the state acts in furthering those
concerns (compare Tully 2009: i, 3). Special issues of justice and legitimacy
arise with those who are not adult or not able-minded, with those who are
not permanent residents, and indeed with those who are not yet born, as
related issues arise, of course, with the treatment of other animals. But I
shall almost exclusively concentrate on the general issues that arise for how
the state should treat current citizens in my broad sense.
I shall take the state and the government to be corporate entities related
in such a way that the state acts through government agents or agencies –
legislative, executive or judicial – and whenever those agencies act, the state
acts through them: they act, as it is said, in the state’s name. The issues of
social justice and political legitimacy that I discuss can be equally well cast
either as issues involving the state or as issues involving government and I
shall make use of both formulations. In each formulation, the issues raised
presuppose that the relevant entity, state or government, is an agent. While
it may involve the participation of many different individuals, it is disposed
75
76 Social justice
and expected to form and avow a coherent set of purposes and judgements
and to live up to them in how it behaves; it counts in that sense as a
corporate agent (List and Pettit 2011).
The state or government in every society imposes an order of law,
however we choose to define law, and the questions of social justice and
political legitimacy involve two different aspects of that order. The social
justice question is: does the state treat its citizens well and equally in
selecting the order that it imposes? The political legitimacy question is:
does the state treat its citizens well and equally in the way it imposes that
order? Each question can be taken as the universal form of a question that
every citizen is in a position to raise in his or her own case: does the state
treat me well in the nature of the order imposed and in the manner of its
imposition? The universal form is generated on the assumption that there is
no reason to privilege any one citizen over others and that the particular
question arises with the same force in everyone’s case.
The order of law that a state imposes will generally include substantive
provisions on how citizens should relate to one another and procedural
provisions on how they should relate to the state: provisions, for example,
on the rights of voting and of standing for office that they should have. Such
procedural provisions are relevant, particularly on the approach taken here,
to the issue of political legitimacy. And so, for reasons of convenience, I shall
take the social justice question to bear only on the nature of the order
imposed by the state as the determinant of people’s relationships with one
another, whether as individuals or as members of various groups. In asking
the social justice question as to whether the state imposes a suitable order,
then, the focus will be on how satisfactory the order is as a social order: as an
order that shapes people’s relationships with one another. In asking the
political legitimacy question as to whether the state imposes the order in an
appropriate manner, the focus will be – at least in part – on how satisfactory
the order is as a procedural order: as an order that shapes people’s relation-
ships with the state itself.
The issue of political legitimacy, which will lead us into democratic
theory, I postpone until the next chapter. The issue of social justice I
address in this.1 I have no quarrel with those who see the questions as
each bearing on a different aspect of justice in the more comprehensive,
social-cum-procedural sense – justice, for example, in the sense of John
1
A concern with social justice, as distinct from political legitimacy, is the sort of concern that Madison
had in mind in Federalist Papers, no. 51, when he said that it ‘is of great importance in a republic . . . to
guard one part of the society against the injustice of the other part’ (Madison, Hamilton and Jay 1987).
Social justice and equality 77
Rawls (1971, 1993) – but I shall continue to describe them in separate terms.
As will appear in the next chapter, I believe that treating the questions
together within a theory of comprehensive justice has led to a misconstrual
of the second question.
The position for which I will be arguing in this chapter is, roughly, that
the republican theory of social justice – depending on context, I shall often
say simply, justice – requires that people should enjoy freedom as non-
domination in their relationships with one another, whether as individuals
to individuals, as groups to groups, or as groups to individuals. And the
position for which I will be arguing in the next is that the republican theory
of political legitimacy – again, I shall often say simply, legitimacy – requires
that people should enjoy freedom as non-domination in their relationships
to their state or government. Republican justice is primarily opposed, on
this approach, to private domination; republican legitimacy to public. In
introducing the approach in the present chapter I shall assume, as men-
tioned, that a state is necessary and has a corporate character. I will try to
vindicate those assumptions in the next.
This chapter falls into four sections. In Section 1, I introduce the repub-
lican ideal of justice as the enjoyment of equal freedom as non-domination
amongst citizens, arguing that this can only mean equality in the enjoyment
of certain fundamental choices, the basic liberties, on the basis of a guaran-
tee of public resources and protections. When choices are safeguarded by
public resourcing and protection in this way I will often describe them for
convenience as being secured or safeguarded or entrenched, though
entrenchment in this sense does not necessarily connote constitutional
embedding. In Section 2, I use two plausible constraints to identify the
basic liberties that ought to be publicly entrenched in that sense. In
Section 3, I give an illustrative sketch of the policies that the republican
theory of justice would require; this is to gesture at an institutional model of
republican justice on a par with the institutional model of republican
legitimacy – in effect, republican democracy – that I seek to provide in
the final two chapters. And then in Section 4, I offer some comments on the
distinctive features of the emerging theory.
1. social justice and equality
The connection
Almost everyone will agree that the challenge in a theory of justice, social or
indeed comprehensive, is to identify in any context the ‘proper balance
78 Social justice
between competing claims’ that arise there (Rawls 1971: 9). In the domestic
context of a particular state I take the citizens to be the relevant claimants,
where the citizenry includes all adult, able-minded, more or less permanent
residents. And I take the state or the government – that is, the agency
whereby the state enacts its business – to be the addressee of their claims and
so the entity that has the job of identifying and enforcing the demands of
justice. So what are the claims that citizens have within the state and what
would constitute a proper balance between those claims?
Almost everyone will agree, not just that this is the central question in a
theory of domestic justice, social or comprehensive, but also that any
plausible answer to the question has to satisfy a certain egalitarian con-
straint. In order to establish a proper balance between the claims of its
citizens, the state must treat them as equals in determining that balance: it
must express or manifest an equal degree of concern for each of them
(Dworkin 1978). And so, whatever benefits the state makes available as a
matter of justice to its citizens, it is subject to the constraint of making them
available in an expressively egalitarian way. This constraint does not neces-
sarily argue that the state should adopt a policy that imposes a certain
substantive equality amongst citizens. In itself it requires only that in
selecting policies, the state should recognize that people count equally: it
should act on the principle that ‘all are entitled to respect’ (Raz 1986: 219).
The connection between social justice and expressive equality is scarcely
surprising, for the very paradigm of injustice is the scenario where those of a
certain caste or colour, religion, gender or ethnicity suffer discrimination
under the institutions established by the state. The just system, so the lesson
goes, cannot be a system that discriminates on any such basis between its
members; it is inherently impartial.
My discussion of social justice in this chapter, and my discussion of
political legitimacy in the next – my discussion, in other words, of each
aspect of comprehensive justice – starts from the normative assumption that
the state ought to be expressively egalitarian in this sense. The assumption
that the state ought to treat people as equals, satisfying expressive egalitari-
anism, implies on the side of the citizenry that they ought to be willing to
live in society under an arrangement where they are treated as equals: they
ought to be prepared, as we may put it, to live on equal terms with others
and not claim a special position for themselves. This means that any
complaints about their treatment that citizens can expect to command a
hearing must be consistent with a willingness to live on equal terms with
their fellow citizens: in effect, they must be publicly avowable in the context
of reciprocal exchange.
Why commit to expressive egalitarianism in this sense? The reason goes
back to a point registered in the introduction to the chapter. Each of us can
reasonably ask about how well the state treats us both in the nature of the
order it imposes and in the mode of its imposition. And the assumption that
we are each in a position to raise those questions with equal validity – an
assumption encoded in our being willing to discuss them with one another –
generates the universal counterparts. Thus a theory of social justice will
address the question as to what order the state must impose if in that respect
it is to treat its citizens well and equally. And a theory of political legitimacy
will address the question of how it must impose that order if in that distinct
respect it is also to treat them well and equally.
The two points of general agreement that we have registered take us only
a limited distance in thinking about the demands of social justice. Particular
theories of justice diverge on the good or goods in respect of which the state
should treat its citizens as equals: that is, on what it is to treat them well.
And they diverge also on what it is for the state to satisfy the equality
constraint in delivering that treatment: that is, on what it is to treat them
equally. They go different ways both on the substantive matter of what
should be pursued in an expressively egalitarian fashion, and on the meth-
odological issue of what expressive equalization entails.
Take the substantive issue, first of all. Libertarian theory holds that
the state ought to treat people as equals in protecting their purportedly
natural rights; depending on how rights are interpreted, this comes in two
versions, respectively right-wing (Nozick 1974) and left-wing (Vallentyne
and Steiner 2000a, 2000b). Luck-egalitarian theory holds that the state
ought to treat people as equals in distributing resources, except when
people can only blame themselves for their misfortune: the inequality
they suffer, as it is sometimes put, is a result of ‘option-luck’ rather than
‘brute luck’ (Dworkin 2000; Fleurbaey 2008). Utilitarian theory holds
that the state ought to treat people as equals in a concern for their utility
or perhaps their opportunity for utility (Arneson 1989; Roemer 1998).
Capability theory maintains that it ought to treat them as equals, rather, in
providing for their basic capacity to function in the local society (Sen 1985;
Nussbaum 2006; Alexander 2008). And mixed theories look for the
treatment of people as equals in respect of different goods in different
spheres (Walzer 1983) or in respect of any of the goods that confer a
presumptive advantage (Cohen 1993).
Rawls’s (1971) theory, to which we will return later, belongs in this mixed
category, for it requires the state to treat its citizens as equals in providing for
each the mix of primary or omni-functional goods – goods that are
80 Social justice
indispensable for people, regardless of their particular goals or wishes – that
are prescribed under his two principles of justice. These goods include
liberty and opportunity, income and wealth and a basis for self-respect.
Rawls formulates the principles of justice that are meant to deliver primary
goods as follows, with the first principle having lexical priority: that is, it
cannot be violated for any gain in the second.
a. Each person has an equal right to a fully adequate scheme of equal basic liberties
which is compatible with a similar scheme of liberties for all.
b. Social and economic inequalities are to satisfy two conditions. First, they must
be attached to offices and positions open to all under conditions of fair equality
of opportunity; and second, they must be to the greatest benefit of the least
advantaged members of society. (Rawls 1993: 291)
Whatever good or mix of goods they favour, theories of justice differ also
on the methodological issue of what it is for the state to treat people as
equals with respect to that good. In order to understand this point of
divergence it may be useful to think of the different ways in which parents
might think of treating their children as equals in providing from a fixed set
of funds for a good like education. Assuming that none of the children
suffers from cognitive or related problems, parents might follow any of five
salient, though not necessarily exhaustive, strategies.
* Equal investments. The parents might take expressive equality to require
providing exactly the same funds for the education of each child; this
would give the children equal opportunities for success.
* Equal direct returns. They might take it to require providing or refusing
such funds for each as ensure that, regardless of their different capacities,
the children all end up at the same educational level or position; this
would give the children equal educational outcomes.
* Equal derived returns. They might take expressive equality to require
providing funds for each that will enable the children to develop their
educational potential and thereby to achieve the same level of fulfilment
in their work and life; this would give the children equal outcomes in a
different currency.
* Equal rates of return. They might take it to require providing funds in
proportion to the children’s different capacities to gain educational

returns from the funds invested in them.
* Equal threshold benefits. Or they might take it to require providing
funds that enable the children each to enjoy a certain threshold benefit,
where this provision may allow some of the children to do better than the
threshold while others barely reach it.
As the parents in our little example might treat their children as equals
under any of these five strategies, so a theory of justice might require the
state to treat its citizens as equals in any of the corresponding ways. Each
strategy is expressively egalitarian insofar as it can be taken to reflect an equal
concern for members of the relevant constituency, children or citizens. But
only the first three can be cast as strategies that reflect a concern that equality
should prevail amongst those individuals (Temkin 1996). Those three
strategies are substantively as well as expressively egalitarian.2
The substantively inegalitarian character of the fourth, rates-of-return
strategy appears in the fact that the utilitarian project of maximizing overall
happiness in a society, however unequal the distribution amongst individ-
uals, exemplifies this approach. It involves investing resources at any time
wherever the rate of return is highest and doing so to the point where
resources are exhausted and the marginal rate of return from investment in
any one person is the same as the marginal rate of return from investment in
any other. The substantively inegalitarian character of the fifth threshold-
benefits strategy appears in the fact that it would allow some individuals to
do significantly better than others in relevant benefits, provided they all
reach the appropriate level. One example is Harry Frankfurt’s (1987)
sufficientarianism, which argues that all should be provided with what by
suitable standards is a sufficient level of resources for social flourishing. The
capability approach adopted by Amartya Sen (1985) and Martha Nussbaum
(2006) is also sufficientarian, on the face of it, since it argues that all should
be provided only with a basic capability for functioning in the local society.3
These observations give us a useful lead on what form we might expect a
republican theory of justice to assume. On the substantive question, they
suggest that it ought to make some version of freedom as non-domination
into the good with respect to which the state is required in justice to treat its
citizens as equals; otherwise it can hardly count as a distinctively republican
theory. And on the methodological question, they suggest that while the
theory has to be expressively egalitarian, it may take any of a variety of
forms, some of them substantively egalitarian, some of them not. I look now
at how best to interpret the goal of republican justice and then at what this
interpretation implies for the equality strategy that the theory ought to
endorse.
2
Another strategy that might be taken to reflect an equal concern for the children, while not being
substantively egalitarian, is prioritarianism (Parfit 2000). I discuss it briefly later in the text.
3
For discussions on how far the approach is sufficientarian see Part 1 of Kaufman (2006).
82 Social justice
The freedom goal of republican justice
In order to build further on these observations we need to identify in greater
detail the ideal of freedom as non-domination that the state should be
required to foster. In principle, the state ought to promote the enjoyment of
free or undominated choice amongst its citizens, under the expressively
egalitarian constraint of treating those citizens as equals. But what does this
prescription entail in practice?
In seeking a more concrete version of the prescription I propose that we
should take as a guiding heuristic the image of the liber, or ‘free person’, in
the republican tradition. That picture claims to represent a status in which
people can all enjoy freedom of choice fully, and yet also enjoy it equally.
Using this heuristic as a guide to what republican justice would require
presupposes that the circumstances of the society are not so dire, nor the
disparities between people so severe, that it is impossible for people gen-
erally to achieve the status of a free person. This is just to require that the
circumstances of justice obtain, in John Rawls’s (1971: 126–30) phrase: that
is, that the society is operating within a more or less normal range of
material scarcity and individual capacity.
In the received republican image, free persons can walk tall, and look
others in the eye. They do not depend on anyone’s grace or favour for being
able to choose their mode of life. And they relate to one another in a shared,
mutually reinforcing consciousness of enjoying this independence. Thus, in
the established terms of republican denigration, they do not have to bow or
scrape, toady or kowtow, fawn or flatter; they do not have to placate any
others with beguiling smiles or mincing steps. In short, they do not have to
live on their wits, whether out of fear or deference. They are their own men
and women, and however deeply they bind themselves to one another, as in
love or friendship or trust, they do so freely, reaching out to one another
from positions of relatively equal strength.4
There are three important lessons that the traditional image of the free
person teaches us about how we might concretize the ideal of freedom as
non-domination as a target of justice. The first offers direction on the range
of free choices that free persons ought to enjoy, the second provides an
indication of the basis on which those choices ought to be secured or
4
Presented in this way, the image of free persons may seem to be silent on the political front, implying
nothing about people’s political rights or responsibilities. But the presentation is adequate for our
purposes, since we are abstracting in our discussion of justice – that is, social justice – from the relation
between citizens and government. I reintroduce that relation in the following chapters when I go on to
discuss legitimacy and democracy.
entrenched, and the third gives us guidance on when such entrenchment is
enough to give people the status of free persons.
The first lesson is that free persons are free in virtue of being secured in
the exercise of a specific class of choices, not in making just any old
choices. They are not to be entrenched in acting violently towards others,
for example, in appropriating as much land as they want, or in pursuing
spectacular adventures. They are to enjoy access to needed resources and
protections in a range of choice where all can operate at once – perhaps
with special assistance for some individuals – without getting in one
another’s way. This is the domain of the fundamental liberties, as they
were described by the seventeenth-century radical, John Libourne (1646).
While these liberties were often cast in his time and place as the ancient,
historically sacred rights of Englishmen, they gave institutional expression
for Libourne to ‘the freeman’s freedom’; they reflected the fact that men
and indeed women ‘are, and were by nature all equal and alike in power,
dignity, authority, and majesty – none of them having (by nature) any
authority, dominion or magisterial power, one over or above another’
(Sharp 1998).5
The second lesson of the free-person image is that in order to enjoy
freedom in the exercise of the basic liberties, people should have a publicly
established and acknowledged status in relation to others; only this could
enable them to walk tall and look others in the eye. Within the sphere of
those liberties people should be entrenched on a public basis against the
incursions of others. They ought to enjoy objective safeguards that apply
regardless of the will of others as to how they should choose in that domain.
And it ought to be a matter of shared awareness in the society that they are
so guarded. They should have an undominated status both in the objective
and the subjective or inter-subjective sense of status.
The public safeguards required for the enjoyment of such status are
traditionally taken to include the laws that provide in a saliently equal
manner for the entrenchment of people’s liberties. But given that univer-
sally beneficial laws are likely to be supported by attitudes of approval for
compliance and disapproval for non-compliance, the safeguards are also
bound to include associated norms or morals. Norms in this sense are
regularities of behaviour such that, as a matter of public awareness, most
members conform to them, most expect others to approve of conformity or
5
The abstraction from issues of legitimacy means that the basic liberties envisaged here do not include
the political rights that Rawls counts as basic liberties in his first principle of justice.
84 Social justice
disapprove of non-conformity, and most are policed into conformity by this
expectation about what will attract approval and disapproval.6
Corresponding to the coercive effect of laws against fraud or violence or
murder, we might expect to find norms that occasion a complementary,
approbative effect, deterring potential offenders by holding out the prospect
of communal disapproval. Machiavelli (1965) remarks in Discourses (1.18) on
the importance of having norms available to support the laws in this way:
‘just as good morals, if they are to be maintained, have need of the laws, so
the laws, if they are to be observed, have need of good morals’.7 We shall be
exploring the connection more deeply towards the end of the chapter.
The first lesson that I drew from the free-person heuristic bears on what
should be safeguarded – the fundamental or basic liberties – and the second
on how they should be safeguarded: by reliance on public laws and norms.
The third lesson identifies the criterion for determining what is enough by
way of safeguarding them – what level of support is sufficient to let us think
that for practical purposes people all enjoy freedom as non-domination in
the relevant choices. The lesson suggests that people should securely enjoy
resources and protections to the point where they satisfy what we might call
the eyeball test. They can look others in the eye without reason for the fear
or deference that a power of interference might inspire; they can walk tall
and assume the public status, objective and subjective, of being equal in this
regard with the best. This eyeball test fits with our discussion at the end of
the previous chapter of the level of prevention and inhibition that might
effectively counter domination. The satisfaction of the test would mean for
each person that others were unable, in the received phrase, to interfere at
will and with impunity in their affairs.
The eyeball test does not require that people should be able to look one
another in the eye, regardless of their personal lack of nerve. It requires that
they have this capacity in the absence of what would count, even by the
most demanding standards of their society, as mere timidity or cowardice.
The reference to timidity is essential, since no public safeguards can
6
The definition follows Pettit (1990) and Brennan and Pettit (2004: Part iii) with one amendment: it
says that a norm is a regularity such that almost everyone expects others to approve of conformity
rather than, in the older formulation, that it is a regularity such that almost everyone approves of
conformity. The change allows us to recognize as norms regularities that, unbeknownst to people, do
not actually attract general approval (Prentice and Miller 1993); for a fuller discussion of this possibility
see Pettit (2008e). While I do not offer a defence for defining norms in this way, it should be noted
that it fits extremely well, particularly in the way it connects norms and approval, with the under-
standing of norms in the larger literature (Hart 1961; Winch 1963; Coleman 1990; Sober and Wilson
1998; Elster 1999; Shapiro 2011).
7
See too Tyler (1990).
compensate for differences between individual personalities and for varia-
tions in people’s capacity to deal with the overbearing assumptions of
others. The reference to the standards of their society is necessary since
there is likely to be cultural variation in what counts as mere timidity rather
than rational fear or deference. People are liable to vary across societies in
the different levels of vulnerability to which they have become inured, in the
probability that they assign to others becoming hostile, and in the levels of
trust that they invest in one another. If there is cultural variation on this
front, then it is clearly local standards that should provide the relevant
benchmark for determining when fear or deference is irrational and when
prudent; there is not going to be any universally valid alternative that might
be invoked in their stead.
The effect of the eyeball test is to require a certain threshold of
resourcing and protection that should be secured for all in the domain
of the basic liberties. We know from the last chapter that other people’s
capacity to interfere may vary in two ways. They may be capable of any
form of interference in a lesser or greater degree, facing lesser or greater
difficulty and lesser or greater danger. And they may be capable in
whatever degree of a more or less serious form of interference: they
may be capable of preventing the choice of an option at one extreme
or, at the other, capable just of imposing a more or less trivial cost. This
variation teaches an important lesson for how the eyeball test should be
understood and applied. Since both dimensions of variation are impor-
tant, the test requires that the level of protection provided in any area of
choice should increase with the seriousness of the interference against
which protection is needed and with the ease of access to that level of
interference.
The third lesson of the free-person heuristic makes the goal of a repub-
lican theory of justice more accessible than it might otherwise be. It means,
as we saw, that you cannot rely on the state to compensate for your timidity
or other failures and that you can enjoy the full benefits of public resourcing
and protection only if you display a certain degree of personal affirmation
and courage. And it also means that having resources and protections
over and beyond what is required to satisfy the eyeball test – say, as a result
of enjoying above-average wealth – is excess to the requirements of your
status as a free citizen. Certain differences of wealth and power may
jeopardize the freedom as non-domination of the less well off, as we shall
see, and be objectionable on that count. But, assuming that they are not
allowed to have this effect, they are consistent with the status of a free citizen
being available to all, richer and poorer alike.
86 Social justice
My characterization of the three lessons of the free-person heuristic
suggests that they apply independently of one another, and for simplicity
of presentation I shall generally continue to write as if they do. But the third
lesson about the eyeball test has an important priority and is bound to
impact on the application of the first two. That test plays a part in
determining, not just the level of resourcing and protection that ought to
be secured for given choices, but also the range of choices that ought to be
entrenched in that way and the nature of the public entrenchment that
ought to be provided.
Take the first lesson, which teaches that free persons should enjoy
resources and protections in an area of choice where all can operate at
once, at least with special assistance for some individuals, without getting in
one another’s way. When is special assistance to be provided for those in
particular need, presumably by the state? Suppose that some people are
impaired or disabled in a way that makes them incapable, without state
assistance, of operating in a certain area of choice; perhaps they need some
prosthetic aid in order to be able to function there and are incapable of
purchasing it for themselves. How are we to determine whether they should
be given that aid and whether the relevant choice should be entrenched as a
basic liberty for all? The answer can only be determined by the eyeball test.
That a choice is such that those who lack access to it will not be able to pass
the eyeball test in the local society makes the canonical case for why the state
ought to resource the needy in a special way and ought to entrench the
choice as a basic liberty for all. Thus, to take a salient example, it may argue
for providing wheelchairs for those who cannot afford them and for build-
ing wheelchair ramps in public buildings.
Or consider the second lesson, according to which free persons should be
entrenched on a public basis in the exercise of the basic liberties. What exact
sort of public entrenchment ought to be provided? Take the presumptively
basic liberty of moving wherever you will within a society. Does that require
just the protection of travellers, without any particular resourcing? Or does
it also require the provision of a public network of road or rail or air travel, as
distinct from some privately provided routes? Or the provision of a public
means of travel, as in a publicly subsidized rail or bus or plane network? Or
even the provision of subsidies for those who cannot afford to use a public or
private network? Here, as elsewhere, such questions teem, and the eyeball
test provides the only compelling basis for an answer. Thus when I speak in
what follows of the need to entrench or secure or safeguard various basic
liberties I shall abstract from the question of whether this means just
protecting them, or protecting them and providing general resources that
everyone needs, or doing those things and, in addition, providing special
resources for some who are particularly needy, whether financially or by
reason of requiring a mechanical aid like a wheelchair. In every case the
question can only be answered on the basis of the eyeball test but I shall not
explore the precise answers that are appropriate in different cases.
In giving a central role to the eyeball test, I follow an approach opened up
by Amartya Sen (1983b) in the development of his criterion of functioning
capability. He suggests that to be poor – to lack a basic functioning capability
in your society – is closely related, as Adam Smith (1976: 351–2) argued, to
being unable to live without shame amongst your fellows: to failing to meet
material standards that others in the society expect all ‘creditable people’ to
attain. Smith’s criterion is less demanding than that employed here but it is in
the same spirit and may even reflect his own allegiance to republican ways of
thinking (Winch 1978). Smith suggests that poverty entails not being able to
look others in the eye without reason for fear or deference: that’s where the
shame lies. The line taken here is that while that is certainly true, satisfying the
test requires full freedom as non-domination and not just the absence of
poverty. Escaping the shame of poverty is only a first step towards the
enjoyment of interpersonal status and the achievement of freedom.
The equality strategy of republican justice

The lessons supported by the free-person heuristic give us helpful pointers
to what a republican theory of justice would require the state to seek, under
the circumstances of justice, in the relations amongst its citizens: that is, in
my usage, amongst the adult, able-minded, more or less permanent resi-
dents of the society. It suggests that the state should be required to treat its
citizens as equals in promoting their freedom as persons, where freedom in
any choice is taken to require non-domination. To the greatest extent
possible, the state should entrench people’s fundamental liberties, on the
basis of public laws and norms, to the point where each is able to pass the
eyeball test in relation to others.8
8
I observed in the last chapter that domination with interference and frustration plausibly counts as
worse than domination with interference but without frustration, and that in turn as worse than
domination without interference or frustration. But that observation will not play any role in this
chapter, or indeed in the next. In guarding against unfreedom in the enjoyment of fully resourced
choices, it is essential to guard against domination, since domination is sufficient, as well as necessary,
for unfreedom. And if we eliminate domination then that will be enough to eliminate forms of
interference and frustration that are hostile to freedom. The observation from the last chapter will
88 Social justice
With these observations in place, we can see that what it means for the
state to treat citizens as equals in this way, satisfying the expressively
egalitarian constraint, is to provide them each with a certain threshold
benefit in the currency of free or undominated choice, as required under
the sufficientarian strategy described earlier. It means providing for such a
level and kind of entrenchment, in such a range of choice, that people each
have the status, traditionally understood, of the free person or citizen. The
choices to be entrenched are the basic liberties; the kind of entrenchment to
be provided is the public sort that enlists laws and norms; and the level of
entrenchment to be secured is whatever is necessary for passing the eyeball
test in the local society. That each is required to enjoy this threshold of free
undominated choice is consistent with some people having such private
sources of power and wealth that they enjoy free undominated choice in a
yet greater range and with yet greater security. That is why the approach can
be cast as sufficientarian.
Since the enjoyment of such a sufficiency of free or undominated choice
means enjoying the status of the free person or citizen, however, we can
describe the strategy in other terms too. We can cast it as an approach that
argues for equalizing derived returns, as in the third strategy mentioned
earlier, where free-person status is a return derived from the required pattern
of resourcing and protecting choice. Although it constitutes a sufficientari-
anism in the currency of free or undominated choice, the republican theory
of justice supports a substantive egalitarianism in the currency of free or
undominated status. Depending on what the relevant currency is taken to
be, it assumes one or the other profile.
In the remainder of the book I shall generally speak of the republican
theory of justice as substantively, as well as expressively, egalitarian, taking it
to argue for promoting people’s equality in freedom as non-domination,
where this is now understood as freedom of undominated status, not
freedom of undominated choice. It may have been the substantively egali-
tarian aspect of republicanism that Cicero (1998: 21) had in mind when he
made his much-quoted comment on the nature of the libertas enjoyed by
the liber. ‘Nothing can be sweeter than liberty. Yet if it isn’t equal through-
out, it isn’t liberty at all.’
As a substantively egalitarian theory, the republican approach to justice
may seem to be exposed to the now well-known complaint that like any
such theory it would argue for reducing the level of the better off in order to
assume relevance only in ranking imperfectly just or imperfectly legitimate regimes. Thus it will count
in favour of one imperfect regime over another that the domination it allows, unlike the domination
allowed in the other, does not promise actual interference or frustration.
achieve equality with the worse off, even when that reduction has no other
beneficial effects; it would support levelling-down (Parfit 2000). This
objection will be irrelevant so long as all actually achieve an undominated
status. While some may have greater wealth or power in such a scenario this
will not give them a greater status freedom, as we saw, and so will not
provide a reason for levelling down. Nor will the objection be relevant
where all can in principle be brought to achieve an undominated status but
not all actually attain it. In such a situation there will be transfers possible
that lift all to that undominated level. But what of the case where it is
impossible for everyone to enjoy an undominated status? What of the case
where some are deprived of status freedom, and deprived in such a way that
their position cannot be remedied by redistribution from the better off?
Would the approach argue that in this scenario the state should worsen the
position of the better off just for the sake of establishing equality: just for the
sake of depriving everyone in the society of an undominated status and
setting up a beggar-thy-neighbour form of equality?
No, it would not. The primary commitment in the republican approach
is to expressive egalitarianism and the first injunction is that the state should
promote people’s enjoyment of undominated choice under the constraint
that it treats them as equals, displaying an equal concern for each. We
argued that this commitment would support an equal free status for all,
invoking the traditional image of the free citizen, on the supposition that
the circumstances of justice prevail and that such a status can in principle be
achieved for all. If that assumption fails, as it is taken to fail here, then the
only commitment remaining in place will be the original, expressive one. It
may be reasonable to assume that short of emergency situations, or grave
disparities in the capacities of different people, there will be an applicable
ideal of free citizenship that all can in principle achieve; this is because the
ideal is differently interpreted in different societies and can allow for
variations in background fortune or misfortune. But if things were so bad
that no such ideal could be satisfied for all, then the only commitment
would be to expressive, not substantive, equalization.
What might expressive equalization in the sphere of free choice support
under such hard circumstances? What might it mean for the state to
promote free or undominated choice amongst its citizens, under the con-
straint of treating all as equals? There are a number of possibilities, one of
which is a version of the prioritarianism that Derek Parfit (2000) supports.
This holds that in the circumstances imagined the state’s equal concern for
each would argue for giving a greater weight – of whatever magnitude – to
bettering the lot of the worse off. The idea is not to prioritize the worse off as
90 Social justice
a way of reducing the relativity between them and others – it is not driven
by a concern with furthering substantive equality – but to prioritize the
worse off because of the poorer absolute level at which they exist and
function. If the state is to care equally for each of its citizens, then such a
weighting of the worse off would make very good sense.
The possibility envisaged in the objection just raised is that the repub-
lican theory of justice might be over-demanding, assuming as it does that
people within a society can all be positioned to achieve the status of the free
person or citizen. But another objection to the theory might be that it is
likely to be intuitively under-demanding, looking for an ideal of equal status
freedom that allows for differences in people’s private wealth and power.
The objection is that the theory may permit such differences in private
resources that it ceases to be intuitively appealing: to recall John Rawls’s
(1971) test, it fails to deliver a reflective equilibrium with our considered
judgements as to how far differences in private resources are allowable.
This objection also misfires. The level of resources and protections
required for undominated status – the threshold of provision necessary –
is determined on a basis that takes into account the resources and protec-
tions available to others. If the state allows excessive disparities between the
endowments of different people, then the less well off are unlikely to be able
to attain that threshold. It is true that equality in status freedom – equality,
as we may say, in freedom as non-domination – is consistent with differ-
ences of private wealth and power and with corresponding differences in
resources and protections. But still the ideal imposes severe constraints on
how large or pervasive those differences can be allowed to be.
The constraining aspect of the ideal comes out in a pair of related effects
that are going to be present in many, although perhaps not all, contexts
(Pettit 1997c; Lovett 2001). Suppose that you have fewer resources and
protections than your neighbour and that we, acting for the state, have a
choice between conferring more on you or conferring more on the neigh-
bour; the choice may arise with providing services, delivering subsidies or
imposing taxes. If we invest in the neighbour rather than investing in you,
then we are likely to do relatively less well in guarding against domination,
since you are in more danger of domination and so more likely to be in a
position to benefit from the extra investment. And if we invest in the
neighbour rather than investing in you, then we are likely to worsen the
danger of domination in absolute terms, since the neighbour is more likely
to be enabled by the extra investment to dominate you or others.
The first of these effects means that investing resources or protections in
the better off has diminishing marginal productivity; as it targets the better
and better off, it is less and less likely to be productive – that is, less and less
likely to increase non-domination. And the second of the effects means that
such investment also has increasing marginal counter-productivity; as it
targets the better and better off, it is more and more likely to be counter-
productive – that is, more and more likely to reduce non-domination. The
effects combine to give us reason for thinking that if the state seeks to
promote equal freedom as non-domination – that is, to make the status of
free citizenship available to all – then it will be systematically programmed
to reduce material inequalities in people’s resources and protections.
The reason why the ideal of equal status freedom supports an antipathy
to material inequality is that it is essentially social in character. You would
not enjoy freedom as non-domination in a universe where there were no
others, as by traditional accounts you might enjoy freedom as non-
interference. To enjoy this freedom presupposes relationships with others
and consists in relating to them on a pattern that rules out domination. It
requires the absence of domination, not as such, but in the presence of
relationships that make domination saliently possible and non-domination
correspondingly desirable. To enjoy the relevant freedom of non-
domination is to be someone who commands a certain standing amongst
your fellows.
The social nature of the ideal of equal freedom as non-domination means
that it has an interactional character of a kind that some recent theorists of
equality have celebrated (Anderson 1999; Scheffler 2005; O’Neill 2008).
What matters is not that individuals are allotted the same good in a certain
dimension, independently of how this affects their relations with one
another. What matters is that they have the same good in a dimension
that impacts on how they can interact with one another and on what
standing they can command in one another’s eyes. Thus it is entirely
intelligible that how far you enjoy undominated status depends, not just
on the resources and protections at your disposal, but on how they compare
with the resources and protections at the disposal of others. Let them
compare too unfavourably and you are very unlikely to enjoy an undomi-
nated status.9
9
A nice test that emphasizes the importance of interactional equality, built on an idea taken from Derek
Parfit (2000), is this. Imagine a world of mutually insulated societies and suppose we have a choice
between two egalitarian policies, A and B. The policies would produce the same average equality score,
measured by whatever index we favour, but policy A would do so while achieving a high degree of
equality within each society and policy B would leave as much inequality within societies as between
them. From the point of view of equality which policy should we prefer? If we prefer A, which seems
like the obvious choice, then that shows just how far we value equality for its impact on interactions
between people.
92 Social justice
The discussion in this section gives us a good idea of how to go about
elaborating a republican theory of justice and I try to provide a model of its
institutional implications in Section 3 of the chapter. But before coming to
that sketch we need to devote some time to a preliminary task. This is to try
to work out a conception of the fundamental or basic liberties that, as we
have seen, are going to require safeguarding under a republican theory.
2. the basic liberties
Two criteria
One approach to promoting people’s equal freedom as non-domination
might be to look for a system under which people have the same level of
freedom in different sets of choices, where the set of choices available to each
gives no one reason to envy the position of another. I might be free in a
domain of choice where it really matters to me that I have freedom and you
might be free in a distinct domain that answers better to your concerns.10
But even if we could find customized domains of choice that satisfied the
envy test at any moment, they would not continue to satisfy it as the
membership of the society changed or as the preferences of members
altered. And in any case it is extremely hard to see how the system imagined
could be implemented. In seeking to equalize freedom as non-domination,
the only feasible project is to look for a system under which people
approximate the same level of freedom in the same choices.
But what choices should people be put in a position to enjoy equally?
They must be choices that people can each have the wherewithal to make, at
least when some are given special assistance by the state; whether special
assistance is appropriate, as we saw earlier, will be determined by the eyeball
test. And they must be choices, of course, that the state can protect,
shielding people from the domination of others. So what choices should
the state safeguard if it is to provide people with the equal freedom that
justice would seem to require? What are the choices that people ought to be
enabled to make, and to make without the domination of others – without
dependency on their goodwill and forbearance?
10
We might take this to be what Hobbes recommends when, as we saw, he thinks that free-men will
each be unhindered in doing the things that by their strength and wit they are able to do; we might
take him to suggest that native wit and strength determine the choices in which each will want to
avoid hindrance.
The basic liberties 93
The free-person heuristic provides us with useful guidance in dealing
with this question (Pettit 2008a).11 Since freedom on that heuristic is an
ideal that all the citizens in a society ought to be able to enjoy under the law,
the basic liberties ought to be construed so as to make such universal
enjoyment possible. The exercise of the basic liberties ought not to be an
impossible ideal that is accessible to none; an elite ideal that only some can
access; or a competitive ideal such that if some access it, that reduces the
chances of others doing so. John Locke (1960: ii.57), who remained faithful
in this respect to republican thinking, spells out the message clearly.
Arguing that ‘where there is no law, there is no freedom’, he says that
everyone should be provided only with a liberty to act ‘within the allowance
of those laws under which he is, and therein not to be subject to the
arbitrary will of another, but freely follow his own’.
This observation argues in favour of two basic criteria that candidates for
entrenchment as basic liberties ought to meet. First, the choices to be
entrenched ought to be capable of being exercised by each, consistently
with being exercised by all. And second, they ought to be capable of
satisfying or fulfilling each, consistently with satisfying all (see Anderson
1999). I call the first the criterion of co-exercisability and the second the
criterion of co-satisfaction. If co-exercisability were breached, then people
would have to compete for being able to exercise the basic liberties, so that
the status of the free person would not be equally accessible to all. If co-
satisfaction were breached, then the basic liberties would not enable people
to achieve the level of fulfilment we naturally associate with having the
status of a free person.
The basic liberties ought to include only choices that meet these two
criteria. And they ought to include all of the possible choices that do so. Any
shortfall in the choices that the state safeguards would be an unnecessary
restriction on the freedom that can be enjoyed by citizens and would not fit
well with the traditional image of the free person. Thus we may identify the
basic liberties with all, and only, those choices that are co-exercisable and
co-satisfying in a society.12 I now proceed to look in greater detail at the
choices that are likely to qualify for being basic liberties in that sense.
11
While I follow the lines developed in Pettit (2008a), I make some small amendments and abstract
from much detail.
12
The extension of a set of choices in this manner will be uncontentious in the case where there is no
impact on the existing choices by the addition of more. But there will certainly be issues of how to
measure and compare liberties when two or more extended sets differ in the precise choices
entrenched. I ignore those questions here, however, as I think that there is no easy answer in the
abstract, no algorithm for determining the relative merits of contending sets; the comparison between
such sets can only be carried forward in context-bound judgements about their relative attractions.
94 Social justice
The two criteria that I employ in identifying basic liberties, it should be
noted, are not value-laden and should be capable of being applied in
common by thinkers with very different commitments. The assumption
that the basic liberties should be as large a class as possible makes almost all
potentially free choices into candidates for consideration and then the co-
exercisability and co-satisfaction constraints act as filters to narrow down
that class. The approach breaks with two more standard lines. One offers a
few examples and directs us to the class of basic liberties without the help of
any criteria, only a gestural ‘and so on’. The other identifies the basic
liberties by a value-laden criterion: say, as liberties that are essential for
the development of the moral personality (Rawls 1993). Our approach is
more directive in the guidance it provides than the first alternative and it is
less dependent on a particular moral vision than the second.13
Co-exercisable choices
The first criterion holds that any choice that deserves to be safeguarded
should be the sort of choice that all the citizens of a society – roughly, all
adult, able-minded, relatively permanent residents – can exercise, and
exercise at more or less the same time. This criterion can be taken to outlaw
two categories of choice. As a constraint of individual exercisability, it
outlaws choices that are inaccessible to particular people, even with special
state provision. And as a constraint of collective exercisability, it rules out
choices that people cannot exercise together – roughly, at one and the same
time – even with suitable state resourcing.
Taking up the first constraint, it should be clear that there are many
choices that not everyone is able to enact or exercise. Not everyone can decide
what you think or say or wish for; only you can do that. So by this require-
ment there is no possibility of entrenching for all a basic liberty of making up
your mind on such matters. All that might be entrenched is the liberty for
each to make up his or her own mind on the issues: your liberty to make up
your mind, my liberty to make up my mind, and so on. What goes for
making up your mind goes for any action that involves you, or any particular
individual, in the position of agent, and this suggests that the basic liberties
must all have a self-referential or agent-relative character. They must be
restricted to liberties on the part of A to determine how A acts, liberties on
the part of B to determine how B acts, and so on. They cannot extend to
liberties on the part of anyone to determine how others act.
13
My thanks to Annie Stilz for forcing me to see this point.
The individual exercisability constraint imposes this restriction on the
basic liberties as a matter of logical or metaphysical necessity. But it imposes
other restrictions as a result of contingent facts about what most of us can
and cannot do. It is a contingent fact that not everyone can climb Mt
Everest and that not everyone – perhaps not anyone – can jump ten feet.
Since these limitations are not ones that the state could plausibly enable
people to overcome, the basic liberties must operate within the boundaries
of exercisability that they establish. They must be limited to choices that fall
within everyone’s competence, at least when that competence is enhanced
by the state. Everyone might have a basic liberty of movement within the
public territory of the state – and a basic liberty of leaving that territory if
another state allows this – even though such movement might require the
state provision of transport facilities for those who cannot walk or cannot
afford the cost of transport. And while that basic liberty would enable some
individuals to exercise their special capacity to climb mountains, climbing
mountains is still not a basic liberty. At most it is a derived liberty that is
available, in virtue of their having the basic liberty of travel, to those with
mountain-climbing skills.
The individual exercisability constraint does not only restrict basic
liberties to choices we are individually capable of exercising, at least with
public support. It also restricts them to choices in the exercise of which we
do not depend on the voluntary cooperation of others. None of us can be
sure of being able to tango or to sing in a choir or to incorporate with others
in a group agent, since such actions require others to be willing to join with
us in their performance. And so choices involving those activities cannot
count as basic liberties that ought to be secured by state resourcing and
protection. The only choices that might call here for entrenchment as basic
liberties are the choices of tangoing, or associating in other ways, with those
who are willing to tango or associate with you.14
The collective, as distinct from individual, exercisability constraint impo-
ses further restrictions on candidates for the choices that the state ought to
secure.15 Clearly, it would rule out entrenching options that as a matter of
necessity not everyone can adopt at once. Thus no option that requires
achieving a superior position to others could be a basic liberty. You cannot
have a basic liberty to achieve an above-average score in a test, or to win a
14
Nor, it may be added, should the state secure the choices of existing groups, since this would favour
those who already happen to have formed such groups and would fail to treat people equally.
15
This constraint is close to the compossibility constraint that Hillel Steiner (1994) imposes on basic
rights.
96 Social justice
citizen-of-the-year award, though, of course, you may have the liberty to try
to attain such a distinction (O’Neill 1979–80). Apart from ruling out such
positional options, the constraint would also rule out options that are
available only conditionally on others not availing themselves of that
option. As things now stand, you may have a choice between selling your
house at current market value or continuing to own it. But not everyone
who owns a house can exercise that choice at once. Let all house-owners try
to sell their houses at current market value and the inevitable consequence
will be that the market value will fall well below its current level.
The collective exercisability constraint imposes these particular restric-
tions as a matter of logical necessity. But it imposes even more interesting
restrictions on the basis of contingent fact. Thus the fact that many of the
resources required for certain choices are scarce rules out making those
choices into basic liberties. In these cases, as Herbert Hart (1955: 175) puts it,
‘owing to scarcity one man’s satisfaction causes another’s frustration’. The
farmer and the cowboy may be friends if there is country enough for them
each to find land that they can use as they wish; the farmer will fence in one
region, the cowboy let cattle roam in another. But if there is only so much
land to go around, then under that condition of scarcity it will be impossible
for them each to use the land as they wish. Thus the collective exercisability
condition implies that the freedom to use land to your personal taste cannot
be a basic liberty.16
Hart himself invokes the unrestricted use of land to illustrate the fact that
there are some liberties that cannot be simultaneously enjoyed by all; there
would be physically inevitable conflicts amongst people who sought to
exercise that liberty at once (Hart 1973: 546–7). Another example, as he
suggests, would be the action of travelling by one’s preferred mode of
transport, since there would be a similar ‘conflict between pedestrians’
freedom of movement and the rights of automobiles’ (Hart 1973: 546,
n. 49). And another might be the action of withdrawing one’s money
from a bank; let everyone try to do that and the institution will break
down. Further examples can readily be imagined. The principle generating
them is captured nicely in G. A. Cohen’s (1979) observation that while
everyone in a room might be individually able to leave by the doorway, they
need not be able to do so together: the door may be just too narrow.
16
The farmer and the cowboy will be familiar from Rogers and Hammerstein’s musical, Oklahoma, but
the predicament they exemplify was already a matter of human experience in clashes between farming
and foraging peoples, some as early as the fifth millennium bce; see Morris (2010: 112–14, 127–8, 271).
No plausible sort of state resourcing could overcome problems of collec-
tive exercisability that are based in logical necessity.17 But is there anything
that the state might do to overcome the problems raised by contingent facts
of scarcity? Are there any legal or political measures that might enable
people to co-exercise options otherwise precluded by scarcity, or at least
to enable people to co-exercise options that are closely related to the
precluded options? If there is nothing the state can do on these fronts,
then the alternatives are bleak. We will have to give up on the idea of
establishing basic, co-exercisable options in the use of land or transport, for
example. We might then have to acquiesce in an anarchic free-for-all where
the spoils go to the victor. Or we might have to adopt a system of central
rationing under which people live with whatever they are centrally allocated
in the way of permits for land-use or transport. Neither sort of alternative is
appealing.
Fortunately, however, there are measures that the state can take to
resource collectively exercisable choices in such areas. It can introduce
rules of coordination that would eliminate the problem of competition in
the use of land and transport and under those rules it can define or identify
choices that all can exercise at one and the same time. Let the state set up
common rules of ownership, for example, and it will be possible for every-
one at once to own and use land – or any other commodity – according to
those rules. Thus there can be a basic, rule-dependent liberty of owning and
using property. Let the state set up rules of the road and it will be possible
for everyone at once to use his or her preferred mode of transport; drivers
will take one route, pedestrians another. And so, again, there can be a basic,
rule-dependent liberty of travelling under the rules of the road by whatever
means you prefer.
Regulatory or coordinating initiatives of this kind would enable a society
to resource choices in ownership or travel that can then be candidates for
protection as basic liberties. And parallel initiatives could allay a vast array of
similar problems. The rules that govern banking can make it possible for
people to have regulated or coordinated access to their funds. And any rules
that coordinate access to something that cannot be accessed at once by all –
this, on the model of exit from a crowded room – can make it possible for
people to have a basic liberty in the exercise of the corresponding rule-
dependent choice.
17
I ignore rather implausible measures whereby, for example, everyone might have a choice of being top
dog for a day.
98 Social justice
Co-satisfying choices
If certain choices are worthy of being entrenched as basic liberties then not
only should they be individually and collectively exercisable – not only should
they be capable of being exercised by each, consistently with being exercised
by all – they should also meet a further condition. This is that they should be
sufficient to satisfy each consistently with satisfying all. As the criterion of co-
exercisability imposes a constraint of individual exercisability and a constraint
of collective exercisability, so this criterion imposes two constraints also, one
of which requires individual satisfaction, the other collective satisfaction.
The criterion of individual satisfaction argues that the choices we
entrench as basic liberties should be ones that by received social criteria
promote the enjoyment and welfare, over the long term, of those who make
them. In principle this criterion leaves room for denying protection for
certain choices that are judged to be harmful for the individuals exercising
them, even for criminalizing them as victimless offences. In practice, how-
ever, the republican approach is unlikely to argue for establishing such
intrusive criminalization, as we shall see later. Doing so would impose
collective judgement and preference on any individual in choices that are,
by hypothesis, co-exercisable. To treat such choices as victimless crimes, or
just to leave them unprotected, would be to impose society’s interpretation
of their interests on people and would represent a sort of paternalism that is
inconsistent with freedom as non-domination, as we saw in the last chapter
(Shiffrin 2000).
This is a point at which the eyeball test serves an important role, as
foreshadowed earlier, in determining how choices ought to be publicly
entrenched. When choices have the potential for harming the adult, able-
minded agents who exercise them, the eyeball test argues against criminal-
izing them or leaving them unprotected but does not support any more
active provision in their support. Take the choices associated with the use of
recreational drugs or with certain forms of gambling. The eyeball test
suggests that such choices do not call for any active form of resourcing,
and may even deserve to be subjected to heavy taxes.18 But it strongly
supports the view that since any other approach would be paternalistic,
failing to acknowledge the status of the agents, still the choices ought to be
protected. They ought to be given at least that minimal form of
entrenchment.
18
Even John Stuart Mill (1978) argued that such a line would be perfectly defensible, suggesting in some
cases that a society should seek to reduce the attractions of choices that can threaten long-term damage
to the chooser by imposing relatively higher taxes on the resources that they require.
Turning to the other aspect of the co-satisfaction criterion, the constraint
of collective, as distinct from individual, satisfaction implies that in order to
count as entrenchable basic liberties, choices should not be such that if some
or all individuals exercise them then many people, perhaps including those
individuals themselves, will cease to enjoy the choices. Candidates for
entrenchment as basic liberties should be choices of such a kind that even
if everyone takes them, everyone can be more or less fully satisfied. This
constraint rules out choices in three broad categories: counter-productive
choices, over-empowering choices and harmful choices.
Taking harmful choices first, it is worth noting that there are many
choices that include the option of intentionally harming another person
and yet pass the constraint of co-exercisability. There is nothing in the
constraint of co-exercisability to stop people being able to lie and to steal
and even be violent to one another. But clearly establishing the choices of
inflicting such harms as basic liberties would in no way contribute to the
satisfaction of people overall. And so, unsurprisingly, the constraint of
collective satisfaction would argue against giving such choices an
entrenched status. The choices to be established as basic liberties should
all be innocent choices, as we might put it.
Over-empowering choices, to move onto a second category, are choices
whose entrenchment might allow the emergence of serious asymmetries in
the relative balance of resources and power amongst people. We saw earlier
that there is good reason to establish rules governing the use of land and, by
extension, property in general. Such rules would introduce rule-dependent
choices in the domain of ownership, exchange and bequest that people can
each exercise without coming to blows in the manner of the cowboy and the
farmer who struggle for possession of the same land. But it should be clear
that the rules of property may enable some individuals to gain such
economic power that they are bound to dominate others in certain con-
texts – for example, in employment or legal action or competition over
access to health resources. The constraint of collective satisfaction would
argue against introducing a system of choices that is liable to have such an
over-empowering effect. It would support any less dangerous alternative
that may be available.19
19
How might we cope with this problem, giving effect to the requirement of collective satisfaction?
There are a variety of measures that might be explored. Introducing progressive taxation as part of the
property system would be one possibility (Murphy and Nagel 2004). Another might be a system of
support or subsidy, in relevant contexts, for the economically weaker. Or yet another might be a
restriction on the comparative advantages that money can buy.
100 Social justice
The problem illustrated by harmful and over-empowering choices
derives from the fact, as Herbert Hart (1973: 550) observes, that entrenching
a choice as a basic liberty ‘necessarily does two things: first, it confers on
individuals the advantage of that liberty, but secondly, it exposes them to
whatever disadvantages the practice of that liberty by others may entail for
them’ (550). But this observation also explains why choices in a third,
counter-productive, category should be ruled out by the constraint of
collective satisfaction. These are choices whose exercise by a number of
people – at the limit by all – undermines the benefit that gives the choice its
characteristic rationale and appeal for some or for all (see Parfit 1984: Part i).
Hart (1973: 543) illustrates the case where the exercise of a choice by any
number of people undermines the benefit that it promises for each. Let
people all have a choice of addressing a group at will: say, a group compris-
ing their fellow citizens. It will clearly be possible for them each to address
the group and to do so at the same time, so that the earlier constraints raise
no problem here. But still we might pause over thinking that this choice
should be established as a basic liberty. For if a number of members try to
speak to the assembly at one and the same time, no one will be heard. And
so no one will be able to enjoy the characteristic benefit of speaking to the
group. Other examples of this sort of case arise wherever the individual
choice of a certain option promises a certain reward for each but the
aggregate result of a number of people taking that choice is that that reward
is not delivered. People may each wish to own a gun for their own defence
but if everyone owns a gun then, plausibly, no one is defensively better off.
People may each wish to drive into the city centre, but if everyone does so
then the point of driving there may be undermined for all.20
Another sort of case that fits in the counter-productive category arises
with choices that impose external costs on others, often triggering claims in
tort law. Let people have the liberty of making whatever use they wish of a
river that runs through their land and there is bound to be trouble in
prospect. The use which those upstream make of the river may severely
limit the use to which those downstream wish to put it; for example, the
20
In these cases, as in the case of speaking to the group, it is generally true, first, that everyone has a
reason to pursue the activity if no one else does; and second, that everyone prefers that no one pursues
it than that everyone pursues it. But it is worth noting that in some of the cases, unlike the group case,
a third clause is satisfied too: everyone may have a reason – a new reason – to pursue the activity if
others all do so. Setting virtue aside, no one will relish being the only person without a gun in a gun-
toting society. This makes these particular examples into cases of a broadly free-riding character
(Pettit 1986). In these cases, people each have a reason for pursuing the activity even if all others do –
they will not want to be made a sucker, as it is sometimes said – but that reason is not the
consideration that originally gave the choice its appeal.
upstream factory can make it dangerous to allow downstream cattle to drink
the water. If any choice that is to be established as a basic liberty has to be
co-satisfying for all, not just co-exercisable by all, then this sort of choice
cannot be entrenched in that way.
Can the state help to overcome the sort of problem raised in our three
sorts of harmful, over-powering and counter-productive choices? Yes,
clearly it can. As with some of the problems raised by the constraint of
collective exercisability, it can introduce rules under which people are given
options that are close to the original, problematic options but are capable,
unlike them, of meeting the constraint of collective satisfaction. Thus the
problem illustrated by the case of speaking to a large group can be solved
under rules such as Robert’s Rules of Order (Robert 2011). These allow people
to take turns in speaking, dictating a pattern under which they can each
make proposals, suggest amendments to the proposals of others, and debate
and vote on the various issues that arise in their discussion. As such rules
might resolve the debating predicament, similar rules might resolve the
problems illustrated with gun ownership and car use. People can be given
the rule-dependent option of owning guns on condition of passing certain
tests or of using free or cheap parking facilities and taking public transport
into city centres.
The case illustrated with the upstream and downstream use of a river is
typically resolved in a different manner. The law of torts allows plaintiffs to
appeal to the courts for case-by-case judgements on whether someone
should be allowed to exercise such a choice and, if allowed, whether they
should be required to adopt precautions against damage to others. The
Hand test, named after the US Judge, Learned Hand, offers useful guidance
in the area. The idea, roughly, is that if the expected cost of effective
precautions to an agent in the exercise of some choice is less than the
expected cost to others of the harm associated with the choice taken without
precautions, then the choice should only be allowed when relevant precau-
tions are in place. Applied to the river case, it would require those upstream
to make use of the river only under limitations or precautions that involve a
lower expected cost for them than the expected cost to others of the
upstream activity taken in the absence of limitations or precautions.
Towards a meaningful life

Consider the range of choices that present themselves at this point as
candidates for entrenchment as basic liberties. They are all individually
exercisable, at least with the help of state resourcing, and all individually
102 Social justice
satisfying. And equally they are all collectively exercisable and satisfying.
Perhaps as a result of suitable coordinating rules, they are designed so that
any number of individuals may adopt them without this affecting their
exercisability as choices or the satisfaction that they deliver across the
community. Would the availability of those choices – of all the choices
that satisfy such conditions – make a full and meaningful life available for
the free citizen (Wolf 2010)? Or would the range of choices entrenched still
fall short in some way of what is needed?
The question is hard to address without a good sense of the sorts of
choices that the criteria would require us to entrench. And, on the face of it,
this raises a problem, for there is no obvious end to the choices that satisfy
our constraints. Happily, however, there is a way around this problem, for it
transpires that amongst the choices that meet the criteria we need only
concern ourselves with a small sub-set. The choices that prove co-
exercisable and co-satisfying may vary in two dimensions: in the distality
and generality of the options involved. And it turns out that it will be
enough to entrench only the more distal and general choices. Let the state
entrench them and it will automatically entrench all the other choices too.
One option will be more distal than another when, despite the fact that it
can be realized by one and the same action, it involves effects at a greater
distance from the agent: effects, in other words, that reach out further into
the world. I realize progressively more distal options as I open and close my
mouth, make sounds that you hear, and let you know my holiday plans. But
options that are relatively distal in this sense may vary in the separate
dimension of generality, even when they can also be realized by one and
the same action. Telling you about my holiday plans is distinct from the
progressively more general options of giving you information about my
private life, giving you any information whatsoever, and speaking to you in
any modality, whether of information-giving or not.
Given this account of distality and generality, it should be clear that if we
secure the more distal and general choices, then we will automatically secure
the less distal and less general counterparts as well. If I am free to tell you
about my holiday plans, I will be free to make certain sounds – maybe these,
maybe those, depending on the words I employ – in your presence. And if I
become free to communicate with you in any modality, then I will be free to
tell you about my holiday plans or to talk to you about any topic what-
soever. If we entrench freedom of speech as such, in other words, then we
will have entrenched the freedom to make the various sounds associated
with speech, to discuss the various topics that speech can engage, and to
exercise speech in addressing one or another audience.
Generalizing from this observation, the lesson is clear: if we entrench the
most distal and general of the choices that pass the constraints on basic
liberties, we will have entrenched all the related choices that pass those
constraints. That proper subset of distal and general choices is all we need
think about, then, and all we need to entrench. Indeed, it makes good sense
to count only members of that set as the basic liberties proper and in what
follows I shall stick to that convention.
What are the general, distal choices that are likely to require entrench-
ment, on the proposal we have been exploring? The following list may not
be exhaustive but it is surely indicative:
* The freedom to think what you like.
* The freedom to express what you think.
* The freedom to practise the religion of your choice.
* The freedom to associate with those willing to associate with you.
* The freedom to own certain goods and to trade in their exchange.
* The freedom to change occupation and employment.
* The freedom to travel within the society and settle where you will.
Back now to the crucial question. Suppose that the largest possible set of
such basic liberties is secured by public resourcing and protection in a
society. Does the entrenchment of liberties like these suffice on its own
for making a full and meaningful life possible for individuals? Suppose you
have the resources for making such choices and are not dominated in any of
these broad areas: you are able to exercise your own discretion, by intuitive
criteria, without worrying about the goodwill of others. Would the enjoy-
ment of the unvitiated, uninvaded capacity to exercise choice across that
range of opportunities suffice to give you access to the life that we would
expect a free person to be able to achieve?
The enjoyment of such liberty might not suffice for a meaningful life
under a Romantic or post-Romantic image of the fulfilment required – for
example, the Nietzschean image of life as an Übermensch. But the assump-
tion in the republican tradition is that it certainly can. It is possible to be as
free as a full and meaningful life requires, so the idea goes, and yet only to be
as free as other people: only to enjoy the same freedom across the same
choices and on the same public basis. It is possible to be personally fulfilled
without being socially privileged.
I go along, unhesitatingly, with this traditional view. Looking at the list
of entrenched liberties, I am inclined to ask a rhetorical question. If the
availability of these choices is not enough to make a meaningful, independ-
ent life possible, what else is needed? If not this, what? Perhaps the best way
of defending an affirmative attitude, however, is to look at what individuals
104 Social justice
would have to enjoy in the way of resources and protections if they really
had access to a maximal set of entrenched basic liberties. That takes us to the
topic of the next section. I sketch a model there of the institutions and
policies that the republican theory of justice would require – a model of how
we might safeguard the basic liberties equally for all – and suggest that the
model fits well, as reflective equilibrium requires, with our considered
judgements of justice.
Variations in basic liberties

Before moving onto that model, however, it may be useful to make a general
observation on the nature of the basic liberties, as that has emerged in our
discussion. This is that while the basic liberties may seem to have a universal
cast, at least when they are formulated abstractly, they can vary in more
concrete interpretation and practice across different regimes; they are not a
natural kind. They can vary, broadly speaking, in either of two ways: in a
radical fashion that divides off past from present cultures, for example, and
agricultural from industrial societies; and in a less radical manner that
appears in differences between otherwise quite similar dispensations, even
between contemporary advanced democracies.
The more radical divide shows up in the fact that the choices that it
would have seemed important to entrench as basic liberties in classical
Rome or medieval Italy or seventeenth-century England – and the choices
that are judged important to secure in extremely poor countries today – are
very different from the choices that we would expect to be established in a
contemporary liberal and democratic regime. The technology and affluence
of advanced societies makes it possible for people to be able to co-exercise
with satisfaction a much richer range of choices than those that would be
available elsewhere. This is true for two reasons. First, everyone in such a
society can do things, and can be secured in the choice of doing them, that
would be unavailable elsewhere; think of how we can travel in an advanced
society, or express ourselves, or win employment. And second, those who
are limited in their capacities can be enabled in an advanced society to
exercise choices that would be unavailable elsewhere: think of the impact of
eyeglasses or hearing aids or wheelchairs or various prosthetic devices.
It may be useful in this connection to go back to how Amartya Sen
(1983b, 1985) has suggested we should conceive of what it is to count as poor,
or as non-poor (see too Nussbaum 1992, 2006). Poverty should not be
conceptualized on a relative basis, he says, so that by definition a certain
percentage of the population in any society are bound to be poor. Nor
should it be conceptualized on an absolute basis, so that the resources that
would put someone amongst the poor of an advanced democracy would
also put that person amongst the poor of an undeveloped, agricultural
society. To be poor is to be lacking in the resources required for being
able to function at a basic level in your local society and to be able, in Adam
Smith’s terms, to live without shame amongst your fellows. And to avoid
poverty is to enjoy a basic functioning capability under the local conditions
of that society. While functioning capability is a universally intelligible
ideal, on this approach, it may require quite different resources in the
context of different cultures, economies and societies.
What goes for functioning capability goes for freedom as a person and for
the basic liberties with which it is associated. The idea of having such
freedom is a universally intelligible ideal, directing us to the standing
someone enjoys when they satisfy the eyeball test and can look their
compatriots in the eye without reason for fear or deference. But freedom
as a person in this unchanging sense may require the safeguarding of
different ranges of choices in different societies. The Roman commoner,
the medieval burgher, the modern yeoman and a contemporary professional
might each well pass in their own society as enjoying the freedom of a
person. But the choices we should expect to be secured for each may differ
quite deeply across the historical divides.
Not only may the basic liberties vary radically insofar as different cultures
offer different possibilities of action and impose different conditions for
passing the eyeball test; they may also vary less radically across quite similar
societies, even societies as similar as contemporary advanced democracies.
There are three sources of this less radical variation, all of which turn on the
fact that basic liberties have no natural interpretation; they depend for their
articulation on the introduction of conventions that are bound to vary
across political divides.
The first of our three sources of variation is the fact that co-exercisability
requires many basic liberties to be rule-dependent in the fashion of the
liberties of ownership and movement. There are many different sets of
property conventions, varying in how far they recognize collective or
communal property and in the different titles and rights and tax liabilities
they associate with private ownership. And as there are differences in the
conventions governing property rights, so there are great differences across
societies in conventions bearing on rights of way and rights of residence.
There may be considerations that argue against some such conventions on
the basis of the demands of non-domination, as we saw – some may allow
over-empowering choices, for example – but it is very likely that certain
106 Social justice
variations are not determinately better or worse than others. Thus we may
expect to find differences in the conventions adopted, and the liberties
entrenched, even between quite similar and equally commendable societies.
The second source of variation in the basic liberties, even across similar
societies, is that culturally variable rules are needed, not just to make certain
choices co-exercisable, but also to make them co-satisfying. These rules are
required to make it possible for people to enjoy certain liberties, even when
all or many exercise them at once. Thus regulations like Robert’s Rules of
Order (Robert 2011) are required to create the basic liberty of addressing
your compatriots in any group. And parallel sets of rules are required to
solve problems such as those with the use of cars or the use of guns. While
some such rules may be better than others in republican terms it is likely
that many different sets can serve the promotion of freedom as non-
domination equally well. Thus this factor, too, explains why we may expect
to find variation in the basic liberties established in similar, equally attractive
societies.
A third reason for expecting such variation derives from the difficulty of
identifying a suitable set of basic liberties. The constraints on basic liberties
cannot always be identified in advance, as indeed the case of the upstream
and downstream use of a river already illustrates. There is bound to be a
problem in definitively establishing, in the abstract, liberties such that their
exercise by some is not going to impact on their exercisability by others or
on the satisfaction available to others; there is bound to be a problem in
finding a set that precludes all such negative interactions (Sen 1970; Dietrich
and List 2008). The only feasible solution is to establish in law – not just in
the law of torts but also in criminal law, contract law and constitutional
law – a way of handling different problems as they arise, leaving room for
the sort of judicial intervention, or even legislative or constitutional amend-
ment, that would interpret and revise the basic liberties so as to reduce
unwanted interactions (Zucca 2007). As we recognize the necessity of
subjecting any system of basic liberties to such dynamic adjustment, we
can see that two similar societies, even societies that begin from the same
specification of basic liberties, are likely to come apart in the course of their
development.
The three factors invoked explain why there should be acceptable varia-
tions in the interpretation of the basic liberties across societies as close as
contemporary advanced democracies. Such variation is very likely to materi-
alize in view of the fact that such societies often come apart in the cultural
expectations that people bring to bear on public matters. In one society,
there may be a very lax view of the background tests that gun owners should
Sketch for a model of justice 107
pass; in the other, the tests may be so strict that most citizens are given no
access to guns. Or in one society people may take a very tolerant view, and
in the other a very strict view, of how far offensive speech is damaging and
should be restricted. The possibilities are legion.
The three sources of variation just reviewed show that the basic liberties,
as we understand them here, are deeply dependent on interpretative rules.
This rule-dependence does not get in the way of our identifying in the
abstract the set of basic liberties that ought to be entrenched in any
contemporary society. But it does mark an important complexity. The
promotion of equal freedom as non-domination is bound to mean subtly
different things in different social contexts, even in contexts that differ in
the relatively small ways that distinguish contemporary, advanced societies
from one another. Not only do the basic liberties have to be secured by the
resourcing and protection of the law, then. The precise choices that are to be
entrenched as basic liberties depend on the laws even for being identified.
3. sketch for a model of justice
Beyond Rawls
How plausible is it to claim that what justice requires amongst citizens is
equal freedom as non-domination, as we have articulated that ideal, and
nothing more? The claim may seem downright incredible in virtue of the
fact that whereas John Rawls’s (1971, 1993, 2001) theory of justice imposes
two principles, one prescribing equal freedom and the other a close cousin
of material equality, this theory imposes the first alone. Rawls’s theory, after
all, is the very paradigm of what we ought to expect of a theory of justice,
and if ours departs significantly from it, then that may seem to raise a
question about its plausibility.
It turns out, however, that the republican account cannot be cast as an
attempt to espouse Rawls’s first principle and to dispense with his second.
There are two reasons for this, even assuming that the same basic liberties
are targeted by each.21 First, unlike the republican approach, Rawls’s first
principle does not require full resourcing for the basic liberties; and, second,
it does not require their protection against domination, only their protec-
tion in a weaker sense.
21
Rawls’s basic liberties are not derived on a systematic basis, unlike those we target. In any case they
include certain liberties associated with political rights, such as the freedom to vote, since Rawls does
not distinguish the issue of legitimacy, as I understand it here, from the issue of justice.
108 Social justice
In Rawls’s usage people are free to do or not do something just insofar as
it is legally permissible for them to do it or not do it: they have a legally
protected right in respect of the choice or, equivalently, ‘government and
other persons . . . have a legal duty not to obstruct’ (1971: 203). ‘Liberty’, as
he says, ‘is a complex of rights and duties defined by institutions’ (Rawls
1971: 230). But the liberty to choose in this sense between two options
is quite consistent with an inability to make the choice: having that ability is
not required for the liberty as such, only for its worth or value (Rawls 1971:
204–5). Thus, for all that Rawls’s first principle requires, the choices that are
established as basic liberties need not be secured for all by public resourcing.
There is a direct contrast with the republican approach.
The second way in which the principle falls short of the republican
requirement is that it does not support an equally high level of protection
for the basic liberties. Rawls assumes, in line with standard practice, that
insofar as others are meant to have a legal duty not to interfere with one of
his basic liberties, they are subject to penalties. But he thinks that such
penalties may be scaled to the probability that others will indeed interfere.
Thus he says that while penalties may always be necessary for purposes of
giving people confidence that others will respect their legal duties, in the
sort of regime where there is general compliance – the regime that he seeks
to characterize in his ‘ideal theory’ – they will not be required for any other
reason: ‘in a well-ordered society sanctions are not severe and may never
need to be imposed’ (Rawls 1971: 240).22
This makes clear that for Rawls domination is not a problem as such. The
society envisaged in his ideal theory may be well ordered just because the
powerful generally display goodwill towards others. And so protecting
people against them only by such penalties as are required to create
confidence that they will not interfere – a confidence that is easily achieved,
given the evidence of goodwill – may leave the powerful effectively
22
Rawls’s views on freedom are quite complex. He begins with the idea that any form of freedom is a
freedom from constraint: ‘this or that person (or persons) is free (or not free) from this or that
constraint (or set of constraints) to do (or not do) so and so’. On this account, ‘many kinds of
conditions’ may serve a constraining role, including ‘duties and prohibitions defined by law’ and
‘coercive influences arising from public opinion and social pressure’, so that freedom presumably
consists in the absence of such constraints (Rawls 1971: 202). But he quickly goes on to give freedom
an institutional cast, as in my presentation here, holding that you are free to do something just insofar
as you have the legal right to do it or, equivalently, just insofar as ‘government and other persons . . .
have a legal duty not to obstruct’ (Rawls 1971: 203). ‘Liberty, as I have said, is a complex of rights and
duties defined by institutions’ (Rawls 1971: 230). On this institutional account to be free to do
something is not necessarily to avoid vitiating factors, for you may not avoid natural limitations,
which are now said just to deprive your liberty of its worth or value (Rawls 1971: 204–5); and
presumably you may not avoid the ‘influences arising from public opinion and social pressure’.
unbound; it may leave them able to interfere at a relatively low cost. Were
domination taken to be a problem by Rawls, then he could not adopt this
line. He would have to recognize that even in a well-ordered society, it is
essential to have suitably protective blocks and burdens in place in order to
establish that no one depends on the goodwill of others for avoiding
interference.
What would make for a suitable level of protection – and, more generally,
entrenchment – against others? Some protection will come about via the
denial of resources or opportunity to the powerful – this amounts to
disarming those individuals or groups – and some via the defence of
individuals against them, whether at the time of the attempted offence or
by means of later requital. But no matter what form it takes, it is likely to
come in degrees. So what level of protection is going to be adequate to deal
with the danger of domination? The republican answer, of course, is that
protection ought to extend to the baseline dictated by the eyeball test: a
threshold, recognized in public awareness, such that for those at or above it,
a tendency to worry about the power of others – and so a tendency, for
example, to try to keep the powerful sweet – will be taken to express an
unwarranted, perhaps even irrational, degree of timidity or cravenness.
Whatever the level of protection that all should enjoy under such a test, it
is certainly higher than anything envisaged by Rawls in his vision of a society
that satisfies only his first principle of justice.
Because it does not guarantee the resourcing of the relevant choices, or
provide for their protection against domination, Rawls’s first principle
leaves many intuitively unjust inequalities in place. Not guaranteeing
resources required for exercising the relevant basic liberties, it may leave
some people in a position where they are unable to exercise those choices.
And failing to protect the liberties against domination, it may not register
the same degree of vulnerability, or the same need for safeguards, that would
appear in the republican project. It is because of the inequalities that this
principle would leave in place that Rawls introduces the second principle of
justice, seeking a fit – perhaps after some adjustment on the two sides – with
his considered judgements of justice; such a fit would help to establish the
reflective equilibrium with the judgements that his methodology requires.23
I argue that things look very different if we articulate the freedom-centred
concern in the republican way, relying on the eyeball test for interpreting it.
Let the resourcing of basic liberties, not just their protection, be important
23
For an explicit comment on how fulfilment of the second principle is required to provide in this way
for the basic liberties – in Rawls’s terms, to give them value. See Rawls (2001: 177).
110 Social justice
for freedom as non-domination, as the first chapter showed. And let the
protection required be sufficient to guard against domination under the
eyeball test. My claim is that in that case, the cause of entrenching the basic
liberties will come intuitively close to providing for what most of us will be
happy to think of as justice. It will not leave in place the shortfall that
prompts Rawls’s recourse to his second principle.
In order to see that this is so, it will be useful to look briefly at a model of
the policies that the equal entrenchment of the basic liberties is likely to
require. This is a slightly hazardous enterprise, for two reasons. Policies have
to be guided as much by empirical assumptions as by philosophical princi-
ples, so that principles alone will not lead us firmly to policies. And in a real-
world scenario of limited resources any policy programme is going to have
to require a variety of trade-offs, some of which are hard to track in advance.
The following sketch of the sorts of policies that a republican theory of
justice would support is bound, therefore, to have only a tentative and
illustrative status. For reasons of simplicity I set it out without much
argument over detail, though at almost every point I rely on the intuitive
guidance provided by the eyeball test.24
There are three main areas of policy-making that any plausible project of
entrenching the basic liberties would have to address. To describe them
alliteratively, the first of these has to do with public infrastructure, the
second with public insurance and the third with the public insulation of
people against danger from others. Infrastructural programmes would
establish a framework for the enjoyment of freedom as non-domination
amongst all citizens, insurance programmes would guard each against some
of the misfortunes that can undermine the achievement of equal freedom as
non-domination and insulation programmes would provide for their pro-
tection against the dangers that others can represent within specific relation-
ships and on a more general front.
Infrastructural programmes
There are three kinds of infrastructural programmes that will be required in
a republican theory of justice. These are necessary to establish a suitably
24
Notice, however, that republican policies will not have to alter with every change in actual circum-
stances. No matter what the actual circumstances, they are required to promote non-interference
across a range of possible circumstances; they have to make non-interference robustly available over
variations both in people’s own preferences for what they should do and in other people’s preferences
for what they should do. I return to this point in the concluding section of the chapter.
broad range of choices in which people can hope to enjoy equal freedom as
non-domination in relation to one another.
The first requirement is that children in the society each have access to
the sort of education necessary to provide them with essential skills, to bring
their particular talents to fruition, to give them a full sense of the rights and
responsibilities of citizenship and indeed to let them see how bad it is for
anyone to suffer domination in the sphere of the basic liberties. Let people
be lacking in such developmental ways, and they will be incapable of
asserting themselves with others, or assuming the status of free persons;
indeed they may even be a danger to others, not recognizing the reciprocal,
freedom-based claims that are made upon them.
The second and third programmes required by way of infrastructure are
environmental in character, not developmental. One bears on the institu-
tional environment in which people operate, the other on the physical
environment, natural and constructed, in which they live.
In order for people to be able to enjoy the basic liberties equally, there has
to be a legal dispensation in place that provides a suitable institutional
environment for people’s lives together. That dispensation would establish
the property conventions and other arrangements required for the avail-
ability of certain co-exercisable, co-satisfying choices; provide rules for the
definition and redress of torts or wrongs that people may suffer as a result of
the negligence, if not the intrusive ill will, of others (Goldberg 2005: 6);
enable people to associate and incorporate with one another on a suitable
contractual basis, whether commercially, in employment, or in marriage
(Cohen 1933); and allow them to argue through the courts or other channels
for adjustments to the basic liberties that are protected. Beyond that, it
would enable people to enjoy the extra choices that may be made available
for entrenchment as a result of stable economic and financial institutions
(McGilvray 2011; Tomasi 2012). And at the same time it would establish
restrictions and regulations that guard against financial crisis, economic
recession, uncoordinated competition, commercial monopoly and
uncosted externalities, as well as the inequalities of power and influence
that allow some to lord it over others. All of these ills are likely to limit or
frustrate people’s equal access to basic liberties.
The third domain of policy-making that the state must address, if it is to
promote the widespread enjoyment of freedom as non-domination,
involves the material rather than the institutional environment. There will
be fewer basic liberties available for people, at least in the long run, the less
sustainable the natural environment, the less efficient the energy network,
the less reliable the organization of transport and information, the less
112 Social justice
effective the public health system and, indeed, the less secure the territory of
the state. Thus it will follow under the republican programme, as it will
follow under any plausible agenda for the state, that there have to be policies
in place to promote corresponding, broadly ecological goals.
Catering for the development of citizens, and for the maintenance of a
suitable environment, institutional and material, is one basic part of what is
going to be domestically required of a state if it is to promote equal freedom
as non-domination, establishing a suitably broad range of choices that can
be entrenched as basic liberties. The other two parts require arrangements
for the public insurance of individuals against certain ills, and for their
public insulation against the power and ill will of others.
Insurance programmes
There are a variety of conditions in which people are going to be unable to
exercise some of their basic liberties or are going to be exposed to the
domination of others. People will suffer such vulnerabilities if they do not
have sure access to shelter and nourishment; to treatment for medical need
or support for disability; or to representation in appearing as plaintiffs or
defendants in the courts. It is going to follow straightforwardly, on the
republican approach, that they should therefore be publicly insured – or
publicly required and enabled to have private insurance – against such
possibilities. They should be provided with social security, medical security
and judicial security, whether by means of a system of social insurance,
national health and legal assistance, or by any of a number of alternatives:
say, the provision of a basic income for each citizen (Raventos 2007). They
should be provided in these domains with Sen’s (1985) basic capabilities for
local functioning (Nussbaum 2006).25
Why require insurance against the ills listed rather than just putting
arrangements in place that make it likely that those in need will be helped
out? Why not just provide incentives – for example, tax concessions – that
will motivate the wealthy and powerful to help the needy? Or why not just
rely on people’s natural philanthropy to cater for the needs of others?
Perhaps those in physical need will be better off if they can have recourse
to privately funded kitchens and shelters, and perhaps those in medical and
25
This policy recommendation presupposes certain empirical assumptions, as I said that almost all such
recommendations do: in this case, the assumption that markets will typically leave people in need of
such insurance. For a defence of that assumption, and a critique of the ‘great risk shift’ whereby
markets were supposed to be capable of providing suitably, see Quiggin (2011).
legal need will be better off if they can enjoy the pro bono services of
philanthropic professionals.
Republican theory is bound to reject the idea of forcing the less well off to
have to rely in this way on voluntary forms of philanthropy. The needy will
depend on the goodwill of voluntary benefactors for not being exposed to
the interference of those who are in a position to interfere in their lives,
which already comes dangerously close to being dependent on their good-
will in a dominating way.26 And once the needy are incorporated into
relations of such dependency, the baseline of expectation is bound to shift
so that any withdrawal of philanthropic assistance will count as the denial of
an established option: that is, as a form of interference. At that point the
needy will be exposed to a salient power of interference on the part of their
benefactors – an unvitiated, uninvaded capacity to interfere – and they will
be straightforwardly dominated. Or at least that will be so in the event,
surely quite likely, that benefactors are not pressurised to provide their
services or there are not so many benefactors lined up to provide help that
the needy depend on the goodwill of none in particular.
Not only should the needy be insured against the sorts of misfortune we
have been considering, they should be publicly insured in this way or, at the
least, publicly required and enabled to have private insurance. The need for
public insurance derives from the republican requirement that people
should be guarded against private domination by their enjoyment of a
public status as citizens who can access needed resources and protections.
If people enjoyed only publicly unmarked safeguards against private dom-
ination then they could not hope to live up to the traditional image of free
persons. Public insurance does not require universal provision for social,
medical and legal security but it does require that if people are unable to
provide for themselves in that manner, the public purse is available for their
support.
Many advanced societies fail to provide properly for the insurance of their
members on the social, medical and legal fronts, but it is worth noting that
almost all admit the desirability of public insurance on other fronts, so that
they can hardly object in principle to the argument here. No advanced
society is ready, for example, to deny help to those in a particular area who
are exposed to a natural catastrophe such as an earthquake, a hurricane or a
26
This goes back to the issue at the end of Chapter 1 as to whether failing to guard against the invasion
of freedom is as bad as invasion and whether the power of so failing is as bad as the power of invading.
I am assuming here that any philanthropic agency is going to be in a position, should it suit its
purposes, to withdraw benefits from a particular individual or at least from a particular area, whether
of geography or activity.
114 Social justice
tsunami. Very few are willing to deny special assistance to those unlucky
enough to be born with certain disabilities, or to develop them in later life.
And few are ready to allow depositors to lose all their savings when a major
bank is in difficulty. Such catastrophes expose people to the possibility of
domination in just the way that more personal crises can do. If it seems
reasonable for government to guard against their effects, then it should seem
equally reasonable for it to guard against the effects of personal crises as well.
But what if the personal crises are self-induced, as we might say? What if
the people who suffer them are to blame for falling on such bad times; they
gambled their money away, they smoked excessively or drove recklessly, or
they engaged in legally risky ventures? What, in short, if their problems are
due to bad option-luck rather than brute, bad luck? Does that imply that
there is no ground for state provision?
This implication is not supported by a republican approach to these
issues. For whatever the origin of the problem in any such case, the crisis still
exposes people to domination and still calls for remedy (Scheffler 2005). But
isn’t there a certain moral hazard in providing the benefits of a public
insurance for the bad results of an event for which a person holds a certain
responsibility? Might it not lead to reckless risk-taking on the part of some?
It might but it probably wouldn’t, for two reasons. First of all, the crises
involved are such that few would willingly run a salient risk of suffering such
a problem. And second, the insurance benefits available are not likely to be
such as to compensate fully for the loss endured in such a crisis. In any case,
it would seem perfectly reasonable for a community or state that is prepared
to insure its citizens in certain ways to take measures that guard against
reckless risk-taking. Such measures might take the form of obliging people
to wear seat belts while driving or of imposing restrictions on their oppor-
tunities to gamble or to incur excessive contractual obligations.
Insulation programmes: special protection

The third area where the republican agenda of entrenching basic liberties
would prescribe a distinctive pattern of policy-making is in protecting or
insulating people against specific and generic dangers from others. I look
first at issues of special insulation and then at the questions raised by general
cases.
Special insulation is the sort of protection that is required in relationships
like those of wife and husband, employee and employer, debtor and
creditor, where there are often asymmetries of power. A standard approach
in such areas is to legislate for equality, imposing legal duties on the
presumptively stronger party – typically, the husband, the employer and the
creditor – and thereby establishing the corresponding rights of the weaker.
But often such rights, however well supported, are frail reeds, for they have
to be triggered by the weaker party and the act of triggering them can have
costs of its own. The wife who calls in the police against an abusive husband
may find herself exposed to anger and further abuse as a result. The
employees who complain to an inspector about working conditions may
find that they consequently get all the rougher jobs. And so on.
The measures that have to be put in place to give even a modicum of
protection within relationships of this sort have to go beyond formal, legal
rights. They may include screening in various options for the weaker, as in
making provision for the abused wife to take refuge in a women’s home, or
as in legalizing the unionization of employees and recourse to strike action.
They may include screening out various options for the stronger, as in
imposing a court order against an abusive husband or in restricting or
regulating the right of an employer to fire at will.27 And they may include
creating new opportunities for the weaker, as in allowing a spouse to sue for
no-fault divorce, or in establishing the possibility for the unemployed to be
able to rely on public support.
No such measures are likely to be effective, it should be said, unless the
actions of the state are complemented by supportive norms of civil society.
If women are to be rid of the domination of abusive husbands, then there
must be a women’s movement that provides them with the support and
solidarity of fellow spirits, should they fall on bad times. And if employees
are to be rid of domination at the hands of employers, then there must be a
willingness to organize and stand up for one another amongst the employ-
ees. There is only so much the state can do on its own.
The relationships within which people need special protection are not
limited to person-to-person relationships. One of the most dramatic devel-
opments of recent times is the proliferation of corporate entities, commer-
cial and otherwise, and these group agents pose a great danger of
domination in how they relate with individuals and with collections of
individuals (Coleman 1974). They act only via their own individual mem-
bers but usually those members play different parts, take turns – on office
hours, as it were – in playing their parts, are shielded from scrutiny in the
27
The policy likely to be supported would run directly counter to ‘the traditional negative libertarian
“at-will” doctrine that, consistent with contractual obligations, an employer may fire an employee for
“good cause, no cause or even for cause morally wrong”’ (Levin 1984: 97). For historical background
see Cornish and Clark (1989: 294–5).
116 Social justice
parts they play, and act to an aggregate effect that may not be clearly visible
to any one of them (List and Pettit 2011). Consequently, the corporate
agents they bring to life lack the vulnerability and the capacity for empathy
that individuals generally display (Bakan 2004). Compared with individual
human beings, such organizations will generally have an indefinite time-
horizon, a systematic indifference to life’s anxieties, and a more or less
bottomless purse.
Given the rights that they now enjoy in most societies, corporate agents
represent salient sources of domination. Think of how a multinational
corporation can treat its workforce – say, in its determination of working
hours or conditions – when it can move base without excessive cost and the
workforce are heavily dependent on it for employment and welfare. Or think
about how it may treat the local population or the local environment when
the community would be put in serious economic jeopardy by the corpo-
ration’s departure. Or think, finally, about how it may treat those who suffer
as a result of its negligence, whether as individual customers or as the victims
of an explosion or oil-spill, when it has access to better legal counsel, can drag
out court cases for years, and is subject to none of the pressures that shape
individual mentalities, not even the pressure of prospective death.
This rehearsal of problems is confined to the commercial sphere, ignoring
the issues raised when the victims of child abuse try to take on a church, for
example, or disaffected students or their parents attempt to achieve redress
from a school or university, or indeed a small business tries to negotiate terms
with a powerful union. But the problems should be more than enough to
illustrate the difficulties that have to be overcome if domination is to be
reduced within relationships between individuals and corporate bodies.
I do not mention the problems now in order to list some pat solutions.
What I mainly want to do is to stress the depth of specific safeguards that the
republican programme would call for and the challenge that it raises for the
design of suitably protective institutions (Lovett and Pettit 2009). But I
cannot resist making one suggestion. This is that corporate bodies of the
kind reviewed might be subjected in their own right, and on various fronts,
not just to the tort law that is universally allowed, but also to the discipline
of the criminal law that I shall be discussing shortly (Pettit 2007d, 2009a;
List and Pettit 2011).
The criminal law presupposes a capacity for understanding the merits of
the options presented in any choice, in particular the prohibition to which
relevant options are subject, and a capacity to respond appropriately to a
judgement of those merits. But there is every reason to think that a suitably
organized corporate entity will have, or can come to have, such capacities.
And equally there is every reason to think that it makes sense to hold such an
entity responsible for any acts that it sponsors, organizing things in a way
that allows members to perform the acts in its name. Holding such an entity
responsible in its own right is consistent with holding those members
responsible at the same time for playing the parts required of them.
If we do expose corporate entities to criminal liability, then we can
impose penalties that ought in the longer term to deter offenders. While
the penalties may have to be mainly financial – we can’t put corporate
entities in prison, though we can disband them – they will inevitably carry
associated reputational costs: punishments in the currency of esteem
(Brennan and Pettit 2004). Such costs are bound to be particularly prob-
lematic for any corporate agent. They will affect the capacity of the
commercial corporation to attract or retain customers and the capacity of
non-commercial corporations like a church to attract or retain members.
They can chasten and discipline the most powerful organization and they
direct us to a way of establishing significant political control over corporate
behaviour.
Insulation programmes: general protection

Let me turn, finally, to the measures of general protection that the entrench-
ment of basic liberties, as understood in republican terms, would require. I
focus on general protection within the state, not internationally, and will be
looking only at issues of criminal justice, therefore, not global security. I will
address four questions in particular that arise within the theory of criminal
justice. What is it for the state to criminalize an act? Why should it
criminalize any acts? What acts should it criminalize? And how should it
criminalize them?28
Q. 1. What is criminalization?
To criminalize certain types of act is to introduce a system of law that is
designed to regulate – in particular to minimize – their occurrence by the
imposition of penal costs. Regulation by such penal costs is a distinctive
form of regulation. It contrasts in particular with opportunity-cost regu-
lation, on the one side, and admission-cost regulation on the other.
Opportunity-cost regulation would rely on regulating a particular sort of
action by rewarding the alternatives to such action. Admission-cost
28
I draw freely in the following discussion on Braithwaite and Pettit (1990) and Pettit (1997b). I have
been deeply influenced by Braithwaite’s work in this area.
118 Social justice
regulation would regulate the action by permitting people to perform it, but
only on condition that they pay a certain absolute cost: in effect, it charges
them for the right to perform the action.
By contrast with opportunity-cost regulation, criminalization imposes
absolute rather than relative costs: it makes threats against those who perform
a criminalized act rather than offering rewards for those who avoid it. And by
contrast with admission-cost regulation, it does not exonerate criminal
offenders who are willing to accept the penalties attached to their behaviour
and pay the cost; it continues to condemn the acts performed, communicating
a presumptive disapproval on the part of the community in general of anyone
who would perform them (Duff 2001). In making these points, I draw on what
I consider as a more or less commonly shared understanding of what crimi-
nalization involves. I take myself to be explicating the meaning of criminaliza-
tion rather than prescribing anything about the form that it should assume.
Q. 2. Why should the state criminalize any acts?

It goes without saying that the state will have to regulate against certain
hostile types of act that citizens may perpetrate against one another. But
why should it regulate in the penal, condemnatory fashion of criminaliza-
tion? The answer from a republican perspective is straightforward. While
opportunity-cost or admission-cost regulation might reduce the probability
of interference with people’s basic liberties, it would not indict or reduce
domination in itself. Opportunity costs – in effect, refusable rewards –
would allow potential offenders to retain a power of interference. And
admission costs would allow them to interfere without shame on condition
that they are prepared to pay those costs.
Severe admission costs might work reasonably well in inhibiting potential
offenders, of course. But lacking the element of condemnation, even
extremely severe admission costs would be unlikely to do as well as costs
that have a condemnatory aspect. They would not serve in the same way to
indict offences and to affirm the protected status of victims against such
offences. And they might well fail to have the same deterrent impact.
Criminological evidence suggests that what keeps most people on the
right side of the criminal law is the fact that that law communicates public
disapproval of offenders and leads most people to put most offences off the
menu of possibilities that they consider as genuine options (Tyler 1990).
Q. 3. What acts should the state criminalize?

Criminalization involves the state in an oppressive role in which it can use
force to prevent criminalized acts and impose punitive retaliation for the
performance of such acts. The danger of allowing a state to become too
powerful, constituting a threat to domination, argues for restricting the
criminal law to acts of unlicensed interference with people’s basic liberties,
as those liberties are defined in the society, and to acts that make such acts of
interference more likely in various ways. Acts may make incidents of
interference more likely by undermining measures for their criminalization,
as in contempt for judicial proceedings; by putting preparatory steps in
place, as in purchasing materials required for crime; by inciting others to
such acts, as in offences against public order; by creating a slippery slope, as
in minor environmental offences that can spread to devastating effect;
and so on.
The restriction of the criminal law to acts in these categories does not
mean that all such acts should be criminalized; only those whose criminal-
ization promises a net benefit should be dealt with in this way.29 There
would be no useful point, for example, in criminalizing deception in
general, only deception perpetrated under well-defined circumstances, as
in the case of perjury. The restriction of the criminal law to acts in the
specified categories argues that many of the acts that count as criminal
offences in current legal systems, particularly in America and other
common-law regimes, ought to be designated as breaches of a non-criminal
sort; they ought not to attract the full condemnation of the criminal law
(Husak 2008). Crimes ought to be confined to the paradigms of violence,
fraudulence and theft that are recognized everywhere as offensive and to acts
that clearly open the way to such offences.
Q. 4. How should the state practise criminalization?

This is a large and difficult issue, bearing on arrangements for the pursuit,
prosecution, judgement and sentencing of offenders. But there are three
plausible principles that should guide policy in this area on the republican
approach (Braithwaite and Pettit 1990; Pettit 1997b). The first principle is
that people should be liable to conviction or sentence only to the extent that
they are culpable: only to the extent that they are fit to be held responsible
for an offence. Were the state entitled to impose sanctions independently of
culpability then it would be given an extraordinary degree of dominating
power. This principle requires departures from normal penal practice when
29
In taking this restrictive line I adhere to an implication of John Stuart Mill’s (1978) harm principle:
‘the only purpose for which power can be rightfully exercised over any member of a civilized
community, against his will, is to prevent harm to others’. That this is the only purpose for which
power can be rightfully exercised is consistent with its not being a purpose for which power is always
rightfully exercised.
120 Social justice
duly convicted offenders are deemed to be less than fully culpable.30 And it
supports a tough standard of due process, arguing that no one should be
convicted of an offence unless there is evidence of culpability that stands up
under arrangements that put the onus of proof on the prosecution.31 This
standard should reduce the capacity of authorities to scapegoat an innocent
party, thereby protecting against a salient possibility of domination. If a
police force or a court system can scapegoat opportunistically, then no one,
however innocent, will be guarded against such domination. And if this is a
matter of common awareness, then everyone will be likely to be intimidated
by the presence of such a capricious power in the criminal justice system.
But while the republican approach guards in this way against the punish-
ment of the innocent, or the punishment of the less than fully culpable, how
does it say that those convicted as guilty should be punished? Here a second,
obvious principle applies: that the penalty imposed should seek to rectify
the crime, vindicating the status of victims and potential victims, and yet at
the same time that it should keep open the prospect of reintegrating the
offender into the community of citizens. There are a number of require-
ments built into the rectificatory goal. First, to reassert the victim’s status as
a free person, condemning the offender’s assumption of power and, ideally,
eliciting a credible apology. Second, to compensate to the greatest extent
possible the victim – or in the case of a capital offence, the victim’s family –
for any material or related loss. And third, to reassure the community – in
particular, those for whom the victim’s status is indicative of their own
status – that they are no worse off as a result of the offence: the probability of
suffering such an offence has not increased.
The third principle supported by the republican approach is that while
the threat of domination argues against giving the criminal justice author-
ities a right to impose an exemplary punishment – this right could easily be
30
Where there is a shortfall from full culpability, different cases may require different departures from
normal procedure. Many forms of shortfall mean that the offence will not involve the first of the three
elements of harm that the text goes on to discuss, or at least not in the full degree; it will not impose
the will of a fully conversable agent. That argues that the departure from normal procedure should be
towards greater leniency. But the case of the highly dangerous, if not fully culpable, offender – say, a
certain sort of psychopath – suggests that this need not always be so. While such an offender ought
not to attract the condemnation attached to the criminal law, it may be necessary to impose severe
restraints for the sake of public protection. There is a difficult issue as to how such offenders ought to
be treated under a republican dispensation but I cannot explore it further here. I am grateful to Ben
Ewing for discussion of these issues.
31
For reasons advanced elsewhere, I believe that corporate bodies may be fit subjects for the criminal law
and that in their case it may make good sense to require, not actual guilt or mens rea – not full
culpability – but the capacity to develop procedures for guarding against the offence in question
(Pettit 2007d, 2009a; List and Pettit 2011).
abused – there is no parallel argument against giving them a right to exercise
mercy under specified circumstances. Thus if an offender becomes disabled
in the course of committing an offence, there may be no pressing need to
reassure the community, and independent considerations may argue for
clemency. The right to exercise mercy would occasion domination only if it
were used at the unchecked discretion of authorities to treat like cases
differently, or used in such a widespread fashion that offenders can generally
expect to escape the advertised rigours of the law.
Any sentence that incorporates rectificatory features will have a deterrent
quality, since it is bound to impose costs on the offender. These include the
reputational costs of condemnation and the material costs of compensation.
But they include in particular the costs that must be paid for communal
reassurance, since these are determined precisely by the requirements of
effective deterrence. A criminal justice system is designed in the first place to
provide people with a suitable level of public protection against crime, and
hence assurance that they are not likely to be subject to the will of criminal
offenders (Kelly 2009).32 The threatened penalty required by such protec-
tion and assurance is designed to deter potential offenders and it is the
imposition of such a deterrent penalty – the implementation of the threat –
that provides the reassurance that rectification of the crime may require.
Since reassurance will be less readily forthcoming with chronic offenders,
this consideration also allows for the possibility of increasing penalties for
repeat offenders (Braithwaite 2002).
This connection with deterrence does not mean in itself that the
approach taken here has a consequentialist character, since retributivists
can also argue that while penalties should be scaled with offences on
independent, non-consequential grounds, the nature of the penalties
employed at different levels should be determined by deterrent consider-
ations.33 But the approach sketched, like the approach adopted throughout
32
This means that the severity of sanction will often have to be designed to compensate for the difficulty
and unlikelihood of detecting a relevant offence and not just a function of the degree of harm caused
by the offender and their degree of culpability.
33
Retributivism, as I understand it, is defined by commitments like the following.
1. There are certain constraints, which are not fully sensitive to consequences, that apply to criminal
punishment.
2. These include a ceiling constraint and perhaps also a floor constraint; the first argues against
exemplary punishment, the second against mercy.
3. Punishment should involve hard treatment, and not just because hard treatment has a deterrent
effect or is required if criminal penalties are to have a protective purpose.
4. Penalties should be proportional, ordinally or otherwise, to the degree of severity of the harm
combined with the culpability of the act, and not to anything else (with a possible exception if
mercy is allowed; see commitment 2 above).
122 Social justice
this book, is indeed teleological or consequentialist in character. The argu-
ment is that judicial and penal practice in the state ought to be designed on
such a rectificatory pattern that overall it makes for the promotion of equal
freedom as non-domination. This sort of practice-wide consequentialism is
consistent, of course, with requiring individual judges and courts to abide
more or less rigidly by the requirements of the practice, treating those
requirements as unconditionally binding (Rawls 1955). Thus, while the
practice described would allow judges and courts to be merciful in certain
cases, provided they can justify the exercise of such discretion, it would not
allow them ever to impose an exemplary punishment, making a generically
deterrent example of a particular offender.
4. the character of republican justice

Our observations in the preceding section are broad-brush strokes but they
should provide at least a rough picture of where republicanism leads in the
theory of justice. It is important to have a sense of the republican require-
ments of justice before we approach the discussion of legitimacy. The
theory of justice is meant to provide an image of how people’s social
relations ought to be organized, if they are each to enjoy equal freedom as
non-domination: if they are each to avoid private domination. The theory
of legitimacy, to anticipate later discussion, is meant to provide an image of
how the state ought to make its decisions about justice and other matters, if
citizens are to enjoy equal freedom as non-domination in relation to
government: if they are each to avoid public domination. Unless we have
a sense of what private non-domination requires, we will be ill positioned to
make a judgement on the requirements of public.
In concluding this discussion, I would like to draw attention to some
characteristic features of the republican theory of justice, as it has been
outlined here. The theory is distinctive on three broad fronts: one, in the
principle or standard of justice that it deploys; two, in the demands of
justice that it licenses; and three, in the importance that it accords to the
public recognition of the laws – and the corresponding norms – that secure
the basic liberties.
5. Merely ordinal proportionality would require that penalties and crimes should be ordered from
toughest/worst to least tough/least bad and then paired off appropriately, with the toughest
applied to the worst crime, the next toughest to the next worst, and so on. Non-ordinal
proportionality would require that something more is possible in either or both of the two
scales; the limit case would be a belief in the lex talionis: an eye-for-an-eye, a tooth-for-a-tooth,
and so on.
The character of republican justice 123
The principle of republican justice
The principle or standard that the theory deploys, as we have seen, is simply
that arrangements between citizens should be designed to promote people’s
equal enjoyment of freedom as non-domination: that is, the equal entrench-
ment of their basic liberties on the basis of public laws and norms. This
principle is modest in its suppositions, goal-oriented in form, and general in
scope – in particular, general enough to be able to serve in either ideal or
non-ideal theory. Corresponding points apply to the principle of republican
legitimacy, as that emerges in the next chapter, but it will suffice to show
how they apply in the case of justice.
The republican principle is extremely modest in characterizing justice by
the requirements of equal freedom alone, without reference to any of the
richer and more controversial values invoked in many other theories of
justice; it is minimalist in the terms required to justify the policies supported
(Cohen 2004). What enables it to achieve this modesty is the independently
plausible argument that for relevant, unvitiated choices freedom requires
non-exposure to the interference of others, not just the absence – the
fortuitous absence, as it might be – of that interference. This construal of
freedom is plausible and also, as we saw in the Introduction, one with a long
historical pedigree. It was only in the eighteenth century, in the work of
figures like Jeremy Bentham and William Paley, that freedom came to be
routinely characterized as requiring non-interference rather than non-
domination. Contemporary theories of justice all give importance to free-
dom but take equal freedom to require so little that its requirements have to
be supplemented from other sources – as, for example, Rawls’s second
principle – in order to fix the demands of justice.
The principle of justice deployed in the republican theory, to move to a
second characteristic, offers a goal-oriented or consequentialist criterion of
evaluation. It says that a socio-political arrangement – a basic structure, as
Rawls calls it – will be just to the extent that it promotes free or undomi-
nated choice amongst its citizens, under the expressively egalitarian con-
straint of treating them as equals. More concretely, assuming that the
circumstances of justice obtain, it prescribes the promotion of the substan-
tively egalitarian goal of equal status freedom for all. This does not mean
that agents within the socio-political system are generally licensed to behave
opportunistically or strategically, as already mentioned, adjusting their
responses according to what they happen to think is best for promoting
that goal. It is the structure itself that is designed to promote the goal and it
is only likely to do this if the agents acting under the institutions established,
124 Social justice
be they officials or citizens, are generally required to act in accordance with
the duties and rights that they have under that structure (Rawls 1955).
Not only is it a practice-wide form of consequentialism; it is also a
constitutive, as distinct from a causal, variety of the approach (Pettit
1997c: Chapter 3). The goal of equal freedom as non-domination is modally
demanding, as we saw at the end of the last chapter, requiring a pattern of
non-interference that is robust across scenarios in which there are variations
both in what people prefer to do and in what others prefer that they do. But
people will enjoy the robust non-interference associated with justice – the
status of a free person – as soon as there are institutions of resourcing and
protection in place to safeguard them. The existence of those institutions
will not provide for people’s free status by way of a downstream causal effect
that takes some time to materialize. It will establish them in that status
without waiting on the result of any contingent causal process. The insti-
tutions will relate to the free status they establish in the way that the
antibodies in your bloodstream relate to the immunity against a certain
disease that they constitute. We can understand what such freedom requires
without knowing which institutions are required to support it – thus
freedom is not defined in terms of those institutions – as we can understand
immunity without knowing anything about antibodies. But still the con-
nection between the institutions and the freedom, or the antibodies and the
immunity, is not of a causal character. As the antibodies promote the
immunity by constituting it, so we can say that the institutions promote
freedom and justice in a parallel, constitutive fashion.
We have seen that the principle of republican justice is modest and goal-
oriented. A third feature worth mentioning is that it is available for the
applied exercise of measuring and comparing regimes for how just they are,
not merely for the theoretical exercise of identifying the perfectly just
society. It is a general theory of justice, as we might say, not the sort of
transcendental theory, as Amartya Sen (2009) describes it, which can rule
only on matters of perfect justice. While we can use the principle for
identifying an ideally just society – say, a society that is just enough to
pass the eyeball test – we can also use it to evaluate and rank the rather less
than ideal systems with which the real world presents us and to track
piecemeal progress within them (Marti and Pettit 2010: Chapter 5).
There are two reasons why the theory can serve us in the exercise of
measuring and comparing actual, imperfect regimes for how just they are.
The first is that it is a consequentialist theory that associates justice with
approximation to a certain goal. And the second is that it does not – or does
not necessarily – make any idealizing assumptions about human nature or
compliance with the law that would render it difficult to apply in the
ordinary world. These two features combine to ensure that in principle
we can measure and compare actual regimes for how just they are by seeing
how close they get to the relevant goal. In both respects the approach
contrasts with Rawls’s theory of justice. This insists in a non-
consequentialist spirit that justice is associated with instantiating a certain
pattern – the two principles – not with promoting it. And, perhaps even
more restrictively, it operates with the idealizing assumption that in ranking
regimes by reference to this pattern, we should assume that there is full
compliance in each with whatever laws are established there.34
But to say that the republican theory of justice allows for the applied
exercise of measuring and comparing justice in actual regimes is not to say
that our discussion has done enough to equip it for such application. There
are different dimensions on which a society’s degree of approximation to
establishing equal freedom as non-domination might be charted and a fully
applicable theory would have to provide weightings for these. As between
two different societies, for example, people might differ in enjoying freedom
in different sets of choices or in enjoying it with different degrees of
entrenchment. And where they enjoyed the same low degree of entrench-
ment for the same choices, they might differ in the likelihood of achieving
preference-satisfaction: they might differ in enjoying entrenchment, here
for exercising options they are likely to prefer, there for exercising options
they are unlikely to prefer. The republican theory can be used in practice for
measuring and comparing justice only under stipulations as to the relative
importance of these and other dimensions of variation in the measurement
of freedom and other values. And if no plausible stipulations are available on
the relative importance of two dimensions, that will mean that societies that
differ insofar as one does better in the one dimension, the second in the
other, may have to count – perhaps quite intuitively – as displaying justice
in the same degree.35
34
Rawls (1971: Part iii) argues that his two principles will tend to be internalized by those living under
them and to constitute a stable structure. But it would seem to be more satisfactory to consider
different principles, without assuming universal compliance, and then to rank them by reference,
among other standards, to how far they are likely to attract general compliance and display stability.
35
A full and applicable political theory would not only have to allow for variations in different
dimensions of social justice but also for variations in the relative degree of political legitimacy or
international sovereignty that societies enjoy. While there are reasons for giving political legitimacy a
basic importance, as mentioned briefly in the Introduction, the issue of weighting between these
different aspects of performance is too complex to be addressed fully here.
126 Social justice
The demands of republican justice
The republican theory of justice displays characteristic features, not just in
the principle that it deploys, but also in the demands that it supports. The
demands of the theory are quite substantive, despite the modesty of the
principle from which they are derived, as our institutional model of
republican justice indicates. They are designed to put resourcing and
protective measures in place that guard against personal domination in
the sphere of the basic liberties to the extent of enabling people to pass the
eyeball test. They are meant to reduce the incidence of domination
between people and to nullify the institutional or structural factors that
facilitate it (Hayward 2011).
The demands made by republican theory are certainly not as extensive as
those of egalitarian theories that seek the elimination, for example, of all the
effects of brute luck on people’s fortunes.36 Nor are they even as extensive as
Rawls’s requirements. His second principle of justice would look for mate-
rial equality, after all, except when the second principle of justice moderates
it, allowing a degree of relative inequality on condition that it improves the
absolute returns to the worst-off position. But those theories often seem like
moral fantasies: manuals for how God ought to have ordained the order of
things – or manuals for how we ought to rectify God’s failures – rather than
real-world manifestos for what the state should do in regulating the affairs of
its citizens.37
Such comparators aside, the republican theory of justice would still look
for quite extensive, even radical measures; while minimalist in its justifica-
tory base, it is not minimalist in substance (Cohen 2004). Beyond the
infrastructural requirements on which all sensible theories ought to con-
verge, as we have seen, it would seek a high level of social insurance, a firm
basis for the insulation of people in vulnerable relationships and a suitable
basis for the general insulation provided by the criminal justice system. It
would provide for the basic functioning capabilities of all citizens, as Sen
(1985) and Nussbaum (2006) describe them, and would enable people to
look others in the eye, without reason for fear or deference – or at least
without any reason related to the danger of interference.
36
For a recent interpretation and defence of luck egalitarianism see Tan (2008) and for a response that I
find congenial see Sanyal (forthcoming).
37
Hegel (1991: 80) has a salutary remark on the topic: ‘One cannot speak of the injustice of nature in the
unequal distribution of possessions and resources, for nature is not free and is therefore neither just
nor unjust.’
In arguing that the just state ought to promote equal freedom as non-
domination amongst its citizens, the republican theory of justice does not
suggest that freedom as non-domination is the only value that matters, as a
utilitarian theory might argue that utility is the only relevant value. What it
holds, however, is that if we look after the requirements of equal freedom as
non-domination, then we will have looked after the requirements of many
other values as well: for example, the value of enjoying functioning capa-
bilities. Look after equal freedom as non-domination in the relations
between citizens and you will have looked after an intuitive, left-of-centre,
account of the demands of justice. And, to anticipate the argument of the
next chapter, look after equal freedom as non-domination in the relations
between citizens and state and you will have looked after an intuitive,
democratic account of the demands of legitimacy. Freedom as non-
domination is not the only value in politics but in the account defended
here it serves a gateway role: if we pay the price of securing freedom as non-
domination in a suitable measure, we will have paid enough to secure social
justice and political legitimacy (Pettit 2005a; for a dissenting view see
Markell 2008).
The role of law and norm in republican justice

How far are the demands of republican justice likely to be met without
establishing a high degree – perhaps an unattainably high degree – of
material equality? Their extra material resources may enable the richer to
give themselves extra protection, and the protection provided publicly will
have to be good enough to make any such private measures seem redun-
dant. Again, their extra resources may enable the richer to breach the
defences of the poorer, and the protection provided publicly will have to
be good enough to protect against such private assaults. Can the public
entrenchment of the basic liberties guard against such dangers?
One reply to this question might be that the state may regulate the uses of
wealth so that there is a limit to the extra protections or powers that the rich
can enjoy. That is certainly true, but what I want to emphasize here is
something distinct: that there is a quality to the public resourcing and
protection that law can provide, at least when things go well, which
provisions of private wealth can do little or nothing to match. Insofar as
the law is the work of a legitimate state, it will connect with public habits of
mind and find support in communally endorsed norms. And in doing this it
can establish for citizens an entrenched status – their public status as free
128 Social justice
persons – that suffices as a bulwark against the advantages on which the rich
can draw.
Norms, by our earlier account, are regularities of behaviour in a society
such that, as a matter of shared awareness, most members conform to them,
most expect others to approve of conformity or disapprove of non-
conformity, and most are reinforced in this pattern of behaviour by that
expectation. Not every regularity will be a norm, then: like the regularity
whereby people eat so many times a day, it may not be taken to attract
approval or disapproval. And not even every regularity that is taken to
attract approval will be a norm: like the regularity whereby people wash
themselves regularly, the expectation of such approval may play little or no
role in supporting it. In order to count as norms, behavioural, presump-
tively approved, regularities have to obtain as a result, at least in part, of
being expected to attract approval. And for full measure, all of this has to be
a matter of shared awareness: each is in a position to recognize that the
conditions are fulfilled, to recognize that others recognize this, and so on.
By this definition, any society is going to abound in norms, since
human beings everywhere are disposed to approve and disapprove of the
behaviour of others and are equally disposed to cherish approval and
shrink from disapproval (Brennan and Pettit 2004: Part i; Appiah 2010).
Adam Smith was particularly emphatic on the point. ‘Nature, when she
formed man for society, endowed him with an original desire to please,
and an original aversion to offend his brethren. She taught him to feel
pleasure in their favourable, and pain in their unfavourable regard. She
rendered their approbation most flattering and most agreeable to him for
its own sake; and their disapprobation most mortifying and most offen-
sive’ (Smith 1982: 116).
The norms that we might expect to emerge in any society will include
some that apply, perhaps with an unsavoury, self-seeking effect, to specific
groups: those in a particular business or political party or criminal gang. But
they are also likely to include norms of a socially beneficial kind that serve,
on the one side, to govern the behaviour of individuals in particular
positions – say, parents and children, teachers and pupils, the governing
and the governed – and on the other to indict activities that are damaging to
people in general such as deception and violence, infidelity, fraud and
free-riding.
The secret strength of law is that if it is well shaped and well supported –
if it is relatively just and legitimate – then it can recruit such beneficial,
communal norms to the cause of its enforcement (Richardson 2002;
Bohman 2007). Properly promulgated and defended, indeed, it can also
reshape any existing norms that do not serve the cause of justice. Once law
gains normative reinforcement it no longer has to rely on the strength of the
public sword, in the old metaphor, for winning compliance amongst the
citizenry. Being disposed to approve of compliance and to disapprove of
deviance, citizens can become non-intentional sources of enforcement, as
they observe one another’s behaviour, and form attitudes of approval and
disapproval in response (Pettit 1997c: Chapter 7). And of course that
enforcement will be further strengthened by people’s disposition to talk
to one another about the behaviour of offenders and perhaps to talk to the
offenders themselves about their misdeeds.
The fact that law can enjoy this sort of support means that when it secures
the resourcing and protection of individuals then it can be powerful
enough, in principle, to achieve a balance in people’s relative positions,
even when those individuals vary considerably in the private resources at
their disposal. Thus, despite continuing variations in such resources, it can
do much for the equal enjoyment of freedom as non-domination across the
society. It is for this reason indeed that we can expect that, short of material
equality, people may still be able to pass the eyeball test. They may still be
able to live up to the image of the free person – the liber, or ‘free-man’ – that
is central to republican thought.
chapter 3
Political legitimacy
In the last chapter we focused on issues of social justice: that is, justice in the
relations between people within a state, including relations mediated by
their belonging to groups and bodies of various kinds. According to the
theory developed, justice requires the state to promote the freedom as non-
domination of all its citizens – broadly, all adult, able-minded and more or
less permanent residents – safeguarding their fundamental liberties on the
basis of public laws and norms. This focus on relations amongst citizens
leaves out of consideration the relations between citizens as a whole and the
state itself. It ignores the question of whether the state operates with
political legitimacy in imposing a social order, however just that order
might turn out to be. It is one thing to argue that the social order imposed
by a state is just, it is quite another to argue that the political imposition of
that order is legitimate.
Social justice does not entail political legitimacy, by this account, nor
does political legitimacy entail social justice. Thus, to take the second
dissociation first, a state might be fully legitimate, by whatever criterion,
and yet not succeed in furthering the cause of social justice very well; it
might support misconceived, if not ill-willed, policies. It is a failure of this
kind that Rousseau had in mind when he acknowledged the possibility that
the perfectly legitimate regime – in his terms, the regime that seeks to enact
the general will – may still go astray: ‘By itself, the people always wills the
good, but by itself it does not always see it’ (1997: ii.6.10).
Turning now to the other dissociation, it also seems that the just society
might fall well short of being fully legitimate. The traditional paradigm of
the illegitimate regime is that which is controlled by a despot or by a foreign
power. But we can imagine a benevolent despotism, or an enlightened
colonialism, under which people’s relations with one another are ordered
in a socially just manner. That order might not have the robust entrench-
ment that we would associate with justice, especially from a republican
viewpoint. The will at its source might not have the reliability – the
130
Political legitimacy 131
resistance to despotic or colonial discretion – that the Roman Digest of law
requires when, quoting Ulpian, it says that ‘Justice is the steady and
enduring will to render unto everyone his right’ (Watson 1985: Book i,
Part 1, para. 10).1 But in other respects, it could be beyond reproach.
These remarks are meant to underline the conceptual divergence
between the demands of social justice and the demands of political legiti-
macy, not to suggest that in practice they can come apart very deeply. It is
unlikely that people who fared badly in justice terms could do much about
constraining the state to satisfy legitimacy. And it is unlikely that the
unconstrained state would do much to establish justice amongst its citi-
zenry. But the fact that political legitimacy can come apart in any measure
from social justice means that it makes a distinct demand and that we have
to begin afresh in thinking about what it requires.
Not only does political legitimacy make a distinct demand, it makes a
demand that we cannot ignore. In any society there are going to be different
views as to what social justice requires in the organization of people’s
relationships. Thus there is going to be a question as to whether it is
appropriate for people, recognizing their differences with one another, to
accept the decisions of the state as binding on them all and to submit to the
coercive application of those decisions by the state. And it will be appro-
priate for them to do this, presumably, only insofar as there is something
about the standing of the state in relation to them as a whole – something
about the relations between the state and its citizens – that makes it into the
legitimate arbiter and decider in their lives. The question for a theory of
legitimacy is to identify the factor that can give such a standing to the state
(Nagel 1987).
My discussion in this chapter falls into five sections. In Section 1, I look at
the legitimacy issue itself, asking after the precise source and nature of the
problem, and seeking to combat the tendency to let it drop from sight. In
Section 2, I argue for a tight linkage between legitimacy and freedom and
show that under freedom as non-domination legitimacy has a straight-
forward requirement: a form of popular or civic control over the state. In
Section 3, as in an intermission, I analyse the notion of control in general,
contrasting it with consent. In Section 4, I argue that if it is to support
1
The Latin is: Justitia est constans et perpetua voluntas ius suum cuique tribuendi. The view that justice
does not require such constancy of will is expressed in Leo Strauss’s (2000: 75) reading of the Greek
tradition, which he may or may not endorse himself (see Vatter 2011): ‘The just man does not hurt
anyone, but helps everyone who has dealings with him. To be just, in other words, simply means to be
beneficent. If justice is then essentially translegal, rule without laws may very well be just: beneficent
absolute rule is just.’
132 Political legitimacy
legitimacy, then popular control must be individualized, unconditioned
and efficacious in character, satisfying a rich version of the democratic ideal.
And then in a short Section 5, I make some observations on the desirability
of such a democratically controlled state, drawing attention to differences
between the perspective adopted here and other approaches.
The chapter provides a theory of democracy, arguing that the job of
democratic institutions is to ensure the form of popular control required for
political legitimacy. But the discussion does nothing to identify the sorts of
institutions that might actually do this job. That task is taken up in chapters
4 and 5, which provide an institutional model of democracy that is meant to
illustrate what the republican theory requires.
We argued in the last chapter about the requirements of republican
justice, starting from the expressively egalitarian assumption that the state
ought to treat its citizens as equals. As explained in that earlier argument, the
same egalitarian assumption will direct the argument in this chapter, when
we look for what is required by republican legitimacy. The normative
assumption is, on the one side, that the state ought to treat its citizens as
equals and, on the other, that citizens ought to be willing to accept this and
to live on equal terms with one another. This assumption is intuitively
compelling and is not specific to the republican approach; it is endorsed in
every plausible political philosophy.2
1. the legitimacy question
The source of the problem

We assumed in our discussion of social justice that the state is required for
promoting justice and it is this assumption that raises the question as to
whether the state that imposes a social order – ideally, a just social order –
imposes it in a legitimate manner. But is the state necessary for the
promotion of social justice? Is it required for entrenching a suitable range
of basic liberties, as republican justice prescribes? There are a number of
considerations, rehearsed in Section 4 of this chapter, which suggest that the
state is a more or less inescapable institution, not something we might
choose to have or not to have. But suppose for the moment that the state
2
The egalitarian aspect of the theory pursued in this book is emphasized by the extent to which it
coincides with the democratic egalitarianism advocated by Sanyal (forthcoming). He looks broadly for
equality of non-domination in relations between citizens and for what he calls equality of autonomy in
relations with the state; but the latter is close to equality of non-domination in relations with
government.
The legitimacy question 133
were not inescapable. Would the cause of republican justice – or indeed the
cause of any plausible view of social justice – still require us to invent it? I
argue that it would.
The state, as I conceive of it here and throughout this book, is not just an
impersonal apparatus of rules and routines, dictating how officials operate
within its domain; in this respect it is unlike the market, for example. The
state is an agent or agency that espouses any of a variety of purposes and
pursues those purposes according to reliably maintained representations of
the opportunities and means at its disposal. And, most important of all, it is
an agent that presents itself to other agents – its own citizens, the organ-
izations that operate within its boundaries, as well as other states and
international bodies (McLean 2004) – as an entity with which it is possible
to do business: an entity that is conversable in the manner of an individual
human being. In this self-presentation it operates like a legal person, making
avowals of attitude and promises of action to which, on pain of failing as a
state, it allows itself to be held (List and Pettit 2011). While the state cannot
speak or act except on the basis of the words or actions of different govern-
ment agencies and officials, those agencies have to be coordinated with one
another so that the state speaks with a single voice and acts to a coherent set
of ends. If certain agencies or officials do come apart on either front, making
inconsistent avowals or promises, or acting in ways that undercut one
another, then they have to be sensitive to the need to restore harmony
and they have to be able to mobilize certain procedures for doing so.
The state is necessary for implementing a conception of justice like the
one outlined in the last chapter, because only an agency of that kind would
be capable of discharging the many varied and demanding tasks involved:
maintaining the developmental, institutional and material infrastructure
that justice requires; establishing and adjusting the laws required to identify
substantive and coherent basic liberties; ensuring that those liberties are
resourced on the basis of any needed conventions and subsidies; and
protecting people against the invasion of those liberties, whether in parti-
cular relationships or on a more general front. The tasks involved here are so
complex, interconnected and dynamic that no abstract apparatus of rules
could plausibly ensure their fulfilment. There can be no effective system of
justice, so it appears, in the absence of a state.
The fact that a corporate, conversable entity is required for establishing a
system of justice not only means that there has to be a state. It also means,
more specifically, that the state has to have two characteristic features. The
state must have the power of coercively imposing the order it establishes,
threatening and implementing penalties for those who disobey, without
competition from rival bodies. And, given that this is going to be manifest
to all, the state must as a matter of its practice make an explicit or implicit
claim to having the exclusive authority to use coercion in that manner.
But isn’t this argument for the necessity of a corporate, coercive state too
quick? Might a just order not be capable of emerging, under suitable
conditions, as the by-product of a system of communal norms – a network
of rules that no single agency like the state is needed to create or trigger,
monitor or shape? Communal norms, as we saw in the last chapter, are
regularities of behaviour that are maintained in part as a result of the fact,
registered in common awareness, that people generally expect the approval
of others for conforming to the regularities and/or the disapproval of others
for not conforming; the background assumption is that, for intrinsic or
instrumental reasons, people will generally seek the approval of others and
shun their disapproval. Suppose then that, independently of state or law,
communal norms emerged that had the effect of exposing all forms of
deception, fraud, manipulation, intrusiveness and violence to the condem-
nation of the community. Might such a system of norms not prove
sufficient for establishing a regime of justice?
The possibility of a rule of impersonal norm is intriguing from a
republican point of view. The norms envisaged would be established by
individual people, and the associated penalties would be imposed by
individual people. But they would emerge and stabilize as by an invisible
hand, since neither the introduction of the norms nor the imposition of
costs would require a state. An effective regime of such non-intentionally
policed norms should be very appealing for anyone concerned with free-
dom as non-domination. By hypothesis it would serve to protect people
from the domination of others, imposing costs that undermine the power
that the powerful might have had to interfere. And yet it would protect
them without imposing the potentially dominating will of a protective
agency; it would protect them in the benign manner of a will-independent
force for good.3
3
The ideal of a rule of impersonal, will-independent norms may have had a certain presence in the
history of republican thought. It often seems to lurk in the background when writers emphasize that
the great protection against the tyrannical and, more generally, the illegitimate regime is an ‘empire of
laws and not of men’, as James Harrington (1992: 8) described it, invoking Aristotle and Livy in his
support. The tradition led John Adams (1776) to write that ‘the very definition of a republic is “an
empire of laws, and not of men”’. This line of thought is also present in many authors who enthuse
about the common law, on the grounds that it emerges without a concerted, potentially dominating
will from decisions over particular cases. See for example Hayek (1988), who may have been influenced
by Kant’s (1996: 294) belief that ‘the sovereign, which gives laws, is, as it were invisible; it is the
personified law itself, not its agent’. For a commentary on the Kantian influence on Hayek, see
But however attractive the idea of an impersonal, will-independent
regime of protection, there is little hope that it could operate effectively in
the absence of a state. There are three salient problems that it would
confront. A first is that spontaneous norms would be unlikely to identify
basic liberties of an appropriate kind, particularly in view of the differences
between people on relevant matters and the difficulties of sorting out a
suitable set of liberties. As we saw in the last chapter, a variety of more or less
conventional rules, and provisions for the continuing adjustment of those
rules, are required in order to define suitably co-exercisable and co-satisfying
liberties.
A second problem with the proposal is that even if spontaneous norms
were satisfactory in other respects, they would be unlikely to mandate and
support full resourcing of basic liberties, especially when such resourcing
requires a degree of redistribution. Unregulated by the agency of a state,
wealth and power tend to accumulate in fewer and fewer hands. As by an
‘iron law’, to quote a recent history of political order, ‘the rich tend to get
richer, in the absence of state intervention’ (Fukuyama 2011: 368). It is
extremely unlikely that any spontaneous norms could resist the effects of
growing economic accumulation and ensure the resourcing of basic liberties
for the poor as well as the rich.
An order of spontaneous norms would be unlikely, then, either to
identify or to resource a suitable set of basic liberties. A third problem is
that it would also be unlikely to provide suitable protection for all.
Communal norms may assume any of a variety of objectionable profiles,
as they reflect divisions within society and impose patterns that are highly
injurious to those on the weaker side of gender or religious or ethnic or other
divides. It was for this reason that John Stuart Mill (1978) railed against the
despotism of custom. Norms of the kind envisaged may not impose the will
of a potentially dominating agency; after all, they are sourced and supported
by non-intentional attitudes. But they will almost certainly fail to protect
the members of certain groups against the most rampant domination on the
part of others.4
Kukathas (1989). A more plausible view of the laws is found in Demosthenes (1939: Section 224) in
Against Meidias: ‘And what is the strength of the laws? If one of you is wronged and cries aloud, will the
laws run up and be at his side to assist him? No; they are only written texts and incapable of such
action. Wherein then resides their power? In yourselves, if only you support them and make them all-
powerful to help him who needs them. So the laws are strong through you and you through the laws.’
4
See Fukuyama (2011: Chapter 17) for a related, salutary discussion of how the emergence of the rule of
law typically depends on the support of a state; it does not appear as by an invisible, non-intentional
hand.
I conclude that the cause of social justice requires the corporate agency
of a coercive state and that no apolitical order could serve in that role. To
assume that the state is essential for the promotion of social justice, of
course, is not to assume that any particular sort of state is essential, and
certainly not to assume that the existing pattern of states is ideal. But the
assumption that any sort of state is necessary generates the problem that
is now at the centre of our concern. Given that a state is necessary for
justice, there is an issue about how it ought to relate to its citizens, as
distinct from the issue of how the citizens ought to relate to one another.
And that, as I conceive of it, is the issue of legitimacy. Where the issue
of social justice is a matter of the horizontal relations of citizens to one
another, political legitimacy is a matter of their vertical relations to the
state that rules over them.
What turns on legitimacy?

Where the issue of social justice bears on the nature or character or content
of the social order established amongst people in the state, then, the issue of
social legitimacy bears on its source or origin: that is, on how far the state
that coercively imposes that order relates appropriately to the people on
whom it is imposed. Does it relate to people in such a way that however well
or badly it does by anyone’s view of social justice – and people’s views of
justice may differ considerably – they each have to admit that it operates
with a suitable licence or pedigree and that its presumptive authority to
identify and impose the social order is unobjectionable? To give a positive
answer to that question is to affirm the legitimacy of the state; to give a
negative answer is to deny it.
What turns on whether a state is legitimate or not? What implications
does its legitimacy have for how the citizens of the state ought to behave?
And in what way do they differ from the implications of justice? As I
understand the two ideals, they each impose a different set of moral
obligations on citizens. The obligations imposed are pro tanto obligations
that may not apply in special, emergency circumstances, as when the life of
an innocent party is at stake. But they are none the less important for that.5
5
That the reasons are pro tanto in character means that they can be overridden by countervailing
considerations, as when the red lights go on. But consistently with having this pro tanto status, the
reasons might function under normal conditions – that is, in the absence of red lights – in an
exclusionary manner that precludes exercises in the weighing of pros and cons (Raz 1986). I am
grateful to Caleb Yong for a discussion on related matters.
Suppose that the social order imposed by a state is just and that the state
itself, by whatever criteria, is legitimate. In that case we will naturally say
that in view of its legitimacy people are morally obliged to accept the
regime, in some sense of acceptance, and that in view of its justice they
are obliged to endorse and comply with the laws. But what does it mean to
accept a regime? And how does such acceptance come apart, if at all, from
endorsing and complying with its laws?
The answer appears once we consider a case where the regime continues
to count as legitimate but certain of its laws are unjust. The injustice of
those laws will mean that people are not subject to the same content-
dependent obligation to endorse and comply with them, although they
will certainly be subject to a fall-back, pro tanto obligation that applies in all
half-reasonable systems of laws, just and unjust. This is the content-
independent obligation to comply that derives from the fact that once social
regularities are established by law people generally coordinate their activities
around the expectation of mutual compliance (Raz 1986). But the injustice
of certain laws will also mean, plausibly, that people are morally permitted,
perhaps even obliged, to try to change those laws. That raises the question as
to what is allowed to people in their attempts to change unjust laws. And it
is here that legitimacy, and the acceptance that I have associated with
legitimacy, are relevant.
The acceptance of the regime means, I propose, that attempts to change
unjust laws should be restricted to measures that are consistent with the
regime’s remaining in place. It requires you to acknowledge the state as the
appropriate arbiter and decider of legal issues, rather than taking the law
into your own hands, and to campaign for change on the assumption that
the state is to retain that role. Legitimacy imposes a pro tanto moral
obligation, then, if you oppose certain laws or measures – and given differ-
ent conceptions of justice, everyone will be disposed to challenge some – to
oppose them in ways allowed by the system: to stop short of revolution or
rebellion or, in an older word, resistance. It makes it permissible, invoking
justice or some other virtue, to oppose certain laws within the system: in a
word, to contest them. But it makes it impermissible to reject or resist the
regime itself.
Most regimes will offer some ways of opposing its laws that are clearly
within the system: appealing to the legislature, taking the government to
court, speaking out in the media, demonstrating in the streets and, of
course, challenging the governing party at election time. But the act of
breaking the law – notwithstanding coordination-based reasons to comply –
may also count as a mode of contestation, a way of opposing laws within
the system.6 This is exemplified in campaigns of civil disobedience where,
for example, campaigners break the laws – perhaps the laws they oppose,
perhaps other laws – in order to display opposition. When brought before
the courts, campaigners typically acknowledge the authority of the courts to
penalize them, thereby displaying their acceptance of the regime itself and
disavowing revolution or resistance. They invoke their willingness to accept
the authorized penalty for whatever abuse they have committed to draw
attention to the injustice of those laws.7
The legitimacy of a social order is often thought to correlate with the
political obligation of citizens. Such political obligation, on the proposal I
am making, is not the pro tanto obligation to obey the laws but the pro tanto
conditional obligation, if you oppose the laws, to oppose them within the
system.8 This conception of political obligation is independently plausible,
since political obligation is traditionally taken to bind people insofar as they
are citizens proper, not merely temporary or transient inhabitants. Even
those passing through a society have a pro tanto obligation to obey the law;
like others, they will have a content-dependent obligation to obey just laws,
and may have a content-independent obligation to obey any law, just or
unjust. But only citizens are likely to have substantive rights to oppose the
law within the system and only they can be meaningfully bound to limit
their opposition to intra-systemic contestation.
This comment squares with the broad conception of citizenship with
which I have been working. All adult, able-minded, more or less permanent
residents count as citizens, on this conception, not just those with the right
to vote and stand for office. While not all citizens in my broad sense will
have electoral rights, they will all have formal and informal rights to oppose
the law of a kind not readily given to those merely passing through. And in
effect most will have the right to seek electoral rights – and to expect to gain
them – by applying for formal citizenship. They will be in a position akin to
those enfranchised citizens who choose not to enrol on the voting lists, or
not to take part in voting, and who thereby deny themselves the oppor-
tunity to exercise their electoral rights.
6
The fewer the means of opposing a regime within the system, of course, the less likely it is that the
regime will count as legitimate: that is, the more likely it is that there are going to be some unjust
laws such that the available ways of opposing them within the system are so few that people are no
longer obliged to confine themselves to such opposition. But the availability of civil disobedience
means, at least in theory, that there is always at least one way in which a regime can be opposed
within the system.
7
For a broadly congenial – purportedly ‘republican’ – account of the sort of civil disobedience that
might be warranted see Markovits (2005).
8
For an intriguing but rather different view of political obligation, see Gilbert (2006).
The discussion so far speaks as if it is only the state, or the social order
it imposes, that counts as illegitimate. But this is for convenience only.
When a state or regime is illegitimate then the laws it upholds are
individually illegitimate and the branches of government that run the
state are illegitimate too: they are tainted by the illegitimate routines of
appointment employed. But there is also room for claiming in the same
sense of the term that while a regime is generally legitimate, certain laws
or appointments – including the appointment of an executive or other
branch – are illegitimate: they happen to breach conditions of legitimacy
that the regime generally respects. While I shall continue to concentrate
on the legitimacy or illegitimacy of regimes, I do not mean to deny that
there is room for invoking the concept in more local complaints as well.
And of course I allow that illegitimate laws, being objectionable like
unjust laws, may be suitable targets for contestation within a legitimate
regime.
Is it appropriate, however, to treat the legitimacy of a regime as an on–
off matter, as in effect I have been doing? I believe that the factors that
make for political legitimacy, like the factors that make for social justice,
come in different degrees. But as I am happy to treat social justice as an
on–off matter, given the eyeball test for determining where the justice
threshold lies, so for similar reasons I am happy to treat political legitimacy
as an on–off matter. As we shall see, there is an intuitive tough-luck test for
determining in a parallel way where the legitimacy threshold falls.
I said earlier that while the justice of the laws gives people a content-
dependent if pro tanto reason to conform, there will almost always be a
content-independent reason to conform to any reasonable system of laws,
just or unjust, viz., that they serve as coordinating devices around which
people build mutual expectations. Given that the laws of a legitimate
regime may not themselves be legitimate, and given in particular that the
regime itself may fall short of legitimacy, it is natural to ask whether there
is any corresponding fall-back reason why people should still not resist
the regime: any reason why they should restrict themselves to opposing it
within the system.
The issue is not central to our concerns but, for the record, I think that
there may be a very good reason – and not just a reason of personal
prudence – to stop short of resistance to a certain sort of illegitimate regime.
This is that however illegitimate it may be, the regime is still capable of
being made legitimate by being treated as if it were legitimate: that is, by
being opposed only within the system. Without being fully legitimate, the
system may in that sense be legitimizable. And my proposal is that
legitimizability can plausibly count as a fall-back reason to avoid outright
resistance and oppose a state only from within the system.9
The virtue of patriotism, traditionally associated with taking sides with
your existing state in international conflicts, might be taken in the domestic
context to argue for this sort of fidelity to an existing state and constitution,
even in the face of the state’s failure to be legitimate (Mueller 2007). I shall
not have much to say about legitimizability or patriotism in the remainder
of this book. But it should be clear that the issue is of immense importance
when it comes to drawing practical implications for how far people are
politically obligated to the states they live under. Few actual states may
count as legitimate, even by the realistic, tough-luck test of legitimacy
introduced later, but many are likely to count as at least legitimizable.
We have seen that according to any theory, the legitimate and just order
imposes a moral obligation to accept the regime and to comply with the law –
as always, in a pro tanto sense – and that the legitimate but unjust order
imposes an obligation to accept the regime, opposing unjust laws – and, of
course, illegitimate laws – only within the system: that is, contesting those
laws but not resisting the regime. The illegitimate and unjust order will not
require acceptance or compliance on the same basis; putting aside fall-back
reasons, it will permit citizens to resist the regime and, using whatever means
are available within, or presumably without, the system, to try to alter the law.
What of the fourth, unlikely case, where there is an illegitimate, yet just order?
That will allow resistance to the regime but offer reasons of both a content-
dependent and content-independent kind for compliance with the law; it will
authorize a mix of challenging the authority of the state to impose the laws
combined with recognizing that the laws imposed are themselves unobjec-
tionable. The possibilities are charted, however roughly, in Table 3.1.
Table 3.1 The demands of justice and legitimacy
Just social order Unjust social order
Legitimate social order Should accept, should comply Should accept, may contest
Illegitimate social order May resist, should comply May resist, may contest
9
As legitimizability offers a reason for treating a state as if it were legitimate, so I have argued elsewhere
that a corresponding form of responsibilizability offers a reason for treating someone as if they were
responsible: that is, for holding them responsible. See Pettit (2001e, 2007d). The responsibilizable
agent will be capable of being made responsible – fit to be held responsible – by being treated as if they
were responsible – that is, by being held responsible. The legitimizable state will be capable of being
made legitimate – fit to be opposed only within the system – by being treated as if it were legitimate –
that is, by being opposed only within the system.
Legitimacy in its heyday
One reason for speaking of political legitimacy and social justice, rather than
of justice in a comprehensive sense, is that it may help to guard against the
danger of not recognizing how distinct these questions are.10 The social
justice question bears, as we know, on the character of the rules that
determine the claims that citizens have, relative to one another, within
the state. The legitimacy question bears on the way in which that set of
rules – and any other associated rules – is imposed on citizens. It concerns
the actual world and the contingent relationship between the people in a
regime and the social order that obtains amongst them; in particular, the
relationship whereby that order is sustained under the regime. By all
accounts the social order will be legitimate insofar as it is sustained in an
appropriate way, illegitimate insofar as it is sustained inappropriately: say, to
invoke paradigm cases, sustained independently of what citizens think, by
the will of a local despot or a foreign power.
Theories of the requirements of legitimacy assumed a remarkably
strong form in seventeenth- and eighteenth-century political thought,
though always against a background assumption that limited the people
or citizenry to mainstream, usually propertied males. The theory of the
divine right of kings had suggested that only the actual, if presumptive,
blessing of God could establish the legitimacy of a sovereign and of the
order that that sovereign chose to impose. Thomas Hobbes (1994b)
directly challenged that theory in arguing that it was the implicit, con-
tinuing consent of subjects – albeit a form of consent that could be
extracted under fear or pressure – that made the rule of a particular
sovereign legitimate; postulating the harsh alternative of a lawless state
of nature, he thought that rationality required individuals to give their
consent to whatever individual or body had the power to keep the peace.
John Locke (1960) argued that two conditions gave legitimacy to a
commonwealth: first, the unanimous, historical consent of individuals,
guided by a desire to establish a fair arbiter of disputes, to exit the state of
nature and set up a state; and two, the support of the existing government
by majority will, as evident in the failure of a majority to be moved to
rebellion by any alleged abuses of the arbiter role. Finally, Jean-Jacques
Rousseau (1997) held that legitimacy required that the majority voice of
the assembled citizens should rule on issues of general law, where once
10
The person who has done most to underline the distinctness of these questions in the recent literature
is A. J. Simmons (1976, 1979, 1999).
again such majoritarian decision-making had been unanimously and
freely accepted, in exiting the state of nature and forming a state.
Thinkers like Hobbes, Locke and Rousseau certainly had ideas about the
social order that they thought government should establish, but their works
were primarily addressed to the issue of legitimacy, which they made vivid
with the theatrics of an imagined state of nature. Each was intent on
identifying those conditions under which they thought people were obliged
to accept a regime and conditions under which they were entitled to rebel:
that is, to reject the state or sovereign whereby the existing order is
sustained. The Hobbesian and even Lockean theories offered very limited
grounds for rebellion, although it is worth recalling that Bishop Bramhall
(1658: 515), his opponent in debates about freedom, thought that Hobbes’s
Leviathan was ‘a rebel’s catechism’. Rousseau’s Social Contract was more
radical in intent, suggesting that many existing states were not involved in a
legitimate exercise but merely in ‘subjugating a multitude’ (Rousseau 1997:
i.5.1). It is no wonder that he attracted the hostile attention of the French
and other authorities, although he was prudent enough not to call explicitly
for rebellion against their regimes.
The eclipse of the legitimacy issue

These earlier figures provide vivid examples of a concern with political
legitimacy rather than social justice, but this concern was eclipsed in the
later development of political thought. This is clearly true of the utilitarian
tradition that derives from Bentham, which dominated political theory
down to the mid-twentieth century in the English-speaking world and
continues to have a commanding presence. Looking for the sort of regime
that maximizes expected happiness, utilitarianism focuses on the issue of
how far the social order treats citizens well and equally in this dimension
and it pays little attention to the issue of whether the order is suitably
supported by those who live under it. Utilitarianism has been effectively
challenged over the past half century or more by a variety of broadly Kantian
approaches. But while these often seem favourable to the legitimacy con-
cern, they all endorse a move that banishes it from consideration.
Kant himself exemplifies the pattern. He begins from the thought that
the will imposed in the enactment of law – and so the will that sustains the
social order created by law – should be ‘the general (united) will of the
people’ (Kant 1996: 295): the will associated with the ‘original contract’
(480). This looks favourable to the Rousseauvian approach from which
Kant begins. But the appearance is misleading, for he insists that the original
contract is ‘only an idea of reason’, not something ‘presupposed as a fact’,
and argues not that the state should actually be grounded in such a will but
that ‘every legislator’ – in effect every state – should establish laws that ‘could
have arisen from the united will of a whole people’ (Kant 1996: 296). That
the laws could have arisen in that way is ‘the touchstone of any public law’s
conformity to right’, he says; ‘a rational principle for appraising any public
rightful constitution’ (Kant 1996: 297, 301).
With this shift, the issue of the legitimacy of a social order, on which
Rousseau was primarily focused, is put aside in favour of the issue of the
nature or quality of that social order: its comprehensive, social-cum-
procedural justice. The fact that the laws could have been chosen in an
original contract – their contractual eligibility – is now a sign of their having
a certain intrinsic character, not an indication of their being sustained in a
legitimate manner. It may not be surprising, then, that Kant shows scant
regard for questions of legitimacy. While he favours a representative regime
in which there is a separation of powers (Kant 1996: 324, 481) – a republic, as
he calls it – he holds that even if a government should ‘proceed quite
violently (tyrannically), a subject is still not permitted any resistance by
way of counteracting force’ (298). No established state can be illegitimate in
our sense, then, since this ‘prohibition’ against ‘resistance to the supreme
legislative power’ is ‘unconditional’: it applies even under the most appal-
ling regimes.
No contemporary theories follow Kant in prohibiting resistance to any
established regimes, but many take his lead in making some version of
contractual eligibility the criterion of whether a social order ought to be
accepted, thereby downplaying the distinctness of the legitimacy concern.
John Rawls (1971) does this in arguing that his two principles would have
been chosen in an original position of contract where people operate under a
veil of ignorance as to their own particular prospects in this or that social
order. And he remains faithful to that same, broadly Kantian, approach
when he later replaces the test of contractual eligibility under a veil of
ignorance with a looser, associated test of civic justifiability: the justifiability
of a legal order, on the basis of reasons that ought to be publicly acceptable
to the citizens who live under it (Rawls 1993, 2001).11 A concern with
11
The approach that he exemplifies in this later turn is shared loosely among a large range of
contemporary political and moral thinkers, including Charles Beitz (1989), Thomas Pogge (1990),
Thomas Nagel (1991), Brian Barry (1995), T. M. Scanlon (1998) and Rainer Forst (2002). For a radical
critique see Simmons (1999) and for a critique more closely related to the Rawlsian starting point see
Reidy (2007). Anna Stilz (2009) represents a rather different approach that also starts with Kant – in
particular, with his observation that the right to external freedom presupposes membership in a
justifiability in this sense is a concern that that order should have a certain
intrinsic character – in particular that it should be comprehensively just –
and is distinct from the concern for whether it is appropriately supported
that we find in Hobbes, Locke and Rousseau (Simmons 1979, 1999).
The elision of the problem of legitimacy under the contractualist way of
thinking may be due to a confusion that is easily made. Legitimacy is the
ideal, under a natural formulation, of having a social order that is imposed
only insofar as it satisfies terms that people actually endorse. Under this
ideal people effectively dictate the terms on which a social order will be
accepted and, on any interpretation of that claim, they can be said to
support the order imposed. Justice, in the new contractualist formulation,
is easily confounded with this. It is the ideal of having a social order that is
imposed only insofar as it satisfies terms that people could have rationally
endorsed, even if they didn’t: in Rawls’s (1995: 148) way of thinking, ‘terms
that all reasonable parties may reasonably be expected to endorse’. The
difference in formulation is subtle but it may mark a deep divide. Even an
order that is not suitably supported by its people, as they are actually
disposed, might satisfy terms that they would endorse if they were properly
rational or reasonable.
As I mentioned, contractualist theory focuses not only on social justice
but on comprehensive or social-cum-procedural justice, so that Rawls (1971:
2003) includes procedural or political rights amongst the basic liberties
required by justice. Does the wider focus of contractualist and similar
theories mean that the criticism offered here is unfair? I do not think it
does mean this, but postpone an account of why not until the end of this
chapter. At that point we will be in a better position to see both the strength
and the weakness of the approach that contractualism typifies.12
I have been arguing that perhaps the two major strands of political
thinking, respectively utilitarian and contractualist, tend to neglect the
issue of legitimacy, ignoring the importance of how people actually relate
political society: this, because there can be no property, for example, without a coercive state (Kant
1996: 409). She argues that given our joint engagement in creating our local political society, it will
have a distinctive claim on us – in our terms, a claim of legitimacy – and we will have a correlative
obligation of loyalty to it; or at least this will be so insofar as the society does actually provide for our
enjoyment of external freedom.
12
If the procedural rights that count for Rawls among the basic liberties were required to ensure people’s
control over the social order, then on the view to be defended here, Rawlsian justice – that is, his two
principles – would require legitimacy in our sense. But as a matter of fact Rawls (1971: 233) himself
downplays the political liberties, as we shall see, casting them as ‘subordinate to the other freedoms’,
and does not suggest that they ought to ensure control in that sense. His focus on justice is mainly a
focus on social justice, as we have been describing it, and it does not bring issues of legitimacy into
prominence.
to the regime under which they live. But the oversight is not confined to
these two approaches. Egalitarians of various stripes also tend to ignore the
question, concentrating on the extent to which a social and perhaps proce-
dural order displays the particular sort of equality they favour, be that
equality in resources, utility, capability, or whatever. Other very different
writers like David Gauthier (1986) and Robert Nozick (1974) also overlook
legitimacy in favour of something else. Gauthier looks for an order that self-
interested subjects would have rationally agreed on, starting from a bargain-
ing standpoint that it would have been rational to accept. And Nozick looks
in a legal order for the satisfaction of a parallel counterfactual: if people were
situated in a morally acceptable world – as he thinks of it, a Lockean state of
nature – then they could plausibly have generated that order as the bypro-
duct of morally acceptable, self-interested adjustments.
But there are some contemporary figures who may be keyed, however
implicitly, to the legitimacy problem. They generally start with democratic
decision-making, however understood, and argue for the importance of
having a social order that is sustained by suitable democratic mechanisms.
This group will include deliberative democrats who insist on the impor-
tance of having an order that is actually sustained by processes of democratic
deliberation; it will not include those, however, who insist merely on the
importance of having an order that could have been sustained in that way.13
More generally, the group will include those who see importance as such in
the fact that a social order is grounded in a pattern of democratic decision-
making, whether canonically deliberative or not.14 The republican theory of
legitimacy, as we shall see in later chapters, connects with these democratic
traditions of thought.
Legitimacy in other senses

In concluding this discussion of legitimacy we should note that my use of
the term should not be confused with three other uses that figure in the
literature of political theory and political science. One is the common
13
Juergen Habermas (1995), Amy Gutmann and Dennis Thompson (1996), Joshua Cohen (2009) and,
in a different key, John Dryzek (2003) are salient figures in this category. David Estlund (2007: 87–93)
argues that deliberative democrats often slip into endorsing the hypothetical test mentioned in the
text and thereby move closer to Rawls. Thus Habermas (1995: 458) appears to go hypothetical in
formulating his discourse principle of legitimacy: ‘The only regulations and ways of acting that can
claim legitimacy are those to which all who are possibly affected could assent as participants in rational
discourse.’
14
This group includes Jeremy Waldron (1999b), David Estlund (2007), Thomas Christiano (2008) and
Amartya Sen (2009).
sociological use, deriving from Max Weber (1947), in which to say that a
social order is legitimate is just to say that it is widely accepted amongst
the people: accepted, perhaps, as a result of being believed in our sense to be
legitimate. According to Charles Taylor (1985b: Essay 10) this is the sense
of legitimacy in play when people speak of the legitimacy crisis of contem-
porary states.
A second sense of legitimacy that figures in common usage refers us to
international, as distinct from domestic, legitimacy (Buchanan 2004). A
regime might not be legitimate in the way in which it relates to its citizens
but might still count as a legitimate member of the international commun-
ity: it may be a state that satisfies human rights in such a measure, for
example, that it is not exposed to certain sorts of complaints in international
forums (Beitz 2009). Such a state might not be democratic or liberal, in any
standard sense, but in a phrase used in this context by John Rawls (1999) it
might count as a decent regime that merits full respect and incorporation on
the international scene.
There is also a third sense of ‘legitimacy’ now often invoked within
contractualist circles, which equates legitimacy with civic justifiability
(Simmons 1999: 756–7). Rawls (1993: 224) illustrates this when he says
that ‘the principle of political legitimacy’ requires that the ‘basic structure
and its public policies are to be justifiable to all citizens’.15 By this account,
the legitimacy of a social order does not require a contingent relationship
between the subjects and the order imposed that might enable us to say,
for example, that they support the order. It depends only on the fact that
the order imposed is such that it could be justified to them. In our terms it is
not a conception of legitimacy but a reformulation of the standard con-
tractualist conception of justice.
2. legitimacy as popular control
A question of freedom
What might make it the case that the state that imposes a social order on its
citizens does so legitimately? What might ensure that while citizens may not
all approve of what the state actually does in establishing laws, imposing
15
Despite his opposition to contractualism, Bernard Williams (2005: 4) adopts a usage that is very close
to this when he says that the ‘basic legitimation demand’ requires the state to ‘offer a justification of its
power to each subject’. Since the justification need not be accepted by each subject, this conception of
legitimacy is also distinct from that invoked here.
Legitimacy as popular control 147
policies or levying taxes, still they do not have good grounds for complaint
about its presumption in taking those actions? They remain obliged to
accept the state as the relevant arbiter and decider of legal issues, restricting
themselves to opposing laws only within the system set up by the state.
The primary reason that the state raises a question of legitimacy of this
kind is that in pursuing its distinctive tasks, it assumes and exercises a
presumptively unchallenged right to coercion: it brooks no opposition.
People are not given an individual choice on whether or not they will
have to follow the laws imposed, fall in line with the policies pursued, or
pay the taxes that are levied on them. They are required by the state to do so,
on pain of enduring the rigours of punishment. The problem of legitimacy
is how to reconcile such political submission with personal freedom, identi-
fying a sort of regime that can coerce citizens without depriving them of
their freedom. The question, in Rousseau’s (1997: iv.2.7) words, is how ‘a
man can be both free and forced to conform to wills which are not his own’.
Or in a contemporary variant, ‘How might a person be self-governing
through institutional enactment of a law to which he is opposed?’
(Michelman 1999: 23).
On this freedom-centred interpretation of legitimacy, the question in
republican terms is whether a state can impose coercively on citizens with-
out dominating them. We saw that in order to promote social justice the
state must treat citizens well and equally in ordering their relations with one
another; in republican terms it has to provide for their equal undominated
status in those relations. What we now have to see is whether in a corre-
sponding sense it can treat citizens well and equally in ordering their
relations with the coercive state itself; whether it can provide for their
equal undominated status in those relations.
It is worth noticing that, phrased in this way, the problem of legitimacy
can be detached from ideas of the state of nature that provided a stage-
setting in the heyday of the problem. Once it is granted that freedom can be
compromised, not just by the social order imposed by the state, but also by
the mode in which the state imposes it, the problem becomes inescapable.
And the problem remains inescapable even if we reject the idea that in order
to solve it we have to show how it could have been rational for people to
opt for political organization from within an imagined scenario of a pre-
political, natural kind. Arguably, the legitimacy problem was at the centre of
concern from well before the advent of state-of-nature thinking; it assumed
that place in the Roman origins of republican thought and remained at the
centre throughout the period of Renaissance and modern republicanism.
Roman republicanism was grounded in the claim that people’s freedom
would be compromised under any form of monarchy and that it was only
citizenship in a suitable republic that could ensure freedom. This same
thesis is present in Machiavelli’s Discourses and in the literature of the
seventeenth- and eighteenth-century tradition, inspiring a search for repub-
lican devices whereby a state might impose on people without compromis-
ing their freedom (see Richardson 2002).
There may be grounds for arguing that despite taking away people’s
freedom, a state can claim the authority that goes with legitimacy on distinct
grounds and that legitimacy does not have to be interpreted in freedom-
centred terms. Perhaps the state is successful in epistemic terms, letting
people reliably identify the requirements of justice without subjecting them
to the rule of others (Estlund 2007).16 Or perhaps it enables them to live
together without civil war, as a pragmatic justification might suggest
(Williams 2005). Perhaps it is the only sort of state that coheres well with
an assertion of the fundamental equality of human beings (Buchanan 2002).
Perhaps, to invoke a broadly Confucian approach, it puts people of virtue
and talent in power; it is distinctively meritocratic (Bell 2010). Or perhaps,
as Ronald Dworkin (2011: 321–2) has recently argued, it manifestly strives to
promote social and procedural justice, ‘recognizing that the fate of each
citizen is of equal importance and that each has a responsibility to create his
own life’.
Such theories of legitimacy will only have plausibility, however, on the
assumption that no effective state can preserve people’s freedom. It would
hardly make sense to invoke an epistemic, pragmatic, egalitarian or meri-
tocratic feature – or the goodwill that Dworkin invokes – in arguing for the
legitimacy of a freedom-denying regime, if there were an alternative regime
available that could claim to preserve people’s freedom. Why should people
accept a freedom-denying state, whatever its other merits, when there is an
alternative available that can play the role of common arbiter and decider
without compromising their freedom? It is only if we have to despair about
assessing legitimacy in terms of freedom that we will be required to explore
alternative approaches. Thus we are returned to the freedom-centred
version of the legitimacy issue. How can the citizens of a state be free and
yet subject to state coercion?
This question, as might be expected, is going to assume a different
form, depending on how freedom is understood. I look in this section
at the way it will present itself, first, under the conception of freedom as
16
For another broadly epistemic approach, although one inspired by a pragmatist perspective see Talisse
(2007).
non-interference, and then under the conception of freedom as non-
domination. I shall argue that while freedom as non-interference cannot
support any plausible conception of legitimacy, republican theory can;
and that under a republican conception, what legitimacy requires is
shared, popular control of the state.
Legitimacy under freedom as non-interference

By the account offered earlier – and indeed by most established accounts –
interfering with a choice may involve intentionally replacing, removing or
misrepresenting any of the options, preferred or unpreferred, by which the
choice is defined. That means that all laws take away from the freedom of
subjects in at least some of their independently available choices. The point
was emphasized, almost with relish, by Bentham (1843: 503) himself. ‘As
against the coercion applicable by individual to individual, no liberty can be
given to one man but in proportion as it is taken from another. All coercive
laws, therefore . . . and in particular all laws creative of liberty, are, as far as
they go, abrogative of liberty.’
This means that under the conception of freedom as non-interference,
there is no way in which the subjects of the law can enjoy full freedom of
choice. They will be subject to state interference, and so to a loss of freedom,
insofar as the state decides what laws to put in place, including laws that
establish the basic liberties; enforces those laws with the threat of penalty;
imposes penalties on those who offend; and taxes its subjects with a view to
maintaining the system.
It is true, of course, that in a well-run society the interference that the
state perpetrates in imposing laws and taxes and penalties may be less, by
whatever measure is taken to be appropriate, than the interference that the
state prevents. But even in that ideal event, it will remain the case that the
state does indeed perpetrate interference and that in doing so it fails to
preserve the freedom of citizens in their dealings with it; they are subject to
its will in a way that takes from their freedom under this conception. Thus
Isaiah Berlin (1969: 3) writes, in elaboration of what he takes to be a lesson
from Bentham and indeed Hobbes: ‘Law is always a fetter, even if it protects
you from being bound in chains that are heavier than those of the law, say
some more repressive law or custom, or arbitrary despotism or chaos.’
There are two reactions that these blunt observations might prompt
amongst those who equate freedom with non-interference. One is to
think that the concern with having a state that does not impinge on the
freedom of subjects is simply misplaced and that we should give it up.
According to this view, the cause of freedom argues for a state that furthers
the cause of non-interference overall – that may be thought to make for the
justice of the social order imposed – but it says nothing in and of itself on
how the state should be supported by the people: nothing, for example, on
whether democratic or non-democratic rule is better on this count. The
cause of freedom will argue for one or the other sort of rule, depending only
on which is best for serving non-interference or some other value overall –
which, for example, achieves the best balance between interference perpe-
trated and interference prevented.
This is the view adopted by Paley (2002: 314) himself, as we saw in the
Introduction. ‘Were it probable that the welfare and accommodation of
the people would be as studiously, and as providently, consulted in the
edicts of a despotic prince, as by the resolutions of a popular assembly,’ he
says, ‘then would an absolute form of government be no less free than the
purest democracy’.17 And essentially the same view is supported by Berlin
(1969: 7), when he suggests that the cause of democracy cannot be grounded
in the cause of freedom as non-interference and that it has to be defended on
some other basis: ‘Freedom in this sense is not, at any rate logically,
connected with democracy or self-government. Self-government may, on
the whole, provide a better guarantee of the preservation of civil liberties
than other regimes, and has been defended as such by libertarians. But there
is no necessary connection between individual liberty and democratic rule’
(see too Berlin 1969: 130–1).18
The other possible reaction to the recognition that the state inevitably
perpetrates interference against its citizens, and thereby offends against their
freedom as non-interference, is not to ignore or dismiss the legitimacy
question in this manner but to argue that there is a special circumstance
in which a state may yet be legitimate. This response draws on the idea of
freedom of contract that Hobbes introduced in the seventeenth century and
that became the catch-cry of classical liberal thought in the early nineteenth
17
It is worth noting that unlike Bentham, Paley (2002: 312) argued that no coercive laws that promoted
the common good should be regarded as reducing the freedom of subjects.
18
As I argue in Pettit (1997c: Chapter 1), a friend of Bentham’s, John Lind, used these ideas to
propagandize on behalf of the British government against the American colonists in the 1770s.
Drawing on Bentham, he argues against the republican Richard Price, that liberty is ‘nothing more
or less than the absence of coercion’ and therefore that law inevitably takes away liberty since ‘all laws
are coercive’. And then, with those claims in the background, he asks why the rule of the British
government in America is problematic, given that it imposes laws at home as well as in the colonies.
The idea is that the Americans have no particular complaint since the law-making power is ‘exercised
by the same persons over all the subjects who reside in all the other parts of this same empire’ (Lind
1776: 16, 24, 114).
century (Atiyah 1979).19 The idea is that if someone consents to an arrange-
ment with another under which they suffer the other’s intervention, then
such an invited form of intervention does not count as interference and does
not take away from freedom. In a legal, Latin tag that was often invoked in
this connection, injuria non fit volenti: ‘no injury, and no interference, in
the presence of consent’ (Hobbes 1994a: 21.3, 1998: 3.7).
This response need not imagine a population-wide agreement to esta-
blish a political regime of the kind invoked in the state-of-nature tradition.
What it says is that the state will not count as interfering with its citizens,
even as it imposes coercively upon them, so long as the interference
practised attracts the consent of each individual subject. The consent
postulated may be prior consent or consent of a continuing character. But
in neither case can the postulate serve the purpose for which it is designed.
The consent you may have given to the state’s coercion in the past is quite
consistent with that coercion’s being unwelcome and frustrating at later
times in the characteristic manner of interference. So there is no hope on
that front. The continuing consent you may be thought to give to the state’s
coercion offers a more promising ground for claiming that the coercion does
not constitute interference. But this possibility also turns out to be a dead
end. The problem is that it is utterly implausible to postulate continuing
consent on the part of citizens generally, however implicit or tacit the
consent may be, to the coercion of their state. People may welcome the
regime they live under but, as David Hume (1994) had already argued in
the eighteenth century, they will not consent to it in the sense of adopting it
voluntarily: that is, adopting it in preference to an otherwise acceptable
alternative (Olsaretti 2004). And it is only consent that is voluntary in that
way – and not just voluntary in the Hobbesian sense in which submission
out of fear counts as voluntary – that can make the coercion of the state
count as something less than interference.
If these observations are sound, then the conception of freedom as non-
interference raises a dilemma for understanding legitimacy in a freedom-
centred way. Either, to take the first response, legitimacy in that sense ceases
to count as an ideal that a state or social order ought to satisfy, which is
problematic. Or, to take the second, it becomes an ideal that no actual state
19
I bracket the role of the monarchomach (or king-killing) tradition of thought as that is represented,
for example, in the late-sixteenth-century tract, Vindiciae Contra Tyrannos (Languet 1994). Adherents
of this tradition, which became an established school of thought in many European countries, argued
against absolutists that any people that is ruled by a monarch must be supposed, as a collectivity, to
have made a contract with their ruler under provisos that would allow them to dismiss that ruler –
literally or figuratively to kill the king – in the event of certain forms of injustice or tyranny.
or social order is capable of satisfying, which is equally problematic. Either
the ideal is irrelevant, in effect, or it is infeasible. Things look very different,
however, from a republican perspective in which freedom is cast as non-
domination.
Legitimacy under freedom as non-domination

Let it be granted that the state inevitably interferes in people’s lives both in
establishing the basic liberties that are to be publicly entrenched, and in
pursuing the taxation, coercion and punishment that is required for such
entrenchment. And let it equally be granted that if the state is going to count
as legitimate then it should be supported by the people in such a way that it
does not take away from their freedom. How could a state that inevitably
interferes in the choices of people manage to be supported in such a way that
its interference does not take away from their freedom?
Under the conception of freedom as non-domination, as we saw in the
first chapter, domination alone is sufficient to reduce the freedom of the
dominated. That means that freedom can be reduced in the presence of
domination, even if there is no interference or frustration. But, under the
conception of freedom as non-domination, domination is necessary as well
as sufficient for a reduction of freedom. And that means that if there is no
domination involved, freedom is not reduced in the presence of either
interference or frustration.
In order to illustrate the possibility of interference without domination we
mentioned the scenario in which you want to reduce your drinking. In order
to promote that end, you give me the key to the alcohol cupboard with
instructions not to heed a request for the key except at twenty-four hours’
notice. When you suffer a rebuff to a request for the key now, then you will
certainly endure interference and frustration; the element of consent, contrary
to the freedom-of-contract idea, does not change the fact that you want the
key now and I am refusing to hand it over. But you will not suffer any loss of
freedom in the presence of such interference, for while I impose a will that is
hostile to your current wishes, this imposition is subject to your control. The
interference you endure, as it is traditionally phrased, is of a non-arbitrary
form; it does not express my will and does not reflect my arbitrium.
The scenario with the alcohol cupboard is a variant on the classical
scenario of Ulysses and the sirens. In that story, his sailors bind Ulysses to
the mast of the ship and keep him bound even as he begs them, under the
seductive lure of the siren voices, to let him go. But they do so under his
instructions and, presumptively, under his ultimate control. And so the
The nature of control 153
interference they practise, like my interference in denying you the key to the
alcohol cupboard, is not dominating. It does not impose an alien will on
Ulysses, being exercised only on terms that he himself lays down.
The idea of controlled interference provides us with the core element for
a republican theory of political legitimacy. It suggests that if the people
governed by a state control the interference practised by government – if
they control the laws imposed, the policies pursued, the taxes levied – then
they may not suffer domination at the hands of their rulers and may
continue to enjoy their freedom in relation to the state. A state that was
suitably controlled would be legitimate in the required sense of not exercis-
ing domination over its people. It would practise interference, for sure –
think about how frustrating laws and taxes can be – but it would only
interfere with them on their terms, not at its own will or pleasure.
We shall be expanding on this core idea in Section 4, when we look in
detail at exactly what sort of popular control over government might
guarantee the legitimacy of the state. We will look there at the domain in
which people have to exercise control over the state. And we will explore the
nature of the control required, arguing that it must be individualized,
unconditioned and efficacious. Just to spell out these requirements, the
system of control must be equally shared amongst citizens; it must not be
conditioned on the willingness of the government, or of any other agent, to
humour the citizens; and it must score well enough in the level of control
provided to ensure that even unwelcome government intervention does not
provide citizens with any evidence of an alien will at work in their lives: it
can come across as just tough luck.
But control may not seem so very different from consent and before
getting to that discussion it will be useful to spend some time on the analysis
of the idea of control and the related idea of influence. We have been
invoking the idea of control at various points in the text and it is important
that we be clear about what exactly is involved. The points that emerge in
the course of this analysis will be of importance in later discussion.
3. the nature of control
The idea of control

To have a degree of control over a result two things are essential. First, you
must have some influence over the process leading to the result. And
second, you must use that influence to impose a relevant direction on the
process, helping to ensure that a suitable result transpires.
The need for influence is obvious from the fact that you could not claim
to have any control over a result that you merely observed as a bystander,
having no capacity to make an input. Influence will not give you control,
however, just on its own. Imagine the effect you will have on the traffic at a
busy intersection if you play police officer and give hand signals in the
usual manner, inviting the cars to ignore the lights. In all likelihood some
cars will take their lead from your signals, others not; and amongst those
that do not, some will try to drive quietly by, others protest with honking
horns or exasperated gestures. You will certainly have an influence in such
a case, making a difference to how the cars behave; you will probably
create utter chaos. But will you have control? Well, if your aim was to
create chaos, you may be said to have exercised control in a certain sense.
But assuming that you wanted the cars to follow your signals, as they
might follow the signals of a police officer, you will be a dismal failure. You
will have made a difference to how the cars behave but not a difference
that imposes any desired direction or pattern – not a difference that serves
any identifiable end or goal.
What will be required in order for your influence to give direction to a
process like the flow of cars in this example? The influence must give rise
to a recognizable pattern in the process and that pattern must be one that
you seek; the influence, as we say, must control for the appearance of a
desired pattern. The influence must not only make a difference, as any
form of influence will do; it must make a designed difference. There will
be a range of ways in which you can vary your input to the process, since
there are different hand-signals you can give. And for each of those inputs
there will be a corresponding output: the traffic will alter in response to
your signals. In the case where you take the police officer’s place at the
intersection this condition will not be fulfilled: there will be a more or less
random correlation between how you move your hands and how the cars
adjust. Were a police officer to be in your place, however, then things
would certainly be different. The officer’s hand-signals would reliably
generate, now this sort of effect, now that; as we say, they would control
for how the traffic moves.
The lesson of these observations is that there is no control without such
control-for: no control without an influence-bearing input that controls for
the realization of a suitably patterned output. The influential input may
control for the appearance of that pattern rather than none at all. Or, if it
can vary, as in the case of the traffic signals, it may control for the
appearance of one pattern rather than any of a range of alternatives for
which it might have controlled; in that sense, it may assume a richer form.
The paradigm case of control is the intentional control in which you
exercise influence as an agent and this leads to the realization of some
desired end. You act according to your beliefs for the satisfaction of your
desires and you control for your actions and their effects in the sense that as
your desires or beliefs change so the outputted actions and effects will
change in response. This is the sort of control that a police officer will
exercise over traffic. And it is the sort of control that you will exercise in the
arrangement about the key to the alcohol cupboard. Your desires and beliefs
will lead you to issue appropriate instructions and those instructions will
lead me to uphold the arrangement you propose.
But while intentional control provides the paradigm case, control need
not be intentional in every case. Purely mechanical, non-intentional mech-
anisms can have such an influence and impose such an identifiable direction
that we happily say they exercise control; the influence exerted by the
mechanism makes a designed difference but without any intentionality on
the part of the mechanism itself. An example might be the cooling-heating
system that keeps the temperature in a room within a certain range, coming
on as the temperature climbs too high or falls too low and exercising an
influence that restores it to the pre-set range. And a parallel example, drawn
from the natural, rather than the artificial, domain, might be the homeo-
static system that keeps the internal temperature of an animal like you or me
within a biologically functional range.
But even the control exercised by intentional agents may be non-
intentional. It may be like the control illustrated in Amartya Sen’s (1983a)
example of the patient in the coma whose wishes, thanks to family or
friends, control for the treatment provided by the surgeons at the hospital.
Given the form taken by the patient’s wishes, the surgeons provide one sort
of treatment. And had the wishes taken a relevantly different form – had
that input varied – then the treatment would have been different too; at the
least it would not have been of the form that the patient’s current wishes
generate. And yet the patient does nothing intentionally with a view to
revealing or implementing those desires; they are effective by virtue of the
efforts of others, not his or her own.
Whenever a system exercises control over a process, then, there will be
systemically shaped inputs that control for the form of corresponding out-
puts. In the richer form of control, there will be influence-bearing inputs
that vary depending on the state of the system and, as they vary, those inputs
will determine, at least probabilistically, the direction taken by correspond-
ing outputs. This structure will apply to all control systems, intentional or
non-intentional, and amongst intentional systems – say, agents like you and
me – to all modes of control, be those modes intentional or not. Hence, to
take the case that interests us, it will apply to any system of control that
citizens might exercise over the state. We will be examining that popular
control in the next section and we will be looking at the institutional ways in
which it might be realized in the two chapters remaining.
The varieties of influence

It may seem that the influence exercised in control has to be the active sort
that involves a positive input on the part of the controller. But it should be
noticed that there are two other forms of influence possible, which I shall
describe as virtual and reserve influence. And each of these can support
control just as effectively as the active variety.
Think of how I may control a horse that I ride. I may actively pull on the
reins, now steering the horse in this direction, now in that. Or I may let the
horse follow its head, given that it is moving in the direction I want to take.
Or I may let the horse have free rein, given that I am happy for it to go
wherever it wishes. The first is an example of active influence, the second a
case of virtual influence, and the third an instance of reserve influence. In
active influence I intervene in order to get the horse to move as I wish. In
virtual influence I am poised to intervene, but only if an intervention is
needed to keep the horse on track. In reserve influence I am equally poised
to intervene, but only if my wishes change and an intervention is needed to
satisfy them.
We have already seen examples of active, virtual and reserve forms of
influence in our discussion of freedom of choice in the first chapter. If I
frustrate you by obstructing a preferred option then I actively intervene in a
choice. If I would have interfered with an option you might have chosen but
do not interfere with the option you actually choose, as in interference
without frustration, then I intervene in a virtual manner. And if I am not
currently disposed to interfere but would do so if I changed my preference
over how you should choose, as in domination without interference –
invigilation – then I intervene in a reserve way.
Active, virtual and reserve forms of influence all serve to put a factor in
place that raises the probability of a certain result: a certain patterned or
designed result, as it will be if the influence mediates control. That factor
consists in an actual event in the case of active influence and in a suitable
disposition in the case of virtual and reserve influence. The reason for
speaking of influence in the three cases is that the probabilification is
effect-based in each of them, not merely evidence-based. It is not like the
evidential probabilification involved when we say, for example, that your
consuming an unusually large quantity of grapes – the typical hospital
visitor’s gift – makes it more probable that you are ill.
Thus, to return to the earlier illustration, I ensure that the horse I ride
goes where I will, whether or not the influence I exercise is of the active,
virtual or reserve variety. When I steer the horse, my action ensures that it
goes where I will. When I give the horse its head in the manner of virtual
influence, my disposition to pull on the rein should it change direction
ensures that it follows my will. And when I give it wholly free rein,
allowing it to go where it itself happens to want, my disposition to pull
on the rein should my own wishes change in that regard ensures also that it
goes where I will.
Patterns of virtual and reserve influence are as important as patterns of
active influence, not just in toy examples like this, but also in the world that
is tracked in natural, and particularly social, science. No inventory of the
connections it is important to recognize for purposes of prediction and
intervention could afford to overlook them (Pettit 2007c). They are parti-
cularly worthy of notice here since, as we shall see, there is every reason to
think that the modes of influence whereby people might exercise control
over government include virtual and reserve influence, as well as active.20
Control and consent

Control in the sense explicated here is quite different from consent, for there
may be consent without control and control without consent. On the one
side, you may consent to a form of interference that you do not control:
once consent is given, you may have no means of calling off the arrangement
to which you consented; at the limit, you may have consented to a slave
contract – the bête noire of traditional republicanism – that binds you
indefinitely. On the other side, you may control a form of interference to
which you never gave your consent. Suppose you were born into a society
where by common convention parents propose marriage partners for their
children. While you go along with the system without ever having con-
sented to it you may still exercise a good deal of control over it. You may be
in control to the extent that you can opt out of the arrangement altogether
20
Why not describe the three sorts of influence as causal, virtual and reserve? Two reasons. One is that
virtual influence may be causal in all relevant senses, as when I now put an obstacle in the way of your
choosing Y and you are independently disposed to choose X. And a second is that, the first possibility
aside, virtual and reserve influence might be taken by some as examples of negative causation: cases
where the absence of a certain factor can be cast as a cause.
or, short of opting out, can continue to turn down proposals until your
parents propose an acceptable partner.
The divergence between consent and control remains in place even
when, breaking with the standard notion of freedom of contract, we
stipulate that consensual interference means interference to which you
currently consent, not just interference to which you consented in the
past. True, you cannot give your consent proper to a form of interference
you do not control; at best you can welcome it, while being unable to call it
off.21 But while there cannot be current consent without control, you may
control a form of interference to which you do not give your current
consent. Amartya Sen’s (1983a) example of the comatose patient whose
wishes are given effect by family or friends illustrates the possibility. Their
intervention empowers the patient’s recorded wishes, forcing the surgeons
at the hospital to conform to them. And it does this without the current
consent of the controlling agent.
These observations should help to explain why thinkers in the republican
tradition do not give much weight to consent, or at least to a form of
consent that leaves you under a will that you do not control.22 Such consent
can play no role in guarding against arbitrary or uncontrolled interference.
It ensures freedom of entry but it is quite consistent with the absence of a
power of exit or of any other way of checking others: say, the power
associated with having a voice. Perhaps it was for that reason that John
Milton (1953–82: ii, 137–58) argued so fervently that marriage could be free
only if divorce was possible: that is, only if there was a possibility of release
from the marriage bond – and from subjection to the rights and powers of a
spouse – in the event of estrangement between the parties.
Those who cast freedom as non-interference do not treat consensual
intervention as interference and so they would regard a consensual govern-
ment, however unlikely the possibility, as a legitimate, freedom-compatible
agency. This approach is implausible, as noticed earlier, since your past
consent does not mean that my present intervention lacks the frustrating
potential that is the hallmark of interference. But could those who cast
21
John Rawls (1993: 222) appears to envisage this form of current consent when, having dismissed the
possibility of past consent, he writes: ‘we may over the course of life come freely to accept, as the
outcome of reflective thought and reasoned judgement, the ideals, principles, and standards that
specify our basic rights and liberties, and effectively guide and moderate the political power to which
we are subject. This is the outer limit of our freedom.’ For a defence of the view that such acceptance
can count as voluntary, see Otsuka (2003: Chapter 5).
22
If you bind yourself only conditionally, retaining the ability to call off the arrangement in the event of
the other person breaching some conditions, you will retain a degree of control and may not be
subject in a relevant sense to the domination of the other.
freedom as non-interference make use of the idea of controlled interference,
joining republicans in the claim that a suitably controlled but in no way
consensual government – this is the possibility explored in the next section –
might be a legitimate, freedom-compatible agency?
Might they argue that controlled intervention is not really interference,
to take a first possibility, as they seek to argue that consensual intervention is
not really interference? No. Controlled but non-consensual intervention
must count as a form of interference, since it may also be characteristically
frustrating. Moreover, unlike consensual intervention, there is no sense in
which it will count as invited or condoned. Thus, even if the intervention of
the state in people’s lives takes place under their control, there is very little
plausibility in saying that it does not really count as interference. It may be
extremely frustrating and in general there is no sense in which citizens need
invite or condone it.23
Might those who cast freedom as non-interference accept this but still
argue, to take a second possibility, that controlled interference does not take
away freedom in the same way as uncontrolled? Again, no. Under the
conception of freedom as non-interference, you are going to have freedom
of choice in relation to me just insofar as I do not interfere with you: I do not
intentionally remove, replace or misrepresent any option. You will suffer a
loss of freedom as a result of any such interference on my part and you will
suffer a loss of freedom only as a result of such interference. On this
approach, then, the issue of whether I have a power of uncontrolled
interference in your choice is irrelevant to whether I take away from your
freedom. And equally, the issue of whether the state has a power of
uncontrolled interference in the choices of citizens is irrelevant to whether
it takes away from their freedom. Freedom just is non-interference, on this
approach, and the issue of whether the interference is controlled or not is
simply beside the point.
There may be a powerful intuition, as I think there is, that the interfer-
ence you control – the interference exemplified in the alcohol example –
does not take away from your freedom of choice. But in order to give
countenance to that intuition, members of the opposed school of thought
would have to give up on their central commitment to the idea that
interference and interference alone reduces freedom. They would have to
endorse the republican idea that it is domination that really matters in the
23
The only plausible case in which we might want to deny the name of interference to a form of
intrusive intervention is where the intervention is both consensual and controlled. It genuinely
attracts the voluntary, continuing consent of the subject of the intervention.
freedom stakes. Thus the idea that a suitably controlled government might
be a freedom-consistent agency is simply unavailable under the identifica-
tion of freedom with non-interference. It is exclusive to the republican way
of thinking about freedom.
4. the requirements on popular control

We have argued for the very abstract conclusion that if the citizenry control
state discretion in a suitable manner – in a way that parallels your control
over the person who holds the key to the alcohol cupboard – then the
imposition of a social order on those citizens will not take away from their
freedom and will count as fully legitimate. But what is the nature of the
control required? And what is the domain in which legitimacy requires
citizens to exercise such control over the state? We will consider those
questions in reverse order, looking first at requirements on the domain
where citizens must control their government and then at requirements on
the nature of the control that they must exercise in that domain. It turns
out, according to this argument, that the domain requirements are less
demanding and the nature requirements more demanding than might have
been expected.
Requirements on the domain of popular control

The idea of having control in relation to the state – having an uncondi-
tioned form of directed influence – may seem preposterous. For the state,
you may well say, does not give you or anyone else control over whether you
are to live in political society rather than in a state of nature, or in one state
rather than another. Moreover, if you live in political society, it does not
give you or any other citizen, no matter how virtuous and compliant,
control over whether the laws are to be coercively enforced against you;
uniform coercion is the rule of the political game. So surely the republican
proposal, regardless of how it is to be elaborated, is completely hopeless.
It is true that neither you nor anyone else has a choice over whether to live
under a state or in a natural condition; that neither you nor anyone else has
an assured choice over whether to live in one state or another; and that if you
live in political society you do not have any choice over whether to comply
with the law on a non-coercive basis: that is, without being exposed to
penalty. But these constraints on your life raise an issue of legitimacy for the
state only if they are imposed by the state itself and represent subjection to
an alien will. And it turns out that they are not imposed by your state – or
The requirements on popular control 161
indeed by any other state – in the manner of a decision to interfere rather
than not interfere; they materialize on an independent, unwilled basis. This
means that the domain in which republican legitimacy requires civic or
popular control does not include the imposition of such constraints.
It is an unintended precipitate of human history, and in particular of the
mutual adjustments of different populations, that the earth is now a state-
bound planet: a place where there is no inhabitable area that is free of the
rule of some state. The fact that you do not have a choice between living
inside a state and living outside a state is not the product of interference or
domination on the part of your state. It is a historical necessity on a par with
the necessity of living under the laws of physics.24 You may think that your
own state, or indeed some other state, could renounce its standing as a state
and establish a stateless zone. But states are locked into a pattern where none
can resign without exposing its population to the will of rivals, creating a
vacuum where other states will battle for control. You may bemoan the
necessity of living in a state-bound world, then, even long for a period in
history or pre-history when it did not obtain. But you cannot think that
because you are constrained to live in a state, you are dominated by the state
under which you live, or indeed by any other state.
As historical necessity means that you have to live in one or another state,
so political necessity means that in general you have no choice over whether
to live in your current state or in some other. Assume that your state allows
you a right of emigration and does not confine you within its boundaries; if
it did, it would certainly dominate you.25 It is still going to be the case that
other states cannot guarantee you entry, given the political necessity for
states to maintain their borders and disallow open access. The fact that you
have no choice over whether to live under another state is not going to be a
product of domination by your own state, then, only a result of how other
states behave.
Does this mean that your lack of choice is due to domination on the part
of those other states? Not necessarily. No state can open its borders to non-
residents in general, on pain of internal malfunction or collapse; as a matter
of political necessity, every state has to place limits on who can enter and in
what numbers. This means that states have to institute selective policies
about who to admit. Let us suppose, then, that a state adopts a policy that is
24
Nor of course can you escape the deeper necessity of living in community with other people. For an
extended argument that this should be regarded as a matter of more than causal necessity see Pettit
(1993).
25
I ignore the question of whether this would hold even if the very survival of the state depended on
keeping you, and others like you, within its boundaries.
not discriminatory and dominating from the point of view of anyone. The
policy will mean that it does not have a choice, as such, between admitting
you and not admitting you, depending on its will. And if under that policy it
refuses you admission, then this should not count as fully voluntary
interference. It will be a by-product of an independently necessitated,
otherwise unobjectionable policy: a restriction on your freedom of move-
ment, to be sure, but one that vitiates that freedom rather than invades it.26
These observations indicate that you need not think of yourself as
unfree – unfree in the sense of being dominated by an alien will – just
because you have to live under a state and perhaps in this state rather than
any other. That you have to live under these constraints is determined by
historical and political necessity. Like having to live in the presence of
gravity, it is a product of the existing world order, not the result of a
dominating presence in your life. The existing world order may be highly
undesirable, of course, and in principle there might be ways of reorganizing
it – for example, ways of incorporating states into a single agent like a world
state or confederation – in which there is less domination overall. But given
the world order we have, it is not due to the dominating power of any state,
and certainly not of your own state, that you have to live in some political
society or other. And it need not be due to the dominating power of any
state, and certainly not your own state, that you have to live in this political
society in particular. These circumstances may be exasperating but they do
not testify to the presence of a dominating will in your life.
An analogy may help to illuminate the role of historical and political
necessities under these arguments. Suppose that you live in a world, how-
ever fanciful, where as a result of past adjustments the space of economic
activity is entirely taken up by corporations. You could not complain of
being dominated in such a world by the fact that you have to work in an
incorporated company, not for yourself or for a private employer, though
you might bemoan the absence of such possibilities. And you could not
complain of suffering interference and domination just because you cannot
change place of employment to another company; no company can afford
to take on all those employees who wish to work for it. There might be
reasons for the government to try to change employment practices in certain
ways, and reforms might hold out the prospect of more freedom as non-
domination overall. But it remains the case that in the world imagined the
restrictions associated with having to work in some company, and not
necessarily in the company of your first choice, do not in themselves
26
There are difficult issues here that I cannot explore fully. See Abizadeh (2008).
constitute domination. The analogy is imperfect but may help to explain
why you and your fellow citizens are not dominated by the historical and
political necessities to which you are subject.
Back to our main theme. As historical and political necessities mean
that the state is not in a position to allow you to live in a natural, rather
than a political, condition, or in any other state you might prefer, so a
functional necessity means that it is not in a position to grant you the
privilege, no matter how virtuous you may be, of obeying the law on a
non-coercive basis: that is, without fear of penalty. This is a third area
where state restriction is independently necessitated and does not entail
state domination.
Assume, in line with common wisdom, that in every society it is neces-
sary to employ coercion against at least some potentially recalcitrant mem-
bers. Coercion may be necessary to get those individuals to comply – lesser
obstacles of shame or whatever might not work – and/or it may be necessary
to provide assurance for the community that they will indeed comply – they
will not free-ride on the efforts of others. But if the state imposes on some in
this way, then assuming as we do that it has to satisfy the egalitarian
constraint, it has no choice but to impose on all in similar manner; doing
so will be a functional necessity. If the state did not impose equally on all
then, by any yardstick, it would offend against that constraint, not treating
citizens as equals. And if you or anyone else looked for the privilege of not
being exposed to the coercive imposition of the law – if you argued for being
allowed to display your virtue in voluntarily obeying the law – then you
would also be in breach of that constraint: you would show yourself unwill-
ing to live on equal terms with others. The normative constraint of expres-
sive egalitarianism is fundamental to the approach adopted here, and to any
half-plausible alternative. Accept that constraint and you cannot think that
it is within the state’s discretion to give you the privilege of not being
coerced to obey the law.27
The three constraints reviewed mean that you cannot regard your state as
dominating you just because you have to live in political society, you have to
live in this state rather than any other, and you have to live under coercive
regulation. It is true, of course, that if your state dissolved itself, then it
would not impose such constraints. But its continuing to exist is neces-
sitated by its role in protecting against foreign domination and does not
represent a power of imposing on you or not imposing on you, depending
27
To put the point in Kant’s (1996: 393) words, each of you has a claim only to a freedom in relation to
the state that ‘can coexist with the freedom of every other in accordance with a universal law’.
on its will. Thus you scarcely have a reasonable complaint against your
actual state. On the contrary, as we suggested, the fact that other states are
poised to assume control of the territory in which you live, and therefore
represent possible sources of domination, means that you would have a
reasonable complaint if your state did dissolve itself. Let the state remain in
place and you can avoid foreign domination: that is, a form of rule which,
by definition – that is, in virtue of what it means to be foreign – is going to
be uncontrolled by you and your fellow citizens, as you actually are. Let it
dissolve itself and foreign domination is unavoidable.28
The background picture to the argument in this section is a bleak but
realistic image of the world of states. Charles Tilly (1975: 42) claims that in
Europe, ‘War made the state and the state made war.’ The idea, which
applies elsewhere as well (Fukuyama 2011: 110–11), is that in a world of
potential conflict, different communities are likely to find political organ-
ization attractive and that the more organization there is in other commu-
nities, the more attractive it becomes in each. States call one another into
existence, in other words, as by an iron law of community competition.
Short of a world-changing, game-changing challenge, I suspect that this law
will continue to dictate the need for political organization, making the state
into a domestically indispensable and internationally resilient institution.
Not only is it necessary for the promotion of any plausible conception of
justice, as we argued earlier; it is necessary on empirical grounds as well.
The world in which states protect their citizens from the threat of other
states may be considerably less attractive than a world in which there are no
states or – more plausibly, in view of our argument that the state is required
for justice – a world in which there is only one state or federation (but see
Pinker [2011]). But in any case, this world represents an equilibrium
scenario where a state will not do better by its citizens, and will almost
certainly do worse, by unilateral dissolution. The state-bound world might
be represented, in a pessimistic analogy, on the model of a racing circuit
where the cars are spread out all around the track and each is travelling at a
highly dangerous speed.29 It would be better if all the cars travelled at a more
moderate pace, as it might be better if all existing states dissolved them-
selves. But each would do worse for itself by slowing down, as each state
would do worse for its citizens by dissolving itself, since that would almost
certainly cause a major pile-up.
28
I am grateful to Niko Kolodny for pressing me on this issue.
29
This is developed from an analogy presented for other purposes in Jackson (1987).
The lesson of this discussion is that while the legitimate state is
required under republican theory to be controlled by its citizens, the
domain of control cannot extend to the point where people are able to
decide whether to live in or out of political society, whether to live in this
state or another, and whether to live under coercive or voluntary fidelity
to the law. No state – or at least no expressively egalitarian state – has a
choice between giving or refusing its citizens control on such issues, and
what legitimacy requires is that it give them control only on the range of
issues where it itself has such discretion. This is the range of decision-
making in which historical, political and functional necessities allow it
to choose between interfering and not interfering and, where it does
actually interfere, to choose between interfering in this or in that manner,
to this or to that effect. Let citizens have control over this restricted
domain and they will have all the control required for avoiding state
domination.
On the account of domination offered in the first chapter, there are
two forms of intervention in your choices that are not going to be
dominating; however, they may cramp your decision-making. One is
the intervention of constraints that do not reflect the will of another
towards you and the other is the intervention of intentional agents who
act under your control. The lesson of this discussion has two parts,
corresponding to these two forms of undominating intervention. The
first is that being required to live with others in a particular, coercively
regulated state is the product of will-independent constraint and not a
dominating imposition. And the second is that the threat of domination
arises only in the area of discretion that the state enjoys – that is, in
choosing to interfere with its citizens on one pattern or another – and that
it is here that you and your fellows have to be able to exercise suitable
control if you are to escape domination.
In the example with the alcohol cupboard that I used to introduce the
idea of a controlled and undominating state, there are two ways in which
you control me. First, you can suspend the arrangement under which I am
required to return the key at your request although only on twenty-four
hours’ notice. And second, you can shape and reshape that arrangement as
you will. We can see now that there is only a partial analogy between this
case and the case with the state. For in the case of the state, the arrangement
under which you and your fellows are subject to state coercion is not of your
devising and is not subject to your suspension; it is imposed under will-
independent necessity. You can only have control in the matter of how the
state operates under that arrangement.
But the disanalogy between the cases is not a problem. For your free-
dom as non-domination does not require you and your fellows to be able
to suspend the arrangement under which you live on a state-bound planet,
you are confined to living under the state into which you are born, and
you are required to live under a coercively imposed regime of law. Those
constraints may be mediated in the actions of states, showing up in the fact
that states claim a monopoly on territory and don’t allow outsiders to
enter at will and impose their laws coercively on all their subjects. But they
derive from necessities that leave states, and your state in particular, no
choice in such matters. Being required by independent necessity to
impose the relevant constraints, the state does not impose them out of a
will to restrict you and your fellows; necessity sidelines any will or
preference it might have in the matter.30
Requirements on the nature of popular control

The upshot of the previous discussion is that the domain in which you and
your fellow citizens have to be able to exercise control over the state is much
smaller than might have been expected; it is much smaller, for example,
than the domain that a state-of-nature staging of the legitimacy problem
would suggest. You will want to be able to exercise control with your fellow
citizens over the state that directs your lives only in the range of decision-
making where the state has a capacity to take one or another approach and
the power, therefore, to form and act on a will or preference as to how
precisely you should be restricted.
So what might it be for you and other citizens to enjoy a suitable form of
control in this domain of state discretion? What sort of control do you and
they have to enjoy over the making, administering and adjudicating of
law – and over related decisions – in order not to be dominated by the state
or by any of the government agencies or officials who act in the name of the
state? What sort of popular control do you and they have to enjoy in order
for the state to count as legitimate?
30
We saw earlier that an approach centred on the conception of freedom as non-interference cannot
take the existence of a system of control to ensure that state interference does not count as
interference. But it is worth noting, as implied in our discussion, that the approach does not have
to treat the actions of the state that are necessitated in the manner described in this section –
necessitated historically, politically or functionally – as instances of interference. The actions may
restrict people’s choices, but since they are not actions in which the state has any choice, they cannot
be taken as acts of voluntary restriction – acts that are chosen over acts that would count as acceptable
alternatives (Olsaretti 2004).
Any system of control, as we have seen, is going to be a system of directed
influence: that is, influence exercised to a certain purpose or direction. This
system of control will have to guard against the domination of individual
citizens, not just the domination of the collective citizenry. But popular
control of government will not ensure someone against public domination,
if they have little or no part in the exercise of that control: the interference of
government will continue in that case to be imposed on them by an alien
will. Thus the first requirement on popular control is that it should be
individualized appropriately, giving each an equal share in the control of
government.
But there are two other requirements that popular control of government
must also satisfy if it is to guard against the domination of the state. Not
only must it be an individualized form of control, it must also count as
unconditioned and efficacious in character. It must be unconditioned in the
sense of being robust over changes in the will of the controlled government,
or indeed of any party other than the controlling people, and it must be
efficacious in the sense of being intuitively sufficient to guard those people
against having to see the coercion of the state as the work of an alien will.
The individualized system of control must ensure a suitable resilience and
level of impact to guard people against state domination.
If a system of popular control meets these three requirements – I discuss
them in greater detail shortly – then it ought to satisfy citizens that they are
not dominated by the imposition of certain laws or by the way those laws are
implemented by the executive or applied by the judiciary. It is true that
citizens will not be able to think of themselves as exercising personal control
over the state. But that they cannot exercise personal control is a by-product
of the historical necessity of living in political society. And that they cannot
demand a form of personal, as distinct from equally shared, control is an
implication of the normative necessity of living on equal terms with their
fellow citizens. No complaint about having to share control equally with
others will be consistent with a claim to be willing to live on equal terms
with their fellow citizens.
The requirement of sharing control equally with others might be
described as a one-for-all constraint, since it means that citizens each have
to accept that they do not have unilateral control over government. But the
constraint is balanced by what we might call an all-for-one compensation.
Under a system of equally shared control, the controlling collectivity is
bound to be much more effective than any single actor could hope to be.
And that agency may be expected to achieve for each whatever it achieves for
any; otherwise the equality constraint would be in jeopardy.
An individualized system of popular control
A system of control will be individualized insofar as it gives a comparable
role to each of the individuals involved in the exercise of control. Strictly,
the system might take either of two forms. It might enable people each to
exercise control in their own right, where their personal level of control is
equal. Or it might enable them to exercise a form of joint control, where
their share in that joint control is equal. But I see no plausibility in the first
scenario. The salient way to realize it would be by giving everyone an
individual veto over the doings of government – by requiring continuing,
universal consent – or perhaps by giving each an equal right of exit (Warren
2011). But establishing a general right of veto or exit would be inconsistent
with the state’s continued existence as a corporate agent that can reliably
generate and implement law, since it would put it at the mercy of individual
whim.31 Hence the only way in which the state can operate effectively and
yet satisfy the demands of republican legitimacy is by giving each of its
citizens an equal share in a system of joint control.
This observation will rule out a unanimity voting system, amounting as
it does to a regime that gives each a veto, but all other systems of shared
control will remain on the table, at least to the extent that they allow equal
sharing. These candidates will include other systems of voting, whether
majoritarian or super-majoritarian. But they will also include systems in
which voting plays only a subsidiary part, or perhaps no part at all. We
have seen that control may be exercised on the basis of active, virtual or
reserve influence, for example, and that if the influence is active, it need
not even have an intentional character; it may be like the influence of the
comatose patient.
Popular control, by the lesson of our earlier discussion, has to involve
popular influence and popular direction. Thus the requirement of equally
31
There is also a normative argument against a universal veto. If people each had a veto over how the
laws are made, implemented or adjudicated, that would enable a single recalcitrant individual to undo
the social order, create a stateless zone and, to continue with an earlier assumption, make invasion by
other states more or less inevitable. I assume that the obligation of a legitimate state to give citizens
equal control over its doings would put that requirement – that normative necessity – in the way of
this option. If the state has a duty to give its citizens equal control over how it performs, then to allow
any individual the power of exposing their fellows to the jeopardy of foreign intervention would be
quite inconsistent with that obligation. Notice that despite the opposition he would certainly want to
mount against a universal veto, Kant (1996: 393–4) could not rule it out on the basis of his requirement
of ‘innate right’ – the basic right in his political philosophy (Ripstein 2009) – that one be independent
from ‘being bound by others to more than one can in turn bind them’. Notice, too, that this argument
against a universal veto does not carry over to a majority veto, at least not if the majority is not an
incorporated agent with a will of its own (List and Pettit 2011).
shared control entails, on the face of it, that people should share equally
both in exercising influence over government and in determining what
direction that influence is to impose. This entailment has not always been
endorsed within historical tradition. There was an extended debate in the
eighteenth century, for example, as to whether equal liberty requires that all
citizens – in effect, all mainstream men – should each be fully enfranchised,
thereby having an equal influence, or whether equal liberty can be achieved,
and so a suitable direction imposed on government, via the well-disciplined
influence of the few (Reid 1988).
It should be clear, however, and it should have been clear in the eight-
eenth century, that there cannot be equally shared control without equally
shared influence. If some people are to depend on the good offices of an elite
party in order to force government in a direction they all care about – and
not in a direction particularly favourable to the more powerful – then they
are dependent on the goodwill of that party for their very freedom. And
that, in republican logic, is anathema. Richard Price (1991: 80) spoke for the
right side – and he suggested, the traditional side – when he claimed that
‘every independent agent in a free state ought to have a share in the
government of it, either by himself personally, or by a body of representa-
tives in chusing whom he has a free vote’.
What, however, does equality in the exercise of influence require? It
cannot require that everyone should participate equally in the system of
popular influence, since some individuals may choose not to play their part
in the system, whether generally or on specific occasions; they may be happy
to go along with what others decide. Even under a system of compulsory
voting, such as that in Australia, people may spoil their vote and are in no
sense compelled to have a normal electoral input.32 What equally shared
influence requires, therefore, can only be equal access to the system of
popular influence: an opportunity for participation in that system that is
available with equal ease to each citizen.33
32
But it is worth noticing that even when people do not vote – or do not contest government decisions
in any active way – still they exercise some influence: they reduce the majority of the winning party or
they allow government not to have to deal with their contestation. On such matters see Guerrero
(2010).
33
Richard Tuck (2008) makes an intriguing case for thinking that preferring to have a group achieve a
certain effect makes it rational to prefer to play a part – even a palpably redundant part – in bringing
about that effect; see too Goldman (1999). I think that the preference to play a part may be
intelligible – there may be good reasons why you might want to do so – but I do not agree that the
preference over the outcome rationally requires the preference for playing a part. It may be otherwise
objectionable to let others bring about a result that you cherish, even when you know that your help is
unnecessary, but it is surely not irrational. For a similar critique, see Brennan (2011).
What might make for equality, not in the exercise of influence, but in
determining the direction that the influence imposes? Each citizen will
share equally in determining the direction their influence imposes if that
direction is required to be one that each is ready to accept; that each is
disposed to find acceptable.34 This requirement may not be satisfiable
amongst fanatics or zealots who insist on special treatment: say, the pri-
vileging of their religion or ethnicity. But it will be enough by our normative
standards if it is satisfied amongst those, as we put it earlier, who are willing
to live on equal terms with others: amongst those who accept that the state
should treat its citizens as equals.
To sum up the two lessons of individualization in a slogan, then, the
citizens of a legitimate state have to enjoy equal access to a system of popular
influence and that system of influence has to give the state an equally
acceptable direction – that is, a direction that they are all actually disposed
to accept. We will be looking at what might promote equally accessible
influence in the next chapter and at what might promote an equally accept-
able direction in the chapter after that.
An unconditioned system of popular control

In the case of the control that a people have over the state, as in all forms of
control, the controller will have to exercise a certain sort of influence, and
certain variations in that influence will have to correlate with certain
variations in how the controlled agent behaves. There will have to be a
correspondence between the inputs of the controller, on the one side, and
the outputs of the controlled, on the other.
In mechanical cases of control, this correlation is more or less sufficient
on its own to allow us to ascribe control. In particular, it is sufficient to allow
the ascription of control even when the correlation obtains only condition-
ally on the goodwill of a distinct agent. Thus the cooling-heating system
counts as controlling the temperature in the room, even when I remain in a
position to turn it off so that the operation of the system correlates with the
ambient temperature only conditionally on my allowing it to do so. This
observation is important because it marks a contrast with what is required
for ascribing control to one agent or set of agents over another. One agent
34
Saying that something is acceptable often has a normative significance among contractualists,
implying that it is such that people ought to accept it. Here and throughout this book, the word
has a non-normative sense, implying that the object or policy or whatever is such that people are
disposed to accept it; they find it acceptable, as we say.
will count as controlling another only insofar as the influence exercised leads
to the required result independently of the will of the controlled agent, or
indeed of any third party. Inter-agential control will count as control proper
only to the extent that it is unconditioned on any will other than that of
the controller.35
Suppose that you, A, are said to control what another person, B, does
because B obeys your instructions or, to take a case where your control is not
intentional, because B anticipates and acts on your wishes without your
even being aware of this. Imagine now that the correlation between your
instructions or wishes and B’s actions – the inputs on your part, the outputs
on B’s – is contingent on B’s wanting to humour you, or on my requiring B
to humour you. In that case you can scarcely be held to control what B does;
more colloquially, you cannot be said to have power over B. If the corre-
lation depends entirely on B’s will, no one has power over B; and if it
depends on my will, then it is I who has power over B, not you. The
correlation may depend on any of a variety of contingent circumstances:
that B is in thrall to your charms, for example, or that you have greater
physical strength or cultural clout. But it cannot depend on B’s willingness
to play the part, or on the willingness of any third party to make B do so.
Think of the example in which you give me the key to the alcohol
cabinet, instructing me to hand it over only at twenty-four hours’ notice.
Suppose I go along with your instructions, but merely with a view to giving
you the pleasant illusion of control; suppose that I am about to exit our
relationship and think of this as a parting, somewhat sardonic gift. In that
case you do not have the control you imagine; without realizing it, you are
in my hands, at my mercy. So why is there control in the normal case? Why
is the correlation between your instructions and my responses stable in a
way that is independent of my will? Presumably because, by implicit
assumption, the relationship matters to me, whether for intrinsic or instru-
mental reasons, and there are heavy costs associated with refusing to go
along: say, with just giving up the key in blatant disregard of the arrange-
ment, or with refusing to part with it at all.
For a further illustration of why inter-agential control has to be uncon-
ditioned, think of the case of the comatose patient, in Amartya Sen’s
example, whose wishes over hospital treatment are empowered by the fact
that his family or friends force the doctors to abide by them. Although we
did not make the point earlier, we can now register that if it is up to the
35
For a congenial interpretation of related themes see N. Southwood’s unpublished paper, ‘Democracy
as a Modally Demanding Value’.
discretion of a single person amongst the family or friends, or up to their
jointly exercised discretion, to determine whether the patient’s wishes
should prevail, then those individuals are going to be in control, not the
patient. In order for the patient to have control, it will have to be the case
that the responsiveness of the doctors is not conditioned on the willingness
of any single agent to speak for the patient. This requirement might be
fulfilled if there were some pressure on family or friends to intervene in the
patient’s name – if it was not just a matter of their discretion – or if there was
a queue of individuals lined up to play the role, so that the intervention did
not depend on their individual or joint discretion.
We saw in the first chapter that you are not free in a choice between X, Y
and Z if you can act according to your will – if you can control your
actions – only insofar as I allow it. What we are now discussing is a parallel
observation that bears, not on your control of your own actions, but on your
control of the actions of another party. You control your own actions only
insofar as you do not depend on the will of another for being able to satisfy
yourself; you can satisfy yourself over changes in what any other wants you
to do. You control the actions of another agent only insofar as you do not
depend on the will of another – that same agent or a third party – for being
able to get the other to satisfy you, acting on your instructions or wishes;
you can get the agent to satisfy you over changes in what the person wants to
do and over changes in what anyone else wants the person to do. As the
failure of the first condition means that you do not enjoy freedom – power
or control over yourself – so the failure of the second means that you do not
enjoy power or control over the other. Freedom requires a robust capacity to
satisfy yourself; power over another requires a robust capacity to get that
other to satisfy you.
The observation made in these examples applies to the control of any
agent or agency over another and so to the popular control that we want
citizens to exercise over the state. The implication of the requirement in the
political case is that the equally accessible popular influence that is required
to impose an equally acceptable direction on government has to meet a
further demanding constraint. It has to produce that effect independently of
the willingness of government to go along and independently of the will-
ingness of any other agency to have the government go along.36
36
The other agency envisaged here might be an effectively independent army, a group of moneyed
supporters or even a foreign power. The army possibility led most traditional republicans to oppose
the idea of a professional or standing army; many saw the development of such an army – indeed such
armies – as the factor that brought down republican Rome.
In every regime, the government, or at least the executive arm of govern-
ment, is likely to be enormously more powerful than any other individuals
or bodies, having special access to the means of universal coercion. But if the
government is the powerful party in the relationship between people and
government, then how can we expect the people to be able to exercise the
independent or unconditioned form of influence that is required? It may
seem that the people can hope to enjoy only the appearance of a controlling
influence, not the real thing: only the trappings of power that are available
when an indulgent state is willing to humour its citizenry.
The difficulty posed by this observation is real but not overwhelming.37
For the experience of societies over the past couple of centuries, even
perhaps before, shows that the control of the people over the state can be
grounded in a disposition of people to rise up in the face of a government
abuse of legitimacy and a disposition of government to back down in
response to the fact or prospect of such opposition. This is the trump
card that the people are always in a position to play, relying on any of the
various measures, violent and non-violent, direct and indirect, individual
and collective, that can be used to resist a regime. To the extent that the
possibility of popular, successful resistance is on the cards – to the extent
even that it is on the cards as a matter of common belief – the influence of
the people over government can be established on a robust basis and can
constitute a real form of power.
The observation that popular control of government is grounded in the
actual or perceived potential for widespread resistance – people’s presump-
tive power of rebellion – is not new. John Locke embraced the importance
of the possibility in arguing for the right of people to rise up against the
government, should it not be fulfilling its allotted role: as he saw it, the role
of being an impartial arbiter of disputes. In his view, as on the view to which
our considerations lead us, the legitimacy of a government ultimately turns
on whether ‘the community may be said in this respect to be always the
supreme power’ (Locke 1960: ii.149). The people will have to be the
supreme power in any polity that has a claim to legitimacy.
Adam Ferguson, a Scottish adherent of the eighteenth-century republican
creed, gave the idea memorable expression in describing the requirements
37
It is taken to be overwhelming by those like Georgio Agamben (2005) who follow Carl Schmitt in
arguing that even democratic governments fail the requirement. Their argument is that even such
governments freely decide – that is, decide independently of any external pressure such as that which I
go on to identify – that a given case is a non-exceptional one where the rule of law applies or is an
exceptional or emergency case where the rule of law is suspended. In Schmitt’s (2005: 5) words,
‘Sovereign is he who decides on the exception.’
of liberty.38 The liberty of the British people, he suggests, may be manifest
in the fact that laws are formulated under widely accessible influence and
given a widely acceptable direction. But its grounding goes much deeper: ‘it
requires a fabric no less than the whole political constitution of Great
Britain, a spirit no less than the refractory and turbulent zeal of this
fortunate people, to secure it’ (Ferguson 1767: 167). This characteristically
republican theme has a long history. It appears most dramatically in the idea
championed in Machiavelli’s (1965) Discourses that what enabled the citi-
zenry of republican Rome to enjoy their freedom vis-à-vis the state was
something that might appear at first sight to be a source of instability: the
willingness of the plebeian poor to rise up against even the suspicion of an
abuse or usurpation of power by the nobles. The price of liberty, in the
hallowed republican slogan, is eternal vigilance.
It goes without saying that societies differ greatly in the extent to which
this support for popular influence and control over government is avail-
able. It will be available in a measure that reflects the extent to which two
factors are in place and/or are taken as a matter of common belief to be in
place: on the one side, the disposition of the people to resist perceived
abuses of power by the government; and on the other, the disposition of
those in government to be inhibited by the fact or the prospect of such
resistance. These two factors determine how resistive a society is: how far,
in reality and/or perception, the citizens are resistance-prone and the
government resistance-averse.39
We saw earlier that in order to be individualized, popular control must
be grounded in an equally accessible system of influence that operates
in an equally acceptable direction. But we now see that the system of
directive influence to which citizens are given equal access has to be a
system of unconditioned or independent or resilient influence. It must
enable people to enjoy a directive influence over government that is based
on the resistive character of the society, not on the goodwill of government
or of any other agency.
38
As a matter of passing interest, Ferguson opposed Richard Price on the issue just mentioned as to
whether equal control requires equal influence: in effect, the universal franchise. See Price (1991: 80).
39
A society might be resistive in this sense and yet not be very regularly subject to popular disturbance;
the important thing is that resistance should remain a permanent possibility. John Locke may mean
just to register this – and not to strike what may seem like a complacently conservative note – when he
argues that since resistance is difficult to organize, and dangerous if it fails to attract majority support,
it won’t materialize except in the event of extreme abuse: ‘such revolutions happen not upon every
little mismanagement in public affairs’; they are likely to be triggered only by ‘a long train of abuses,
prevarications and artifices’ (Locke 1960: ii.225).
An efficacious system of popular control
But popular control of government does not only require that people should
enjoy equal access to a popular system of unconditioned influence and that
this system of influence should push government in an equally acceptable
direction. If the popular system of unconditioned influence is to ensure the
popular control required for legitimacy, then it must also satisfy a further
condition. It must be effective or efficacious enough to impose a popular
direction on government that nullifies the intrusion of alien will. It may not
reach the upper limit at which it would be as effective as possible, consis-
tently with the equal sharing of that control. But whether it reaches or falls
below that limit, it must be intuitively satisfactory.
This third condition takes us back to the requirement that the state
should treat its citizens well and equally in the manner in which it imposes a
social order, as well as in the character of the order imposed. The state would
treat its citizens as equals if it gave them the same equal share in a system of
unconditioned control. But it would not treat them well if that system of
control were unnecessarily limited. In order to ensure social justice, the state
must secure people’s equal, undominated status vis-à-vis one another to a
suitably high level or degree, as we saw in Chapter 2. And in order to achieve
political legitimacy it must secure their equal, undominated status vis-à-vis
the state to a similarly high level. That is the requirement registered in the
demand that the system of control in which people equally share should be
efficacious.
In theory, the constraints imposed under a suitable system of equally
shared control might bear only on the policies and related decisions imple-
mented by government, reducing the options in each area just to one. Such
policy-centred limitations would clearly deny those in power the opportu-
nity to evade popular control in the decisions they take. But in practice, any
plausible constraints are likely in each area to leave a number of policy-
options in place, not to reduce the options to just one. And so the
constraints imposed on government must also include many that bear on
the processes whereby ties are broken and decisions taken in each domain of
policy-making.
Such constraints, too, can serve the purpose of stopping those in power
from operating outside the boundaries of popular control. They might
require, depending on the issue, that decisions are taken by reference to
expert or impartial or judicial bodies, by a majority vote in the legislature or
in a referendum, or even by some kind of lottery. The assumption is that
while such process constraints will leave the final determination of policy in
any area to the luck of the draw – including, for example, the draw that gives
one or another side a majority in a legislature or referendum – that mode of
determination can be consistent in the relevant area with popular control. If
it allows the wills of some to rule on certain matters, as in giving the majority
party control of various issues, it does so only because such restricted
empowerment is consistent with all that popular direction and control
requires; like the power that I enjoy when you give me the key to the
alcohol cupboard, the power given to the majority party can only be
exercised within accepted bounds and need not count as dominating.
The fact that in practice many decisions will have to be left to govern-
ment under the best imaginable regime of popular control raises the spectre
of abuse in this range of discretion. Abuse might consist in the authorities
smuggling some self-serving candidates into the set of policies between
which a decision has to be made in any area, or indeed keeping some
popularly acceptable candidates out of that set. Or it might involve the
authorities taking advantage of loopholes in the processes available to
resolve policy disputes in any area for their own special benefit or the benefit
of cronies.
These possibilities sharpen the question as to how much popular
control – in particular, how much popular influence – will count as
enough to banish the spectre of uncontrolled or arbitrary government.
How efficacious ought the people’s influence over government to be?
How far ought it to be able to impose a popular direction on government,
removing opportunities for the intrusion of private interest and private
will? The question is akin to the issue, addressed in the preceding chapter,
as to how far people need to be guarded against domination from one
another for the enjoyment of freedom. Here the issue is how far people
ought to be able to control those in power on an equally shared basis, if
they are to enjoy freedom in relation to government. Where the earlier
question asked after the degree of entrenchment required for private non-
domination, this question asks after the degree of empowerment required
for public non-domination.
In answer to the earlier question, I suggested that the best we can say is
that entrenchment is adequate in a given society to the extent that by local
standards people would display excessive timidity if they could not pass the
eyeball test there – if they could not look one another in the eye without
being prompted to fear or deference by the possibility of another’s interfer-
ence. Where the eyeball test is a good index of whether the guards against
private domination are adequate, a parallel tough-luck test is a good index of
whether the guards against public domination are efficacious.
Imagine that the party or personnel in government do not meet with
your approval. What are you to think if they are manifestly appointed under
a suitably efficacious form of popular influence and direction, in which you
have an equal share? You can only think that it was just tough luck that
those appointed are not to your taste; it was not the work of a dominating
will, as it would be, for example, under a colonial administration. Or
imagine that the government passes legislation of which you disapprove:
say, legislation requiring the construction of new prisons. What are you to
think if it does so, again manifestly, under the equally shared, suitably
efficacious control of the people? You can only think that it was just tough
luck for you that the decision went that way; it was not the result of a will at
work in the public sphere that operates beyond the equally shared control of
you and your fellows. Or imagine, finally, that in implementing the
legislation, the government decides to your dismay that one of the prisons
should be located in your neighbourhood – in your back yard. What are you
to think if it manifestly makes this decision under a suitably efficacious form
of control in which you and those in your area equally share? Again you can
only think that it was just tough luck that the decision went against you; it
was not due to the special influence of those who are richer or electorally
better placed or closer to the corridors of power.40
The point of legitimacy is to ensure that you and your fellow citizens are
not subject to an alien, controlling will, despite that fact that there may be a
good deal of discretion exercised by those in power. Such legitimacy will be
adequately ensured, so these observations suggest, to the extent that you and
your fellows have good grounds to think that any unwelcome results of
public decision-making are just tough luck. By local standards of when trust
is well placed and those in government inhibited from exploiting loopholes
in the constraints imposed, you need not view unwelcome results as the sign
of a malign will at work against you or your kind; or at least you need not do
so, assuming you are not subject to excessive anxiety or distrust. You may be
exasperated by what transpires in the formation of government, or in the
creation or administration or adjudication of the law, but if you really share
in equal, suitably efficacious control of government, which is what repub-
lican legitimacy requires, then you will not have a reason to feel resentment
at how the state performs.
You may feel resentment, of course, at how your fellow citizens behave in
private or public roles, even when there is no question of popular control
40
My thanks to Ben Ewing for discussion of the points in this paragraph.
being undermined; you may be indignant about the culpable ignorance or
indifference, for example, that their refusal to support a certain policy seems
to display. Such resentment at individuals is quite consistent with allowing
that the state itself, since it enacts a will over which you enjoy shared,
suitably efficacious control, is not deserving of resentment. That the state
enacts an unjust policy, by your lights, will be a result of the bad luck of your
having many culpably ignorant or indifferent compatriots, not a result of its
harbouring an alien will. Insofar as the tough-luck test is satisfied, the will
displayed by the state is under an efficacious form of control that you share
equally with others in imposing. If that will fails to track the good, then you
and your fellows have only yourselves to blame. And if you don’t blame
yourself, you can only blame your fellows.
A variant of the tough-luck test of efficacy applies, not just in the sorts
of example given, but also in a case where you are charged with a crime,
found guilty and sentenced to some punishment: perhaps a period in
prison. If you are innocent of the crime, there will still be a great difference
between the case where it really is just tough luck that all the evidence
pointed against you, and that the institutions in place were incapable of
registering your innocence, and the case where the system allowed some-
one the power of acting against you: say, a police officer or public
prosecutor who was hostile to you personally, or to those in a particular
subgroup to which you belong. But what if you are guilty of the crime? In
that case, a legitimate system will enable you to think, not that it was
tough luck that you were sentenced to prison, but something related: that
you have only yourself to blame. Knowing the rules under which you live,
and knowing that you share equally in control of them, you cannot
complain about what is imposed under those rules as a penalty for your
breaking them.41
The tough-luck test offers an intuitive way of gauging how far the efficacy
of democratic control over government is satisfactory. It should be noted,
however, that it is only available as a plausible criterion of performance, on
the republican assumption that legitimacy requires people to exercise con-
trol over the will of government, not control over everything that happens.
Thus it is no problem for republican theory that in any area of policy brute
41
There are difficult issues, however, that I am ignoring here. Some arise as a result of moral differences;
for an excellent discussion of such cases see Talisse (2009). Other difficult issues are multi-cultural
questions of the kind that Will Kymlicka (1995) in particular has put on the table; for a broadly
republican approach see Nathan (2010) and for an incisive discussion of one particular issue see
Laborde (2008).
The democratic state 179
chance ultimately determines what exact policy or decision the government
selects under the terms imposed on it by the people.42
The lesson of the two previous requirements on the popular control of
government that legitimacy requires was that people should have equal
access to a popular system of unconditioned or independent influence that
pushes government in an equally acceptable direction. Adding in the third
requirement, the lesson is that popular, legitimating control of government
requires equal access to a popular system of unconditioned and efficacious
influence – specifically, one that imposes an equally acceptable direction on
government. Putting the implications of legitimacy in a slogan, we may say
that people have to enjoy an equally accessible form of influence that
imposes an equally acceptable direction on the state. But it should always
be understood that the directive influence to which people are given access
has to be not only equal or individualized, but also unconditioned and
efficacious.
5. the democratic state
A theory of democracy
Starting from the requirements of legitimacy, understood on republican
lines, we have argued that they can be fulfilled only insofar as citizens enjoy a
suitable range, kind and degree of control over government. There are no
other salient conditions that we might require a legitimate state to fulfil.
And so we may conclude that a state will be legitimate just to the extent that
in a suitable domain people enjoy equal control over government: equally
accessible influence – unconditioned and efficacious influence, as I shall
assume – that pushes government in an equally acceptable direction.
Any system that satisfies such conditions deserves to be described as a
democracy, since it gives the demos, or ‘people’, an equally shared,
42
In this as in other respects, republican theory enables us to support judgements that are independ-
ently plausible: it satisfies the test of reflective equilibrium with such judgements. This is particularly
striking in the present instance, since the plausibility of the tough-luck test is not so readily
established on other accounts of the role of democracy. On many other accounts the people would
only enjoy control to the extent – an impossible ideal – that they controlled everything that transpires
at the hands of government agencies: only to the extent that they determined the content of every law
and policy, every decree and judgement. On the present account they will enjoy control to the extent
that they control the will at the origin of what transpires. And they can achieve this insofar as they can
impose terms that leave little or no room for the intrusion of alien will on the part of government
agencies, even if they leave matters of detail to the luck of the draw. We will be looking in later
chapters at examples of how such control can work.
independently supported and intuitively efficacious degree of kratos, or
‘control’, over the state.43 And so what we have developed can count as a
republican theory of democracy. There is no one proprietary meaning of the
term ‘democracy’, as it is used in ordinary or philosophical or scientific
language; people often divide on what exactly the notion of democracy
connotes (Connolly 1993). Thus the theory according to which democracy
requires a popular system of individualized, unconditioned and efficacious
control does not count as analytical in character; it does not offer an analysis
of the term ‘democracy’ as a theory of causation might offer us an analysis of
the term ‘cause’. What it offers, rather, is an ideal that democracy, at its best,
might be required to achieve or approximate.44
Put in other terms, what the theory offers is an account of what demo-
cratic institutions ought to do: a job specification that they ought to be
designed to implement. But such a job specification naturally raises a
question about its capacity to be implemented. The question is whether
the ideal of a regime in which people share equally in control of government
is institutionally feasible or just a utopian pipedream.
The only way to answer this question is to develop a model of how certain
realizable institutions might bring the ideal into existence, or close to
existence. Having sketched the requirements of republican justice in sec-
tions 1 and 2 of Chapter 2 – having described the job specification under
which a just social order requires the publicly entrenched enjoyment of
basic liberties – we went on in Section 3 to indicate the sorts of institutions
that might realize that ideal. We need to do the same thing for the ideal of
legitimacy or democracy, offering a model of democratic institutions in
which it might be brought to life.
I take up this task in the two chapters following. In Chapter 4, I look at
the civic or popular influence that democracy in a republican sense would
require and in Chapter 5 at the civic or popular direction that that influence
ought to impose on government. The model that emerges in those chapters
need not persuade everyone who accepts the republican theory of demo-
cracy. For all I know, there may be a better, perhaps even radically different,
alternative available. My commitment to the model goes less deep,
43
For a popular but very useful introduction to the origin of the notion of democracy see Dunn
(2005).
44
The conception is close to what Ronald Dworkin (2006: 131) describes as a ‘partnership view of
democracy’ under which ‘democracy means that the people govern themselves each as a full partner in
a collective political enterprise so that a majority’s decisions are democratic only when certain further
conditions are met that protect the status and interests of each citizen as a full partner in that
enterprise’. See too Dworkin (2011: Chapter 18). For a somewhat different account of the linkage
between democracy in this sense and the absence of domination see Lovett (2010: 210–20).
therefore, than my commitment to the theory. I offer it in the spirit of a
realistic proposal, as I take it to be, which shows that the republican ideal of
legitimacy and democracy is not other-worldly; it is an ideal that can inform
politically serious contemporary debate.45
The appeal of the democratic state

On the assumption that the republican ideal of a legitimate and democratic
regime is institutionally feasible, I would like to make some observations on
the appeal that such a state ought to have for us. The democratic state, as
that ideal begins to take shape within republican theory, is not just an
institution that remedies or reduces some of the hardships that derive from
our not being of an angelic nature. So long as it achieves republican justice,
it is the source of a great and otherwise unavailable good. The necessity of
the state for the achievement of this good suggests, contrary to a good deal of
received opinion, that the state is not an institution about which we should
be ambivalent. While it certainly requires high maintenance, as our argu-
ment should have made clear, the well-maintained state holds out the
prospect of our being able to enjoy an important benefit that would
otherwise escape us.
This benefit is the good of status freedom. It consists in the objective and
subjective status of your enjoying freedom as non-domination, equally with
fellow citizens, in the common sphere of the basic liberties and of doing so
without domination by an uncontrolled, if benevolent, government. Status
freedom in this sense has a horizontal and a vertical dimension, requiring
independency on the will of others, as it used to be said, both in your
relations with fellow citizens and in your relations with your state. It is the
good, now extended to you and all your fellow citizens, that was savoured in
traditional republican texts as the freedom of the liber, or ‘free-man’: the
freedom of the fully and openly incorporated member of a society of fully
and openly incorporated citizens.
In order to appreciate the good that a just and democratic state might
provide, it may be useful by way of contrast to consider an ideal stateless
society that is much celebrated in the tradition: the manifestly utopian and
infeasible society associated with the Kantian idea of a kingdom of ends. It
turns out that even the kingdom of ends could not provide the good of
45
The republican model of democracy emerging from this account might figure among the models in
David Held’s (2006) standard account, but it does not coincide with the model that he describes as
republican.
status freedom, at least not with the robustness that a just and democratic
state might underwrite.
The kingdom of ends is a stateless dispensation in which members may
differ in their levels of wealth and power but are each morally committed at
the highest possible level. They are committed, first, to showing respect for
others and second, perhaps going beyond Kant, to showing concern, as well.
Their commitment to respect obliges members of the kingdom of ends to
renounce interference in one another’s basic liberties, so I shall assume, and
their commitment to concern obliges them to provide any resources that
others may require.
Would the introduction of a just and democratic state improve the lives
of the moral paragons who inhabit the kingdom of ends? I believe it would.
In all likelihood, it would provide a much more effective means of identify-
ing the basic liberties that people are to enjoy. But, much more importantly,
it would also provide in a novel way for the entrenchment of those liberties.
Were people moral in the degree imagined, then the more wealthy and
powerful would be disposed not to interfere with others and not to allow
others to go without needed resources. But their acting on that disposition
would depend on their continuing to embrace and abide by the require-
ments of the assumed morality. It would depend, to recall a point from the
first chapter, on their displaying a good will rather than a weak will or a will
to evil. In such a world, therefore, some members would have to depend on
the goodwill of others for enjoying the basic liberties. And so the coercive
laws of a just and democratic state would play a crucial, beneficial role,
establishing external checks and supports to guard each member against
dependency on the will of the more powerful. They would be a means of
providing for people’s enjoyment of status freedom.46
Would this freedom really be a benefit, however, given that the members
of the kingdom of ends are vanishingly unlikely to turn ill-willed? Yes, for
reasons rehearsed already. Seeing and treating one another as agents, not as
pre-set, probabilistic devices, even the members of the kingdom of ends
would have to recognize that whether another relatively powerful member
chooses to interfere with them, or chooses not to help out in the absence of
46
One qualification. In the kingdom of ends, social norms of a particularly respectful, and indeed
benevolent, kind might be expected to emerge and, having emerged, might put suitable checks on
interference with others or even on indifference to their needs; any such failure might trigger costs in
the economy of esteem (Brennan and Pettit 2004). I ignore this here, since my only point is that the
kingdom of ends as traditionally conceived would deliver more benefits if it were complemented by a
democratic state. For a useful discussion of Kant’s kingdom of ends that is in some ways congenial
and in others uncongenial, see Van der Rijt (2012: Chapter 5).
resources, will often be up to that person. And so they would have to
recognize that on one or another front they depend on the person’s good-
will. Thus, as we imagine the perspective of weaker parties, we see that they
would be inevitably and manifestly indebted to the powerful amongst them
for non-interference or for the resources they enjoy. They would be in the
position of dependants, not equals, in relation to those parties.
The failure of the kingdom of ends to provide for the status freedom of its
members highlights the importance of what the just and democratic state
can provide. Introduced to the kingdom of ends, it would give each member
protection against the accessible, if highly improbable, interference or
indifference of the stronger. And in providing that benefit, it would confer
a free status that would previously have been unavailable. The actual
behaviour of people after the introduction of such a regime of law and
coercion might not differ very much from their behaviour beforehand. The
role of coercive law in providing a second line of defence against weakness or
illness of will on the part of the powerful would not be visible in anything
that anyone did – at least not if we assume that prior to the appearance of a
state the powerful didn’t ever display any weak or ill will. But the role of
coercive law would still be of the utmost importance. It would mean that
the non-interference and resources that the weaker enjoy come to them, not
in the guise of a gift for which they should be grateful, but rather in the guise
of an institutionally guaranteed claim.
As mentioned earlier, the coercive state is certainly needed to promote
compliance and to assure compliers that others are complying too. But we
can now see that it is needed on a third count, too. For the just and
democratic state relies on coercion – the coercion involved in protecting
against interference and in redistributing to those with insufficient
resources – to ensure that people’s enjoyment of the basic liberties is
publicly entrenched. This coercive entrenchment provides people with a
freedom as non-domination that is independent of the will of others as to
how they should choose, including the will of government itself. And that
is not something that mere morality can provide.
G. A. Cohen (2008: 148) expresses the orthodox view on the coercive
state, however just and however democratic it may be, when he says that
there are only two functions that it can play: ‘coercion is necessary only for
deviance or assurance reasons’. But this is wrong, according to the picture
emerging here. Provided it is just and democratic, under the republican
specification of those virtues, the coercive state can generate a good for all of
us – the good of status freedom – that we cannot provide for one another on
the basis of our own goodwill. We cannot provide this for one another as we
actually are. And we could not provide it for one another, even if we were as
angelic as the members of the kingdom of ends.
Economists characterize benefits that the private market is incapable of
producing, and that appear to require collective organization, as public
goods. These are goods like defence, criminal justice and public health
that for reasons of psychology or organization – say, the disposition to
free-ride – we would be unlikely to generate in the absence of centralized
coercion. The upshot of the discussion is that the free status that repub-
licans envisage is also a public good. Freedom requires independency on
the will of others, even the goodwill of others. And for that deep and
inescapable reason, free status is something that we can make available to
one another as individuals only by collectively organizing ourselves in a
state – strictly speaking, a just and democratic state.47
Back to Rawls
In introducing the model of social justice that republican theory would
support, I argued that in important respects it would break with Rawls,
replacing his two principles with a single principle that calls for giving
people status freedom in relation to one another. While the republican
theory would undoubtedly overlap with Rawls’s in many of the policies
supported, it would provide support on a different basis. It may be useful in
concluding this discussion of the republican theory of legitimacy and
democracy to sketch a parallel comparison with the approach that
Rawlsian and related theory would adopt.
As we saw earlier, Rawls and others focus on comprehensive, social-cum-
procedural justice in evaluating a legal order – a basic structure, in his
sense – rather than breaking that evaluative task, as we have done, into the
distinct issues of social justice and political legitimacy. Rawls explores the
comprehensively just order by asking which basic structure or structures
would satisfy his contractual eligibility or his civic justifiability test. And in
answering that question he requires, unsurprisingly, that the just structure
should give people certain basic political liberties and rights. The issue for
47
At this point we have reason to think that it is not only good, assuming there are people around, that
no one should dominate others, and that no one should vitiate anyone’s undominated capacity; it is
good that there should be people around who can enjoy mutual undominated relationships. This
good is close to that which is celebrated by recognition theorists (Honneth 1996) and others who are
influenced by Hegel’s account of the status associated with freedom (Patten 2002). Adopting this
view would help republican theory to deal with some of the complex issues raised in population
theory as a result of the non-identity problem (Parfit 1984: Chapter 16). I am grateful to Jake Nebel for
exchanges on this topic.
us, then, is whether the political rights he envisages might be sufficient to
satisfy the demands of political legitimacy or republican democracy:
whether they might be enough to give people an equal share in a public
system of individualized, unconditioned and efficacious control.
The political rights to which Rawls (1971, 1993) directs us include all the
procedural liberties associated with electoral democracy – for example, the
freedom to vote or to stand for office – as well as the rights of free speech,
association and movement that they presuppose. But he does not lay much
emphasis on the procedural liberties, casting them at one point as ‘sub-
ordinate to the other freedoms’ (Rawls 1971: 233). What is important in his
perspective is that if we consider impartially what society ought to provide
for individuals – if we consider this, for example, from the point of view of
the original position, in which none of us know how we will fare under the
basic structure adopted – we must conclude that it ought to furnish them
with procedural as well as more substantive liberties and rights. Those
requirements are benefits that each should fully and equally enjoy, within
a suitable mix of substantive and procedural rights, on pain of suffering
unfairness. But they do not have a special status and may even be sub-
ordinate to their substantive counterparts.
On the republican view, things look quite different. Republican theory
casts the procedural rights that people are to enjoy – together, as we shall
see, with other, less formal, rights – as powers that they must each fully and
equally exercise if they are not to suffer subjection and domination by an
uncontrolled state. It invests the state with heavy responsibilities in the
sphere of social justice, as Rawls does. But it emphasizes that nonetheless the
coercive state is a real danger – a danger to the very freedom that social
justice is meant to promote – and that citizens must share equally in control
of political coercion if they are to enjoy the status of free persons. The
spectre of the despotic and illegitimate state – even the benevolently
despotic state – haunts the approach as thoroughly as the spectre of the
socially unjust regime. And it inevitably gives prominence to the demand
for a regime of shared, popular control.
This may explain why Rawls and others do not actually give the same
standing to the requirements of legitimacy and democracy as republicans.
But are they actually prevented from doing so by their focus on compre-
hensive, social-cum-procedural justice? Could they maintain that focus and
yet be persuaded by the case for republican freedom to give a greater
importance to democracy?
In principle, yes. The tests of contractual eligibility and civic justifiability
on which Rawls relies can be seen as ways of operationalizing the normative
demand for the state to treat its citizens as equals (Dworkin 1978). Accepting
such a test, we can still go along with republican thought and take freedom
as non-domination, understood in a status sense, as the good with reference
to which the test ought to be applied. And if we do this, then no basic
structure will appeal unless it gives citizens a suitable level of equally shared
control over how it is likely to evolve at the hands of government. Let it fail
to give citizens such control and their freedom is going to be jeopardized at a
very deep level. They may seem to enjoy social justice under the envisaged
structure but they will enjoy it only as the result of an indulgence on the part
of those in power. The freedom that social justice purports to give them will
not really be freedom, since it will be tainted by their dependency on the
goodwill of government.
The upshot is that while the Rawlsian approach of articulating the
demands of social and procedural justice in a comprehensive package may
actually lead to a downgrading of legitimacy and democracy, this is not an
inevitable result. It would be avoidable, at least in principle, under the
assumption about the importance of freedom as non-domination that
republicans make. It is certainly useful to guard against the possibility of a
downgrade by distinguishing between the demands of social justice and
political legitimacy, as we have done here. But that way of presenting things
merely facilitates an insight into the importance of democracy that is already
guaranteed by embracing the core republican ideal.
chapter 4
Democratic influence
The idea in this chapter and the next is to explore the institutional
possibility that the people in a polity might have such control over those
who run the state that they are not individually dominated by the interfer-
ence that the state practises in taxation, coercion and punishment. To the
extent that they have a control that makes such interference non-
dominating, the citizens will not lose out in freedom just by the fact of
living within that state and, by republican criteria, the state will count as
politically legitimate. A state that is legitimate in that sense may not achieve
a great deal in guarding against private domination and achieving social
justice, though it probably has to achieve some minimum threshold if
citizens are going to be capable of exercising control over its doings.
Control, as we saw, depends on two distinct elements, influence and
direction. Thus the people will achieve control over the state insofar as they
attain influence, on the one side, and succeed on the other in using that
influence to impose a suitable direction on government. Such popular
control will be suited to republican purposes, guarding against domination,
to the extent that it gives each citizen an equal share in the control,
particularly an equal share in a form of control that is suitably uncondi-
tioned and efficacious. People must enjoy an equally accessible form of
unconditioned and efficacious influence that imposes an equally acceptable
direction on the state. In this chapter I look at how people might enjoy the
required influence and I turn in the next chapter to how this influence
might impose the required direction. In this chapter’s discussion of the
institutional means whereby people might come to enjoy the required
influence I shall anticipate the discussion in the next and assume that the
influence they enjoy can support a popular direction and not merely be
wayward in character.
How to approach the discussion of the influence that the people might
have in a polity? Should I look at utopian, perhaps fanciful ideals of how
popular influence might materialize? Or should I start from democratic
187
188 Democratic influence
institutions with which we are all familiar and begin to chart the influence
that they make possible? It’s a hard choice. If I go the first way, I may be
accused of dabbling in ‘models of speculative perfection’, in William Paley’s
(2002: 319) words, and risking irrelevance in the discussion of obtainable
reforms. If I go the second way, I may be charged with lowering my sights
and paying undue obeisance to the status quo.
I propose to approach the challenge as follows. I begin with a model in
which the citizens gather periodically in a plenary assembly, a committee-
of-the-whole, to determine the laws of the community. This model, which
is reminiscent of Rousseau’s assembly, offers a plausible, initial interpreta-
tion of what it might be for the people to enjoy equally accessible influence
over government. Whatever its other faults, however, I argue that the model
is quite infeasible, even in an electronic age in which people might assemble
virtually. Having identified the problems of feasibility that a plenary assem-
bly would confront, I go on in Section 2 to discuss the rival merits of two
quite different forms of non-plenary assembly, one indicatively representa-
tive and the other responsively representative. And having argued in favour
of the responsive form, I then argue over sections 3, 4 and 5 that in order to
serve the cause of republican legitimacy, this assembly has to be subjected to
a series of important amendments. The model to which these amendments
lead promises to deliver a form of popular influence that is appropriately
individualized, unconditioned and efficacious, as our ideal of legitimacy
requires.
This final model does not conform to the status quo in existing systems of
representative democracy but it is close enough to warrant a claim to
institutional feasibility: it represents a regime we can see how to establish
and stabilize. And yet the model is far enough away from the status quo to
direct us to reforms that are required if our institutions are to support a form
of popular influence that is individualized, unconditioned and efficacious.
The model incorporates versions of the mixed constitution and the contest-
atory citizenry that bulk large, as we saw, in Italian–Atlantic republicanism.
1. the plenary assembly
The origin of the idea

The thesis that democracy requires an assembly of all the citizens to be the
sovereign body, enjoying control over law-making, is often represented as
the Athenian model – the model realized in Athens in the fifth and fourth
centuries bce – probably because of the slanted account of Athens given by
The plenary assembly 189
Polybius (1954) in his history of the purportedly contrasting, mixed con-
stitution that he identified in Rome. But Athens does not conform well to
the plenary model, at least in the form it assumed from the late fifth century
on. In that period, as earlier, the courts, or dikasteria, which involved
hundreds of citizens, selected on the basis of a lottery, could punish some-
one for attempting to change the law within the ecclesia – the ‘plenary
assembly of citizens’ – and for thereby assuming some legislative authority
for themselves.1 And while the issue of whether a law should be considered
as a candidate for revision was determined by the ecclesia, any actual
revisions were decided by another body of hundreds of citizens, again
selected by lot: the nomothetai, or ‘law-makers’ (Hansen 1991; Ober 1996).
Ironically, the plenary assembly model of democracy goes back in its
purest, clearest form to two thinkers who themselves defended an absolute
monarchy. They are the sixteenth-century French jurist, Jean Bodin, and
the seventeenth-century English philosopher, Thomas Hobbes (Pettit
2012b). They suggested that in casting democracy in this way, they were
being faithful to the model of classical Athens in the fifth and fourth
centuries bce. But, mischievously or not, they were actually quite innova-
tive in the image they proposed.2
Bodin and Hobbes thought of law as the command of a supreme power
or sovereign. This was something of a novelty in their time, for the main
body of law that coordinated relationships between individuals and com-
munities in the medieval world – the jus commune, or ‘common law’, as it
was known – did not have any single, centralized enforcer. Derived from
Justinian’s Roman law-books, it was enunciated in scholarly glosses and
commentaries as much as in princely edicts, including the edicts of the
German-based, medieval emperor (Woolf 1913). Assuming that law was
command, however – as it must often have more plausibly seemed in their
divided, post-Reformation world – Bodin and Hobbes argued that there
had to be a single commander and a single will at its origin; there had to be a
sovereign individual or a sovereign body.
Following Aristotle’s (1996) account in The Politics, Bodin and Hobbes
maintained that there were just three possible types of state, corresponding
to three types of sovereign: monarchy, in which the laws are handed down
by a sovereign individual – this was the strong preference of each; aristoc-
racy, in which the laws are issued by a sovereign committee of the few; and
1
The procedure under which this was possible was known as the graphe paranomon: it allowed for the
indictment of those proposing measures contrary to the laws (Hansen 1991; Ober 1996).
2
For an exchange on Hobbes on democracy see Tuck (2006) and Hoekstra (2006).
democracy, in which the laws are imposed by a sovereign committee of the
many. Unlike Aristotle, Bodin and Hobbes ridiculed the idea of a state that
would operate under a mixed constitution. While taking different argu-
mentative routes to the conclusion, they each maintained that because such
a constitution would not identify and establish a single commander or law-
giver, it would undermine the capacity of the state to serve its purpose,
ensuring order and peace in the land; it would be a recipe for civil strife and
political disaster.
But if Bodin and Hobbes were the first to enunciate clearly the image of
government by a plenary, legislative assembly, it was Jean Jacques Rousseau
who offered the first, full-scale defence of the model. Working with the
categories inherited from such absolutist predecessors, as we saw in the
Introduction, Rousseau (1997: ii.2.2) rejected the idea of the mixed con-
stitution: it would turn the sovereign, he says, ‘into a being that is fantastical
and formed of disparate pieces’. But he could not accept the idea of a
monarchical or an aristocratic state, given his attachment to the broadly
republican ideal that no one should have to live in subjection to another.
And so he had little option but to argue in support of what Bodin and
Hobbes would have called a democracy. He held that the people should
gather periodically in a sovereign assembly to deliberate and decide upon
the laws and that they should then give the job of implementing and
adjudicating those laws to appointed magistrates. He himself thought of
this design as republican but not democratic, since he used the term
‘democracy’ in a somewhat idiosyncratic manner to describe an arrange-
ment in which the assembly would implement, as well as make, the laws
(Rousseau 1997: iii.17.5).
Following Hobbes, Rousseau holds that while there has to be unanimity
in the contract to enter political society – ‘the act by which a people is a
people’ (1997: i.5.2) – that contract makes it alright for further decisions to
be taken by majority vote: ‘the vote of the majority always obligates all the
rest; this is a consequence of the contract itself’ (iv.2.7).3 But he breaks with
Hobbes in hailing a deliberative ideal, according to which the members of
his assembly should vote on the basis of the common good. They should
each think as citizens, focused impartially on their common interest, and
they should vote for any measure they support, whatever their personal or
factional interests, on the grounds that ‘it is advantageous to the State’
3
He recommends that on some important matters, it may be useful to require a super-majoritarian
approval: this could be introduced, presumably, by a prior majority agreement. See Rousseau (1997:
iv.2.11).
(Rousseau 1997: iv.1.6); this is required if the majority vote they cast is to
reflect the general will of the community, as he calls it. We shall return to a
consideration of the deliberative constraint in the next chapter, when we ask
how popular influence might be directed in a manner acceptable to all. In
this chapter, our focus is on the prior question of how to organize a system
of popular influence so that everyone has an equal share. And in discussing
that question we can abstract from whether or not people should be
required to exercise their collective influence under a deliberative constraint.
The discursive dilemma

Can we look to the plenary assembly as a model on which to design a system
of popular influence? Many argue not, on the grounds that almost all
societies are too large to allow citizens to come together in regular assembly.
Rousseau (1997: iii.15.12) himself seems to have recognized the problem
and it became a standard observation amongst succeeding thinkers like
Immanuel Kant (1996: 296) and James Mill (Lively and Rees 1978: 7). But
even if the size problem could be overcome, say by resort to virtual
assemblies on the Internet, there is no real prospect of enabling the citizens
as a whole to establish an effective law-making body. They will confront a
problem illustrated by the discursive dilemma.
The most basic requirement on any law-making assembly, acting in
concert with the executive and judicial bodies that complement it, is that
it should be able to promote consistency in the law and, if certain laws are
shown to be inconsistent, should be capable of recognizing and responding
to the problem. If the state were insensitive to challenges of consistency
then, as we saw, it could not constitute an agency capable of pursuing justice
or discharging any of its presumptive duties; it would be an unconversable,
unreachable entity. This would be a disaster, not just from the point of view
of republican justice, but also from the point of view of republican legiti-
macy: you and your fellow citizens could not hope to use your influence to
impose any sort of direction on such a state. You can’t do business with a
body that finds nothing to apologize for in the fact that it upholds a number
of inconsistent propositions or instructions.
The problem with the plenary assembly is that while the participants
might be individually consistent in their final judgements and votes on a
range of connected issues, they would be liable to generate an inconsistent
body of judgements if they relied on majority voting for the aggregation of
their individual sets of judgements into a common, shared set. Suppose
there are just three of us in the assembly: I, A; you, B; and a third party, C;
think of us, if you like, as standing in for three distinct subgroups. And
suppose that we want to form judgements on three connected issues:
whether p, whether q and whether p-and-q. The proposition ‘p’ might be
the claim that we should increase defence spending, ‘q’ the proposition that
we should increase other spending. I, A, and you, B, might think that p,
while the third party, C, rejects it. And you and C might think that q, while
I reject it. In that case ‘p’ will be supported by a majority of members – you
and me – and ‘q’ will be supported by a distinct majority of members – you
and C. But you will be the only one to support p-and-q; I will not support it
because of rejecting ‘q’, and C will not support it because of rejecting ‘p’.
Thus, despite endorsing ‘p’ and ‘q’ separately, we as a group will not endorse
the combined package ‘p-and-q’; indeed, under standard rules of voting, we
will reject it. We will be committed, under majority voting, to thinking that
we should increase defence spending, we should increase other spending,
and yet we should not increase both defence and other spending.
While we will individually hold consistent views on the relevant issues,
then, we will be collectively led under majority voting to adopt an incon-
sistent set of group views, as Table 4.1 illustrates.
Majority voting ensures that the corporate judgements are responsive to the
judgements of individuals, as seems appropriate, and the problem is that such
individual responsiveness makes it difficult to achieve collective rationality.
There is a discursive dilemma in prospect, as I have described it elsewhere
(Pettit 2001c, 2001e: Chapter 5; List 2006a).4 The dilemma consists in the fact,
broadly stated, that members can opt for individual responsiveness, in which
case collective rationality is jeopardized, or for collective rationality, in which
case individual responsiveness is in danger. There is no middle way, so that
either individual responsiveness or collective rationality has to go.
Table 4.1 Illustrating the discursive dilemma
p? Increase defence? q? Increase other? p-and-q? Increase both?
I, A, vote Yes No No
You, B, vote Yes Yes Yes
C votes No Yes No
We vote Yes Yes No
4
The discursive dilemma is a generalization of the doctrinal paradox that arises for collegial courts; this
was first properly identified and analysed by Lewis Kornhauser and Larry Sager (1992a, 1992b, 1993).
The problem, as a number of results have shown, is not specific to
majority voting. It is liable to arise under a variety of interpretations of
the constraints of individual responsiveness and collective rationality.5 But
it is sufficiently well illustrated for our purposes by the infeasibility of
majoritarian decision-making.
The problem means that it is not going to be enough for a group,
however deliberative it may be, just to rely on majority voting – or on any
of a variety of such bottom-up voting procedures – to aggregate its views.6
The group must keep track of where the accumulating commitments
generated by individual rounds of voting are leading it – its members
must gather feedback on whether the commitments are generating incon-
sistency – and it must be able to respond appropriately to that information.7
One strategy it might adopt is the straw-vote procedure. Under this proce-
dure, a majority vote in support of some judgement will be treated as a straw
vote until the result is tested and shown to be consistent with judgements
supported previously; if it fails that test then either it or some prior vote is
rejected (List and Pettit 2011: Part i). The lesson is that in order to guard
against inconsistency in its group views, the group has to be able to access
and process feedback on where the votes of members are leading it and,
when those votes lead to inconsistency, it has to make revisions that restore
consistency – or at least insulate any inconsistencies that prove recalcitrant.
A group might conduct this reflexive exercise in a mechanical way,
establishing a routine whereby any later vote that proves inconsistent with
earlier commitments is nullified: this would force our A–B–C group to reject
the negative vote and endorse ‘p-and-q’, for example. But that would scarcely
be satisfactory, since the recognition that prior votes require accepting a result
such as the combined package might be thought by members to provide
reason, not for revising the negative vote on the combined package, but
for rejecting the prior, positive vote on one of the elements in the package.
A satisfactory, decision-making body would have to be able to conduct
5
For an initial result see List and Pettit (2002) and for a survey of later results see List and Polak (2010)
and List and Pettit (2011).
6
Some authors like to contrast aggregative and deliberative processes, as if deliberation meant that
aggregation was unnecessary or unproblematic. But it is agreed on all sides among deliberative
democrats that deliberation will not generally lead to consensus and that the differences which
deliberation still leaves in place have to be resolved by a voting process. The discursive dilemma
shows that when the issues to be voted on are logically connected, as they generally will be over any
stretch of time, then there may still be a serious problem of aggregation.
7
Those members, as Niklas Luhmann (1990: Chapter 9) argues, have to keep track in an exercise of self-
reference on how under a common name – in his argument, that of the state – they are getting to be
committed.
second-stage, post-feedback deliberation – for short, reflexive deliberation –
as well as the basic deliberation conducted prior to first-stage voting (Pettit
2007a, 2007b). This is the sort of deliberation required under the straw-vote
procedure. The group votes that p, votes that q, and, on finding that it also
votes against p-and-q, deliberates about which vote to reverse.8
This takes us to the denouement (Pettit 2003b). No plenary assembly
could be expected to be able to conduct this reflexive form of deliberation,
assuming that its numbers are large. It may be possible to get members to
deliberate in a basic way about a single issue – say, whether to increase
defence spending or to increase education spending – and then to resolve
remaining differences by voting. But what can they do to deliberate in the
wake of feedback on the inconsistency of the various measures they have
supported – this might be made salient by a suitable feedback committee –
and on the comparative merits of resolving it in one or another manner? It is
unlikely in the extreme that any large body could discharge this reflexive
task satisfactorily. The consideration argues powerfully against the feasibil-
ity of the plenary assembly model and, more broadly, against any wholesale
reliance on popular referenda for determining the conduct of government.9
One response to this critique might be to suggest that apart from the
plenary assembly, there should be a body that is charged with restoring
consistency whenever plenary voting generates a problem. But this body
would have enormous discretion and power, since the resolution of an
inconsistency could take any of a number of directions: the most recent vote
might be nullified or, alternatively, any of the previous votes that are relevant
to the inconsistency might be set aside. To empower such a body in this
measure would undermine the rationale for the plenary assembly; it would
make a mockery of the claim that the assembly gives power to the people.10
8
Might it help if the group voted only on logically independent propositions that are fit to count as
premises, allowing judgements on other propositions to be determined by the votes on the premises?
No, because there is no prospect of agreement on which propositions are fit for the role of premises,
and which not. See Harman (1986).
9
Why do I concentrate on the problem of aggregating judgements, illustrated by the discursive
dilemma, rather than on the problem of aggregating preference orderings, illustrated by the
Condorcet paradox, that Kenneth Arrow (1963) made famous and by related results (Riker 1982)?
Three reasons. One, it is easier to appreciate the problem illustrated by the discursive dilemma. Two,
the Arrow impossibility can be presented as a special case of the impossibility illustrated in the results
on judgement aggregation; see List and Pettit (2004) and Dietrich and List (2007). And three, the
possibility of reflective deliberation, unavailable in a plenary assembly, also offers the only half-
satisfactory response to the Arrovian problem. For a debunking account, however, of the Arrovian
challenge see Mackie (2003).
10
This critique of the plenary assembly depends on the assumption that the members of such an
assembly are going to have to make their decisions by casting on–off votes in expression of their
The responsively representative assembly 195
2. the responsively representative assembly

Given the failure of the plenary assembly to provide a satisfactory model of
how to organize a system of popular influence, where are we to turn next? I
assume that any plausible alternative is going to require an assembly that
represents the people in some way. In taking this line, I reject the view that
representation is in any way inimical, as is sometimes said (Manin 1997), to
the ideal of democracy; like Nadia Urbinati (2006) I see that as an error with
Rousseauvian origins.
It turns out, however, that there are two distinct sorts of bodies that
might be held to serve a representative role. One of them is representative in
the indicative way in which a painting is representative of its subject; the
other is representative in the responsive way in which an attorney may be
representative of a client, or perhaps an actor representative of a character.
The rival concepts answer to the principal metaphors that were invoked in
traditional discussions of representation (Skinner 2005). In this section, I
look at the indicative and responsive assemblies in turn, argue that no body
can be fully representative in both ways at once, and then provide some
considerations in favour of going with a primarily responsive assembly.11
The indicative assembly

The first sort of non-plenary assembly is envisaged as a simulacrum or
likeness of the people that promises to satisfy the following condition: if it
decides in favour of something, then that is good evidence that the people as
a whole would make the same decision, were they capable of deliberating
judgements on particular issues. But that assumption is realistic (List and Pettit 2011: Part 1). It is not
going to be possible to pre-package all the issues to be decided, only allowing consistent packages to
come up for assembly vote, since new issues are bound to arise all the time. And when it comes to
endorsing a response to any issue, members are going to have no option but to record a positive or a
negative vote on each possible response. There is no feasible way, for example, in which they could
record a probabilistic degree of credence in each response and then look for some method whereby
their different probability distributions might be aggregated into a group distribution. There may be
abstract theories in which such aggregation can be modelled but it is entirely unclear how it might be
achieved in actual institutions.
11
For a more developed account of the responsive-indicative distinction on which the following
discussion draws see Pettit (2009b, 2010b). Jane Mansbridge (2009) documents the history of a
closely related distinction between a sanction and a selection model of representation and establishes a
range of points bearing on how the selection model may be developed. That article is broadly
congenial to the viewpoint I defend here, though it is more sanguine about the possibility of
combining responsive and indicative representation. For a different way of generalizing the concept
of representation beyond responsive representation see Rehfield (2006) and for a very useful dis-
cussion of the history and concept of representation, see Vieira and Runciman (2008).
together about it. The body is meant to be a microcosm of the society both
in the way it is composed and in the way it operates.
In order to be a compositional microcosm this assembly will have to be
statistically representative of the community as a whole, with the main
communal categories being represented proportionally within it. And in
order to be an operational microcosm its members will have to guard against
any pressures that would lead members to behave out of character: that is,
on the basis of anything other than the dispositions that make them typical
of their category. Those requirements suggest that the assembly envisaged
should comprise a random sample of the population that is small enough to
make deliberation possible and that it should be given authority for a period
that is short enough not to expose members to the warping incentives and
pressures of office. It might comprise five hundred or fewer members,
selected on the basis of chance, and it might hold power for at most a
year or two. Given the information and leisure required for public decision-
making, the resolutions of such an assembly could reasonably be expected to
be indicative of the decisions that the people as a whole would support, were
they able to assemble and deliberate appropriately.
The idea of an indicative assembly of citizens has good historical creden-
tials. It was this sort of assembly that traditional institutions like the lottery
were designed, wittingly or not, to establish.12 While the ecclesia of Athens
in the fifth and fourth centuries bce was open to all citizens, the member-
ship of equally crucial bodies was determined by lottery. Thus the judicial
panel of about six thousand citizens was fixed on an annual basis by lot and
the case-by-case membership of the courts and other smaller bodies –
including the ‘nomothetai’, who came to have responsibility for recom-
mending on any proposed change in the law – was fixed by a second, ad hoc
lottery amongst the members of the panel (Hansen 1991; Ober 1996).
A similar lottery system played a part in the constitutions of many medieval
Italian city-states, where appointments to law-making and related bodies
were often made by lot (Waley 1988; Dowlen 2008). And the lottery idea
also had an implicit presence in the medieval development of the jury
system: to be subjected to the judgement of your peers, whether in deter-
mining that there is a legal case to answer, or that you are legally liable, was
12
It is unlikely that the design was witting. There were other purposes that might have been more
prominent in the minds of traditional communities: for example, that lottery ensured against the
excessive influence of any particular clique or faction. See Dowlen (2008). My thanks to Matteo Faini
for discussion on this point.
to be exposed to a body that was meant to be indicative of the citizenry as a
whole (Abramson 1994).
There are also contemporary cases of indicative bodies being put to work
in place of the population as a whole. The British Columbia Citizens’
Assembly is a good example, though only one of many (Sintomer 2007).
A body of 160 citizens, selected on a statistically representative basis, it was
entrusted by the government of British Columbia in 2004 with producing a
recommendation on the voting system to be used in the province (Warren
and Pearse 2008). Another example is provided by the deliberative opinion
polls championed by James Fishkin (1991, 1997). These comprise fewer
numbers and meet over rather shorter periods, but are also designed to make
recommendations on public issues that reflect what the community as a
whole would recommend, were it possible for the whole population to
assemble and deliberate.
The responsive assembly

The salient alternative to having a non-plenary assembly that is a faithful,
microcosmic reflection of the people as a whole is the assembly that is
elected by the people to debate and enact laws on their behalf.13 Under this
arrangement the people would select those who are to serve for any period
on the legislative assembly in an open, competitive exercise. That exercise
may involve a single population-wide district or a number of geographically
or otherwise distinct districts. It may implement any of a variety of voting
systems in order to determine which of those candidates who stand for
office are to be successful. And it may not only allow for the election of those
who belong to the law-making assembly but also, as in the US presidential
system, for the election of the head of the administration (with or without
some share in law-making); this appointment, in other words, may be taken
out of the hands of the assembly. I abstract for the moment, however, from
such details.
Where the microcosmic body is designed to be representative in a
statistical, indicative sense, the elected body is meant to be representative
in a different, responsive manner. The members are appointed to office on
the basis of their electoral promises and may be presumed to be responsive at
13
It is quite possible for an assembly and a government to be responsive without being elected. However
appointed to office, for example, it may depend on being suitably responsive to people’s manifest
expectations for not being faced with public outrage and uproar. While this possibility is important in
looking at how far unelected governments can still be accountable to their peoples, it is not relevant to
our current discussion.
some level to what they see as the wishes of their constituents. Such
responsiveness will often be required in order to satisfy their electoral
commitments and it will certainly be required, if that is relevant, in order
to facilitate their re-election.
The members of the indicative assembly count as proxies for the people
as a whole, standing in for them on the basis of their aggregate likeness to
the people. The members of an elected assembly count more naturally as
deputies: figures whose job it is, on pain of losing power, to be responsive to
their constituents. Deputies may be required to be responsive to the more or
less explicit wishes of constituents, in the manner of instructed delegates. Or
they may be given a large degree of latitude on this front, being allowed, in
the manner of trustees, to make their own judgements on what the
presumptive interests or principles of their constituents require (Pitkin
1972). In what follows, I shall speak simply of deputies, focusing on the
contrast with proxies, and generally ignore the distinction between deputies
of a delegate and deputies of a trustee kind.
The lottery-based and election-based systems each serve to give members
of the relevant assembly the status of authorized representatives. But they
rely on quite different mechanisms to ensure that how the members are
disposed to behave connects with the dispositions of the population at
large.14 Under the mechanism at work in the indicative case, the connection
is evidential. Because the members of the assembly are selected to be
indicative of the people as a whole, the fact that they are disposed to support
this or that policy is a reliable sign that the people are similarly disposed.
The responsive mechanism establishes a causal rather than an evidential
connection. Because the members of the elected assembly depend on
responding to popular dispositions in order to be re-elected – and more
generally to satisfy electoral pressures – the fact that the people are disposed
thus and so tends to cause a similar disposition in the assembly.15
The two mechanisms are diagrammed in Table 4.2.
14
My discussion is simplified insofar as I do not look at responsive and indicative relationships in which
there is a correlation not only between the dispositions that are present in people and representatives –
for example, their beliefs in a certain domain – but also between the dispositions that are absent on the
two sides: for example, their failures to believe certain things in that domain. For a treatment in which
this simplification is lifted, and probabilities of correlation are introduced, see List and Pettit (2011:
Chapter 4).
15
The difference between the two modes of political representation corresponds to a distinction drawn
in epistemology between two modes of epistemic representation, sensitive and safe (Sosa 2007). A
responsive body will offer a sensitive representation of popular dispositions: as those dispositions
change under certain inputs, so its dispositions will tend to change too in sensitive response. An
indicative body will offer a safe representation of popular dispositions: as its dispositions change under
Table 4.2 Mechanisms of representation
People’s disposition <––––– Representatives’ Indicative

evidential dispositions representation
People’s –––––> Representatives’ Responsive
dispositions causal dispositions representation
An indicative-cum-responsive assembly
Although supporters of the French Revolution often invoked Rousseau’s
name as a source of inspiration, they uniformly went along with the idea
that any governing assembly would have to be elected. But the general
assumption was that such an electoral body would also have an indicative
status, as a microcosm or model of the whole population. Thus, in an
address to the French Constituent Assembly, Honoré Mirabeau main-
tained: ‘the representative body should at all times present a reduced picture
of the people – their opinions, aspirations, and wishes, and that presentation
should bear the relative proportion to the original precisely as a map brings
before us mountains and dales, rivers and lakes, forests and plains, cities and
towns’ (Pitkin 1969: 77). This opinion was widely shared amongst enthu-
siasts for assemblies of elected representatives. About that same time, the
American anti-federalist, Melanchton Smith, could write: ‘The idea that
naturally suggests itself to our minds, when we speak of representatives is,
that they resemble those they represent; they should be a true picture of the
people’ (Ketcham 2003: 342).
But the shift from a statistical to an electoral selection system inevitably
makes for the diminution, even the elimination, of the indicative status of a
governing assembly. The assembly that is elected may be compositionally
indicative of the population as a whole, at least if the electoral system is
designed, as John Stuart Mill (1964) thought it should be, to ensure a high
degree of proportionality. But there is every reason to think that the
assembly will fall short of being operationally indicative in any significant
measure. It is unlikely to be a body such that if it votes for a certain measure,
that is a sure sign that the population as a whole, were it able to assemble and
deliberate, would vote for that measure.
certain inputs, that is a fairly safe sign that under those inputs the dispositions of the people would
change too. Responsive representatives are reliable trackers of popular dispositions, so we might say,
where indicative representatives are reliable mirrors of popular dispositions.
The assembly could make a good claim to be an operationally indicative
status if the interests on which its members acted combined in aggregate to
model the interests distributed across the population as a whole. But unlike
statistically selected members, those who are elected are extremely unlikely
to meet this modelling requirement. The condition on which they can be
expected to be responsive representatives, as we saw, is that they will
generally want to be re-elected, or to satisfy other electoral pressures, and
will therefore seek to provide what their constituents and supporters are
disposed to welcome. But to the extent that they pursue this goal, it is
unlikely that they will mirror in aggregate the interests distributed across the
population.
Seeking re-election as they do, they will each favour measures that appeal
especially to more pivotal or marginal voters in their constituency or to
the supporters on whom they particularly depend. And equally they will
tend to strike idealistic postures that cast them in a good light as individuals –
say, moralistic postures on prostitution or taxation or crime or indeed
capitalism – rather than endorsing the compromise solutions that they
might be otherwise disposed to support (Brennan and Lomasky 1993).
A body of such electorally interested representatives is not going to act like
a faithful microcosm of the society as a whole. The point was made by
Benjamin Constant (2003: 387) in 1810. ‘You choose a man to represent you
because he has the same interests as you. By the very fact of your choosing
him, however, your choice of placing him in a different situation from yours
gives him a different interest from the one he is charged with representing.’
It is important to register that any representative body will tend to be less
indicative of the population as a whole, the more its members are electorally
responsive. But the observation does not mean that there is no point in
trying to combine indicative and responsive elements in the representative
status given to any body or indeed, as we shall see, any public office. Thus an
indicative body might be usefully forced to be responsive by being exposed
to review or challenge, if not to election; the British Columbia Citizens’
Assembly was exposed, perhaps for this reason, to a referendum on its
recommendation.16 And an electorally responsive body might be more
acceptable to a population to the extent that it satisfies a degree of propor-
tionality and can claim in some respects to be indicative.17
16
For the record, its recommendation for a change in the voting system in British Columbia received a
little short of the 60 per cent support that it was required by government decree to receive.
17
The very common practice of dividing a country into distinct districts might be taken to promote
proportionality in a geographical dimension. Is a wish for a certain proportionality behind this
practice? It is hard to believe that it did not play some role in justifying that practice but the evidence,
In favour of the responsive assembly
There is surely a useful role that indicative bodies such as the British
Columbia Citizens’ Assembly can fulfil in a contemporary democracy.
But if we are looking for a permanent assembly to govern a society, then
the electoral, responsive assembly scores decisively above it. A fairly obvious
consideration in its favour is that it allows for the same members to be
returned to office, thereby making for a continuity of experience and
expertise across different assemblies. The indicative assembly does not
promise such continuity, at least short of radical amendment, and it is
hard to see how government could operate effectively if a completely new
legislative body had to take over every year or so. But apart from this
consideration, there are three other reasons why the electoral model ought
to have much greater appeal.
The first is much cited in the literature on representative – that is,
electorally and responsively representative – democracy, including in
authors as different as John Hart Ely (1981), Joseph Schumpeter (1984),
Juergen Habermas (1995), Stephen Holmes (1995) and Ian Shapiro (2003).
A system of open, competitive, periodic election would require the satis-
faction and reinforcement of a number of basic liberties, depending for its
very operation on their successful exercise. But a system that was designed to
facilitate indicative rather than responsive representation – or, for that
matter, a plenary assembly system – would not be likely to have the same
salutary impact.
The liberties I have in mind are the freedom of speech, association and
travel that are required if people are to be able to stand for office, combine in
parties, promote their policies and expose other parties, including the
governing party, to criticism between and at the time of elections. It is
only in a society where individuals have these freedoms – and, I would say,
have them in the rich republican sense of escaping domination – that they
can hope to continue to share equally in a system of independent influence
over government. Let any of the freedoms be curtailed and the prospects for
equally accessible or individualized influence are in immediate danger.
Those who are exposed to any possibility of check or coercion in response
to speech may not speak their piece in criticism of this or that proposal, for
example. And their failure to speak will not provide default support for the
proposal, as it might have done in various contexts. No matter what the
according to Rehfield (2005), is against this hypothesis. Still, districting does induce a similarity in one
dimension – if, nowadays, a fairly unimportant one – between the population as a whole and the
legislature that represents it.
context, people’s silence will communicate nothing, since it can always be
taken to reflect their lack of freedom (Pettit 1994).18
Let any of these freedoms be curtailed and not only does the prospect of
equal, individualized access to influence decrease; so does the prospect for an
unconditioned sort of influence. We saw in the last chapter that the people’s
influence will be unconditioned, and not granted as an indulgence of the
government or the army or whatever, to the extent that the society is resistive:
as a matter of actual fact and/or as a matter of common perception, people are
disposed to resist any governmental abuse of power and government is
disposed to respond to such resistance. It is only in a society where people
actively exercise freedom of speech, association and travel – and are allowed to
do so by government authorities – that these dispositions can assume the
status of unchallenged facts, registered in the shared consciousness of the
population. And so it is only in such a society that people can have influence
over government, regardless of whether those in government – or those in any
other agency – are happy to give them influence.
Where an electoral, responsive regime would require – and as it continues
in existence, reinforce – a culture of freedom in the domain of speech,
association and travel, the statistical, indicative regime would run the risk of
failing to nurture such a culture. It might be introduced in the context of
such freedoms, but its operation would not require their continuing exer-
cise, at least outside the assembly itself. With any such freedom, it is at least
likely that if you fail to use it, you lose it: your successful exercise of the
freedom plays a crucial role in checking and confirming the absence of a
power of interference on the part of others. Thus the statistical, indicative
regime would expose a community to a very great danger that the electoral,
responsive regime would actively serve to combat.
Apart from this consideration – and the consideration of continuity in
office that we mentioned earlier – there is a second reason for preferring the
electoral, responsive regime to the statistical, indicative one. I assume, on
the basis of the argument in the last chapter, that a deliberative, democratic
assembly, be it indicative or responsive, should mediate the influence of
individuals in a particular direction or to a particular purpose; we postpone
discussion of the required direction to the next chapter. The second con-
sideration in favour of the responsively representative system is that it is
more likely than the alternative to make it possible for people to impose a
direction on government.
18
For an exchange on this issue see Langton (1997) and Pettit (2007a).
A system of influence that is required to further a given purpose may fail
in either of two ways. One involves false negatives, as they are called; the
other false positives. False negatives consist in oversights: failures to propose
or generate all the candidate policies that might promote the direction or
purpose sought. False positives consist in mistakes: failures to test and filter
out proposals that do not promote that direction or purpose, perhaps
serving an inimical end instead. This double possibility of failure suggests
that the ideal channel for popular influence will involve, first, an initial
process that guards against oversights, generating an ample supply of
candidates for consideration; second, a process that guards against mistakes,
testing the candidates with a view to filtering out those that prove unsat-
isfactory; and third, a repeating sequence in which these interact in the
identification of a final set of successful candidates. It will involve a complex
process of a familiar generate-and-test character.
In a generate-and-test process, every element generated is tested before
being allowed into the evolving product; it is incorporated in that product if
and only if it passes the test (Dennett 1996). The procedure will be familiar
from a variety of contexts. At one end of the spectrum of possible examples,
it is illustrated by the process in which authors generate texts and editors test
and filter them, so that what gets published is the joint product of both
processes. At the other end it is illustrated by the process in which random
mutation generates variations in existing genomes and natural selection tests
the variations for whether they are adaptive – whether they increase the
inclusive fitness of organisms – letting them survive only if they are.
Returning now to our two representative assemblies, it should be clear
that the electoral, responsive assembly does considerably better on this
count. The elected legislators have every incentive to search out policies
that are likely to go down well with voters, competing as they do for voter
support, so that they may be expected to generate a rich supply of candi-
dates. And when they go to the electoral polls – and when they record voting
intentions in opinion polls – the electing voters will have every incentive to
scour both the policies proposed and the policies implemented to see how
far they measure up to whatever metric they impose. Assuming that there is
a suitable metric in operation – this, in effect, is the topic of the next
chapter – the process has both of the elements required in a generate-and-
test procedure.
Things are bound to be very different, however, with the statistical,
indicative assembly – and indeed also with the plenary assembly. Such an
assembly would not offer separate locations at which policies can be
generated and tested and it would not provide actors with distinct
incentives, on the one side, to generate an ample supply of candidate
policies and, on the other, to test those candidates for how well they
measure up to relevant desiderata. It promises a calm, smooth process of
legislation, in which impartiality ought to prevail quite easily. But that
promise is much less likely to deliver people from the false negatives and
positives that threaten them than the two-dimensional, back-and-forth
dynamic that we should expect to find in the presence of electoral
competition.
The problem is made vivid in reflection on how we in a general pop-
ulation might feel about being governed by an indicative assembly. How
would we be likely to react to being governed, this year by one contingently
selected group of fellow citizens, the next year by another? We would surely
want each group to be held to account for the policies it introduces in its
period in office. It would seem utterly complacent to let an indicative
assembly act as seems best without having to face any challenge from
outside. True, the members would have to live under whatever laws they
introduced and that has always been taken as an important check in
republican thinking.19 But is it check enough? Does it provide against the
possibility of the group’s being somewhat unimaginative or somewhat
thoughtless, for example, in the decisions it makes? The answer, plausibly,
is that it does not. The absence of a group of electors to whom they are
answerable means that the members of an indicative assembly will not be
kept on their toes in the same manner as the members of a responsive
body.20
The two considerations just put forward for preferring the responsive to
the indicative assembly will carry weight, quite independently of republican
commitments. But there is a third consideration that carries particular
weight from within our republican perspective.21 This is that an indicative
body, unlike a responsive, is necessarily linked with the spectre of a certain
sort of domination. If such a body is given charge of the full range of
legislative issues, then it will have a power to determine both the issues it
considers and the resolutions that they are given. Operating as a corporate
19
Algernon Sidney (1990: 571) endorses this constraint in the seventeenth century when he writes of
members of Parliament: ‘They may make prejudicial wars, ignominious treaties, and unjust laws. Yet
when the session is ended, they must bear the burden as much as others.’ Joseph Priestley (1993: 140)
endorses it a century later when he raises and answers a question about the American colonists: ‘Q.
What is the great grievance that those people complain of? A. It is their being taxed by the parliament
of Great Britain, the members of which are so far from taxing themselves, that they ease themselves at
the same time.’ See too Locke (1960: Section 138).
20
This consideration is also relevant in assessing the use of lotteries defended in Burnheim (1985).
21
I am grateful for observations on this issue to Niko Kolodny and Jake Zuehl.
agent, it will have to form a will of its own and exercise that will under few, if
any, constraints. Not being subject to electoral challenges, however – or
perhaps to the challenges that presuppose a vibrant, electorally nurtured
culture of freedom – the assembly’s will may be difficult for the people to
control and render undominating. The assembly may have been chosen for
the prospect of forming an indicatively representative and congenial will – a
will that the population as a whole might well have formed in their place –
but it will not be forced in any way to remain congenial; it will have the
standing of a benevolent despot.
The lesson of this observation is that while we may rely on indicatively
representative bodies to make decisions on particular issues, or particular
ranges of issues, it would not be a good idea to give such an electorally
uncontrolled body the open-ended authority of a legislative assembly.
There may be no objection on republican grounds to a statistically repre-
sentative body like the British Columbia Citizens’ Assembly, which was
charged with making a judgement and recommendation on the best voting
system for their Province and on nothing more. But there would certainly
be grounds for objecting to a statistically representative body that would
have the open-ended discretion of legislature.22
Respresentative assemblies and assemblies of representatives

While these considerations argue strongly in favour of a primarily responsive
assembly, it is important to recognize that depending on how things are
otherwise organized, a responsive assembly may take either of two broad
forms. To mark the distinction in perhaps too sharp a manner, it can be
primarily a representative assembly or primarily an assembly of representatives.
In the case of the representative assembly, the aggregate decisions of the
membership are responsive to dispositions in the electorate as a whole. In
the case of the assembly of representatives, the individual decisions of
members are responsive to the dispositions of their particular constituency
or support base. In the first case, the assembly has the form of a corporate
agent that primarily tracks the electorate in the decisions it takes as a body
22
Might a non-agential collection of individuals, combining in the formation and imposition of a joint
judgement on a particular issue, or a particular range of issues, still impose a freely formed will on the
people as a whole? This is indeed a possibility but it is not a great danger, provided the group remains
unincorporated and its members do not form a will that extends over an open range of issues. But the
danger identified may still argue for guarding against the possibility of a conspiracy to dominate, as we
might describe it. And the safeguard required is obviously that any policy supported by an indicatively
representative assembly should be subject to an electoral check: endorsement by an independently
representative body or, as in the case of a referendum, by the population as a whole.
(List and Pettit 2011). In the second case, the assembly is a collection of
individuals who primarily track their own constituencies or supporters in
the votes they cast within the assembly. And in this case, therefore, assembly
decisions may not support a pattern that corresponds to any recognizable
profile of demands in the electorate as a whole; they may be the product of
ad hoc log-rolling and deal-making amongst the different members.23
This somewhat artificial distinction maps the difference, in broad terms,
between the Westminster and Washington systems. In the Westminster
system of parliamentary, non-proportional democracy, government is
generally under the control of one cohesive party or alliance of parties.24
In the Washington system of presidential democracy, government is not
subject to a similar cohesive control. Because the executive in the
Westminster system is upheld by the legislature, those legislators who
support the government have to close ranks and vote according to an agreed
programme, else the executive will fall. Because the executive is elected
independently in the Washington system, members of the legislature – even
members of the President’s own party – are not subject to the same pressure
and can operate more or less independently. Thus the main Westminster
parties can go to the electorate with detailed programmes that they are in a
position to implement, so that government policies are determined in good
measure before the election. But Washington representatives are not in a
position to do this, since government policies will be determined in one-by-
one deals between members of the two houses and the administration;
policies will be determined post-election, not pre-election.
The Westminster Parliament can be responsive to the electorate as a
whole insofar as it implements policies that have electoral support. But the
dark side of that responsiveness is that individual members of the assembly
will be able to do little in response to the demands of their particular
constituents; whether they belong to the government party or the opposi-
tion party they will be expected to vote in general on party lines. The
Washington Congress allows individual members of the House and Senate
to be responsive to the demands of their particular constituents. But again,
that responsiveness has a dark side, since the system as a whole will often be
23
In his famous account of how he thought democracy ought to work, Edmund Burke (1999) derided
this sort of assembly as a ‘congress of ambassadors’.
24
It is important that the system I have in mind here is not only parliamentary in character, with the
legislature electing the administration, but also that it is non-proportional. Where many parties have
to combine to form government, as is typical of highly proportional systems, then the system will be
more like Washington than Westminster. Thus, to anticipate the text, so many parties will be needed
to form government that policy is mainly going to be determined post-election in inter-party deals.
unresponsive to the country-wide demands of the electorate. While both
count as deputies, the members of the representative assembly are likely to
behave as trustees, to invoke our earlier distinction, and the members of the
assembly of representatives to behave as delegates (Pitkin 1972).
This is not the place to make a judgement on the merits of these two very
different systems of responsive representation, though the Westminster
system seems likely to serve republican goals more reliably. In any full-
scale assessment of democratic designs it would be essential to take account
of the differences between the systems and to judge on their specific
advantages and disadvantages, but this book does not attempt that sort of
detailed assessment. Abstracting from detailed institutional recommenda-
tions, the aim is to explore how far any democracy that is organized to broad
republican specifications can ensure the legitimacy of the state.
Remedying electoral shortfalls

This extended discussion leads us to go along with common sense in making
electoral institutions into a centrepiece of any democratic system. But I hope
that seeing those institutions against the alternatives of the plenary assembly
on the one side and the indicatively representative assembly on the other will
enable us to recognize the special merits of having a body of elected,
responsively representative legislators. Such a body ought to be capable,
unlike the plenary assembly, of exercising the reflective deliberation needed
to guard against voting up proposals that are inconsistent in logic or in
practice. And when they operate well, the institutions required for the
existence and operation of an elected legislature ought to ensure that the
society is marked by freedom of speech, association and travel, and that it
recruits ordinary electors to the interrogation of public policy.
The overall thrust of the argument in this book, as will become clear, is that
democracy is defined, not by the presence of electoral institutions, but rather by
the fact that the people exercise control over government, enjoying equally
accessible influence in the imposition of an equally acceptable direction. There
may be possible scenarios where people achieve democratic control in the
absence of electoral institutions and that already means that democracy cannot
be defined by reference to election. But it should be clear nonetheless that on
the account adopted here, the tie between the two is very tight indeed.25
25
Notwithstanding other differences on issues of constitutional design, I hope that these remarks will
signal a deep level of agreement with Richard Bellamy (2007), who also works within the republican
research programme.
Electoral institutions are necessary in practice, if not in strict logic, for
giving influence and control to the people. They force people to exercise
their freedoms in political action, whether as candidates or supporters,
critics or activists, and they thereby reinforce the enjoyment of those
very freedoms. They push political parties into exploring the space of
possible policies and they require ordinary citizens to form and express
their views in response. They constitute the catalyst that triggers and
sustains the process whereby people may hope to gain influence over
government.
But, however essential in practice, electoral institutions do not suffice
on their own to maintain a suitable system of popular influence. There are
serious problems that need to be addressed if an electoral system is to
provide the form of popular influence that is required for democratic
control in the republican sense. It is only if we can identify institutional
means for remedying such electoral shortfalls, therefore, that we can hope
to achieve democracy under the republican specification of what it
requires.
The main requirement on a system of popular influence, as we saw in the
last chapter, is that it must be equally accessible to all citizens. This parallels
the requirement, to be explored in the next chapter, that the direction it
imposes should be equally acceptable to all. But not only should the system
of influence be individualized in this sense; it should also be unconditioned,
as we put it, and efficacious. Assuming that it imposes a popular direction,
as it must do in order to constitute control, its operating to a directive effect
should not be conditioned on the willingness of the government, or any
other party, to go along; and it should not allow for the influence of private
will in public decision-making beyond what the tough-luck test would
allow.
An electoral system of popular influence raises problems on all three
fronts, since it might fail to be individualized, unconditioned or effica-
cious. I shall review these over the final three sections, making sugges-
tions about the institutional measures that might provide remedies. My
institutional suggestions are meant to be illustrative of the initiatives
that might be taken and should not be treated as definitive recommen-
dations for action; such recommendations would have to be guided, not
just by philosophical analysis, but also by empirical observation and
modelling. What the suggestions show, I hope, is that there is no reason
to despair about the possibility of setting up a suitably democratic
system of influence. All that is needed is imaginative institutional
design.
An individualized system of influence 209
3. an individualized system of influence
Individualizing electoral influence

In order for citizens to control the state in such a way that it is not a
dominating force in their lives, it is not enough for them to enjoy collective
control over those in power. Such collective control would be consistent
with the domination of individuals who fail to go along with the collectively
expressed wishes of the group. What is required, as we saw, is a system of
collective control in which individuals equally share. The influence medi-
ated by the system must be equally accessible to all and the direction
imposed must be equally acceptable to all.
Here we focus on the influence that is generated under an electoral
system of government. The question is whether an electoral system could
give all individuals the same share in determining the impact of the
influence it mediates. Under an electoral system, individuals may vote in
a referendum to determine a particular policy issue. Or they may directly
vote to elect members of the legislative assembly and indirectly elect the
administrative and judicial authorities that the assembly is responsible for
appointing. Or, as in the United States, they may even have a direct role in
electing certain members of the administration and judiciary. The question
we face is whether such an electoral process is enough in itself to ensure
people’s equal access to influence – whether it is enough to give people an
equal share in determining the impact of electoral influence.
In some systems of collective influence the marginal contributions of
individuals who make the same inputs are bound to be the same. Consider
the simple physical system of influence constituted by a group of individ-
uals pushing on a giant billiard ball – say, a sphere six feet in diameter –
where there are differences in the direction in which they wish to move it.
Let them each push in their preferred direction, assuming the optimal
position for doing so, so that the direction the ball eventually takes is
determined by the resultant of the different forces they constitute. Under
such a system of continuous influence, as we might describe it, if all
individuals contribute the same inputs – if they enjoy the same degree
of strength and they each gain the same traction on the ball – then they
will each make a difference of the same magnitude as others to the exact
trajectory of the ball; they will share equally in determining the impact of
their joint influence on that trajectory. In our terms, the system of
influence will be perfectly individualized; the influence it implements
will be equally accessible to each.
Under a system of electoral influence, the requirements for ensuring that
all individuals control the same inputs are straightforward. Every individual
should have the same voting power, access to voting should be equal for all,
and votes should be aggregated in a way that privileges no voter over others;
for short, every vote should have the same value. People’s votes will not have
the same value, and people will not have equal access to influence, if for
example the more educated have extra votes, as John Stuart Mill (1964)
recommended. Equally, people’s votes will not have the same value if it is
more difficult for some individuals to vote, due to poverty or other pres-
sures.26 And, of course, people’s votes will not have the same value if the
voting systems under which they operate are not appropriate: if, for exam-
ple, the districting system allows electorally equivalent districts to vary in
the size of population. But these difficulties are straightforwardly remedi-
able, if not always easily remedied, and we may assume that the institutions
in place do not offend on this count.27
Would the fact that people can cast equally valuable votes ensure that each
has access to the possibility of sharing equally in determining the impact?
Clearly not. Any electoral institutions will implement a system of discontin-
uous, as distinct from continuous, influence, whether in the selection of a
particular representative, in the determination of who is in government, or in
the resolution of a particular policy issue. In such a system there will always be
winners and losers, with the relevant output satisfying the wishes of the
winners and frustrating the wishes of the losers. Thus there is going to be
little or no sense in which the losers can feel that, equally with the winners,
they contributed to determining the direction in which things were moved on
26
Examples of such exclusion occur in many parts of the United States, where the papers that voters
have to produce in proof of their identity at the polls are not always in the possession of the poorer; or
where those who are convicted felons lose their right to vote, even when they have completed their
prison sentences. Such abuses argue strongly in favour of the compulsory system of voting employed,
for example, in Australia. This system imposes a small fine on those who do not register at the polls,
and have no independent excuse to offer, but it ensures a voter turn-out well above 95 per cent; more
important, again, it ensures that those who do not turn up at the polls are a more or less random
group, not a specific class. See Hill (2000) and A. Fowler’s unpublished paper, ‘Turnout Matters:
Evidence from Compulsory Voting in Australia’, for relevant background on the effects of compul-
sory voting. Under non-compulsory systems it is always possible, perhaps even likely, that the poorer
and underprivileged will be systematically ignored in political circles; see Bartels (2008).
27
Many systems give rise to this last difficulty when there is a second house, such as the Australian or the
US Senate, to which states of different sizes elect members in equal numbers. This might not be a
problem, if the second house had a distinct function from the lower house in those countries, but
other things being equal it is a problem as things actually stand and it does call for reform. Other
things may not be equal, as suggested in the next chapter, to the extent that apart from a norm of
equal influence, people in a society like Australia or the United States also endorse a norm – perhaps
for contingent historical reasons – to the effect that states should have equal senatorial representation.
any issue, whether moved in the immediate fashion of a referendum or in a
manner mediated by the decisions of elected representatives.28
The discontinuous nature of electoral influence raises a serious problem
for how we might individualize the influence mediated under an electoral
system, letting it resemble the influence achieved under a continuous
system where, as the example with the ball shows, individuals can share
equally in determining the impact of their joint influence. Individuals will
not share equally in determining the impact of any particular electoral
exercise and so it may seem that they cannot be provided with individu-
alized influence of the kind we are seeking.
But while this is a serious problem, there is a more or less salient solution on
offer. Although individuals cannot share equally in determining the impact of
any particular electoral exercise, it might be that with each of the issues
resolved in such an exercise, they have an equal chance of sharing equally in
the determination of the impact – an equal chance of being on the winning
side.29 And if they had an equal chance of winning, then we could scarcely
hold that the system of influence was not appropriately individualized.
Things might be arranged, following this proposal, so that there is no ex
ante reason to think that for any randomly chosen pair of individuals, there
will be a difference between their chances of being on the winning side on any
randomly chosen issue. This condition will be satisfied under a system where
each person has the same vote, and each vote has the same value just so long as
one further proviso is fulfilled. This is that the issues over which they vote are
ones where there is no ex ante reason to think – no reason grounded in their
identity or character, as that is determined independently of their disposition
to vote – that some are bound to vote for a particular side. But this proviso,
alas, is not always going to be satisfied. It will flounder on a problem that has
become famous as the problem of an ‘elective despotism’ or a ‘tyranny of the
majority’ (Madison, Hamilton and Jay 1987: no. 10).
The tyranny of the majority

Suppose that a group of five people meet weekly to play cards or chess and
dine together afterwards at one of two local restaurants, one specializing in
28
The only sense in which they might each make a contribution is in determining the size of the
winning majority – and in some circumstances, the size of that majority is politically significant. See
Guerrero (2010).
29
Why not require that they have an equal chance of being decisive? Mainly because the chance of any
individual’s being decisive in a majority vote among a large number is close to zero. I am grateful for
comments on this point from Niko Kolodny.
vegetarian food, the other not. And suppose that with a view to giving each
equally accessible influence over the choice of restaurant, they let the choice
be determined by majority vote on every occasion. This system will work
well to the extent that the majority rotates across weeks – to the extent, in
other words, that the majority is mobile. But it will work very badly if there
are two people, say two vegetarians, who invariably vote for a vegetarian
restaurant while the other three invariably vote for the other. In the case of
such a sticky divide, the system of majority voting will continue to give each
an equal vote but the voters will not have an equal chance of being on the
winning side. The stickiness of the divide between majority and minority
means that equal votes are not enough to ensure equally accessible influence
for the different voters; those in the minority are more or less permanently
locked out of the possibility of being amongst the winners.
Much has been written for and against various voting systems, whether as
systems for the election of representatives by the electorate or for decision-
making in a legislature or in a referendum. But whatever voting system is
introduced in such a context there is always a possibility that for similar
reasons it will not provide people with an equal chance of being on the
winning side and so not ensure the equal access to influence that republican
theory would require. On one or another range of issues there may always be
a more or less sticky divide between a majority and a minority and, if there
is, then on that range of issues people will not enjoy equal access to
influence, not having the same ex ante chance of being on the winning
side; the patterns of electoral or legislative voting may shut out the minority.
Those of a minority religion are going to be in this situation with an issue
like whether to set up the majority religion as an established church; those of
a minority culture with an issue like whether to authorize only a mainstream
pattern – say, the use of the majority language – in public life; and those of a
minority sexual orientation with an issue like whether to give full civic
recognition only to heterosexual unions.
The problem of the sticky divide is that there are independently identi-
fied individuals who, on certain issues, are more or less bound to be on the
losing side. It is important that the individuals who are bound to lose are
independently identified – say, identified on the basis of creed or colour,
race or sexual orientation – not identified just by their disposition to vote a
certain way on those issues. Otherwise we would have to say that just by
being unlucky enough to choose minority views on various issues people
would be deprived of access to equal influence. Otherwise, indeed, we
would have to say that just by being contrarian enough always to choose
views that look likely to be in the minority, people would be deprived of
access to equal influence.30 There will be no problem if individuals who are
ex ante as likely to choose one as another side on given issues happen in
general to choose the losing side or aim in general to choose the losing side.
The problem is confined to the case where there is ex ante reason, associated
with their independently fixed identity, to think that certain individuals will
be in the minority on given issues: their identity pre-commits them, as we
might say, on those issues.
How serious is this sort of problem likely to be? Our interest in having
individuals share equally in the system whereby the people influence
government derives from the fact that otherwise they will fail to share
equally in control of government and will be exposed to a degree of
discrimination and domination at the hands of those in power. The prob-
lem raised will certainly be serious if a sticky minority is shut out from
having equal influence on any issue of significance: for example, any issue
where being shut out means that they do not share equally with others in
determining the basic liberties that all are to enjoy, or do not share equally in
controlling the sort of resourcing or protection that is secured by the state
for those basic liberties. When members of a minority group have an
identity fixed by something like creed, or colour, or race, and there is an
issue to be decided by majority voting on which that identity pre-commits
them, then that issue is very likely to have such significance. And so the
inequality of influence suffered by a sticky minority is bound to be of
concern (Guinier 1994).
Beyond the tyranny of the majority

How to respond to the problem raised by sticky minority–majority divides?
I see only one realistic possibility, which is to introduce a system of
individualized contestation that parallels the collective challenge that elec-
tions make possible.31 There ought to be openings for particular individuals
and subgroups to test the laws or proposals for how far the process in which
they are generated respects the value of equal access to influence and, more
generally, the value of equal status. And those openings ought to hold out
30
There is a scope ambiguity, as logicians describe it, between the characterization of the case where
there is a sticky-divide problem and the characterization of the case where there isn’t. In the former
case we say that there are individuals such that there are issues on which those individuals are in a
minority; in the latter we say that there are issues such that there are individuals who are in a minority
on those issues.
31
Electoral challenge can be seen, of course, as a form of collective contestation; on this point see
McCormick (2011: 152).
the prospect of an impartial judgement on the question and, if the judge-
ment goes in favour of the challenges, the prospect of an adjustment that
satisfies them.
Under plausible assumptions about our little group of five, the members
will be easily granted such individualized, potentially effective contestation.
The vegetarians will be free to point out the effect of majority voting and to
argue on grounds of equal access to influence for an adjustment to the
pattern. Here there is an obvious alternative: to let the members take turns
in dictating where they are to eat (Risse 2004). The ethos of any plausible
group is likely to support the egalitarian objection, and to prompt the
adoption of the alternative mode of decision-making.
As with the vegetarians in this example, people whose religious or cultural
identity pre-commits them to taking a particular side on certain issues can
enjoy equal influence over government only insofar as they are able to
contest the appropriateness of majority voting in a referendum or via a
legislature for determining such issues. They have to be able to make the
case for why majority voting on those issues ensures that they do not share
equally with others in controlling for the direction taken by government.
And they have to be able to expect that the case they make in contesting the
law or proposal receives an impartial hearing and is dealt with appropriately.
Allowing for the contestation of the results of a majoritarian voting
system – or of any comparable system of individual inputs to aggregative
decision-making – certainly makes things messy. But it is hard to see how
anyone who is concerned with giving people equal access to influence over
government could disallow it. Jeremy Waldron (1999a, 1999b) offers power-
ful support for majority voting on the grounds that it does better than
alternatives in the promotion of equal influence. But how could we invoke
the value of such equality in the exogenous design of a voting system
without allowing that if there is ground to argue that the system is not
actually delivering equality, then voters should be able to argue for one or
another amendment to the system (Beitz 1989)? To deny voters the right to
invoke the value of equality in critique of a system that is chosen for its
promise of delivering equality would be to fail to honour that very value.32
32
Waldron (2006: 1364) takes a more congenial line in arguing that although the members of a
democratic society, on his assumptions, ‘believe in majority rule as a rough general principle for
politics, they accept that individuals have certain interests and are entitled to certain liberties that
should not be denied simply because it would be more convenient for most people to deny them.
They believe that minorities are entitled to a degree of support, recognition, and insulation that is not
necessarily guaranteed by their numbers or by their political weight.’ I am grateful to Caleb Yong for
drawing this to my attention.
When a system of influence allows individuals to invoke in criticism of its
operations the very equality of influence it is designed to promote, then it
counts in an established term as a recursive procedure (Benhabib 1996). It
exemplifies a system that may be put into operation, not just in shaping
distinct, external issues, but in shaping the very form that it assumes itself.
To canonize majority rule – to give it sacred and unquestionable status –
would be to deny the possibility of recursion and, inevitably, the possibility
of reforming the system itself. To allow majority rule to be subject to
contestation and amendment, and to keep the possibility of continued
contestation and amendment alive, is to keep future developments open.
Institutional implications
In analysing how a representative assembly might operate – a responsively
rather than an indicatively representative body – I abstracted from the
precise details of the electoral system that ought to be introduced. This
argument for allowing the individualized contestability of the assembly has
been put forward under a similar abstraction from institutional detail. But it
should be clear, in broad outline, what sorts of measures are likely to be
required. A system that allows potentially effective, individualized contest-
ability will have to provide, not just for contestability proper but,
pre-contestation, for transparency in the decisions contested and, post-
contestation, for impartiality in resolving the charges raised. The require-
ments are as follows.
* Transparency: the capacity of members of the society to know what
proposals are under consideration and what measures have passed.

* Contestability: the opportunity for members to challenge overtures both
in advance of legislation and after they have passed into law.

* Impartiality: the availability of forums in which challengers can expect an
impartial assessment and, ideally, resolution.

What these measures require at a more specific level is an issue for
detailed institutional design, as with the more specific requirements in a
districting or voting system. According to standard thinking, often influ-
enced by the long republican tradition, transparency will be enhanced by
forcing the legislative assembly to debate its measures in public; to give
recognition and exposure to the minority of legislators who constitute an
opposition; to impose constraints on themselves, and on all public
authorities, to justify their decisions in public; to ensure the existence of
independent, potentially critical media; to allow for the formation of civic
watchdog bodies that can track what is done in government; and so on.
Equally, according to this traditional body of thought, the opportunity
for challenge argues for opening up channels of consultation and appeal
between the public and the legislature; for establishing a system of courts
and other tribunals in which challenges to laws can be heard and decided;
for appointing independent officials and bodies in ombudsman or auditor
roles, enabling them to raise challenges in the name of individuals; for
recognizing constraints of rule of law and due process that authorities ought
to be held to account for breaching; and for allowing the sorts of public
meetings and demonstrations that may be necessary to muster more radical
challenges to the status quo. Only in the presence of arrangements like this
will the government be contestable.
And finally, so traditional wisdom goes, the need to ensure an overtly
impartial system of adjudication and resolution makes a case for appointing
the members of courts and tribunals, and the officials on ombudsman or
auditor bodies, under constraints that give them general credibility; for
establishing the independence of those authorities; and for providing plain-
tiffs with the opportunity to launch appeals against the decisions of those
authorities themselves. This requirement of impartiality will only be capa-
ble of being fulfilled, of course, in a community where people are genuinely
willing to live on equal terms with all others, even those in a perhaps
recalcitrant minority. If such a spirit of accommodation and compromise
is lacking, then there is very little hope of achieving the ideal envisaged here
(Bellamy 1999; Margalit 2010). The best prospect for satisfying the ideal
may be to allow the society to divide into two or more states or to organize
itself as a federation of relatively independent regions.
How might issues of equal access to influence be resolved as a result
of contestation? One resolution might be to find an alternative mode of
decision-making, parallel to the resolution of the problem with our little
club, which does not offend in the same way as majority voting. This sort of
resolution might take any of a number of forms, ranging from giving
members of the minority a special weight in the decision-making to relying
on indicative assemblies or deliberative opinion polls to advise or decide on
the issue, to creating special, presumptively impartial commissions to
adjudicate on the matter, to establishing target quotas that ought to be
satisfied in any area of decision-making.
Another resolution would be to take the issue out of the domain of
popular voting altogether. Such a response would disallow the use of regular
voting procedures to make the majority religion – or on similar grounds any
minority religion – into an established church. And more generally, it
would disallow the use of such voting procedures to give any significant
privileges to those who belong to a more or less fixed majority – or, of
course, to those who belong to any fixed minority. In effect, it would
preserve the possibility of people’s sharing equally in influencing and
controlling government by denying government authority over issues
where the equal sharing of influence and control is just not possible.
There is an obvious danger that a system of individualized contestation
may prove unworkable, creating a build-up and backlog of endless com-
plaints. This argues for establishing standardized ways of dealing with
different types of issue where people’s equal access to influence is generally
agreed to be in danger. Those standardized responses might authorize
appeal to special decision-making forums or routines of the kind at which
we gestured. Or they might establish individual rights against majority-
based decision-making. Such rights would have to be laid out in a formal or
informal constitution that was not itself subject to simple majoritarian
amendment. The effect of constitutionally protected rights would be to
put various issues off the popular agenda, as with familiar measures that
disallow majority voting on whether to set up an established religion, on
whether to give special rights to heterosexuals, on whether to privilege those
in a particular ethnic group, and the like. The intent of the protection, as I
interpret it here, would not be to put brakes on the influence of the people,
compromising democratic process. Rather it would be to ensure that the
democratic system satisfies two widely recognized, democratic desiderata:
on the one side, that people should have truly equal access to influence; and
on the other, that the channels through which popular influence flows
should not be allowed to clog up (Ely 1981).33
Serious exercises in institutional design are well beyond the scope of this
book and these comments are not meant as contributions to such a
project.34 I make them only to illustrate in a broad way the sorts of processes
that are going to be mandated once we recognize that people’s equal access
to influence over government requires the introduction of procedures for
33
I assume that a system of rights can be made to work productively, without cramping democratic
initiative (for such concerns see Schwartzberg 2007; Tully 2009).
34
The fact that I avoid making any detailed institutional proposals may suggest that in the perspective
adopted here, the status quo in most democracies is pretty well OK. I do not support that suggestion.
I believe that there is much work to be done in theoretically elaborating and empirically testing the
institutional requirements of republican ideals and that this work may support quite radical measures,
including measures of which we may currently have no inkling. Take a proposal, then, like that which
John McCormick (2011) makes for a people’s lottery-selected tribunate. It would be a serious mistake
to think that because I do not endorse it here, there is a deep rift between the radical republican
democracy for which he looks and the image of republican democracy developed in this book. There
is no necessary divergence between us, at least not on this ground; I am conducting an argument at a
level of abstraction that makes it impossible to explore such proposals.
the individualized testing of any policies generated and of any processes
used in the generation of policies. It requires giving people an editorial, as
well as an authorial, role – a role in testing as well as generating policies –
and, in particular, an editorial role that has an individualized and not just a
collective dimension.35
4. an unconditioned system of influence
The desideratum
We assume that the influence people have under an electorally centred
system is supposed to impose a certain direction on government; this
assumption will be at the centre of concern in the next chapter. It is
important to recognize that people might be able to use their electoral
and contestatory influence to elicit such a direction on the part of those in
authority, without this system of influence being unconditioned or inde-
pendent. They might be able to prescribe for such a direction only because
those in power, wanting perhaps to think of themselves as democrats, like to
let their policies correspond to popular demands. Or they might be able to
prescribe for the direction only because some background power such as a
military service or a moneyed elite or an imperial overseer happens to favour
government policies that are popularly satisfying.
If the people had a directive influence over government only by this sort
of courtesy or connivance, then that influence would not give them control
or power over government. Control requires that the correlation between
popular demand and government supply should be robust across variations
in the will of government or in the will of any third party. Let the correlation
be contingent or conditioned in this way, and it will be the agency on which
the correlation depends that enjoys power or control, not the people whom
that agency chooses to benefit.
One of the main concerns that recurs in the history of Italian–Atlantic
republicanism is the fear that the independence of the people as a source of
directive influence on government is going to be very difficult to maintain.
That independence will be jeopardized under two conditions, so Richard
Price (1991: 78–9) argued. One is when a foreign state allows a people to
govern itself but reserves the right to suspend the arrangement so that it is
‘liable to be altered, suspended or over-ruled at the discretion of the state
35
For earlier treatments in which these two dimensions of democracy are emphasized see Pettit (1999,
2000a, 2000b).
An unconditioned system of influence 219
which possesses the sovereignty’. And the other is when the representatives
of the people operate under a ‘higher will which directs their resolutions,
and on which they are dependent’. In this condition, as in the first, govern-
ment acts under the influence of the people but because that influence is
conditional on the acquiescence of a higher will, the government is subject
to ‘no controul from the people’.36
What measures would be likely to give an independent, unconditioned
character to the system of popular influence available under an electoral and
contestatory regime? I ignore the problem on the international front, since it
would take us into the domain of international relations. The question we
confront is, what measures might guard the system against exhibiting a
domestically conditioned variety of popular influence: a sort of influence
that remains in place only to the extent that government, or some collateral
power, is content for it to remain in place?
In discussing this issue in the last chapter we saw that the only assured
safeguard is a resistive community: that is, a community in which, as a
matter of fact and/or common belief, people are disposed to resist govern-
ment, should it ignore popular influence, and government is disposed to
avoid triggering resistance. Just the presence of an electoral system, even one
that is adjusted in a contestatory way so as to ensure individualized influ-
ence, is obviously not enough to guarantee such a resistive culture. So what
initiatives might we take to promote this desideratum?
36
The concern with being subject to foreign control was central to the complaints of the American
colonists after the repeal of the Stamp Act in 1766. While the repeal seemed to recognize the right of
colonial subjects – or at least propertied, mainstream males – to govern themselves on a more or less
popular basis, it came with a sting in the tail. The Westminster Parliament did indeed withdraw an act
that had seemed to violate that right but it insisted at the same time, in the so-called Declaratory Act,
that it did so as matter of grace, not because of any obligation. Parliament, so it was declared, ‘had,
hath, and of right ought to have, full power and authority to make laws and statutes of sufficient force
and validity to bind the colonies and people of America, subjects of the crown of Great Britain, in all
cases whatsoever’. See the unpublished paper by J. Rakove, A. R. Rutten and B. R. Weingast, ‘Ideas,
Interests, and Credible Commitments in the American Revolution’, and, for theoretical background
to that paper, Weingast (1997).
This sort of concern was widespread in that period. Thus in 1782 the Westminster Parliament was
pressed into repealing a law of 1720 under which the Dublin Parliament – not, of course, a very
representative body – could have its resolutions nullified in London; this law had been designed, in its
own words, ‘for the better securing the dependency of the kingdom of Ireland on the crown of Great
Britain’ (Stewart 1993: 26). But critics protested that repeal was not enough, since it did not entail a
renunciation of Westminster’s right to overrule them. And then, perhaps because of the lesson
learned in America, Westminster responded by passing such a renunciation in 1783. See Stewart (1993:
Chapter 4). Perhaps it was for that reason that in seeking to end home rule in 1800, London did not
suppress the Dublin Parliament but induced its members, largely on the basis of bribes, to dissolve the
body.
I propose that two absolutely crucial measures for promoting it correspond
to the two institutional elements that characterize Italian–Atlantic republican-
ism: on the one side, the mixed constitution and, on the other, the contestatory
citizenry. If the constitution is mixed, then government is likely to be resistance-
averse, and if the citizenry has a contestatory character, then the people are
likely to be resistance-prone. The society, in short, is likely to be resistive.
The individualization of popular influence already argues for a degree of
mixture in the constitution, for example in the separation of adjudicative
power from other forms of power. And the individualization of popular
influence requires people to be contestatory in the assertion of their individual
and minority claims against majoritarian neglect. The best hope for making
popular influence unconditioned or independent is to push for further devel-
opments on these fronts. In outlining the developments needed, as I now go on
to do, I shall assume that the requirements for an electoral system in which
influence is individualized are already in place. The need for transparency,
contestability and impartiality has already been institutionally satisfied.
The mixed constitution

Jean Bodin and Thomas Hobbes inaugurated the assault on the mixed
constitution, arguing that there could be no functionally satisfactory state –
no state capable of effectively formulating and enforcing law – in the
absence of a single, absolute sovereign, be that sovereign a monarch, as
they each preferred, or an aristocratic or democratic assembly. This assault
should not be surprising, given the concern of each with having an abso-
lutely sovereign government that is capable of establishing peace and order
in a world of deep religious and civil conflict. They each saw that the price of
having such a sovereign was the exposure of the people to the whim of that
individual or body. But while they thought that it was desirable for the
sovereign to abide by customary and other constraints, and not to act on
whim, they were each prepared to pay that price. They each held that a
sovereign could breach such constraints without ceasing to be sovereign.
Thus Bodin (1967: i.8.26) held that, notwithstanding ‘manifest tyranny’,
the ‘tyrant is a true sovereign for all that’, and Hobbes (1994b: Review and
Conclusion) argued that ‘the name of tyranny signifieth nothing more nor
less than the name of sovereignty, be it in one or many men’.37
37
One way of marking their position is to say that they were positivists about sovereignty. Just as
positivists about law argue that even bad law – even law that breaches natural law – should still count
as law, they argue that even the bad sovereign – even, for example, the tyrannical sovereign – should
Bodin and Hobbes adopted an antagonistic stance towards the mixed
constitution, because they saw quite clearly what that arrangement, as it was
characterized and celebrated by Polybius (1954), Cicero (1998) and
Machiavelli (1965), was designed to do. It was meant to ensure that the
state could not have the sort of power that would allow authorities to
impose their own, arbitrary will on the people they govern. And so it was
straightforwardly inconsistent, as Bodin and Hobbes saw it, with the sort of
role that they envisaged for sovereign and state.
While traditional defenders of the mixed constitution present it in
received terms as a mixture of the three pure constitutional types, mon-
archy, aristocracy and democracy, this rhetorical trope serves to encode
straightforward institutional constraints. They divide into constraints asso-
ciated with any constitutional order, on the one side, and constraints
associated with the mixing of powers on the other. The constitutional
constraints require that government should operate in accordance with
due process, not ruling by ad hoc decree but via public, general and
prospective regulations (Fuller 1971; O’Donnell 2004). They impose what
has come to be known as a rule of law under which, as Cicero (1998: 151)
expressed the ideal, ‘a magistrate is speaking law, and law a silent magis-
trate’. The mixture constraints require something designed to be supportive
of such a rule of law: that different governmental powers be held in different
hands; that those different powers should be shared out amongst different,
mutually checking agents or agencies; and that the centres of power should
be designed to give all sectors of the people a fair presence or representation
in the exercise of power. To put the lesson in a slogan, the mixture of the
mixed constitution requires a separation of powers, a sharing of powers and
a balancing of powers.
What are the powers of government that should be separated under the
mixed constitution (Gwyn 1965; Vile 1967)? The answer that became
standard under the influence of the Baron de Montesquieu (1989) is, the
powers of legislation, administration and adjudication. But that may be
over-inclusive in one way, under-inclusive in another. It may be over-
inclusive in calling for the separation of legislation and administration,
since the experience of Westminster systems of government is that there
is no real danger – and there may be some real benefit – in not separating
still count as sovereign. And just as positivists take law to count as law in virtue of serving a certain
function – for example, that identified in Hart (1961) – so they take the sovereign to count as
sovereign in virtue of serving the function they identify as crucial: that of establishing peace and order
in the society. For some cautionary remarks about the interpretation of Hobbes on this point, see
Hoekstra (2001).
those branches. And it may be under-inclusive in failing to call for the
separation of different branches of the armed and police forces, the separa-
tion of secular from religious authorities, and indeed, the separation of
centres of political power from those in control of commerce and business.
Such divisions are liable to be just as important as more formal devices in
guarding against the abuse of public power.38
How should any power be shared under the mixed constitution?
Bodin’s early critics pointed out that there were many ways in which
law-making might be shared amongst different individuals or bodies, or
divided out amongst them (Besold 1618: 279–80; Cabot 1751; Franklin
1991). The same remains true today. Thus the legislative power might be
dispersed by giving the right to propose laws to one body or official, the
right to decide on the proposal to another; by having two or more
legislative houses, as in standard bicameral systems; by exposing the
power of those houses to the constraint of the courts, as the individual-
ization of influence requires; and/or by giving a formal or effective veto
to the administration over what laws are to stand or be enforced.39 This is
not the place, however, to discuss the relative merits of such arrange-
ments or the virtues of combining them in different ways.
The balancing of powers complemented the required separation and
sharing of powers in the traditional image of the mixed constitution
(Richter 1977). The constitution was to combine the inputs of nobility
and commons, as in the traditional picture, and thereby ensure that no one
was excluded from influence. This requirement was interpreted in a
demanding, democratic spirit by writers like Machiavelli (1965) in the
sixteenth century – see McCormick (2011) – and by radical commonwealth
writers in eighteenth-century England (Lieberman 2006). That interpreta-
tion emphasizes that popular acquiescence in how a regime operates is the
ultimate guarantor of a constitution and that this gives the citizenry as a
whole a special role in the maintenance of the regime.
38
For an original and congenial development of this encompassing view of the separation – and also the
sharing and balancing – of powers, see Braithwaite (1997). And for a consciously republican
application of the view in the analysis of policy-making in an emerging democracy, see
Braithwaite, Charlesworth and Soares (2012).
39
Thus it might be worth noting that Athens shared out legislative power between the ecclesia, the
courts and the law-amending body, the nomothetai (Hansen 1991; Ober 1996) and that among other
measures Rome had distinct law-making bodies such as the tribal and centuriate assemblies (Millar
1998). This particular Roman arrangement creates different possible sources of law where most
arrangements create different possible veto points. For a nice discussion of the futility of looking
for which institution was really the sovereign in ancient Athens, see Ober (1996: 120–1).
Bodin and Hobbes and like-minded opponents of the mixed consti-
tution would not have objected greatly to the separation of powers as
such, since they were happy to allow the administration of government
to be delegated to other hands. And neither would they have railed
against the balancing of powers, even interpreted radically, since they
allowed that an inclusive, majoritarian democracy was a legitimate, if not
a desirable, regime. Their real opposition would have been to the sharing
of legislative power, since it appeared, particularly in combination with a
sharing of administrative and judicial powers, to ensure that government
would be unable to close ranks and assert its authority in unruly and
divisive times.
From our point of view, of course, it is good that government should be
unable to close ranks in this way. To the extent that the agencies of
government are unable to make common cause against popular pressure,
they will be more disposed to try to avoid triggering popular resistance. Any
one governing agent or agency may be tempted to challenge popular feeling
and to face down popular resistance. But the very fact that the state acts only
when its component bodies all come into alignment – and that it is forced to
act under the constraints of a rule of law – means that the government as a
whole is less likely to be disposed to display such bravado. That government
will not be as ready as any individual part might be to ignore or push back
against popular resistance.
In concluding this discussion of the mixed constitution, it may be useful
to point out that apart from being institutionally worrisome, the absolutist
critique launched by Bodin and Hobbes – and supported in different ways
by Rousseau and Kant (Pettit 2012b) – is philosophically misconceived.
Bodin and Hobbes assume that in order to be a source of coherent law the
state has to perform as an agent or agency with a single mind – a person,
they often say – and that in order to be a source of effective sanction it has to
have absolute power. The first claim is the most relevant in the argument
against the mixed constitution. Building on the assumed need for an agent
or agency with a single mind, they each maintain that this would be
unavailable if the constitution allowed for independent centres of power,
however coordinated. ‘It is impossible that the commonwealth, which is
one body, should have many heads,’ Bodin (1967: vi.4.198) writes. If it were
organized in that way, Hobbes maintains, then we would have ‘not one
independent commonwealth, but three independent factions; nor
one representative person, but three’ (Hobbes 1994b: 29.16). And, as we
saw in the Introduction, Rousseau (1997: ii.2.2) takes up the theme with
relish, arguing that under the mixed constitution the sovereign is turned
into ‘a being that is fantastical and formed of disparate parts’.40
We can agree for reasons already rehearsed that it is essential, as all
thinkers assume, that the state or commonwealth should operate like a
single agent or person in delivering and applying law. To be an agent in the
appropriate sense is to be an individual or body that can recognize demands
like those of consistency and prove responsive to them. It is to be susceptible
to the constraints and challenges of reason in the adoption of ends and in
the formation of judgements as to how best to promote those ends. These
critics argue against the mixed constitution because they assume that in
order for a state to assume such an agential or personal status, there has to be
a spokesperson available to speak for it. This spokesperson may be a king or
queen, they say, or an aristocratic or democratic assembly that operates
under majority rule. More generally, it has to be a spatio-temporally
concrete entity, not an entity that exists on the basis of how different
individuals or bodies operate and coordinate. But this is sheer dogma.
There is absolutely no reason why the state should not be a distributed
agency that is answerable to the demands of reason, like any agent, but
answerable in virtue of the rules of coordination under which distinct
component parts cooperate (List and Pettit 2011).
We saw earlier that as a matter of fact an assembly that operates blindly
under majority rule won’t be able to live up to the demands of reason and
agency and that for every candidate, law or policy generated there should be
a test to determine if it is likely to induce inconsistency. This testing might
clearly be done under a system of mutual checking between different houses
in a legislature and under a constraint of engagement between those houses
and a constitutional court. Such an arrangement, representing a simple
form of mixed constitution, would give a voice to the people, as Bodin and
Hobbes require. But that voice would emerge from the interaction between
different bodies in the complex whole; it would not be the voice of a single
body, authorized to act as spokesperson for the whole.
Why did Bodin, Hobbes and Rousseau overlook this possibility? Apart
from the fact that it suited their political purposes to neglect it, they may
well have been moved by what A. N. Whitehead (1997: 51) calls the fallacy of
misplaced concreteness. This fallacy might have led them to think that if
40
The theme routinely appears also in the long tradition on which Rousseau had a shaping influence.
See for example Hegel’s (1991: 308) critique of the idea that the different powers in the state should be
independent: ‘it is plain to see that two self-sufficient entities cannot constitute a unity, but must
certainly give rise to a conflict whereby either the whole is destroyed or unity is restored by force’.
there is a mind in the commonwealth then this corporate mind must be
located in a single place within the commonwealth: in the head of the
monarch, quite literally, or in the forum where the assembly meets. But the
purposes and judgements on which the state acts – the mind that informs its
behaviour towards citizens and towards other individuals and bodies – may
emerge from the interaction of distinct, separated parts within the state.
They do not have to be located in any one of these concrete parts.
Gilbert Ryle (1949) tells a story about the visitor to Oxford who is taken
around the colleges and, thinking that it is something equally concrete, asks
also to see the University. What the visitor fails to register is that the
University exists and performs as a unified agency in virtue of the way
that the colleges interact; it exists in and through them, emerging as a
superordinate entity, rather than consisting in something coordinate and
additional. What is true of the University in relation to the colleges can be
equally true of the state in relation to its constituent parts. In order to have a
state that speaks with one voice and displays one mind, it is not necessary
that one of those parts should be sovereign. The state may be a super-
ordinate entity that exists on the basis of a sustained, disciplined form of
interaction between the parts. That is the possibility that critics of the mixed
constitution wittingly or unwittingly neglected.41
The contestatory citizenry

There are two complementary requirements for making popular influence
suitably unconditioned or robust. First, as we have just seen, there must be a
mixed constitution in place that disposes government to avoid triggering
popular resistance. And, second, there must be a contestatory culture
amongst the citizenry that disposes people to resist any governmental
abuse. These two elements must not only be present in order to trigger
suitable responses, they must be present as a matter of common awareness; a
matter manifest to all. But this extra condition is likely to get fulfilled as a
matter of course. It is hard to see how a mixed constitution and a contest-
atory citizenry could be present without its being manifest that they are
present.
41
In contemporary philosophy of mind, the critique of the Cartesian idea that the mind or soul of the
human being condenses in a particular entity – the res cogitans or ‘thinking thing’ – indicts Descartes,
so we might say, for having fallen also into the fallacy of misplaced concreteness (Dennett 1992; Clark
1997). There is a nice parallel between concretist assumptions about the mind of the individual
subject and concretist assumptions about the sovereign in the state.
Taking up the need for a contestatory citizenry, we should note that the
individualization of power already requires people to be disposed to contest
laws and policies that seem to undermine their equality of influence. But the
need to ensure that popular influence is not conditioned on the goodwill of
government, or that of any third party, supports the requirement independ-
ently. People must be on the watch for proposals or measures that are not
suitably supported – more on this in the next chapter – and they must be
ready to organize in opposition to such policies. It is only in the presence of
concerted, sustained oversight of government activity – only in the presence
of civic vigilance, in the old term – that we can have any assurance that
government will be forced to remain responsive to popular inputs.
But what is required to ensure an appropriate form of vigilance? The
freedoms of speech, association and travel that any effective electoral system
presupposes must be registered in common consciousness. And citizens
must regularly exercise and put those freedoms to the test in an active,
engaged style of politics. While not everyone need be an activist, vigilance
requires a high aggregate level of civic engagement. Numbers of people
must be there to manifest an interest in every initiative of government and
numbers must be there to insist on the government’s justifying the initia-
tives it takes. Democratic life, as it is sometimes put, has to have an
agonistic – better perhaps, an antagonistic – character.
Is this ideal romantic rather than realistic? It would be romantic if it
required political activists and public-interest groups to keep individual tabs
on everything, as in some traditional images of civic virtue (Montesquieu
1989). But contemporary states are too complex to allow such panoramic
scrutiny and interrogation. What is needed, obviously, is specialization and
organization: in short, a division of labour in the exercise of civic vigilance.
And that ideal is scarcely unrealistic, since contemporary democracies natu-
rally give life to watchdog, activist bodies – non-governmental organizations –
that operate locally, nationally and internationally across the various
domains of political life. These include bodies that specialize, for example,
in consumer issues, people’s working conditions, women’s rights, environ-
mental sustainability, racial equality, opportunities for the disabled, the
conditions of prisoners, gay and lesbian rights, health provision and public
education.
The specialization in civic vigilance and contestation that public-interest
bodies make possible is desirable on two fronts. It does not only make it
possible to have government more effectively invigilated by the citizenry
than would be feasible under a regime of panoramic monitoring. It also
makes it possible to recruit people to the exercise of civic vigilance on the
basis of their particular concerns and passions; recruitment does not have to
appeal merely to the abstract call of virtue. There will be no problem with a
civic vigilance that is exercised on the basis of such particularistic enthusi-
asm and attachment, provided that it appeals to a public standard like the
ideal of equal influence and that appeal is made in a public forum. It can be
sourced in the divisions and resentments emphasized in the long republican
tradition by Machiavelli and by contemporary writers who identify with his
themes (McCormick 2011). It can be nourished just by what Adam
Ferguson (1767: 167) described as the ‘refractory and turbulent zeal’ of
ordinary people.
Not only does the activism envisaged here reject the romantic idea of
each citizen’s exercising a panoramic, altruistic form of oversight; it also
rejects the equally romantic idea of a participatory, Rousseauvian engage-
ment. The participatory approach is so other-worldly that it is likely to
demoralize activists, not inspire them: its ‘insistence on the identity of the
ruler and the ruled has the effect’, as one author puts it, ‘of removing
democracy from the pages of history and of restricting it to an ideal that
has never before been realized, not even in ancient Athens’ (Green 2004:
748). The engagement that is required to make popular influence robust is
rather the activity of the radical social movements that offer an account of
common concerns, articulate a suite of popular demands, and challenge
government for its failures to recognize or reflect those demands in its
policies (Young 1990; Honig 2001). While it gives people an active part in
the political system, it does not pretend that politics is a matter of the people
coming together in a grand, will-forming, law-making exercise.
But if the contestatory model of the democratic people is not romantic in
character, neither does it have the cynical cast assumed by political realists.
Thus it contrasts just as sharply with views according to which democracy
requires the stabilizing effect of widespread apathy and, as by a happy fault,
actually generates such apathy (Lipset 1960). If democracy is taken to consist
in nothing more than the electoral assignment of governmental responsibility
to a particular party or group, then it may operate best if there is little
turbulence amongst the population. But if it consists in the individualized
and unconditioned control of government by the people, then it must wither
away in the presence of popular apathy. Let people allow themselves to
become apathetic and the business of government is bound to be consigned
to hands on which they have no influence and can impose no direction.
Calling for contestatory vigilance is not redundant, then, any more than
it is romantic. People do have to make an effort to overcome political
apathy, so that the habit of making such an effort deserves the name of
virtue. It constitutes a motivated variety of virtue – a sort of virtue that is
independently reinforced by personal interest and spontaneous invest-
ment – as distinct from virtue of a pure, moralistic kind. However demand-
ing, such motivated virtue is certainly within people’s reach. And so people
may be held responsible for failing to exemplify vigilance and contestation.
Let them fail to display it and ‘they may thank themselves’, as James
Harrington (1992: 20) put it in the 1650s; the failure will stem from a
weakness of will for which they only have themselves to blame.
These observations show that the contestatory spirit required under a
republican regime counts as a form of civic virtue, albeit virtue in a sense in
which it ought to be readily achievable. But there is one other aspect of this
virtue that we ought also to stress. This is that, while it consists in a willingness
to challenge public proposals and policies, alleging that they are not supported
by the people, or not supported in a manner that treats all members as equals,
it presupposes a commitment to living under an arrangement where all
members of the community can share in a system of equal popular influence.
Contestatory virtue is not the sort of contrarian or sectarian disposition that
opposes compromise and accommodation. It is the virtue of citizens who
embrace the ideal of a republic in the community in which they happen to
live – they are in that sense patriots (Viroli 1995) – and who are willing to do
all that is required for realizing the ideal. Not only are they patriots in
supporting their country internationally, to invoke the usual application of
the idea. More relevantly for our concerns, they are also patriots at home.
They are committed to establishing an undominating government in their
country, just as it is, and rather than yearning or working for a land governed
by their particular tribe or creed or colour, they accept that this will require
accommodation on all sides, their own included.
The ideal of the contestatory citizenry that we have been describing is
intimately connected with the ideal of the mixed constitution. Given that the
voice of the people is meant to emerge from a process of interaction between
different bodies, there is every reason to think that individuals operating in
contestatory, as well as electoral, guise should constitute one of those parts. But
whereas the idea of a contestatory citizenry coheres in this way with the mixed
constitution, it is utterly at odds with the Rousseauvian image of a law-making
assembly that speaks with the voice of a uniquely authorized spokesperson for
the people. If the law-making assembly is the spokesperson that speaks with
unique authority for the public or the people, as in the image that Rousseau
inherited from Hobbes, then individuals cannot be allowed in their private
capacity as subjects to contest that voice. Were they given rights to question
the dictates of the sovereign assembly, then that assembly could not speak with
An efficacious system of influence 229
the requisite authority. Thus, as we saw in the Introduction, Hobbes (1994b:
18.4) says that if subjects could ‘pretend a breach of the covenant made by the
sovereign . . . there is in this case no judge to decide the controversy’. And
Rousseau faithfully echoes that thought when he maintains that ‘if individuals
were left some rights . . . there would be no common power who might
adjudicate between them and the public’ (Rousseau 1997: i.6.7).42
5. an efficacious system of influence
The dimensions of popular influence

Assuming that there is a direction it imposes, we have been looking at how
far a popular, electorally centred system of influence can be unconditioned,
imposing that direction independently of the will of government or of any
third party. Continuing with the assumption that there is a direction that it
imposes – I try to vindicate the assumption in the next chapter – we must
ask, finally, about how efficacious an electorally centred system of influence
can be. Are the constraints it might impose on government sufficient to
enable the system to pass the tough-luck test? Are they enough to allow
people within the system to think of unwelcome decisions as a matter of
tough luck, not evidence of a malign will at work against them or their kind?
The influence that people might command under an electoral system
that is suitably individualized and unconditioned is enormous. We can
begin to appreciate its dimensions as we recognize that just as a power of
interference makes three distinct forms of impact possible – frustration,
interference and invigilation – so the same is true more generally of the
power of exercising influence. As I put it earlier, the influence exercised may
be an active, virtual or reserve form of influence.
Suppose we, the people, enjoy an individualized, unconditioned form of
electoral influence over government. That puts in our hands a vast array of
inputs that we may actively exploit with a view to pushing government
42
Kant follows Rousseau in this image of a uniquely authorized spokesperson and elaborates a similar
thought when he asks who is to decide the sort of issue that might arise between subject and sovereign
under a contestatory arrangement. Envisaging an individual rather than a collective sovereign, he
gives the following answer. ‘Only he who possesses the supreme administration of public right can do
so, and that is precisely the head of state; and no one within a commonwealth can, accordingly, have a
right to contest his possession of it’ (Kant 1996: 299; see too 463). It is this consideration that leads
Kant to the view, mentioned in the earlier discussion of contractualism, that even should the
sovereign ‘proceed quite violently (tyrannically), a subject is still not permitted any resistance by
way of counteracting force’ (Kant 1996: 298). For an argument that still Kant could have found
grounds for condemning the Nazi state see Ripstein (2009: 341–2).
policy in one or another direction. Where that policy goes will depend on
how we cast our votes and on what we choose to contest as individuals or
subgroups, whether via the legislature, in the courts, through ombudsman
or related channels, or, of course, in the media and on the streets. The
system will be set up in such a way that the actions we take individually and
in aggregate ought to be capable of making a huge impact. There will be
nothing to inhibit us from seeking such an impact, as we operate in a space
of free speech and association and are in no way dependent on the acqui-
escence of a higher will. And the system will be designed under such
constraints – under such checks and balances – that the inputs we deliver
are transmitted smoothly, perhaps even with some amplification, to the
points where they can have an effect on policy.
But these comments address only active efforts at influence and miss out on
influence of other kinds. For if we, the people, have the power of exercising
such active influence, whether electorally or otherwise, then we can practise a
corresponding form of virtual influence on how government behaves. Without
intervening when we are happy with government performance, we can remain
poised to intervene should things not transpire to our satisfaction. We can ride
herd on government, to return to the cowboy metaphor, wittingly or unwit-
tingly ensuring that it does not take a policy-making direction to which we
object. Without actually doing anything active, we can stop government going
against our dispositions, as the cowboy who rides herd on his cattle stops them
from going against wishes he has actually formed.
Apart from active and virtual influence, an electoral-cum-contestatory
system will also allow us to practise reserve influence. We exercise this form
of influence when we do not have formed dispositions in regard to the
direction government should take on a certain issue of policy, but we
remain disposed to intervene should such a disposition materialize. If we
form a disposition that rules against a certain policy, then if the policy has
actually been adopted we will use our active influence to combat that policy,
and if it remains a merely possible option, we will rely on our virtual
influence to keep it from being adopted.
Where we can command a virtual or reserve form of influence over
government, it is very likely that those in power will become aware of the
fact. And if government is aware of our power in this regard, then it is likely
to be suitably inhibited.43 Conscious that we would oppose the adoption of
43
To be aware of the virtual interference of another agent is to be exposed to coercion; to be aware of the
reserve interference of another agent – to be aware of the invigilation practised by that agent – is to be
exposed to intimidation.
a certain policy, it is likely on that account to be slow to pursue it. And
conscious that we might turn against a certain policy that we do not
currently oppose, say as a result of seeing how it works in practice, govern-
ment is likely on that account to be hesitant to introduce it. Thus we can
enjoy influence over government just by being saliently disposed in this or
that direction. Without raising a hand or uttering a word, we can put
government in a position where it will do our work for us in acting
according to our actual or likely dispositions.
These comments should make clear that under the system described so
far the river of popular influence can run wide and deep. It can flow into
every corner and cranny of government decision-making, imposing con-
straints on what decisions the different agencies reach and on how they
reach them. The influence may emanate not just from how we act in general
elections and in individualized contestations but, as in the virtual and
reserve cases, from how we are just disposed to act on those fronts. It can
enfranchise our dispositions, as well as our deeds.
Given the multi-dimensional manner in which our dispositions can
impinge on government performance, there is every hope that the influence
we exercise can constrain government to the point where private will is only
allowed an impact that is consistent with the direction the influence is
designed to impose. Assuming that there is an equally acceptable direction
in which this popular influence can push government – that is the topic of
the next chapter – the influence ought to be efficacious enough, other things
being equal, to ensure the satisfaction of the tough-luck test.
But are other things going to be equal? Is the river of popular influence
going to be allowed to flow wide and deep? Or will it be excessively diverted or
diluted by the other streams of influence that can inundate the public world? I
discuss that possibility in this final section, addressing three dangers in partic-
ular and arguing that none is inescapable. The first is the danger of elected
politicians usurping the influence of the people under motives of self-interest.
The second is the danger of private lobbies usurping that influence out of a
desire to push government in a direction that does not necessarily have popular
support. And the third is the danger that unelected authorities, including
authorities established to counter the earlier dangers, might gain a hold over
government policy that is not sensitive to popular demands.44
44
Many of the points made in the following discussion draw on earlier arguments in Pettit (2002a,
2004b). Where I wrote there of the need to depoliticize democracy, however, I now tend to avoid this
talk. It has helped to bolster the criticism that republicanism, as I interpret it, does not give people a
proper, democratic role. See McCormick (2011: 155–7).
The danger from elected politicians
Benjamin Constant (2003: 387) identifies the first danger crisply when, as
we saw, he observes that if you elect someone to a representative office then
you thereby give ‘him a different interest from the one he is charged with
representing’. By electing representatives, at least in the normal run of
things, we give them a special interest in being re-elected and a powerful
concern to take steps that will facilitate re-election and to avoid steps that
will stand in the way. This observation points us immediately to areas of
policy-making in which elected politicians would have to be saintly not to
be moved by their own interests, regardless of what popular influence
supports – or indeed of what the preconditions for popular influence
require.
The prospects of re-election are going to depend for many politicians on
the voting and districting system under which they operate. A first lesson,
then, is that they should not be given management of such electoral issues.
But the prospects of re-election are also going to depend on how informa-
tion on government performance is shaped and communicated. So a second
lesson is that elected politicians should not be given command over the
release of social and economic data and statistics. Again, to go to a more
general matter, the prospects of re-election are going to depend most
saliently on the experience of voters in the present, so that there is going
to be a powerful incentive for politicians to make that experience sweet and
to downplay future costs, where these are costs that electors may be
presumed to care about, but to be easily led into ignoring. This last
consideration argues for a number of lessons: for example, that government
policies on interest rates, on energy and environmental issues, and perhaps
on criminal sentencing ought not to be left entirely in the hands of the
elected. Politicians, being focused on the electoral short term, are always
likely to favour lower interest rates, easier energy and environmental
demands, and tougher, retributively satisfying sentences. And however
attractive in the short term, such policies can be very destructive and costly
over the longer haul.
How to cope with such dangers? The most obvious response, and one
taken in many democracies, is to put these areas of policy-making at arm’s
length from elected politicians. This response would argue for establishing
an independent electoral commission with responsibility on districting and
other electoral decisions; for introducing a body like the British Columbia
Citizens’ Assembly to make recommendations on some particular issues; for
setting up independent, reputable authorities to gather, analyse and release
publicly important data and statistics; for investing an economically
informed central bank with responsibility for interest rates and related
issues; and for having similar bodies make recommendations, and effec-
tively constrain policy, on issues of energy and the environment and on
matters of criminal sentencing.45
These comments are made on the assumption that the self-interest of
politicians will be exclusively electoral in character. But we should also
recognize the danger that once they occupy positions of power, politicians
will be exposed to a special temptation to accept hidden payments for the
favours they can do for certain individuals, corporations and other bodies.
This danger argues too for the introduction and empowerment of certain
unelected authorities, with a power of auditing the public books and, where
required, sponsoring charges of corruption.
In dealing with the requirements for individualizing popular influence I
argued that there would have to be channels of challenge, and tribunals of
adjudication, that are independent of government. This requirement was
further supported by our discussion of how to make popular influence
unconditioned. For all we said in those discussions, however, judicial and
cognate authorities might be elected in their own right or appointed with a
certain arm’s-length independence from government. The pressures
reviewed in this section argue against election. And so they point us to
yet another area – the conduct of the courts and other tribunals – where
the regulation of public life should be assigned to appointed, independent
authorities, not to figures who are required to keep an eye on the electoral
significance of their various decisions. But does the investment of such
unelected authorities itself raise a problem for the influence of the people?
I turn to that question after discussing the problem raised by private
lobbies.
The danger from private lobbies

Private lobbies are groups of people that constitute factions in the
eighteenth-century sense of that term. They have selfish or sectional inter-
ests such that their satisfaction is not necessarily going to enjoy popular
support and, to anticipate discussion in the next chapter, is not necessarily
45
There is great room here for institutional innovation. For a recent proposal in this spirit of innovation,
see the proposal for a Citizen’s Council in California at http://berggruen.org/files/thinklong/2011/
blueprint_to_renew_ca.pdf. This council would be appointed on the basis of expertise, experience
and presumptive impartiality and would have the authority to make constitutional and legislative
proposals, as well as making recommendations on proposals emanating from other quarters.
consistent with the direction that popular influence would ideally impose
on government. They are private in the sense that they seek either to
sideline popular influence, looking for an inside channel to government,
or to mobilize popular influence on the basis of deception or manipulation
or just the threat or power of inflicting harm: say, the harm to a neighbour-
hood that a corporation can inflict by moving elsewhere, or the harm that
unions can impose by calling a general strike. In this respect, they contrast
with public-interest groups of the kind invoked in the last section. Such
organizations count as public-interest groups just insofar as they seek to
mobilize popular influence without resort to under-handed tactics or
strong-arm measures.
Private lobbies, so understood, may achieve influence by making politi-
cians financially dependent on their support in electoral campaigns. Or they
may achieve influence by being able to do harm to politicians, whether in
the manner of the print or electronic media that can create a lot of
electorally adverse publicity, or in the manner of the corporation or
union, or indeed the foreign power, that can hold out a prospect of action
that is bound to hurt the country and be electorally damaging. Whether a
foreign power or a multinational corporation is capable of being resisted will
depend on the international order and takes us beyond our concerns in this
book. But what measures might be taken to combat more domestic kinds of
private lobby?
Combating the influence of private media is a challenge for every
democracy, and the only response, which may or may not be available
under a given constitution, is to establish such a variety of universally
accessible media that misinformation and manipulation is readily identified.
That is particularly difficult, however, in the contemporary world, where
different sectors of the political electorate can tune into broadcasters of their
choice, opting only to hear the voices of those with whom they are
independently primed to identify: those in their comfort zone (Sunstein
2009). Perhaps the most promising remedy is for government to establish a
publicly funded, politically independent broadcaster like the BBC, where
this broadcaster achieves such prominence and credibility that no one can
ignore it.46
How to combat the influence of private campaign supporters? This is a
particularly powerful challenge in presidential systems like the US, where
46
One of the first policy changes introduced by President Zapatero of Spain, as he sought to follow a
republican agenda, was to put the national broadcaster at arm’s length from the control of the
government. See Marti and Pettit (2010).
there is no governing party or coalition whose members have to close ranks on
pain of allowing the administration to fall (Lessig 2011). In such systems,
legislators can vote independently of party and, while that sounds attractive, it
means that they can be exposed to enormous financial pressure to vote in this
or that manner on particular issues. Limiting private campaign finance,
providing public funds to match private campaigning funds, and restricting
the use of expensive advertisements on the electronic media are all possible
modes of dealing with the problem. But unfortunately such steps are some-
times not possible under received and difficult-to-amend constitutions. Thus
they would be very hard to implement in the US, since the courts treat money
spent on elections as protected political speech, arguing that the more speech
there is, the better; and put few restrictions even on essentially self-interested
bodies like commercial corporations (Bakan 2004).
It is generally easier to deal with this particular problem in parliamentary
democracies. Not only do most parliamentary regimes allow the use of
independent regulators to impose restrictions on campaign finance; when
they are not so proportionally elected as to put very loose coalitions in
power, these regimes offer an inbuilt, though only partial, safeguard against
the influence of private lobbies that seek to buy favours. The members of
the majority party in the legislature have to vote together on major issues,
since the administration is liable to fall if they do not close ranks in this way.
And that means that in order to influence what happens in government,
private lobbies have to buy over the whole party or couple of parties in
power; they cannot concentrate, as in the US system, on buying off
individual, more or less pivotal figures.
The danger from unelected authorities

In exploring how to deal with the previous two dangers I have been
suggesting that there must be unelected authorities established in political
life, side by side with those who are elected. These may be particular officials
such as judges or ombudsmen, statisticians or auditors or regulators; bodies
like an electoral commission, a central bank, a budgeting office or a food
and drug administration; a media organization like the BBC; or indeed, a
temporary, advisory assembly such as the British Columbia Citizens’
Assembly.47 While they may serve in primarily executive, contestatory or
47
Pierre Rosanvallon (2006: 240) treats the Interstate Commerce Commission, established in the
United States in 1887, as one of the first examples of an ‘independent institution’, ‘shielded from the
direct authority of executive power’.
adjudicative roles, they have a distinctive independence in common. They
are appointed to their offices by elected officials rather than being elected
themselves but they do not serve as functionaries of those officials; they are
appointed to serve for fixed or open terms, not at the pleasure of those who
appoint them.
The last question we need to confront is whether such unelected author-
ities constitute a source of influence that is independent from the influence
of the people, undermining the regime of popular influence for which we
are looking. The existence of these authorities may cause less of a problem
for a system of popular influence than the dangers they are designed to
combat. But the question is whether nevertheless they are a problem:
whether they represent a current of influence that dilutes the stream that
originates with the people.
The question is easily answered in the case of the statistically or indica-
tively representative assembly that is charged, like the British Columbia
Citizens’ Assembly, with giving advice on the decision to be taken on a
particular policy or set of policies. Such a body does not adopt initiatives in
causal response to the dispositions of people in the manner of electoral,
responsive representation. But that it takes one or another initiative is good
evidence that the people as a whole would adopt that sort of initiative, under
suitable information and deliberation. The intentional empowerment of
such an indicatively representative body, then, can be seen as a way of giving
influence to the dispositions of people in the domain where it operates.
Certainly it can be seen in this way, if there is a safeguard against the body
assuming a dominating role in its own right; this will be in place, for
example, when its recommendations are subject to independent review –
for example, by a referendum of the kind required in the British Columbia
case. The existence of such an advisory body is not an obstacle to the
influence of the people, but rather a means of channelling that influence.
I suggest that if other independent, unelected officials and bodies are
appointed under suitable conditions, and are forced to operate under suitable
constraints, then they too may have a claim to serve the people in the same
indicative way. Suppose that a judge or court, an electoral commissioner or
commission, a central banker or a central bank are selected on the publicly
contestable grounds of scoring well on experience, expertise and impartiality.
Suppose, second, that they are exposed to incentives of professional esteem to
show themselves to be experienced, expert and impartial in the way that they
resolve different issues (Brennan and Pettit 2004). Suppose, third, that they
are given a brief that has presumptive popular support: a brief to interpret and
apply the law impartially, to set up district boundaries and voting rules that
maximize electoral competition, to establish interest rates that guard against
inflation or unemployment. And suppose, finally, that they can be forced to
operate under conditions of such publicity, and such exposure to criticism
and challenge, review and censure – in brief such contestation – that they face
serious costs if they fail to live up to their brief. In that case, the decisions they
make are likely to be ones that the people, individually or collectively, would
make or approve if they had all the relevant information and expertise. And
this is not just a happy result of how they are independently disposed: it is a
pattern that is more or less imposed on them by the contestatory pressures to
which they are subject.
In our earlier picture, electorally responsive officials and bodies are
representative of the people in virtue of being responsive and are forced to
perform appropriately by the individualized contestation to which they are
exposed. In the picture I am sketching here, unelected authorities have a
complementary profile. They are indicative officials and bodies as a result of
the basis on which they are selected, the incentives to which they are
exposed, and the briefs they are given. And like elected authorities they
are forced to perform appropriately by the individualized contestation to
which they are exposed. They are proxies for the people who serve a crucial
representative role in complementing and containing the people’s elected
deputies.
The unelected authorities to which this analysis applies are not limited to
those in official roles of the kind illustrated. Any system of individualized
contestation requires particular individuals or groups to be ready to bring
charges against those in government, elected or unelected, whether in the
courts, in the press, or on the streets. What gives such contesting parties –
such private attorneys general, as they are sometimes known – the right to
speak, as they generally claim to do, in the name of the people? I think we
can cast them as indicative representatives who are subject to such con-
straint and contestation that they are bound to enact a brief, as we might
think of it, that they are given by the people. The brief is provided implicitly
in the fact that a popularly approved constitution allows private attorneys
general to mount arguments in the courts or other forums against new laws
and other decrees. The rationale behind the brief is to ensure that those in
any potentially disadvantaged sector of society can be represented by one of
their kind, or someone sympathetic, in challenging and testing the measures
at issue. And those private attorneys are more or less bound to enact their
brief, and honour its rationale, by being required to follow established
channels, arguing their case in a publicly accessible forum on publicly
available and contestable grounds.
To gesture at the possibility of making unelected authorities into indi-
cative, suitably controlled representatives of the people is not to design the
incentives and briefs, the appointment procedures and operating con-
straints that might promote that ideal. Here, as in the other cases reviewed
in this chapter, we have to be content with being able to see in the abstract
how the people might enjoy a system of equally accessible influence that
meets republican constraints. I think that the considerations offered do
support the claim that this is a feasible possibility in the current case – and in
the other cases rehearsed – even if the rehearsal of such considerations falls
well short of the hard work required in institutional design.
The picture emerging from the discussion requires an electoral and
contestatory democracy that is established on a constitutional basis,
written or unwritten. The system envisaged is a form of constitutional
democracy insofar as it combines familiar democratic devices with more
or less familiar constitutional elements. But it is not a system in which
these devices and elements are just added together as independent com-
ponents. And it is not a system in which the constitutional elements are
fixtures that resist democratic control. The constitution is there to facil-
itate democratic influence – that is, an equally accessible form of uncon-
ditioned and efficacious influence – not just to complement it as an
independent factor. And the constitution is going to remain there only
to the extent that the democratic influence it facilitates is not exercised for
purposes of revising it. In view of these features, the picture we have been
led to adopt might be cast in James Tully’s (2009: i, 4) phrase, as
democratic constitutionalism: an approach in which ‘the constitution
and the democratic negotiation of it are conceived as equally basic’ (see
too Mueller 2007; Espejo 2011).
I said at the beginning of the chapter that our discussion would lead us to
a model of equally accessible popular influence that is far enough away from
the status quo to direct us towards reforms, yet near enough to assure us of
its feasibility. I hope that this promise seems to have been borne out in the
picture that has emerged from our critique of the plenary assembly, our
endorsement of an electoral assembly, and our introduction of amendments
to ensure that the popular influence it mediates is suitably individualized,
unconditioned and efficacious. With that picture in place, we turn in the
next chapter to ask how such a system of equally accessible influence –
equally accessible, unconditioned and efficacious influence – might push
government in an equally acceptable direction and establish the popular
control that republican democracy requires.
chapter 5
Democratic control
According to the argument developed in Chapter 3, a state will be legitimate

to the extent that the order it imposes on its citizens is imposed under
popular control. But, so the argument continued, people will control the
state just insofar as they have an individualized, unconditioned and effica-
cious influence that pushes it in a direction that they find acceptable. First,
the people must have an influence on government that is individualized,
unconditioned and efficacious. And, second, they must exercise this influ-
ence to a purpose or direction that individuals – or at least those who are
prepared to live on equal terms with one another – find acceptable. People
must have such a power over government that the regime can be described,
in a rich, egalitarian sense of the term, as democratic: a regime that
establishes the kratos, or ‘control’, of the demos, or ‘people’.
This conclusion gave us the task of spelling out how the institutions of a
society might meet this design specification and implement a republican
conception of democracy. We took up that task in the previous chapter,
when we looked at how the people might be given individualized, uncon-
ditioned and efficacious influence over government. Without being able to
go into detail, we saw that that job specification was likely to require an
open electoral system under which individualization would be achieved by
the possibility of individualized contestation; unconditioned independence
by the resistive character of the citizenry; and efficacy by the insulation of
the channels of popular influence against the distorting effects of electoral
pressures and private lobbies. Under this image, the political institutions
would require many amendments in even the best-practice democracies
today; but they would not be so distant from current arrangements as to
seem utterly infeasible or utopian.
We turn in this final chapter to the parallel task of exploring how such a
system of popular influence might push government in a direction that is
equally acceptable to all and thereby come to constitute a system of popular
control. The task is to identify an equally accepted purpose that the system
239
240 Democratic control
of popular influence, individualized, unconditioned and efficacious, might
be organized to promote. The discussion is in three sections. In Section 1,
I look at two familiar candidates for this role, finding reason to be unhappy
with each. In an extended Section 2, I introduce an alternative candidate –
the dual-aspect model of democracy, as I call it – show how this might play
the required role, and discuss its attractions. And then in Section 3, I look
at the implications of endorsing the dual-aspect model for our view of
the state, the people and the constitution – in short, for our political
ontology – arguing that under this model, government can live up to
Lincoln’s requirements: it can truly be government of the people, by the
people and for the people.
1. two familiar models of directed influence
Influence without control

The main point to register in taking up the task of the chapter is that a
democracy of popular influence need not count as a democracy of popular
control; it may not direct government appropriately, establishing the undomi-
nated status of the people. Even if the system of influence were rich enough to
be capable of imposing a popular direction in an individualized, unconditioned
and efficacious way – surely a high ideal – it might still fail in actual fact to give
government such a direction. It might impose a large range of constraints, but
only with the effect of inducing a series of showdowns and stalemates in the
interaction of people and government. In the event of such gridlock, the
government might be frustrated in its own ambitions, defensible or otherwise,
but it would still dominate the people in particular decisions; it would impose
on them coercively, without the coercion being suitably controlled.
The decision-making of a gridlocked government might be wholly
chaotic in character, undermining the agential status of the state to which
it is ascribed and certainly failing to satisfy people generally. Or it might
display a pattern that allows us to extrapolate from past to future perform-
ance, but not a pattern for which the people plausibly control. The pattern
might be the publicly unwelcome precipitate of individual acts of publicly
constrained decision-making. Or it might be a precipitate that is welcome in
some sectors but unwelcome in others. In either case, it would not be a
pattern that counts as equally acceptable to all the people, even to all who
are willing to live on equal terms with others.
For many theorists of democracy today, particularly in economic and
political science, the democratic ideal is one of popular influence – indeed a
Two familiar models of directed influence 241
form of popular influence that need not be deeply constraining – and
nothing more. This representation of the democratic ideal appears clearly
in the influential work of the Austrian–American economist and political
theorist, Joseph Schumpeter. In his discussion of democracy in Capitalism,
Socialism and Democracy, first published in 1942, Schumpeter presented an
argument for thinking that while democratic government depends impor-
tantly on the influence of the people, it need not be given – and, as a matter
of fact, is not given – any particular direction by that influence.
Schumpeter assumes, reasonably, that any plausible democratic system is
going to involve open, periodic, electoral competition, with different parties
seeking to attract enough support to win office. He is sceptical on a number
of counts about the possibility that the results of such a democratic process
would be ‘meaningful in themselves – as for instance the realization of any
definite end or ideal would be’ (Schumpeter 1984: 253). The people do not
form systematic views that they might impose on leaders; under the influ-
ence of popular pressure and party propaganda, they display only ‘an
indeterminate bundle of vague impulses loosely playing about given slogans
and mistaken impression’ (Schumpeter 1984: 253). And even if they did
form such views, they would not be able to impose them. The political
decisions produced from ‘the raw material of those individual volitions’
(Schumpeter 1984: 254) might take any of a variety of forms, depending on
the initiatives of the party boss and the party machine. Parties and leaders
are primarily committed to keeping office, not to representing any standing
principles, and no matter what the input from the electorate, ‘the pyro-
technics of party management and party advertizing’ will deliver whatever
response promises to serve best in ‘the competitive struggle for political
power’ (Schumpeter 1984: 283). Thus, while the people choose between
parties at election time, they do not otherwise ‘control their political leaders’
(Schumpeter 1984: 272).
In view of his scepticism about the idea that democracy might serve any
popularly imposed purpose, Schumpeter defines it by the process it
employs; it is nothing more or less than ‘that institutional arrangement
for arriving at political decisions in which individuals acquire the power to
decide by means of a competitive struggle for the people’s vote’ (1984: 260).
This definition has been popular with political scientists ever since his time,
because it makes democracy into a workable category in the descriptive
characterization of regimes (Przeworksi 1999). In Schumpeter’s words it
provides ‘a reasonably efficient criterion by which to distinguish democratic
governments from others’ (1984: 269). In this respect it certainly scores well
above the republican conception of democracy. There can be little
controversy about whether the politicians in a given system compete for the
people’s vote in order to attain power. But there is bound to be controversy
about how far in a particular society people rely on their individualized,
unconditioned and efficacious influence to exercise equally shared control
over government.1
But we ought not to allow taxonomic considerations to outweigh nor-
mative concerns in deciding what democracy ideally requires. The real
question for us is how attractive a democracy of influence-without-control
is likely to be, especially from a republican perspective. It has some clear
attractions, as Schumpeter (1984: 269–73) himself stresses. It promotes
interaction between rulers and ruled, for example, requires considerable
freedom of discussion amongst the electorate, and gives people a power of
evicting government which guards against the possibility of dynastic rule.
These virtues may give us reason to think that democracy is a lot better than
dynastic or other systems – that it is the worst form of government, as
Winston Churchill is reputed to have said, except for all the others. But a
democracy of influence-without-control would not establish the legitimacy
of government in the republican sense. Its presence would not entitle us to
say that people have such an individualized, unconditioned and efficacious
degree of control over government that its interference in their lives does not
constitute domination. Schumpeterian democracy is a second-grade ideal
from our point of view, not the sort of arrangement for which we have been
looking.
The system of popular influence described in the last chapter will
underpin a system of popular control insofar as it serves to impose a
direction on government that is equally appealing for all those citizens
who are willing to live on equal terms with one another. The task in the
remainder of this section is to review the standard ends or purposes that
the people’s influence on government might be thought capable of
advancing, and to assess them for plausibility. These candidates divide
into two categories, depending on whether the idea is that people might
knowingly and intentionally seek their promotion or might produce them
without necessarily being aware of doing so: this, in the way in which
people are said to produce competitive pricing, as by an invisible hand, in
the open market. I shall look at these two categories in turn, finding fault
1
For an overview of various attempts to define democracy with a view to characterizing actual regimes
see Tilly (2007: Chapter 1). Given the account in this book of the institutions required for a republican
democracy, it ought to be possible to adapt one of these approaches in developing an audit of
democracy. The Economist Intelligence Unit offers a congenial index of democracy at www.eiu.
com/public/democracy_index.aspx.
with the standard approaches associated with them. That will lay the
ground for introducing a third, more promising candidate that I go on
to explore in Section 2.
The intentional direction of government

If people are to participate within a system of influence that imposes a
certain direction on government, then according to many thinkers they
must implicitly or explicitly endorse the purpose that they want government
to fulfil. They must each articulate that purpose or end and, intending to
cooperate with others in its promotion, they must each assume that every-
one can do this, recognize that others can do it, and so on. And then they
must each take steps that make sense as a contribution to the presumptively
general enterprise of advancing that purpose.
Rousseau (1997: iv.1.6) thinks that this is the approach that the members
of his assembly will take, at least if they are operating properly; they will each
cast any vote on the basis that ‘it is advantageous to the State . . . that this or
that opinion pass’, not on the factional basis that that is ‘advantageous to
this man or this party’. In that respect, the intentional-control model of
popular control has close affinities with the general-will model. But how
might the idea apply in the representative regime envisaged here?
Schumpeter (1984: 269) assumes that under an electoral regime a similar
form of intentional control might materialize, though he himself thinks it
won’t. It might be, as he takes ‘the classical theory’ to maintain, that ‘“the
people” hold a definite and rational opinion about every individual question
and that they give effect to this opinion – in a democracy – by choosing
“representatives” who will see to it that that opinion is carried out’. In this
picture, the people identify suitable policies first and then choose represen-
tatives on the basis of their endorsing those policies.
The view that the people intentionally impose a purpose on government,
however, might not take this policy-first form. For all that the approach
strictly requires, the people might cast their votes for certain representatives,
not on the grounds that they are likely to see to it that the independently
formed opinions of voters are implemented, but on the grounds that those
deputies are intellectually and motivationally reliable and are likely to form
a sound opinion about the matters they deal with (Montesquieu 1989). In
this context the sound opinion might be taken by voters to be the opinion
that they themselves would be likely to form, if they had the time and
information and expertise that representatives may be expected to enjoy.
This approach might be described as a deputy-first version of the theory, as
distinct from the policy-first version that Schumpeter ascribes to the clas-
sical theory.2
The purpose that representatives are supposed to advance, whether in a
policy-first or deputy-first way, might plausibly be identified, in traditional
phrases, with the common good, the public interest or the welfare of the
society. Whether it generates an agenda that people articulate for them-
selves, or an agenda that they rely on suitable representatives to articulate, it
answers to the sort of thing that such phrases are generally taken to identify.
But how should we best interpret the phrases? How should we understand
the public interest, to favour one particular phrase, that people might
plausibly endorse and seek to impose on government?
In one interpretation of the public interest, it is nothing more or less than
the intersection of people’s private interests: that is, the interests that they
bring to society – the interests that they have, independently of their having
to live together with each other. This is not a satisfactory account of the
public interest, however, since there may be only a very small degree of
overlap between such pre-social interests. You may have an interest in
hunting for a living, I in farming; you may want one religion supported, I
another; you may want clean water where you live, I may want to build a
factory that will pollute it. And even if there is a considerable overlap in pre-
social interests, that overlap is going to be unstable, since a change in even
one person’s interests is liable to disrupt it. Such a goal would shift
unpredictably and make an unsuitable purpose to assign to government.
Is there a better version of the common good or public interest that
people might plausibly endorse and seek to impose? An obvious improve-
ment will come with the recognition that there is a difference between the
interests people have independently of having to live in society and the
interests they have, given that they have no option but to live with others –
and, to reiterate a basic normative presumption, to live on equal terms with
others. You may be of one religion and I another and neither of us may have
an independent interest in supporting, or even in tolerating, the other’s
creed. But if we have to live with one another, and live under a collectively
coercive state, then we each have an interest – a post-social rather than pre-
social interest – in establishing a regime under which the freedom of our
respective religious practices is assured. This will be better than allowing no
2
Under the policy-first version, deputies are cast as delegates, in Hannah Pitkin’s (1972) phrase; under
the deputy-first they are cast as trustees. As we saw in the last chapter, deputies of either sort are
responsive representatives and should be distinguished from proxies, or indicative representatives. The
view of the American founders was clearly that elections would serve in the first place to select out
suitable deputies (see too Madison, Hamilton and Jay 1987: no. 57; Herreros 2006).
freedom of religion to anyone, which would unnecessarily restrict the range
of basic liberties. And it will be better than privileging one religion, since
this would reject people’s claim to equality; as we saw in the last chapter, it
would effectively ensure that people did not have an equal share in the
electoral system of popular influence.
What obtains in the religious domain holds also in a range of other
domains. As we have a shared individual interest in freedom of religion, so
we will have such a shared interest in having a regime that allows you to
hunt and me to farm, you to have clean water and me to build a factory, and
so on. The public interest, conceived in this way, is composed of those
goods that anyone who accepts the necessity of living on equal terms with
others is likely to want to have collectively guaranteed or promoted. It
consists in the interests that people are going to share insofar as they have
equal status as members of a polity (see Goodin 1996; Pettit 2004a).
While this account distinguishes the public interest from the intersection
of private interests, it also makes it distinct from what might be described as
the good of the nation or people, considered as a corporate entity. Considered
in a corporate role, a nation survives over many generations and its good may
have little to do with the good of the individuals who make it up at any time.
The nation or people may prosper in terms of international power relations,
for example, without that prosperity making for an aggregate gain amongst
members; maintaining the hegemony may be extremely costly and provide
most members with only the dubious benefit of being able to take pride in
their state. To identify the goal that the people might intentionally impose on
a state as the maintenance of such a corporate performance would be counter-
intuitive, though a certain sort of nationalism might support the line. It would
be to reject Aristotle’s (1996: 2.5) common sense when he argues in The Politics
that while a set of odd numbers may add up to an even sum, a population of
unhappy people cannot add up to a happy polis: ‘the whole cannot be happy
unless most, or all, or some of its parts enjoy happiness’.
If these observations are sound, then the most plausible version of this
first account of popular direction would say that the people exercise popular
control insofar as they intentionally use their influence to force government
to pursue public-interest policies, where the public interest is understood in
a post-social but non-corporate manner. The people may be held to do this
either in a policy-first or deputy-first manner: either by first identifying
suitable policies and then selecting representatives who support those
policies; or by first identifying suitable representatives and then relying on
them to formulate public-interest policies. But in either version, as we shall
see, the theory is less than satisfactory.
The elected representatives in each version may deliberate with one
another before casting their votes on the policies to be put in place. But
despite each endorsing a consistent set of policy judgements, the effect of
aggregating their votes may be to support an inconsistent package; that is
the lesson of the discursive dilemma, as we saw in the last chapter. This
means that the body of elected representatives will have a good deal of
discretion in deciding on their policies, whether they decide on them one by
one, as under the Washington system of deal-making, or in the overall
programme that parties typically present to the electorate in the
Westminster alternative.
The availability of such discretion accentuates a deep difficulty for the
intentional-direction approach. The politicians in the system – and to some
extent the people too – must be virtuous enough to make impartial judge-
ments on what is in the public interest and then to introduce policies that
promote that interest. In particular, they must be virtuous enough to do
this, even when electoral or other private interests – and no system of
popular influence can completely silence these – argue strongly for warping
the judgements or the policies in another direction and, of course, for
representing that redirection as itself required by the public interest.
Indeed, they must be virtuous enough to do this, even when all we know
of cognitive frailty suggests that private interest can delude them into seeing
the redirection as a requirement of public interest; it can enable them to
hide their misrepresentation from themselves (Gilovich, Griffin and
Kahneman 2002; McGeer and Pettit 2009).
The virtue required of people and representatives under the intentional-
direction story, unlike the virtue we ascribed to those who identify with
certain watchdog groups and social movements, is likely to be unmotivated.
There need not be any elements of personal interest or spontaneous
commitment to support it, so that the virtuous performance required may
only be forthcoming in the presence of something close to moral perfection.
That is a problem, for two reasons. One, such virtue is not an assured
commodity and a system that requires it is not guaranteed to be sustainable.
And two, even if people and politicians were generally disposed to be
virtuous, they might not be confident about one another’s virtue, and,
lacking assurance on the point, might fail to display virtue themselves; they
might each look after their own interests, fearing the prospect of being made
a sucker by others.
We can be concerned about this problem, I should mention, without
going to the extreme of adopting the so-called knave’s principle. We need
not assume, in Bernard Mandeville’s (1731: 332) words, that the best sort
of constitution is the one which ‘remains unshaken though most men
should prove knaves’. We can reject the principle, in David Hume’s
words (1875: 117–18), that in ‘fixing the several checks and controls of
the constitution, every man ought to be supposed a knave, and to have no
other end in all his actions than private interest’. The knave’s principle
may be counter-productive, as some empirical evidence suggests; it may
argue for institutions that crowd out whatever virtue is available, say by
triggering resentment and defiance (Pettit 1996b; Brennan and Pettit
2004: Chapter 14). Without embracing any such dubious principle,
however, we can still see good reason to reject any design that would
make unmotivated virtue into a prerequisite for the proper performance
of an institution; we can see good reason to economize on virtue
(Brennan and Hamlin 1995).
The intentional-direction model fails on precisely this count. It requires
politicians to put aside their private concerns – including their characteristic
concern with re-election – in formulating the policies that promise to
promote the public interest. And it requires ordinary electors to be disposed
to demand and respond to policies that succeed in doing this. Politicians
and electors must each form judgements about what the public interest
requires. They must do so in full sensitivity to the range of relevant
concerns, however altruistic. They must use their discretion in aggregating
those judgements without letting private interest or other pressures warp
the exercise. And they must then live up to those judgements – if you like,
those public-spirited preferences – in their performance on different fronts:
at the polls, in the party room, and in Parliament or Congress. O brave new
world that hath such people in it.
If this response seems somewhat cynical, I might add in conclusion that it
is fully in line with the long tradition of republican thought. Amongst
republican thinkers across different periods, it has been a constant refrain
that human beings are universally corruptible if often – thanks perhaps to
supportive institutions – actually uncorrupt (Pettit 1997c: Chapter 7).
Public office is a particular source of corruption, according to the tradition,
since it gives bearers a special opportunity to benefit from evil-doing and
often a special capacity to avoid being caught. It enables them, like the ring
of Gyges, to further their self-interest without having to run the usual risks
of detection and punishment. Thus, as Richard Price (1991: 30) puts the
recurrent theme, ‘There is nothing that requires more to be watched than
power.’ To entrust politicians with the articulation and advancement of the
public interest, regardless of their private concerns, would fly in the face of
this caution.
The non-intentional direction of government
We saw in Chapter 3 that it is possible for a person to control a process
without exercising control intentionally. The example we gave, drawn
from Amartya Sen, was that of a patient in a coma whose wishes dictate
the treatment provided, thanks to the intentional efforts of family and
friends. The wish of the patient has an influence on the treatment
adopted by the hospital, due to the influence of family and friends; it
controls for the treatment that the patient receives. And the connection
between influence and effect, input and output, is suitably robust or
unconditioned: while it depends on one of the family or friends coming
forward, for example, any of a number are willing to do so. Thus if the
wish had been absent, then the hospital would not in all likelihood have
provided that treatment; and if a different wish had been in its place, to
mark a richer possibility, then the hospital would have provided that
alternative instead.
In this example, the purpose effected by the system of influence and
direction is truly a goal or desideratum of the agent: it corresponds, after all,
to a wish. It is because of the influence of that wish that the treatment
assumes a certain form. But the causal mechanism whereby the wish
generates the treatment involves, not the intentions and efforts of the
agent, but rather those of family and friends. The lesson of the example,
then, is that direction and control may be non-intentional as well as inten-
tional. That lesson applies in collective, as well as individual, cases, as we
shall now see, and directs us to a second way in which popular influence
might be recruited to serve a popular purpose.
Consider the invisible hand whereby, under standard economic assump-
tions, the behaviour of consumers in an open market ensures the satisfaction
of the wish on the part of each to be able to buy goods and services at the
lowest feasible price. Consumers are each disposed to buy something of a
given quality as cheaply as possible, thereby displaying that wish, and to
shop around in order to achieve this. Producers of the commodity or service
are disposed to undercut one another’s prices in order to maximize revenue;
if they are not, then others will have an incentive to enter the market and
challenge them. And the effect of the interaction between the dispositions of
the two groups is to push producers to sell at the competitive price: that is, at
the lowest price that is consistent with their remaining in business.
Consumers do not cooperate on the basis of a shared intention or wish to
establish competitive pricing – certainly they need not do so. But they do
each wish to be able to buy what they want at competitive prices – although
not perhaps under that description – and the effect of their each acting on
those wishes is that, as by an invisible hand, the wishes are satisfied.
The notion of an invisible-hand mechanism became widely known and
celebrated in the wake of Adam Smith’s economic theories, as presented in
The Wealth of Nations in 1767 (1976). Unsurprisingly, it was soon adapted
for use in political theory. A compatriot of Smith’s, James Mill, argued in
1819, in a small treatise entitled An Essay on Government, that if we only
design the electoral system appropriately, then we can rely on an invisible
hand to promote the public interest (Lively and Rees 1978). Mill had rallied
behind Jeremy Bentham’s new philosophy and he argued that in a suitable
electoral system we can rely on an invisible hand to promote a utilitarian
version of the public interest: in the popular slogan of the time, the greatest
happiness of the greatest number.
Mill begins from an assumption that all human beings pursue their own
interest-satisfaction, where this is understood as the satisfaction of their
selfish preferences – in brief, their own happiness (Lively and Rees 1978: 5).
This gives him a utilitarian conception of the public interest; it consists in
‘that distribution of the scanty material of happiness, which would insure
the greatest sum of it in the members of the community, taken together’
(Lively and Rees 1978: 5). How then is a community of such individuals to
rely on government – government by a small number of their members, as
he assumes it has to be – to promote the public interest? ‘The community
itself must check these individuals, or else they will follow their interest’
(Lively and Rees 1978: 22) – their ‘sinister interest’, as he also describes it
(25). And how is the community to do this? By relying on ‘that grand
invention of modern times, the system of representation’ (Lively and Rees
1978: 21). Specifically, by relying on a system under which, first, electors are
inclusive enough to exemplify the range of interests in the community as a
whole (Lively and Rees 1978: 22) – Mill thinks that all older males will be
enough (27); and second, the period between elections is as short as
possible, consistently with allowing representatives to get on with their
job (25).3 The idea is that this system will force representatives to reflect
the interests of constituents in their voting, and that with such ‘an identity
of interest’ (Lively and Rees 1978: 34) between electors and elected, the
decisions reached will reflect a compromise that is going to promote the
greatest aggregate satisfaction of essentially self-interested agents. This
satisfaction is what he takes to constitute the public interest.
3
I am ignoring the fact that for Mill that job was mainly to keep an eye on the government exercised by
the King and his ministers.
If Mill’s picture is sound, then those in government will promote the
public interest of the community, in the utilitarian sense in which he
understands the public interest, and do so under the popular influence of
voters. That popular influence will generate the public interest, however,
not via an intention on the part of electors as a whole to let the public
interest have a dictating role, but rather via a mechanism that operates
behind their backs. With electors each voting for policies or deputies that
promise to satisfy their own self-regarding preferences, and with deputies
seeking to track the preferences of their constituents, the aggregate, unin-
tended effect is supposed to be that the interests of electors are maximally
satisfied overall.
Not many contemporary thinkers embrace Mill’s account of the popular,
invisible-hand direction of government that can be achieved under a
responsively representative picture. Whether interpreted on a basis that
gives priority to policies or deputies, it is open to the obvious objection
that for all that he says, electors and elected may form coalitions to impose
particular sectional interests in a way that is unresponsive to minority
interests within the society. Forming on the basis of common ethnic or
religious, commercial or regional priorities, such interest-groups could
systematically warp the laws and policies of government in their own favour.
But though Mill’s specific proposal no longer attracts many supporters,
there are loose relatives of the proposal that have been given some support in
more recent times. One is interest-group pluralism. It would suggest that
while interest-groups would certainly upset Mill’s individualist scheme,
they themselves might compete for influence over government in such a
way that the authorities can be expected, on pain of not being re-elected, to
look for the highest possible aggregate satisfaction of those competing
interests. According to this view, as one commentator puts it, ‘influence
over political decisions is diffused over a host of organized groups – trade
unions, business organizations, churches, societies for the promotion of this
or the protection of that – which express a variety of views and interests’
(Lively 1975: 58). And the idea, at least in some interpretations, is that if
people are generally represented in such interest-groups, then the influence
of those groups over government, proportional as it is likely to be to their
numbers, can be expected to be utilitarian in its overall effect: that is, to lead
to something like maximal preference-satisfaction overall.
This alternative, invisible-hand account of how the people might direct
government is open to even more salient objections than James Mill’s own.
For as many commentators have argued, there is absolutely no reason to
think that the influence of the different groups will be proportional to their
numbers (Connolly 1969). It is much more likely to be proportional to their
power over government, where that power is going to reflect a range of
possible factors: their level of organization and their financial resources,
their potential for campaigning against government or disrupting govern-
ment performance, and their usefulness at election time.
Starting afresh
The discussion of existing proposals for the popular direction of govern-
ment, whether intentional or non-intentional, seems to give credibility to
Schumpeter’s scepticism. There may be ways of establishing popular influ-
ence, even ways of making it into a very rich and penetrating form of
influence, as we saw in the last chapter. But going on the official record,
there appears to be very little prospect of using that popular influence to
impose a suitably popular direction on government.
The first of the two broad approaches sketched would organize govern-
ment around people’s public-interest judgements: their impartial preferen-
ces over what happens to the society as a whole. The second would organize
it under a mechanism designed to make government responsive to more
particular attitudes: people’s partial preferences over what happens to them
and theirs. But there seems to be little chance of arranging things so that
either socially oriented or self-oriented preferences can rule appropriately in
the corridors of power. The view that socially oriented preferences might
rule requires a dubious faith in the reliability of people’s virtue. The view
that self-oriented preferences might rule – that is, rule in a more or less
inclusive and fair fashion – requires an equally dubious faith in the capacity
of society to guard against the effects of unequal power.
Notwithstanding these criticisms, however, each of the approaches dis-
cussed has an appealing aspect. The first is attractive in focusing us on a
plausible goal for popular direction: that of the public interest, understood
in a post-social, non-corporate manner. The second is attractive in directing
us to a plausible means for popular direction: an invisible-hand mechanism
that does not require people to maintain monitoring and control of govern-
ment on an intentional basis. This raises the question as to whether there
might not be a third arrangement with both appealing aspects: a way of
directing government towards something like the public interest, as that is
understood in the first approach, that is implemented by something like an
invisible-hand process, as that is exemplified in the second.
As it happens, I think that this scenario is a real institutional possibility. It
will materialize, so I argue, insofar as two conditions are fulfilled. First,
people come to support certain norms of public policy-making, whether or
not this is something they intend, in virtue of pursuing popular influence
over government. And, second, these norms come to direct public decision-
making, whether or not this is something that people intend, thereby
establishing a popular purpose that government serves.
I look at how these two conditions can be fulfilled in Section 2 and at the
merits of the system of popular direction that this would install. And then in
Section 3 I explore the implications of the system for political ontology – for
our view of the nature and the relationships between people, constitution
and state – arguing that it allows us to think that in a distinctive sense
government satisfies Lincoln’s three demands: it is of the people, by the
people and for the people.
I describe the picture emerging from our considerations as a dual-aspect
model of how popular influence can impose a popular direction. This is
because democratic politics works in two timescales, according to that picture.
In day-by-day policy-making, the people exercise an equally accessible form of
electoral and contestatory influence over public decision-making. And by
means of doing this in the short haul, they manage over the long haul to
impose an equally acceptable direction on the performance of government.
2. a dual-aspect model of democracy
Games of acceptance and acceptability

The best way of introducing the core idea in the dual-aspect model is to
consider the difference between two ways in which a group of people who
have to organize some common enterprise might resolve conflicting differ-
ences of interest and opinion in the formation of policies. I describe one of
these approaches to organized decision-making as the game of acceptance,
the other as the game of acceptability.
Consider the group of people who own apartments in a condominium and
who face the task of organizing the condominium’s affairs.4 And suppose that
the group confronts a policy-making decision in some domain: say, the
decision about how much the owners of different apartments, some more
expensive, some less, should have to pay into the funds of the body corporate.
4
My use of the model of the condominium for thinking about republican political theory, which I
pursue here and later in the text, is mirrored in an intriguing way by a recent set of studies that look at
how far condominiums and related organizations live up to republican expectations in post-Soviet
Russia. See Kharkhordin and Alipuro (2011).
A dual-aspect model of democracy 253
In an acceptance game, the parties each seek to influence others in favour of
this or that policy by announcing the minimal concession that they are willing
to make, by revising their offers or bids in face of a failure to secure con-
vergence, and by going through successive rounds or revisions until conver-
gence is achieved. The drive in this exercise is to identify an arrangement that
each is willing to accept – in that sense, it is an acceptance game – assuming
that they each wish to grant the smallest concession that they can get away
with; they each want to make the smallest sacrifice possible to the interests or
indeed the opinions that they stand for. This exercise might be conducted
without any discussion in a round of bids akin to the bids made in opening a
game of bridge. But if discussion is allowed, then it will consist in self-serving
arguments by the different parties. Those on lower, less expensive floors might
argue, for example, that they are unable to concede further in view of their
relative poverty, that their interest in securing convergence is less than the
interest they are being asked to sacrifice – for example, unlike those on higher
floors, they can live without an elevator service – or that the remaining
concession required from the better off is really quite trivial.
In the acceptability game, things are done in quite a different fashion.
The parties are required to propose the policy that they favour in the
domain of group choice – or perhaps to suggest a novel candidate – and
to present considerations to one another that should count as relevant by
the lights of all. The idea is that any considerations adduced should help to
make the policy acceptable to everyone, given shared assumptions about the
dispositions of each; they should engage with those dispositions and help to
make the policy congenial to each. The exercise is much more demanding
than one in which they each just announce what they are prepared to
accept. And equally it is more demanding than one in which each puts
forward considerations that, as they claim, others morally ought to accept:
considerations that they ought to accept, for example, assuming that they
ought to share the speaker’s religious faith or ethical vision. The consid-
erations adduced should count as relevant according to everyone’s views but
according to everyone’s views as they actually are or can be brought to be,
not according to everyone’s views as in some sense they ought ideally to be.
The participants in the acceptability game, being required to treat others
as equals, will debate with one another on broadly the sort of model
envisaged by deliberative democrats.5 They cannot present considerations
that some subset of the group are bound to regard as irrelevant from their
5
The thinker who has done most to emphasize the potential of the acceptability game, though not in
that phrase, is Juergen Habermas (1984–9). Jon Elster (1986) gives an insightful presentation of the
point of view; certainly they cannot do so and expect to carry others with
them. Willingly or unwillingly, they are required to obey the old principle
of law and rhetoric (Skinner 1996: Part i): audi alteram partem, ‘hear the
other side’.6 The only considerations they can invoke are ones that argue on
all sides – though perhaps with a varying force or weight – for accepting the
policy supported; they must count with everyone as grounds that it is at least
relevant or pertinent to adduce in arguing for or against a policy. Being
presented as considerations that everyone licenses in this way, or can at least
be expected to license, the rival parties offer them as reasons for why
everyone ought to be ready to accept the policy proposed.
The acceptability game is governed, in the nature of the case, by a norm
to the effect that participants should only offer considerations for or against
a policy that all can regard as relevant. On the account adopted earlier, a
norm is a regularity of behaviour amongst the members of a group such
that, as a matter of shared awareness, almost everyone complies with it,
almost everyone expects others to approve of compliance and/or disapprove
of non-compliance, and this expectation helps to keep the regularity in
place. Under the pressures of the acceptability game, it is inevitable that
participants will generally comply with the regularity of seeking out con-
siderations that all others, no matter what their interests or opinions, can
treat as relevant in collective decision-making; else they will have little
impact. And it is equally inevitable that participants will register this fact
in common awareness as well as registering at the same time that any failure
of compliance will attract the inhibiting derision or disapproval of others.
Those who present considerations that can only carry weight with a partic-
ular subgroup will be laughed out of court.
approach that distinguishes it usefully from alternatives. Although the participants in an acceptability
game are required to be willing to operate on equal terms with others – that is built into the game – the
considerations that are likely to be valorized there should be distinguished from the ‘public reasons’
that John Rawls (1999) associates with ‘reasonable political conceptions of justice’ (133). Unlike Rawls’s
public reasons, the considerations endorsed may be context-bound in three distinct senses. They may
pass muster only because of the history of the group: an example, used later in the text, might be the
consideration, licensed in a history of federation, that argues for the equal representation of states with
different population levels. Unlike Rawls’s reasons, they may be required to pass muster at any time
with current members, not to be such that they would pass muster under any feasible change in or of
the membership; this limitation is not objectionable insofar as a change of that type, for example a
change in degree of religious uniformity, would generate a revision of accepted considerations under
the discipline of an acceptability game. And unlike Rawls’s reasons, they may be specifically tied to
particular issues about whether to adopt this or that policy or to endorse this or that process of
decision-making; they may not generalize into any easily formulated principles. For a good account of
Rawls on public reason see Larmore (2003).
6
For a recent psychological argument that this is an important and productive discipline on human
reasoning see Mercier and Sperber (2011).
This general, guiding norm – this norm of norms – does not require that
participants only offer considerations that have already been tried and
tested. It will prime them to search out considerations in any domain of
policy-making that people can be brought to endorse, given their existing
commitments. Suppose people are committed to the high-level principle of
equality in all possible domains. Or suppose, more plausibly, that they are
committed to equality in an area that is distinct from the domain under
discussion. On the basis of such prior commitments, the supporters of a
certain sort of policy in that domain may invoke the consideration of
equality, arguing for its relevance on the basis of the general commitment
to equality or the commitment to equality in a distinct but purportedly
parallel domain. They may or may not be successful in gaining such
acceptance for the relevance of the equality consideration in the domain
addressed, certainly not on the first round. But if the consideration really
does cohere well with the existing commitments of participants, then sooner
or later it is very likely to win acceptance as a relevant coin in the currency of
acceptability debates.
If people abide by the norm of norms in debates with one another – if
they invoke only considerations that may be expected to count as relevant
by all lights – then this should ensure the emergence of more specific norms
amongst them. Let participants be successful in getting the pertinence of
certain considerations accepted by all. That means that there will be
evidence to hand that makes it manifest to all that everyone accepts their
pertinence; manifest to all that this is manifest to all; and so on in the usual
hierarchy associated with common or mutual belief (Lewis 1969). Over
time, then, the considerations will come to constitute points of reference
that are manifestly pertinent or relevant, by everyone’s lights, to issues of
public policy.7 For any accepted consideration, C, it will be true, and true as
7
Does the dichotomy of acceptance and acceptability game neglect a third alternative in which it is
required that people all be able to see proposals as acceptable but not necessarily on the same grounds,
only on their own personal grounds? For a sophisticated development of such an approach see Gaus
(2011), and for a related approach that privileges religiously motivated, doctrinally specific reasons see
Stout (2004). My own view, which I cannot fully defend here, is that no such approach can point us to
a genuine alternative, in view of the following dilemma. If people do not have to defend the personal
grounds on which they require public policies to be acceptable – if, in a term introduced later in the
text, those personal grounds do not have to answer to concordant interests – then the proposal is likely
at best to support a version of the acceptance game, as participants seek to make compromises with one
another and avoid a destructive stalemate. If people do have to defend those grounds at a higher level,
however, on the basis of concordant interests – for example in the way that people might appeal to
religious freedom in defence of their own sectarian right of practice – then the proposal amounts to a
version of the acceptability game. For an illuminating discussion of this and related positions see the
unpublished paper by S. Macedo, ‘Why Public Reason?’
a matter of common awareness, that almost everyone treats it as a relevant
factor in political argument, almost everyone expects others to approve of
this treatment and to disapprove of a refusal to grant such treatment, and
almost everyone is motivated by this expectation to grant it this treatment
themselves.
Specific norms of argument and deliberation will emerge, on this
account, as an inevitable by-product of the successful, continuing use of
the acceptability game: that is, by robust adherence to the norm of norms
associated with the game. Participants will learn that as a matter of common
awareness this or that consideration is going to count as relevant – to pass
muster – in arguing for or against a policy. And with the approval or lack of
disapproval attending such a practice, they will learn as a matter of common
awareness that to reject the relevance of the consideration will earn the
inhibiting disapproval of others. Thus the consideration will be reinforced
by its own specific norm, coming to wear the tag: deny my relevance –
however you weight me – at your peril!
The considerations that an acceptability game is likely to valorize in this
way fall into two broad categories: considerations of convergent, and
concordant, interests. Convergent considerations will point participants
to universal benefits that all take to be relevant, or can be brought to see as
relevant, in matters of shared decision-making. Examples in the political
world would certainly include the benefit of equality or cohesion, pros-
perity or peace and, of course, the corresponding harms of inequality or
divisiveness, poverty or conflict. If you can show that only one of a
number of competing policies in any area promises to provide such a
benefit or avoid such a harm, then you will have registered something that
ought to carry weight on all sides, though perhaps heavier weight on some
sides than on others.
Concordant, as distinct from convergent, considerations point the
participants in an acceptability game to benefits that accrue only to this
or that individual or subgroup. Despite the partiality of their relevance to
people’s fortunes, these considerations will command a following on all
sides to the extent that everyone accepts that it is a matter of convergent
interest that the group as a whole should confer that benefit on the sort of
individual or subgroup favoured. No group can accept that the interest of
each in securing a certain advantage, regardless of the cost to others,
provides a relevant consideration in joint policy-making. But a group
may well accept, for example, that none of its members should have to
suffer the sort of inequality that majority voting would impose on the
vegetarian members in the dining-club example of the last chapter. More
generally, the group may accept that members have a concordant interest
in being protected against any such inequality, or against any of a variety
of misfortunes. Thus a likely harm to one or another member on this front
may pass as a relevant consideration in joint policy-making. And some-
thing similar may hold with a suitable benefit. For example, the fact that a
benefit would help some while not doing any harm to others – the fact that
it would represent a Paretian improvement – may argue on all sides for a
policy that promises that benefit.8
Whether they reflect convergent or concordant interests, the more
specific norms of policy-making will direct participants in any acceptability
game to an evolving body of arguments that they can safely draw upon in
supporting this or that proposal. But they will also provide resources that
participants can exploit in supporting this or that process or procedure for
resolving the differences that remain in place after deliberation. Differences
will almost always remain in place, since the considerations presented on
different sides of a policy debate will rarely underpin unanimous agreement;
they will only reduce the number of candidates on the table. And so it will
be necessary at some point to bring the deliberative exercise to an end and
achieve resolution by recourse to some independent procedure: say, by
voting amongst the remaining alternatives, by selecting between them on
a lottery basis, or by referring the issue to an agreed umpire. Thus, as specific
norms of policy-making will bear on different policies, so they will also bear
on the suitability in a given case of resorting to one or another process of
final resolution.
The specific norms that emerge in any group will never constitute a
closed set, fixed once and for all (Young 1990, 2000; Honig 2001). They will
be subject to constant development, as deliberative innovators manage to
gain acceptance for novel sorts of argument, perhaps by extending the reach
of recognized arguments to new domains, perhaps by coming up with new
arguments that gain acceptance by others. Such innovations are likely to be
triggered by changes in the dispositions of the existing membership and, of
course, by changes of membership that occur at any time and across differ-
ent times. As new norms evolve in this way, others may decay and lose
potential, say because they are not acceptable to some members in the
changed society. But in general we would expect the norms that are given
8
Notice the difference between the following claims. One, the subgroup should not be expected to bear
a certain cost by criteria that no one in the society who is willing to live on equal terms with others
would reject. Two, the subgroup should not be expected to bear the cost by criteria that no one could
reasonably reject: by criteria that no one could reject while counting by some independent standards as
a reasonable person. It is the first claim that is relevant in this discussion, not the second.
countenance to increase in number and application, being laid down like a
sediment deposited by the flow of debate and exchange.
Returning to our earlier example, imagine the policy-making norms that
are likely to gain hold amongst the owners in a condominium of apart-
ments. Some of those norms will be established in advance under contracts
of purchase, but even if they are not pre-established, we can easily imagine
them emerging and multiplying as members of the condominium come
together to work out the terms on which they are to live together. As owners
talk, try out proposals and then vote on remaining candidates, they will
inevitably establish norms that rule their communal life together. These will
determine:
* The reasons why they should collectively organize the servicing of
elevators, the tending of common areas, the maintenance of a sinking

fund, and such endeavours.
* The features of collective decision-making that are acknowledged on all
sides as desirable: openness, consultation, deliberation, efficiency, and

the like.
* The duties of individuals in relation to the group in the matter of body
corporate fees, committee participation, general civility, etc.

* The equality of individuals in collective decision-making, despite the
variations in the fees paid by the owners of different apartments.

* The rights of individuals to furnish and decorate their apartments
internally to their own taste.

* The benefits that individuals can expect to be able to claim from the
group corporate in dealing with particular problems like ground-floor

flooding.
* The penalties that ought to be meted out to apartment owners who
breach commonly established procedures.

* The value of giving authority to an elected committee, while allowing for
challenges by other members to its resolutions.

* The utility of outsourcing certain decisions to independent advisers and
auditors, given possible conflicts of interest.

The fact that a group operates under an acceptability regime in which
certain norms of argument dictate the terms of argument and association
does not mean that members will always behave in a saintly, or even a
salutary, manner. Members may enter public debate and decision-making
only reluctantly, preferring to allow things to get determined in many areas
by default. When they enter debate or register dissent, members may not be
moved by an impartial concern for the common good; so long as they are
disciplined by the rules of the acceptability game, their fundamental
motivations may be more or less self-seeking. And in debating or commu-
nicating with one another about different measures, members need not
argue in the formally correct manner appropriate to a seminar. Thus, to
elaborate on this last point, the arguments in which common norms are
invoked may be implied in the particular observations that people make, in
analogues from other areas that they suggest, or just in the unflattering cast
that they give to the claims of opponents. The long tradition of political
rhetoric shows us how any of a variety of such interventions, ranging from
the wittily inverted phrase to the shaft of irony, to the hyperbolic gloss on an
opponent’s claim, to the subtle recasting of negatives and positives, to the
interjection of a vivid example, even to the use of outright mockery, can
activate accepted norms – perhaps fairly, perhaps unfairly – and provide
support for one or another policy or process.9
But while the operation of an acceptability regime in a group like the
condominium does not mean that things will be done in the most saintly or
salutary fashion, the very fact of adopting the acceptability game does testify
to a commitment to seeking solidarity in collective decision-making. The
members of the condominium must be prepared to make their local
community work, and work in a manner that gives each a role in commu-
nity governance. By their very participation in deciding on common
policies, or in revising the processes under which policies are decided,
they show that they are ready to accommodate themselves to others in
various ways, making whatever compromises and concessions are required.
They are willing to think from the point of view of the group as a whole,
rather than living in isolation and resentment or seeking after a community
that is more congenial to their tastes. This commitment to the group
amounts to a local counterpart of what would count as patriotism at the
national level.
The acceptability game in the public world

Is politics going to generate games of acceptance, or games of acceptability,
or a mix of both? In particular, which sort of game is it likely to favour if a
regime of popular influence is established on the lines described in the
previous chapter? Games of acceptance may play a subsidiary role in any
political context, as when political parties negotiate on such a basis within
a parliamentary context. But if a regime is designed to facilitate an
9
For a good introduction to many traditional styles of rhetorical argument see Skinner (1996); on the
potentially benign use of rhetoric see Garsten (2006) and McGeer and Pettit (2009).
individualized, unconditioned and efficacious form of popular influence, as
our model requires, then it is bound to give a prominent role to games of
acceptability and is bound therefore to generate a range of commonly
accepted norms of policy-making.
The dominant, unifying theme in the system of popular influence that
we sketched in the last chapter is that there must be room at many
institutional sites for challenge and contestation, discussion and decision-
making. We argued for a responsive, as distinct from an indicative,
legislature on the grounds that it would allow the collective interrogation
of elected authorities, reinforcing the freedom of speech, association and
travel that this presupposes. We argued that in order to guard against the
possibility that votes would not have equal value, so that the system
would fail the requirement of individualized influence, this system of
collective challenge has to be complemented by a system of individualized
contestation, allowing individuals and groups of individuals to challenge
any problematic measures proposed or imposed by government. We
argued that in order to give this system of popular influence an uncon-
ditioned or independent character, government would have to be con-
stituted out of distinct, mutually checking or contestatory parts, rather
than having the power of an absolute sovereign, and the people would
have to exhibit the deep, contestatory power – premised on a patriotic
commitment to the success of the community – that specialized public-
interest groups might provide. And we argued, finally, that in order to
make popular influence efficacious, it is essential to guard against the
electoral motives of politicians, establishing independent, indicative
authorities in their place, to guard against the influence of private-interest
groups by using such authorities to constrain their operation, and to
guard against abuse on the part of those authorities by subjecting them to
contestation in their own right.
The multi-dimensional, multi-centred pattern of contestation required
for a suitable system of popular influence ensures that the acceptability
game must dominate the operations of that system. Contestation is possible
within an acceptance game as one or another party digs in on the con-
cessions it is willing to make, relying on its wealth or size, or just its
reputation for dogged intransigence, in order to oppose a resolution sup-
ported by others. But this sort of contestation would privilege the indivi-
duals and groups with greater bargaining power and expose political life to
the chaos of random power plays. The only hope of getting individuals to
share equally in a system of popular control is to organize the influence they
exercise under the protocols of an acceptability game.
The acceptability game is quite consistent, as we saw, with an eventual
resort to voting, or to any similar mechanism of resolution. But wherever it
is in place, it will force people to interact in a manner that gives rise to shared
policy-making norms. We may expect such common norms to get esta-
blished at the multiple sites of opposition and contestation envisaged in the
system of influence described. These sites include electoral campaigns and
debates, judicial and cognate hearings, parliamentary discussions, exchanges
between branches of government, public justifications of policy, disputes
between watchdog bodies and government agencies, interactions in the
press and other media, judgements or opinions offered by independent
authorities, and popular campaigns for constitutional change. At each of
these sites the final resolution of an issue may depend on a blunt instrument
like voting and may be determined by the numbers available on any side.
But the requirement for the parties on each side to defend their partisan
proposals in multi-partisan terms ensures that as they construct their differ-
ent programmes, they will lay down a foundation of common ground
between them. The dissensus with which they end may be a failure in
one dimension but it is going to represent an achievement in another: it will
secure or reinforce the norms of argument that the disagreement drives the
different sides to identify.
Not only will policy-making norms begin to emerge and crystallize at
each site where contestation is brought or heard, answered or adjudicated.
Those sites will naturally be accessible to one another, so that norms
endorsed in one context are always going to be available for trialling, and
acceptance or rejection, at another. The identification of an accepted norm
in one context is going to present the parties who operate in any other with a
potential resource for making their case on this or that question. And so we
should naturally expect every candidate that proves acceptable at one site to
become rapidly available for testing at other sites where government deci-
sions are exposed to challenge. Norms accepted at one site, of course – say,
amongst the members of a particular church or culture – may not prove to
be acceptable at others. But the multiple connections between sites –
connections enforced in a public, electoral culture – ought to ensure the
identification of norms that have society-wide credentials.
There is little prospect of differences running so deep that few if any
norms of that kind emerge.10 Or at least there is little prospect of this to the
extent that people on all sides of the society remain committed, on a
patriotic basis, to making their community work. Patriotism of the kind
10
For a different point of view see Gaus (2011).
envisaged here requires people not to allow their more personal or partial
attachments – say, those centred on religious affiliation, ethnic identity or
geographical location – to undermine their commitment to the larger
community.11 Let patriotism prevail and there are unlikely to be divisions
of the kind that would undermine the emergence of society-wide norms.12
What are the norms that we might expect to emerge under a regime
where things are organized to give people individualized, unconditioned
and efficacious influence over government? A first is a norm, broadly
speaking, of equality of influence. The very possibility of individualizing
influence, as we saw in our earlier discussion, requires a recognition on
people’s part of the relevance of any argument, based on the need for
equal access to influence. If people are to guard against the possibility that
a sticky minority will be deprived of all influence, for example, then they
must be required to recognize the importance of avoiding the margin-
alization of such a minority. The very possibility of a system of indi-
vidualized influence – and so the very possibility of a legitimate
government – turns on the extent to which such a norm of argument
has a hold on people’s imaginations and dictates their responses. People
must be committed to the possibility of their particular community being
governed under a regime of equally shared influence – in that sense they
must be patriotically motivated – and they must be willing to take the
electoral and contestatory steps required to uphold such a regime.
Otherwise, by the line of argument developed here, legitimacy in govern-
ment is going to be an unattainable ideal.
But if people’s claim to equally accessible influence – equal respect in
collective decision-making – is recognized on all sides, then so, plausibly, will
be a more general claim to equal respect in how they are treated under
collective decisions. This claim to equal respect will be registered as a reference
point in public contestation and argument (Beitz 1989). And the acceptance
of the claim is likely then to generate a series of more specific egalitarian
norms, as ongoing exchanges and institutions establish interpretations of the
demands of equal respect in schooling opportunities, conditions of employ-
ment, the operation of the courts, the organization of public facilities, and so
on. These interpretations may vary across societies, of course, and indeed
across eras. Thus the interpretation of the demands of equal respect shifted
11
For an illustration of such a breakdown in a small corporate body, amounting to what I describe as
collective akrasia, see Pettit (2003a).
12
As noted earlier, however, recent changes in broadcasting create a danger that those in different
political parties will only listen in on news and debates in their particular niche (Sunstein 2009).
dramatically in the United States with the rejection of the idea that races
could be separate but equal. And that shift of interpretation released a cascade
of changes in the way things were done in public life.
As there are likely to be various norms of equality under the system of
influence described in the last chapter, so there are likely to be various norms
of freedom. The idea of having a system of popular influence that is indivi-
dualized, unconditioned and efficacious goes naturally with recognizing the
need for each to have a sphere of personal choice and with an interpretation of
the freedom that people should have in that sphere. As with equality, the
demands of freedom may be differently interpreted in different polities and
different periods, but they are likely in any society to enforce some boundary
between the domain where people ought to be able to make their own
decisions and the domain where government or other authorities can rule;
to establish some sense of what is required for people to be able to make their
own decisions in this way, enjoying a relative independence from others; and
to license this or that level of government resourcing and protection as a
means of establishing such independence for all.
A dispensation for ensuring a suitable degree of popular influence has to
assume institutional form at some point, with the introduction of measures
that provide for election, challenge and interrogation; that limit the author-
ity of elected officials and regulate the activities of private lobbies; and that
delineate possibilities of legitimate opposition and contestation. And as the
institutions required emerge and stabilize, they are bound to gain accep-
tance and to license associated norms of argument. Those norms will
support appeals to the principles and precautions that the institutions
embody, such as importance of the separation of powers, the independence
of the judiciary, the transparency of government decisions, and so on. The
precise institutions introduced in any society may vary from the institutions
introduced in others, of course, and as they vary, the corresponding
institution-specific norms will vary too.
Apart from the equality-based, freedom-based and institution-specific
examples just canvassed, a host of independent norms are going to emerge
in any society with growing convergence on the question of what the state
ought to do and with institutions that implement such convergence. The
business that government is charged with taking up will certainly include
the defence of the country, the identification of the basic liberties, the
enforcement of law and order, and the facilitation of industry, commerce
and employment. But in most societies it will also extend, under this or that
interpretation, to the provision of education for the young, the regulation of
various markets, the insurance of communities against catastrophe, and the
insurance of individuals against urgent or pressing need, medical, legal or
economic. The list is more or less open-ended.
I have been speaking about the variety of norms that we might expect to
develop in any society, particularly in any society with a system of popular
influence that is designed to be individualized, unconditioned and effica-
cious. There is a common character that we might expect such democratic
norms to display, as I have stressed, but the precise content they assume will
always depend on contingencies of local culture and development. But I
should not suggest that democracy will be free of faults and fetters. The
norms that emerge in any society, reflecting some perhaps questionable
beliefs and values, may often be deficient, judged from the point of view of
this or that conception of justice or legitimacy.
While the history of a democratic society will typically offer a rich and
appealing set of norms to invoke in public life, it can also saddle those who
live in the society with unwholesome constraints and unwelcome fetters.
Every developed democracy will have evolved along a certain trajectory,
being built upon a tradition of letting government take charge of such and
such matters and being associated with more or less settled institutions and
arrangements. This can put a society in a bind, locking members out of the
possibility of arguing for what by most of their shared norms of argument
would constitute an improvement. Thus, to give a simple example, it was
essential in the founding of the United States, and at the time of Australian
federation, that large and small states were given equal representation in the
Senate; this is unsurprising, since they were the parties responsible for
creating the new political order. This pattern of representation would be
unlikely to command much support in either country today, if the con-
stitution were being redrafted from scratch. The citizens of distinct states
within both the United States and Australia are geographically mobile and
probably do not identify as strongly with their local states as their prede-
cessors would have done. But nonetheless, there is no immediate prospect
of an alteration in either regime. The norm of equal state representation has
a pedigree that puts it beyond effective interrogation.
Deliberative regulation by policy-making norms

These observations support the idea that in any regime that implements
something like the popular system of influence that democracy requires,
there are going to be norms of public policy-making that get to be accepted
on all sides. The question we now confront is whether those democratic
norms might exercise a directive or shaping role over policies. Might
popular influence serve to promote compliance or compatibility with
policy-making norms?13
The policies adopted in different regimes of popular influence, and the
processes under which they are adopted, are likely to vary greatly from one
another. Thus the norms that are endorsed across different regimes, shaping
the policies and processes adopted, may diverge significantly. Again, empiri-
cal assumptions are required to determine the implications of any set of
norms for the actual policies or processes put in place and these, too, may
vary across societies. And the mechanisms of final resolution that inevitably
play a role in determining policies may also push societies in different
directions: resolution by majority voting or an impartial umpire can be as
indeterminate as resolution by lottery.
But consistently with such variation it can still be the case that the
policies and processes adopted in any regime are subject to the constraining
or guiding effect of policy-making norms. Those norms will constrain and
guide policy-making to the extent that their presence more or less ensures
that two conditions are fulfilled. First, a policy-condition: no policies are left
on the table as possible candidates for adoption in any domain, if they are in
violation of the norms. And second, a process-condition: no process for
selecting between candidates that survive that first cut – and no process for
selecting between rival processes – is employed, if its employment in that
domain would be in violation of the norms.14
Assuming that policy-making norms impose effective constraints, as
they are surely capable of doing, the dispensation depicted here is going to
give them a substantial directive role. Going back to the example of the
condominium, we can imagine a regime emerging in which things are
reliably done to the satisfaction of the norm-based expectations of owners.
Under the impact of those norms, the owners come to establish certain
decision-making processes and, invoking relevant norms, to build up an
accumulating body of policies. The organization will run under the
13
The role that norms play in democratic control, in the story developed here, overlaps in part, but only
in part, with the role given to people’s values in other theories. See Brettschneider (2007) for one such
approach and Eisgruber (2001, 2002) for another.
14
We saw earlier that in any domain of decision-making a system of popular influence will require the
generation of proposals on the one side and the testing of the proposals generated on the other; the
generation would guard against false negatives or shortfalls and the testing against false positives or
errors. The policy-condition and the process-condition required for giving norms a directive power,
promoting the goal of policy-making norm-compliance, both bear on the testing rather than the
generation required under such a generate-and-test programme. They would be satisfied by virtue of a
filtering or editorial procedure in which norm-incompatible policies and processes are taken out of
play and only norm-compatible candidates are allowed to survive.
influence of those members – and, as we may assume, their individualized,
unconditioned and efficacious influence – and that influence will have the
effect of imposing compliance with the norms that they endorse. The
members will disagree strongly on particular issues, of course, perhaps
dividing into bitterly opposed factions. But to the extent that their differ-
ences continue to be debated and resolved without offending against any
of the norms they hold in common, the members will have succeeded in
establishing a common norm-based order.
Suppose that one party wants to allow some of the apartments in the
building to be rented out to a nearby hotel and the other party is strongly
opposed to this development. And imagine that they make the decision in a
way that conforms to shared norms. Whichever party wins in that case, the
members of the condominium will have achieved an important goal
together: the compliance of their policies and policy-making with norms
that are shared amongst them. They may not come to a common mind on
the particular issue on hand but that will not be a problem so long as the
final decision is made on the basis of norm-compatible process. Suppose
that the numbers for and against the hotel proposal break even and the
decision is made on the basis of a coin-toss, where this process of decision-
making is itself consistent with the accepted norms. The result of such a
chance device will not undermine the common achievement of norm-
compliance. And neither would the victory of one or the other side under
any other accepted process – say, voting on the committee, or in a refer-
endum. One party will prevail on the issue, and perhaps on a range of
related issues, but at a deeper level it is the members of the condominium as
a whole that rule. Their common norms, not the norms of any particular
faction, shape the direction in which the organization moves.
As the norms emerging amongst the members of a condominium can
shape the policies and processes that prevail there, forcing whatever deci-
sions are made to conform to a template they impose, so we might expect
the norms of a political community to have the same directive impact on the
operation of government. The decisions taken by government may vary
enormously, depending on what particular norms have been endorsed, on
who happens to occupy elected or unelected office, and on what those
officials come to decide on specific issues. But if the emergent norms are
truly effective, as they ought to be in the presence of a suitable constitution
and citizenry, then they should put a directive and controlling stamp on
what is collectively done in the community. They should filter out offend-
ing policies and processes, making room only for modes of decision-
making, and actual decisions, that fit with accepted standards.
The regulation its members impose on their condominium, or its citizens
on their government, can be described as deliberative regulation, since the
norms that they deploy all emerge as the by-product of deliberation within
the organization. But the deliberative regulation of the condominium or
government is not likely to come about as a result of every decision being
taken on an explicitly deliberative basis. Since the presence of public
deliberation on a number of occasions at a number of sites is enough to
generate the required norms, that is all it presupposes. And since it will
equally be enough that the norms have been generated and have come to be
generally accepted, even such episodic deliberation will be less and less
required as the condominium or society develops. The regulation will
operate, not on the basis of continuing, ubiquitous deliberation, but rather
as the product of institutional constraints that reflect deliberative norms.
The decisions the body eventually reaches will have to flow along so many
channels, skirt so many potential checks and obstacles, and avoid flounder-
ing on so many contestatory hazards that they are more or less bound to
satisfy the requirements of deliberatively generated norms.
How distinct is this ideal of deliberative regulation from that of deliberative
democracy? It is built out of the same observation, familiar in particular from
the work of Juergen Habermas (1984, 1989), that when people debate with
one another about what they should combine to do, then they have little
option but to seek out considerations that all can see as relevant to a joint
enterprise (Elster 1986). As we know, these may be convergent considerations
identifying benefits they can all share in, such as safety or prosperity or
cohesion. Or they may be concordant considerations licensing the provision
of benefits for a subgroup in a recognized sort of predicament, as when its
members suffer a deprivation that is seen as problematic on all sides, or are in a
position to improve their lot without doing any harm to others.
But while the approach taken here is deeply continuous with the spirit of
deliberative democracy, there is a break at two distinct levels. At a founda-
tional level, the approach is inspired by the republican insight that people
must share equally in their control of government if they are to avoid
domination, and not by a foundational commitment to the value of
deliberation as such (Cohen 1989).15 At the operational level, it is organized
around the idea that it is the deliberative regulation of public business – that
15
Jake Zuehl has persuaded me that in the later papers in Cohen (2009), the foundational commitment
to the value of actual deliberation gets to be qualified in a contractualist way, as the reasons that are
taken to count get to be indistinguishable from the Rawlsian public reasons mentioned in an earlier
footnote.
is, regulation by deliberatively tested norms – that is essential, not the
deliberative conduct of decision-making at every site and on every occasion.
The difference of perspective and practice between the two approaches
shows up in the fact that whereas dissensus always represents a second-best
for deliberative democrats, it is entirely acceptable, even desirable, within
the present approach. It is the experience of dissensus, after all, that drives
the different sides to identify common policy-making norms, as they try to
make sense of their respective views to one another. And it is those norms
that constrain the policies that they are willing to entertain and the processes
of decision-making they are ready to employ.
The fact that the deliberative regulation of collective decision-making
does not entail a great deal of deliberative conduct also has a further
implication. It means that the members of a deliberatively regulated
group, unlike the members of a group that conducts all its business in
explicit exercises of deliberation, may not always be conscious of the
deliberative regulation under which they operate. The impact of deliber-
ative norms may be elusive, indeed almost invisible, if the system is working
well. Assuming that it is working at its best, the institutions established will
not be called into question; the arguments put forward at various sites will
be weighed differently but admitted as relevant on all sides; the proposals
made by one or another side will be opposed by others but not condemned
as normatively objectionable; and the final resolution of differences will be
achieved under undisputed processes of decision-making. In short, the
conduct of public business will run along straight and even tracks without
any evidence of being directed onto those paths by a body of popular norms.
The direction of public business by popular norms will become salient
only when things do not go so well in one or another forum and participants
launch norm-based objections to the content of a policy or to the process by
which a policy is determined. But since participants will presumably want to
avoid such objections, they have an incentive to stay within limits that keep
them safe from such retort. And when they do genuinely stay within such
boundaries, the system will be working fairly well; it will be displaying a
high level of compliance with policy-making norms.
Although popular norms may not have a salient presence in a system that
is working well, however, this is not to say that they do not have an impact
there. The impact that they will have under such an ideal scenario, to return
to an idea already introduced in other contexts, will have a virtual or even
reserve character. To take the virtual case, it will be like the impact of the
cowboy who rides herd on his cattle. The cowboy may do little or nothing
by way of guiding the cattle; he lets them have their head, given that they are
on the right path. But he clearly has a determining influence on the path
they take because he is ready to intervene effectively if any of them should
wander off the track. The same is going to be true of the influence that
policy-making norms have in the scenario presented. As policies are pro-
posed, defended and challenged, and as they are finally selected under one
or another process, there will be players within the system who are ready to
cry foul if a policy or process looks to be in breach of accepted norms. And
so even when those norms do not play an active role in the shaping of what
transpires, they can still have a powerful effect. They can ride herd, as it
were, on the process of policy-formation.
Consider the condominium example again. As insiders or outsiders
contemplate what happens in the day-to-day working of the organization,
the salient features will be the routines that are more or less unthinkingly
followed, the ways in which members come to differ on various aspects of
policy, the arguments that they each invoke in support of their views, and
the mechanical resolution to which most of their differences are eventually
subjected. Nowhere in this manifold of observations will there be positive
evidence of the role of the norms to which members subscribe. And yet
those norms will be in place to outlaw any norm-incompatible routines, and
to rule out any arguments – and any proposals based on arguments – that do
not satisfy the norms. The eye-catching exchanges will all take place within
a field of interaction that is shaped by the norms, but the impact of the
norms themselves may be close to invisible. Even when the waters of politics
run at their most turbulent within the condominium, the turbulence will be
consistent with the existence, beneath the surface, of a common stream of
influence and direction.
Dual-aspect democracy
We have argued that any regime of suitable popular influence will give a
powerful role to acceptability games; that such games will give rise to
widespread adherence to policy-making norms; that those norms will
tend to shape policy-making by putting norm-incompatible policies and
processes off the table; that they may operate to this effect without a great
deal of explicit deliberation taking place within the system; and that when
they operate well in achieving the effect, their impact will often be hard to
discern: the influence they support may often have a virtual or reserve
character.
The lesson of these observations is that a regime that is deliberatively
regulated is likely to operate in two time-scales, whether it materializes in a
private association like the condominium or in the political community at
large. It will require the short-haul, highly charged process of campaign
and election, proposal and counter-proposal, debate, division and contest-
ation. But it will achieve its signature impact in the long-haul process, as
silently as gravity, whereby participants are wittingly or unwittingly led to
establish only such policies and processes as conform to the norms of
argument and association that prevail amongst them. Over this long haul,
a deliberatively regulated politics will generate and regenerate a supply of
publicly valorized considerations. These will serve at any particular time to
keep an indefinite number of policies and processes off the table, render-
ing them unthinkable and invisible. And they will combine from time to
time to make certain existing policies or processes suddenly seem intoler-
able and in urgent need of repair.
This image of a deliberatively regulated regime constitutes what I
describe as a dual-aspect model of democracy. Apart from the fast democ-
racy of election and contestation that commands most attention amongst
political participants, observers and theorists, it directs us to a slow demo-
cracy that works under suitable forms of popular influence to impose a
distinctive stamp on what government, over the long haul, effects.
In the nature of the case, evidence that a political regime conforms to the
dual-aspect model, constituting a deliberatively regulated democracy, is
going to be hard to identify. The claim that the regime imposes deliberative
norms on government is going to be particularly difficult to establish, given
that such regulation will only be evident over the long haul. In order to
overcome this problem, and illustrate the operation of something like the
dual-aspect model, I must resort to history, in particular the sort of history
that looks at the longue durée.
I think the reality of slow democracy, and the viability of the dual-aspect
model, is well supported in the work of Oliver MacDonagh (1958, 1961,
1977) on the massive shifts in the activities of British government over the
half century or so after the debates about the Great Reform Bill of 1832. I
pick this example for two reasons. One, it is distant enough in time and
culture to allow us to get a clear view of what transpired. And two, it was a
period in which the rise of democracy meant that there was much to
achieve – not just, as is now often the case, much to maintain or recover –
in order to keep the government in line with popular norms. While the
franchise was not greatly extended in 1832, electoral democracy became
more and more established as a guiding ideal in nineteenth-century Britain
and later extensions of the franchise were heralded by an enormous rise in
public petitions and public demonstrations (Knights 2005).
The nineteenth-century transformation took England, and Britain more
generally, from a state that had become minimal by any standards to a
regime in which government assumed responsibility for regulating a great
swathe of social behaviour, establishing offices that imposed strict inspec-
tion and control over such matters as the employment of children, the
treatment of women, the preparation of food and drugs, the operation of
the civil service and, relatedly, the conduct of affairs in mines, mills,
factories, ships, railways and public offices. I do not argue here that this
growth in the administrative responsibility of the state was desirable, though
to my eye much of it clearly was. Drawing on MacDonagh’s work, my only
aim is to use this shift to illustrate the depth at which policy-making norms
can operate in a broadly democratic context to impose a slow, long-haul
direction on government – and often, as in this case, on a deeply resistant,
even resentful administration.
The early nineteenth century in England was a period of enormous social
problems. These were occasioned by a number of connected factors: the
introduction of steam-based industrialization, the vast increase in popula-
tion, the capacity of people to move into new areas, even across oceans, and
the ensuing concentrations in industrial cities and towns; in the first quarter
of the nineteenth century, many towns in Yorkshire and Lancashire grew by
a factor of three or four. With these developments, a more or less pater-
nalistic tradition of management in agriculture and village industry gave
way to the ruthless discipline of Blake’s dark satanic mills. As Sir Walter
Scott wrote in 1820: ‘the manufacturers are transferred to great towns, where
a man may assemble five hundred workmen one week and dismiss them the
next, without having any further connexion with them than to receive a
week’s work for a week’s wages, nor any further solicitude about their future
fate than if they were so many old shuttles’.16
While the problems that arose in the new industrial world were daunting
and varied, the doctrine of laissez-faire individualism that prevailed in
official circles might well have left them unaddressed. As democratization
increased, however, the pressures brought to bear on government pushed it
slowly and reluctantly into becoming something unimagined at the begin-
ning. In MacDonagh’s (1958: 57) words, they occasioned a ‘transformation,
scarcely glimpsed till it was well secured, of the operations and functions of
government’.
The transformation was effected now in the mines, now in the factories,
now in ocean-going ships, having an impact that affected the lives of men,
16
Quoted by MacDonagh (1977: 2).
women and children across the society. In each case, the transformation was
prompted by the same factors and developed along the same lines.
The factors that drove the changes, according to MacDonagh (1958: 57–8),
were ‘the increasing sensitivity of politics to public pressures’ that we
would expect with growing democratization and ‘the widespread and ever-
growing influence of humanitarian sentiment and of stricter views of
sexual morality’. The humanitarianism was particularly important, bub-
bling up in ‘the raw and immediate reactions of a variety of ordinary and
thoroughly representative people’, for it managed to overcome the resis-
tance of laissez-faire doctrine and commercial self-interest: ‘the general
level of compassion in contemporary England overswept policy and
commerce’ (MacDonagh 1961: 330). This was not just a dumb, pre-
conceptual sort of compassion; it was organized around concepts and
norms that had deep roots in the political culture. Many of the abuses
condemned were cast as abuses against freedom, for example, and as
analogues of enslavement. In a famous newspaper letter of 1830, a
reformer, Richard Oastler, could write: ‘Thousands of our fellow-
creatures and fellow-subjects, both male and female, the miserable inhabi-
tants of a Yorkshire town . . . are this very moment existing in a state of
slavery, more horrid than are the victims of that hellish system colonial
slavery. These innocent creatures drawl out, unpitied, their short but
miserable existence.’17
The different changes that occurred in nineteenth-century England
were generally prompted, then, by a democratically empowered humani-
tarianism. But not only did the changes have a common source, according
to MacDonagh, they also evolved under a similar dynamic and in parallel
stages.
In each case the change began with the revelation by newspapers or
reform organizations of just how scandalous conditions were in this or that
domain; in each case this scandal led to a degree of public outrage amongst
the population at large; and in each case that outrage prompted a reaction
from government, in particular an initiative designed to put things right.
‘Once it was publicized sufficiently that, say, women on their hands and
knees dragged trucks of coal through subterranean tunnels, or that emi-
grants had starved to death at sea or that children had been mutilated by
unfenced machinery, these evils became “intolerable”; and throughout and
even before the Victorian years “intolerability” was the master card’
(MacDonagh 1958: 58).
17
The letter is to the Leeds Mercury, 16 October 1830; see www.makingthemodernworld.org.uk.
The dynamic that called for such reform never played itself out in a single
round. Almost invariably it transpired some years after the first initiative
that the law had been ineffective and that the scandal remained. And with
that second revelation the dynamic ran its course again, leading to the
appointment of executive inspectors and officers who were charged with
putting the law into effect. Nor was that all. The reports of these admin-
istrators fuelled yet further developments, as it became clear that problems
were proving recalcitrant, and this third iteration of the dynamic generated
a progressively more professional and better-organized public service. The
culmination of such transformations across a variety of domains was the
slow emergence of a new and characteristically modern sort of state.
The impact that popular pressure can give to policy-making norms is well
illustrated, I believe, in this narrative. The norms had a background
presence in people’s attitudes, which survived ideological and party differ-
ences. And they had a slow, relentless effect on how government operated,
pushing it inexorably towards a certain destination. The direction in which
a gradually democratizing government was pushed in Victorian England is
particularly salient to us, given the contrast, at least in the developed world,
between those times and ours. But I believe that it illustrates the sort of
purpose that democracy can serve everywhere in imposing accepted norms
on government. It shows that the dual-aspect model described here is not
necessarily a pipedream.
No illustration of the model is going to be ideal, of course, and there are a
number of particularities about this example that we should note. One is
that while it exemplifies the role of accepted norms in putting certain issues
on the agenda of government and in prompting improvements to a clearly
imperfect practice in that domain, it does not illustrate the less salient role
of norms – so important in any advanced democracy – in maintaining
improvements already in place. A second is that while it exemplifies the role
of accepted norms in putting certain policies off the table – those, for
example, that would allow for the employment of children in mines – it
does not illustrate their equally important role in putting certain processes
or procedures of decision-making off the table. And a third is that it does
not make clear that the satisfaction of humanitarian norms achieved in the
long-haul development described was a by-product of debates that were
focused on rather more specific, short-haul policies. Those norms had a
slowly mounting impact, not because humanitarian norm-compliance was
a target at which anyone aimed, but because it was a common, barely
recognized constraint on the concessions and adjustments that politicians
were forced to make under the pressures of day-to-day politics.
Notwithstanding these features, the MacDonagh studies serve us well in
illustrating how popular pressures can empower norms, impose them on
government and make them into a more or less fixed feature of how things
get to be done. They display the workings of a more or less deliberatively
regulated democracy, as the various individuals and bodies involved in its
short-haul operations are pushed under the pressure of popular influence to
adopt a discernible, long-haul direction.
The sort of thesis that MacDonagh defends is borne out in other
contexts too. Thus, to mention one important example, William
Eskridge and John Ferejohn (2010) argue in a detailed examination of
cases that popular pressure in the United States has led to the statutory
implementation and effective entrenchment of norms of equal citizenship,
market openness and personal security. In an uncanny parallel to
MacDonagh’s narrative, they describe a process in which a social move-
ment or other pressure creates a demand for state action; publicly sup-
ported legislation generates a statute embodying a new norm; the statute is
administered and expanded with feedback and pushback from various
sectors of the community; the norm is revisited and reaffirmed by the
legislature in face of opposition; and this is followed by further admini-
strative elaboration, further feedback and pushback, and further legislative
revision (Eskridge and Ferejohn 2010: 19–20). Like MacDonagh, they
illustrate the slow emergence and impact of popular norms that I see as
evidence for what democracy is capable of achieving over the long haul in
imposing a direction on government.
Our discussion points us towards a popular purpose or direction that
might be imposed on government under the system of popular influence
described in the last chapter. That purpose is compliance with policy-
making norms, whether they be norms that put policies or processes off
the table, norms that are associated with the agenda of government, the
status of citizens or the conduct of government, or norms that are encoded
in particular institutions and in laws or conventions that have a less formal
existence amongst members of the community. That purpose can be
systematically promoted, so it appears, as the long-term effect of short-
term attempts to exercise influence within the constraints imposed by an
acceptability game.
With this image of dual-aspect democracy in place, we can turn
finally to considering its attractions. I promised at the end of
Section 1 that the dual-aspect model would combine the appeal of
each of the existing models of popular control: in the first of these,
government is under the intentional control of the people and in the
other under their non-intentional control. It is time now to make good
on that commitment.
Combining the attractions of existing models

The first of the existing models of popular control would organize govern-
ment around people’s public-interest judgements – their impartial prefer-
ences over what happens to the society as a whole. The second would
organize it under a mechanism designed to make government responsive
to more particular attitudes: people’s partial preferences over what happens
to themselves. The first is attractive, we saw, in focusing us on a plausible
goal for popular direction: that of the public interest, understood in a post-
social, non-corporate manner. The second is attractive in directing us to a
plausible means for popular direction: an invisible-hand mechanism that
does not require people to maintain control in an intentional manner.
The dual-aspect model has the virtue of combining the respective attrac-
tions of these competitors. To the extent that the policies pursued in
government satisfy policy-making norms, they may be expected to further
the public interest, as that is understood in the first of these competing
models. And to the extent that they are forced to satisfy those norms as the
long-term effect of people’s short-term attempts to influence policy-
making, they promise to materialize as the result of an invisible-hand
mechanism akin to that envisaged in the second competing model.
As a plausible example of a matter of public interest – that is, a matter of the
post-social and non-corporate interest of all – we mentioned the interest that
the followers of different religions are likely to have, assuming they are willing to
live on equal terms with each other, in the freedom of religious practice. But the
assumptions that show freedom of religion to be a public interest show equally
that there is likely to be a norm requiring people to acknowledge the relevance
of invoking freedom of religion as a reason to favour certain policies and oppose
others. The general norm of invoking only considerations that can carry weight
with all would permit this consideration to have a place in public debate. And
assuming that there is occasion to invoke the consideration – assuming, in other
words, that there are religious issues that arise in politics – we can expect to see a
specific norm emerge under which people are required, on pain of the dis-
approval of others, to acknowledge its relevance.
This example helps to show how the dual-aspect model is likely to
promote a version of the public interest associated with the intentional-
control picture of democracy. But while it promises in this way to display
the same appeal as that particular model, it is not subject to the same
disadvantage. The intentional-control model supposes that a democracy
will promote the public interest only insofar as politicians and citizens more
generally display a high level of unmotivated virtue in the judgements they
endorse and the policies they support. The dual-aspect model suggests, by
contrast, that a democracy can promote the public interest under an
invisible-hand mechanism of the kind that makes for the appeal of the
standard alternative to the intentional-control model: that is, the model of
democracy, illustrated in James Mill’s theory of representative government,
on which democracy works like the market.
The participants in a dual-aspect democracy will be involved at any time or
context in supporting a decision on this or that policy, or this or that process
of decision-making, and will be required by the rules of the acceptability game
to identify considerations that can pass muster on all sides. But as this game
continues in a society, now in one domain of policy, now in another, now in
one context of policy-making, now in a second, the participants are bound to
generate a range of policy-relevant norms as a by-product of the process in
which they are involved. And as they narrow down the candidates between
which they have to decide on any occasion, and let the final resolution be
determined by a mechanism like voting, they will ensure that process-relevant
norms have a constraining role on the mechanism used. Thus if all goes well,
the policies and processes adopted are likely to satisfy the demands of the
norms in play. That result will materialize without being something that any
of them explicitly pursues.
The people under this model each have a wish not to have to live under
government policies or processes that breach received norms, as evidenced
in the political arguments they advance and accept. And by each acting on
the basis of that self-regarding desire, they combine to ensure an aggregate
result under which they each achieve satisfaction of that desire. The
desires that drive their efforts lead to a pattern of behaviour that satisfies
the desires without anyone necessarily foreseeing or intending this – that
is, as by an invisible hand. The citizens of a dual-aspect democracy control
for their government’s satisfaction of accepted norms, to revert to the
analogue used earlier, in the same way that the consumers in an ideal
market control for the competitive pricing of available goods. The dual-
aspect democracy gives effect to the political wishes of citizens, as the ideal
market gives effect to the economic wishes of consumers. Each mode of
organization has an empowering effect, helping to ensure that the relevant
desires of the parties, economic or political, get to be satisfied.18
18
Not only is a pattern of competitive pricing, as generated in the ideal market, going to be attractive to
all. It is going to be a stable equilibrium: a pattern such that no one has an incentive to depart from it
The non-intentional control that citizens exercise over government in the
dual-aspect model contrasts with the sort of control that members of a
group might exercise as a matter of jointly intentional action. For an
example of jointly intentional action, consider how people sitting on a
beach can act together to save a swimmer who is having trouble in the
water. There are many analyses of what can support joint action, but one
clear possibility would materialize when it is manifest to members of the
group:
* that they together can save the swimmer by forming a chain into the
water;
* that they each want the group to adopt that strategy and achieve that
result;
* that they are each disposed to play their part so long as others join them;
and
* that others are going to join anyone who takes the initiative.
Under such conditions someone is bound to take the initiative, triggering

the cooperation that promises to save the child. And when cooperation
occurs, then it is going to be natural to say that the members of the group
perform a joint action with a joint effect: they intentionally control for the
desired outcome.19
It may well be the case in a democracy that people act out of a joint
intention to influence government; that would certainly fit with the model
of the last chapter. But it is not the case, under the model of this chapter,
that people act out of a joint intention to control government as the
intentional-control model envisages.20 The collective control for which
the dual-aspect model programmes is generated in quite a different manner.
People combine under that model to ensure that the policies and processes
of government conform to popular, equally acceptable norms. But it need
not be manifest to members of the group – that is, it need not be a matter of
common belief amongst them – that they together can achieve that aggre-
gate, cumulative result, and so it need not be a matter of explicit desire or
intention – it need not be in any sense a general will – that they should do
unilaterally, and if anyone does, then the pattern is likely to be quickly restored. The same is true, I
believe, of the pattern of norm-satisfaction that a deliberatively regulated, dual-aspect democracy
would generate. No one would have an incentive to depart from it unilaterally, at least given the
repercussions that this is likely to have on others. And in the event of a departure, those repercussions
would be likely to restore the pattern.
19
There is a large literature on what it is for people to cooperate in forming such a joint intention and
pursuing such a joint action. See for example Bratman (1987); Searle (1995); Tuomela (1995); Gilbert
(2001). My own preference follows broadly the lines taken by Bratman; see also Pettit and Schweikard
(2006); List and Pettit (2011).
20
For a paper that outlines the form that such a model might take see Bratman (2004).
so. They generate the effect by each looking after their own efforts: they
each seek to support their own arguments in accepted terms, and they each
require others to do the same. And they each have a reason to welcome that
effect, given the desire at the origin of those efforts. But the effect itself
eventuates by a hand that is likely to be hidden from them; it materializes
behind their backs.
The critique of the intentional-control model combines with the defence
of the dual-aspect model to suggest that the promotion of policy-making
norm-satisfaction may be possible only on the basis of the invisible-hand
mechanism described. In Jon Elster’s (1979, 1983) phrase, such norm-
satisfaction may constitute an essential by-product. It may be a result that
can be reliably secured only as a side-effect of people’s pursuing some other
aim. Let people explicitly try to agree in identifying and implementing
common norms and they are likely to recognize the costs that the satisfac-
tion of appropriate norms can impose on them and theirs. And as they
recognize those costs, high virtue is liable to fail them and leave them at
irreconcilable loggerheads with one another.21 The only dependable path to
establishing a regime of norm-satisfaction may be the one charted in the
dual-aspect model.
In the account emerging from these observations, the participants in a
dual-aspect democracy will struggle with one another, and perhaps divide
quite antagonistically, over particular matters of policy. But however adver-
sarial their struggle, they will honour the demands of the acceptability game,
and generate norms that have a double effect of cutting down the policies
treated as potential candidates for implementation and of cutting down the
processes to be employed for resolving disputes over candidate policies. This
double impact means that the policies generated ought to make a plausible
claim to be in the common or public interest, where that interest is under-
stood in a post-social, non-corporate manner. Plausibly, indeed, the policies
might even begin to answer to the demands of justice, as those are under-
stood under republican or related conceptions. Being policies supported by
considerations acceptable on all sides – supported in their own right and via
the support given to the processes under which they are selected – they
promise not to be too far out of kilter with what might be expected to
promote justice.
21
There is every reason to think, of course, that as people pursue this norm-identifying exercise under an
acceptability regime, they will establish higher-order norms amongst themselves. The lesson follows
by an argument that parallels Lewis Carroll’s argument in his classic paper on Achilles and the
Tortoise (1895).
The citizens who play their part under the dual-aspect model will each
have their own particular conceptions of justice, of course, differing on at
least some of the policies that they think justice requires. And so in making
proposals, offering objections and generally contesting how things are
evolving under government, they will be guided by those conceptions,
looking for commonly acceptable considerations by which they can hope
to draw others to their side. While no side can expect a full victory in a
society where plurality rules – the republican side included – they may each
still expect that the policies and processes that get established are not going
to offend too radically against their more central principles. And they may
each even hope that as debates and decisions evolve, the most compelling
conception of justice – their own, as they will each take it to be – will come
eventually to prevail.
This is an optimistic vision of how things can transpire under a dual-
aspect democracy and I hasten to add that it will only materialize under
quite demanding conditions. It is a commonplace in economics that few
markets live up to the demanding conditions laid out in textbooks: most are
affected by a variety of economic distortions, ranging from restrictions on
entry to natural monopolies to collusion amongst producers. And it is
equally a commonplace that competitive pricing will not generally materi-
alize in the presence of such distortions. It is important to recognize that a
parallel observation holds in the democratic case. The dual-aspect model of
democracy is only going to work when there is a properly individualized,
unconditioned and efficacious system of popular influence, and when the
rules of the acceptability game have a sufficient presence to ensure the
deliberative regulation of public business. To the extent that distortions
warp the system of popular influence and exchange, there will be a shortfall
from the ideal we have been describing. And the greater the distortions,
plausibly, the larger that shortfall will be.
This observation serves to underline the critical edge of the model
developed here. The dual-aspect model offers us a picture of a system in
which the demos, or ‘people’, enjoy kratos, or ‘control’, over government.
But the model is only going to materialize under the sorts of conditions at
which we have been gesturing in this chapter and the last. Let those
conditions fail and there will be democratic distortions that parallel the
market distortions of which economists speak. While the conditions will
often fail to be satisfied, however – while they will call for revision to existing
practices – the prospect of their fulfilment is not a utopian pipedream. The
dual-aspect model supports a programme of reform and renovation that is
well within the bounds of feasible achievement.
3. the people, the state and the constitution
Government of the people, by the people, for the people

One of the most widely accepted characterizations of democracy was
offered by Abraham Lincoln in his Gettysburg address when he described
it as government of the people, by the people and for the people. Under a
plausible interpretation, the characterization requires that government be
for the people in assuming a pattern or direction answering to people’s
wishes or needs; that it be by the people in being directly or representa-
tively implemented by the citizenry; and that it be of the people insofar as
the people are in ultimate control: they do not depend on the willingness
of any other agent or agency for their capacity to shape the government
they live under. The dual-aspect model of democracy ensures that all three
clauses are satisfied.
The effect of a system of popular influence in forcing government to
comply with accepted policy-making norms is bound to be equally accept-
able to all participants. And in that sense the system of control and direction
valorized under the dual-aspect model will ensure that government is for
the people. It imposes significant restrictions on the will of government,
although it may leave a good deal to the effects of chance; as we saw in the
last chapter, chance will appear in the processes that happen to be chosen for
resolving policy ties, for example, and in the results to which those processes
happen to lead. And the restrictions it imposes will answer equally to the
concerns of all citizens who deserve a hearing – all those who are willing to
live on equal terms with others – since these will reflect considerations that
have to be relevant, by all lights, to the enterprise of collective decision-
making. Thus the policies adopted and the processes employed in selecting
them must all pass muster by universally endorsed criteria; the will that
guides the generation of laws and other measures has to operate on terms
that reflect an outlook common to all relevant members of the society.
The dual-aspect model of government is also bound to ensure that
government is by the people. It requires the system of popular influence
that imposes an equally acceptable direction on government to be indivi-
dualized, unconditioned and efficacious. And in developing an institutional
model under which this requirement can be satisfied it argues for an
electoral-cum-contestatory regime in which the influence that shapes gov-
ernment emanates directly or indirectly from the people themselves.
Everything done under this system of government is going to be done by
the people themselves, whether this involves their collective presence in
The people, the state and the constitution 281
election or their collective representation by suitably responsive or indica-
tive agents. Their agents will act in the people’s name insofar as they are
suitably appointed and they will act under the people’s control insofar as
they have to discharge their representative roles under a suitable form of
popularly imposed constraint and review.
Finally, government will be of the people insofar as the directive influ-
ence exercised by the people is not exercised at the discretion or by the
permission of any distinct agency: not the government itself, not any power
elite within the society, and not any foreign power. The dual-aspect model
ensures the fulfilment of this final clause in requiring that the influence of
the people should be unconditioned in character, mediating a form of
control that is truly the people’s own. This condition will be breached
under any form of colonial or neo-colonial rule, even a rule that allows the
domestic polity to do business in an electoral fashion. And equally it will be
breached in any regime where the people succeed in putting government
through electoral and contestatory hoops only when this accords with the
desires of the governing authorities, the wishes of the army or police, the
dictates of a priestly caste, or the interests of the very wealthy.
We developed the dual-aspect model with a view to identifying a govern-
ment with claims to republican legitimacy: a government that would
interfere in the affairs of its citizens, as all governments have to do, but
that would interfere under their equally shared control and so without
domination. But the model does not only have good republican credentials
in that sense – the satisfaction of the Lincoln triad shows that by almost any
benchmark it is also going to have excellent credentials as a model of
democracy. A system that met the specifications it lays down would ensure
the sort of self-rule or self-control that Lincoln’s characterization requires.
The self-control that the people secure under the dual-aspect system is
relatively indeterminate, of course, since, as we know, it leaves much to the
operation of chance. This may be the chance whereby a particular majority
governs parliament, the decisions of a court or tribunal or commission take
a certain shape, the vote in a particular referendum goes a certain way, or
indeed a lottery mechanism generates one or another result. But the
indeterminacy of the control that people exercise over themselves in the
dual-aspect model is paralleled by an indeterminacy in the personal self-
control that individuals enjoy under various interpretations of autonomy.
Take the interpretation of autonomy as orthonomy, which was men-
tioned in the first chapter. Orthonomy involves the sort of self-rule in which
you guide what you believe and desire, intend and do, by the values that
you endorse, however valuation is construed (Pettit and Smith 1996).
Displaying a high degree of orthonomy, you may be faithful in your belief-
processing to the value of hearing all sides of a question, in your desire-
formation to the value of looking beyond the moment, and in your actions
to the various claims you recognize in others. But while proving orthono-
mous in that sense, you may be led to embrace any of a number of beliefs
and desires and to take any of a variety of actions; the constraints of the
relevant values will leave a lot of slack or indeterminacy in place. That is
fine, so long as the indeterminacy is resolved by more or less contingent or
chance factors, without any inconsistency with the values that are meant to
be in control.
Just as self-government in the personal case is not jeopardized by the
fact that the factors in control – your values – leave a lot of slack in place,
so the same is true of self-government in the democratic context. The
presence of slack or indeterminacy is no problem for the ideal whereby
the civic or constituting people govern themselves in their life together by
the norms they accept, imposing those norms on the practice of policy-
making. Or at least it is no problem, provided that the slack is taken up in
an appropriate, norm-compatible way: that is, without the unlicensed
intrusion of any alien will.
The role of the state under the dual-aspect model

Not only does the dual-aspect model offer us an intuitive sense in which the
demos, or ‘people’, exercise kratos, or ‘power’, it also gives us the wherewithal
to analyse that claim in more exact terms. It provides the materials for a
political ontology in which the notion of the people is given more precise
reference and the idea of a popular government is spelled out more properly.
I shall focus on this ontology in the remainder of the chapter, beginning
with the role of the state, then turning to the more important issue of the
role of the people and concluding with six principles of political ontology
that the dual-aspect model supports.
No matter how multi-centred and multi-dimensional a political system –
no matter how mixed a constitution – we have seen that the state that it
establishes will have to be capable of satisfying constraints of rationality in
its commitments and actions and displaying sensitivity to the charge of
failing to satisfy them. Without such a mode of organization the state could
not effectively discharge its duties under any plausible account of those
duties. And in particular it would not be an entity that we, the citizens,
could challenge on the basis of common norms of policy-making. Unable to
respond to rational expectations, the state would not be an entity with
which it was possible to do business. It would not be a corporate agent or
person – a conversable body that is active and answerable in the manner of a
single agent (List and Pettit 2011) – but just a site of random, potentially
incoherent, response.
How can the people who impose a norm-complying direction on
government, now in this domain, now in that, ensure that the state will
be rationally sensitive and suitably conversable under collective and
individual challenge? How can they guard against the possibility that
the commonwealth they constitute might be a many-headed beast, an
amalgam of conflicting voices, that sovereigntists like Bodin, Hobbes and
Rousseau associated with the mixed constitution? I suggested in the last
chapter that there is no reason why a mixed constitution should not be
able to support a coherent government and state and I charged sover-
eigntists with falling to the fallacy of misplaced concreteness in thinking
otherwise. But it will be useful to expand a little on why the particular
mixture prescribed under the dual-aspect model is perfectly compatible
with coherence in government.
If the democratic state is to constitute a popular but coherent entity,
according to sovereigntists, then it must be incorporated in the form of a
majoritarian assembly of all its members. Ironically, as we saw earlier, they
were wrong in this since, as the discursive dilemma illustrates, a majority of
individually consistent and conscientious members may vote in a way that
supports an inconsistent set of assembly views. In order for such a group to
be able to incorporate coherently – incorporate in a way that makes the
resultant body capable of registering and responding to rational constraints
like consistency – the members must gain feedback on the corporate
attitudes that evolve on the basis of individual contributions, for example
individual votes, and must be able to adjust as a group so as to avoid failures
like inconsistency. An assembly of the sort imagined in the tradition might
do this, as we saw, by relying on a device such as the straw vote. Under this
procedure the group would check on whether the result of any majority vote
generates an inconsistency with previous votes, and if it does then the group
would go through further rounds of reflective deliberation in order to
determine how best to restore coherence.
The constitution envisaged in the dual-aspect model is a mixed consti-
tution insofar as it lets final decisions on law and policy be determined,
whether simultaneously or sequentially, on the basis of interaction between
different centres, civic and official, executive and legislative, constitutional
and judicial. How can a constitution of that kind allow for the emergence of
a suitably coherent and contestable state?
The discursive dilemma is avoidable in the assembly case on the basis of a
straw-vote approach, because this requires members to keep track of any
inconsistencies arising from majority voting and to decide together on how
best to resolve them. Generalizing from that example, we can see that the
coherence and contestability of what any state does in making law and
policy can be ensured to the extent that similar conditions are fulfilled. First,
there are some points at which the system allows for feedback about how far
candidates for commitment, however they are individually generated, make
a coherent whole. And second, the system distributes authority to some
body or bodies to make adjustments in light of that feedback that can help
to ensure coherence in the evolving body of law and policy.
There are many different ways of designing a political system, even a
system with competing centres of control, so that it meets these constraints.
Any design will be satisfactory so long as it requires relevant agents to keep
track of the attitudes that the state counts as upholding, in virtue of their
individual contributions, and to make sure that those attitudes amount to a
more or less consistent, and otherwise rational, package (Pettit 2007b).
They cannot just look to how they themselves perform in their different
roles, as voters in an assembly might just consider their own votes. They
must act with a conscious, continuing reference, not just to themselves, but
to the state in whose name, and for whose sake, they are enjoined to act
(Luhmann 1990).
Two examples of satisfactory but contrasting constitutions are provided
by the Washington and Westminster systems that we mentioned in the last
chapter, where the first organizes the state’s business around an assembly
of representatives, as we described it, the other around a representative
assembly.
In the Washington system, policy is mainly made in the deals struck
amongst individual representatives elected to office: members of the House,
members of the Senate, and the President. Since no one knows in advance
what such deal-making will support, no one at election time is in a position
to commit credibly to a detailed programme. And so no one works out such
a programme in advance, submitting it for scrutiny at the time of election.
This raises a problem, since the hurly burly of negotiation and accommo-
dation that an assembly of representatives allows may lead to laws and
decisions that are incoherent in various ways with one another or with
constitutional provisions. But the courts can come to the rescue in the case
of any such difficulty, interpreting the emerging dictates in a way that
restores internal and constitutional coherence. The system recruits the
courts, side by side with the Congress and Administration, in the process
of decision-making and it helps to ensure thereby that the state – the
government considered as an encompassing set of agencies – speaks with
one coherent voice.
In the Westminster system, the courts play a less important role than in
the Washington one, because there is not a similar need for judicial
interpretation and shaping under this system – even under a version of
the system in which Parliament is not a sovereignist authority, as it is
sometimes held to be in the British case.22 The Parliament at any time is
subject to the control of a single party, or a close alliance of parties, where
members are required to close ranks in maintaining the executive in power.
This being a matter of common knowledge, each party or alliance is
required to commit in its electoral campaigning to a fairly detailed pro-
gramme of legislation, submitting that programme to electoral scrutiny.
And so the main laws and policies enacted by any government will have
been determined pre-election and tested for coherence by members of the
party itself, by the opposition and the punters in electoral debate, by the
electorate at the polls, and indeed by independent or statutory authorities.
While they may be subject to amendment or rejection by the courts, as
happens routinely in the Australian and Canadian versions of the
Westminster system, those laws and policies are likely to achieve coherence
independently of judicial intervention.
These examples show why the operation of a dual-aspect democracy, and
the implementation of the mixed, contestatory constitution that it presup-
poses, is quite compatible with the emergence of the state as an agent with
which it is possible for its citizens, and other bodies, to do business. The
state in prospect is a legal person to which it will be perfectly appropriate to
attribute attitudes of judgement, desire and intention, as we might attribute
them to an individual agent. It will be an agent or agency that we can see as
the ultimate source of law and policy in the country where it rules.
The role of the people, constituting and constituted

The system of influence and direction associated with the dual-aspect model
gives a networked set of electoral and other roles to ordinary members of
the populace and to those who represent them, whether in a responsive or
22
See for example Goldsworthy (1999). Without going into the debate about whether Parliament is
sovereign in Britain, so that there is not strictly a mixed constitution in that country, it should be clear
that a Westminster-style constitution might not give Parliament such uncontested authority, if only
by allowing for a system of judicial review, as in Australia and Canada. For an insightful case for the
merits of such a constrained parliamentarianism see Ackerman (2000).
indicative way. The people are present or represented at each of the
mutually checking centres whose interaction within a mixed constitution
generates the public laws and policies under which everyone lives. The river
of their influence washes into every corner where decisions are taken in their
name and it has the effect of directing those decisions towards the satisfac-
tion of popular policy-making norms.
The existence of such a multi-dimensional, multi-centred system of
popular interaction and decision-making entitles us to say that the people
rule themselves, as we have seen. But what we can now add is that it allows
us to say this in each of two distinct senses. The existence of the system
means, in the first place, that the people considered as a plurality govern
themselves and, in the second, that the people considered as a singular,
corporate entity is also self-governing.
The plural people are the individuals who play various parts, electoral and
contestatory, in establishing their representatives and in sustaining and
operating the processes that secrete the laws and policies under which
they have to live. This is the civic people or, better, in a phrase borrowed
from the Abbé Sieyes (2003), the constituting people: that is, as I shall
understand the phrase, the members of the population acting as citizens to
determine how things are set up and run. The people in this sense rule
themselves to the extent that it is their influence and their direction that
shapes how they are governed.23
The constituting people assume a direct presence in elections and refer-
enda and a represented presence in the authorities who run the government
and in those private attorneys and social movements who contest govern-
ment proposals and decisions. They rule themselves, in the first place, to the
extent that their influence is present on all sides in the mutual checking and
balancing of these elements within a mixed constitution – in the interaction
between the elements out of which, in one way or another, all political
decisions emerge. And the people rule themselves, in the second place, to
the extent that as a result of these often contrary tides of popular influence a
norm-complying direction is imposed on the policies and processes that
gain political support. The constituting people have a double role in their
self-government, then, as we might expect under the dual-aspect model.
Whether directly or representatively, they are active on the one side in the
short haul of day-to-day, policy-by-policy decision-making. And at the
same time they are active on the other side in the long haul of shaping
government policies to the requirements of commonly endorsed norms.
23
On the background to Sieyes’s work see Hont (1995).
But while the plural, constituting people are the agents of a republican
democracy, exercising individualized, unconditioned and efficacious con-
trol over government, they establish conditions under which it is appro-
priate to posit the existence of a people in a distinct, singular sense of the
term. This, in a complementary phrase from Sieyes, is the constituted
people: the people considered as an entity that we can take to be answerable,
as an individual agent might be answerable, for the range of commitments
supported in the laws and policies of government.
In order to be effective in the imposition of policy-making norms, as we
saw, the people in their constituting role must establish a state that serves as
a corporate agent, responsive to rational desiderata. But if the state is
sustained and operated by the constituting people, then its commitments
as an agent are all going to be shaped by those people. And so it can be
depicted as the people in incorporated form: the people, operating in the
manner of a singular, corporate entity. Where the civic or constituting
people are the people qua organizing, the constituted people are the people
qua organized.24
The constituted people, understood in this way, is just the state and the
state is just the people: in John Rawls’s (1999: 26) words, it is ‘the political
organization of the people’. Thus whenever something is done on a political
basis, we can say either that the state or the people – the people in the
singular voice – performs the relevant action (Pettit 2006b). As we attribute
the laws and policies in any country to the state that enacts them so, under
the dual-aspect model, we will be able in the same way to attribute them to
the people (Kelsen 1961, 1970).
If the constituted or corporate people is indistinguishable from the state
in this manner, then we can say in particular that not only do the constitut-
ing people govern themselves through the state, the constituted people
governs itself in a corresponding sense of self-government. While we may
regret this ambiguity in speaking of the self-government of the people, we
should note that it has long pedigree. It may go back to Aristotle’s use of
‘demos’ in The Politics (1996), but in any case it is certainly present in an
important fourteenth-century author.
In one of the first theoretical articulations of a democratic ideal,
Bartolus of Sassoferrato argued that the independent city-republics of
fourteenth-century Italy, such as his own Perugia, did not lack a prince
and were not exposed, on account of lacking a prince, to the will of the
24
For an account of the emergence of the idea of the state, in an early version of our current conception,
see Brett (2011). On the development of the idea of the state see Skinner (2009).
Holy Roman Emperor; the assumption had been that the Emperor’s writ
did not run within cities that were governed by princes (Canning 1983; see
too Ryan 1999). In the city republic, Bartolus said, the people is a
corporate entity, a legal person, and for all practical purposes a prince;
the people, in his memorable phrase, is sibi princeps, ‘a prince unto itself’
(Woolf 1913: 155–60).25 But who exactly does the word ‘people’ or populus
refer to? Bartolus’s answer is, the civitas. In giving this answer, he gave life
to the ambiguity that haunts us still. For the term civitas may refer either
to the constituting or to the constituted people. It may direct us to the
citizenry that rule in a popular regime – in a regime where, in Bartolus’s
phrase, things are done secundum vices et secundum circulum: ‘by repre-
sentation and rotation’ (Woolf 1913: 180). Or it may refer us to the state
that the incorporating citizenry establish when they operate in that
manner. The civitas as citizenry is the ‘constituting people’; the civitas as
state is the ‘constituted people’.
Six principles of political ontology

It is quite useful to be able to think of the corporate, constituted people as
self-governing. Doing so highlights the fact that the people do indeed
incorporate as a group agent under the sort of democratic arrangements
described and, more specifically, that they incorporate in a way that gives
control to its citizens and its citizens equally; the incorporation is not like
that of a typical commercial corporation or a hierarchical church. The
corporate people in this sense is akin to the membership in a democratically
organized condominium and is indistinguishable from the state in the way
that the membership in such an organization is indistinguishable from the
condominium. As we may say either that the condominium or the member-
ship takes such and such measures, so we may say either that the state or
the people adopts such and such initiatives; in each case, the two ways of
speaking have much the same import.
But notwithstanding the utility of this language, talk of the self-
governing, incorporated people should not blind us to the more important
role under the dual-aspect model of the civic or constituting people. It may
be useful to set out some principles that guard against this danger. The
principles affirm the priority of the constituting over the constituted people,
downplay the capacities of the constituted people in a number of respects,
and show how the emerging picture enables us to dissolve a familiar
25
For further commentary on Bartolus see Canning (1983); Ryan (1999).
constitutional paradox. They help to flesh out the picture of people and
state – if you like, the basic political ontology – associated with the dual-
aspect model.26
The priority of the constituting people

The first principle is that it is only in virtue of the role of the several,
constituting people in establishing and continuing to maintain the state that
the constituted or incorporated people exist, figuring as an agent that can be
relied upon to adjust and act in suitably rational ways, or at least to be
sensitive to the demands of agential rationality (List and Pettit 2011).27 This
prioritization of the constituting people contrasts with the view developed
by Hobbes, and endorsed by Rousseau, under which the incorporated or
constituted people displaces the several, constituting people in importance.
They hold that a people become self-ruling only insofar as they are con-
stituted in an assembly, and operate according to majority rule. It is true
that the constituting people have to vote unanimously to set up this
incorporated entity – this public person – as they each say. But once that
vote is cast, according to this picture, the constituting people have no
further role to play: qua several they amount to nothing more, in
Hobbes’s terms, than an ‘aggregate’ or ‘heap’ of agents (1994a: 21.11), ‘a
disorganized crowd’ (1998: 7.11), a ‘throng’ (1994b: 6.37). Following this line
of thought, as we saw, Rousseau (1992: ii.6.7) denies a continuing voice to
contestatory citizens – in our terms, the constituting people – on the
grounds that if they were allowed a hearing then, absurdly, there would
have to be a further arbiter to determine whether they or the incorporated
body speaks for the people.
The dependent sovereignty of the constituted people

Does the constituted people, in our sense, count as a sovereign? It will
certainly be sovereign in the external sense that, things going well, it can
operate in the international arena as an agent with its own ends, its own
26
The discussion that follows is germane to many debates. It is relevant, for example, to the debate
between the view, associated with Hans Kelsen, that prioritizes constituted power in conceptualizing
the state and the view, associated with Hannah Arendt, that puts much greater emphasis on the
constituting power of the people. Hans Lindahl (2007: Section 1) offers a very illuminating account of
the divergence on this issue. It should be clear that my sympathies lie broadly on the side that he
associated with Arendt.
27
This represents a perhaps surprising point of continuity with Negri (1999), Hardt and Negri (2000)
and Kalyvas (2005). For an interesting, broadly congenial view of the people as a constitutionally
shaped process see Espejo (2011).
views and its own resources. But will it be a sovereign on the internal or
domestic front, vis-à-vis its own citizens? Yes, in the sense that it can be
identified as the source of the laws imposed, to invoke Bodin’s founding
conception of a sovereign, and indeed as an entity that is not itself subject to
the laws. No, in any sense that might reduce the importance of the
constituting or civic people.28 In the dual-aspect model, the corporate or
constituted people is an agent that emerges under the norm-imposing
control of its constituting members, not an entity that can push back against
those very individuals; it is dependent on them for its continued existence
and for the manner of its operation. Again, this observation marks a contrast
with Rousseau, for whom the constituted people – in his story, the popular
assembly – has absolute power over the individual citizens, considered as a
plurality. Each citizen may be independent of others in the Rousseauvian
theory, but they are all required to be ‘excessively dependent on the City’,
where the City is just the people in assembly, the people qua incorporated
(Rousseau 1997: ii.12.3).29
The undominating will of the constituted people

The third proposition, related to the other two, is that while the con-
stituted people is a corporate agent and can be said to express its will in the
laws and policies it imposes, that will is undominating in relation to
citizens. This is because the will of the constituted people is formed on
the basis of the processes of interaction and decision-making that obtain
amongst the constituting people, materializing under the direction of the
norms they endorse at the different centres at which they and their
representatives operate. Given these norm-governed processes of will-
formation, the will of the constituted people is bound to answer to the
individualized influence and direction of individuals; things going well, it
is pre-tested case by case for whether it is suitably controlled and non-
arbitrary. Here, too, there is a sharp contrast with the Rousseauvian
picture. The will of the assembly, assuming it expresses the general will,
28
The ambiguity was used by many, of course, to identify the sovereignty of the people with the
sovereignty of the state, thereby depriving the idea of popular sovereignty of any bite it might have
had. Thus Hegel (1991: 318) writes that ‘sovereignty lies with the people, but only if we are speaking of
the whole’ – that is, as he goes on to clarify, only if we are speaking of the sovereignty that ‘belongs to
the state’. In tones that hark back to Hobbes, he says that apart from the state ‘the people is a formless
mass’ (Hegel 1991: 319).
29
My view is consistent, however, with that which Nadia Urbinati (2006: 223) expresses as follows:
‘Popular sovereignty, understood as an as if regulating principle guiding citizens’ political judge-
ment and action, is a central motor for democratizing representation.’
is supposed to be a will that is present in each of the members. It is meant
to be non-dominating, not because it is controlled by suitable norms, but
because it is part of the person’s own will.
The corporate will is not the general will

The will of the constituted people may look similar to Rousseau’s general
will, but the difference is more important than any similarity. The general
will is supposed to be a will that is present in each of the citizens, side by
side with their particular wills. In our terms, it is a will that is meant to be
common to the constituting people. But there need not be any will that is
common to the constituting people, as we saw, let alone a will to which
they might subject government. All that may be in common is a network
of norms that people impose on government as the unintended conse-
quence of pursuing influence under the discipline of an acceptability
game. The corporate will of the constituted people is the will of an
artificial entity that the constituting people create and, emerging under
the interplay of rival bodies, it need not have any place in their individual
hearts. They may be individually willing that it should emerge from that
process of interaction, but neither the ways in which it operates, nor the
policies to which it leads, have to answer to anything like a personal will or
intention or desire on their part.
The replaceability of the constituted people

The corporate people emerges as a group agent insofar as citizens subscribe
to an organizational structure – broadly a constitution – that generates out
of their interactive contributions a generally coherent and coherently
evolving set of commitments: a profile of attitudes that can be ascribed
to that agent. But this means that if the constitution is rejected by the
constituting people – if the constitution is changed but not under any
existing articles of amendment – then the constituted people ceases to
exist and, assuming a new constitution emerges, is replaced by a distinct
corporate entity. The point was made by Aristotle (1996: Book i) when he
noted that a polis, or ‘city’, is individuated by its constitution and is
replaced by a different city, albeit a city with the same inhabitants,
when the constitution is replaced (see too Rubenfeld 2001). But, contrary
to what Hobbes and Rousseau think, the replacement of the constituted
people need not return people to the state of nature that they imagined – a
scenario in which the people dissolve into a mob – since the constituting
people can retain enough coordination as a set of individuals to manage a
smooth transition to a new constitution. This, arguably, is what happened
when the people of the United States rejected the Articles of
Confederation in favour of the 1787 Constitution.30
Beyond the paradox of constituent power

There is a paradox raised about the relation between people and constitu-
tion, but it is more or less readily resolved under the dual-aspect model
(Lindahl 2007; Shapiro 2011: 37–40). The corporate people comes into
existence only in virtue of the constitution or organization that establishes it
as an agent. And yet it appears, paradoxically, that the corporate people
must also be causally responsible for the constitution, at least in a demo-
cratic society. But in our account there is no paradox here. First of all, it is
the constituting people who are primarily required under democracy to be
causally responsible for the constitution. And in any case, causal responsi-
bility does not require the temporal precedence that would give rise to a
paradox (pace Kalyvas 2005). Both the constituting and constituted people
can be responsible for a constitution that was set up by a particular
individual or group, insofar as they have the power to change it and choose
not to do so; that is why there can be a democracy without a founding,
democratic moment. The constituted people will be responsible for the
shape of the constitution to the degree that they can change it under
constitutional rules of amendment. The constituting people will be respon-
sible for it to the degree that not only can they change it under those rules of
amendment, they can also change it by non-constitutional means: they can
replace the old constitution with a new one, thereby establishing a novel
state and a novel constituted people. Under the dual-aspect model, the
constituting people are paramount in this respect, as in so many others; in
John Locke’s (1960: ii.149) phrase, they continue ‘to be always the supreme
power’.31
30
Akhil Amar (1988) argues that the current US Constitution might be constitutionally amended – say,
by a straight majoritarian vote – and yet not in accordance with Article 5; his surprising claim is that
that article does not identify the only constitutional ways of changing the Constitution. The
motivation for his argument is removed in the viewpoint adopted here, for it would not have to
represent a major democratic crisis for the constituting people of the country to decide unconstitu-
tionally against sticking with the existing Constitution. Rejecting the Constitution in that way would
change the identity of the constituted people, but not the identity of the constituting people. And so
there is no logical difficulty in imagining that one and the same people – one and the same
constituting people – might reconstitute themselves under a new arrangement.
31
In Locke’s view, however, the people – in our sense, the constituting people – play just a remote
standby role, as noted earlier, being ready to intervene only at the extreme where things go quite bad.
The interventions he envisages ‘happen not upon every little mismanagement in public affairs’, he
says; they are likely to be triggered only by ‘a long train of abuses, prevarications and artifices’ (Locke
1960: ii.225).
Conclusion. The argument, in summary
The argument of this book has taken us over a wide terrain, introducing the
republican perspective, traditional and contemporary; presenting the ideal
of freedom that lies at its core; sketching a theory and model of the social
justice that this ideal would support; defending a matching, republican
theory of political legitimacy; and then outlining a model of the democratic
institutions that might be thought to satisfy that theory. In conclusion,
I think the best thing I can do is to provide a summary of the claims
maintained in the development of the argument. While the summary is
inevitably sketchy and inexact, I hope that it will help to facilitate readers in
finding their way through a book that I wanted to make shorter and simpler
than it has turned out to be.
introduction. the republic, old and new

1. The main ideas in the republican tradition are: freedom as non-
domination, the mixed constitution and the contestatory citizenry.
Appearing in the Roman republic, in medieval and renaissance Italy, in
seventeenth- and eighteenth-century Europe and Britain, and eventually in
revolutionary America, they suggest that the state should enable its citizens –
however inclusive – to act as free, undominated persons in the sphere of
the fundamental liberties, being protected under a mixed, contestatory
constitution.
2. This republican tradition came under sustained attack at the hands of
Jeremy Bentham and William Paley in later eighteenth-century England,
as they introduced a theory of freedom as non-interference. They argued
that while the state should cater for the freedom of all citizens – now
understood more inclusively – it should do so with only this less demand-
ing ideal in view.
3. Italian–Atlantic republicanism was also challenged in the late eighteenth
century by the communitarian republicanism of Jean Jacques Rousseau.
293
294 Conclusion. The argument, in summary
While he continued to think of freedom as non-domination, he followed
Jean Bodin and Thomas Hobbes in giving up on the mixed constitution
and the contestatory citizenry. He argued that there had to be a single
sovereign in any well-functioning state and thought that in a republic this
should be the assembled, incorporated people.
4. The aim of this book is to build philosophically on the main republican
ideas, developing a theory of social justice and, in particular, political
legitimacy, where justice governs people’s relations with one another,
legitimacy their relations with the state. The republican theory of legiti-
macy gives the state a democratic job specification, requiring it to operate
under equally shared, popular control.
5. The methodological requirement on these republican theories of justice,
on the one side, and of legitimacy or democracy on the other, is that
they can pass the test of reflective equilibrium, as John Rawls described
it; they combine with empirical facts to support independently plausible
normative judgements.
6. Even the interpretation of freedom as non-domination is guided by the
method of reflective equilibrium. There are a variety of context-bound,
apparently conflicting uses of the notion of freedom, and the republican
interpretation is justified as a regimentation of the idea that plays a useful
role in grounding a plausible normative theory.
7. The republican theory of democracy developed here contrasts with Isaiah
Berlin’s views in tying democracy to freedom; with Joseph Schumpeter’s
views in requiring popular control rather than popular influence; and with
William Riker’s in treating constitutional measures as means of popular
control, not as constraints on its operation.
8. What relative importance should be given to justice and democracy? The
question is not directly addressed in the book, but the republican approach
would naturally treat democracy as the more important ideal. Any com-
promise to democracy will represent a deeper challenge to the robustness
with which non-interference has to be guarded under the republican ideal
of freedom.
chapter 1. freedom as non-domination

1. Assuming that you are able to choose between certain options, as meta-
physical free will requires, how might you fail to enjoy freedom in that
choice? That is the question addressed here, with the issue of the freedom
of a person, which was actually more central in republican thought, being
postponed to the next chapter.
Chapter 1. Freedom as non-domination 295
2. One issue raised by the question is whether freedom in a choice might be
reduced by a hindrance to any option or only by a hindrance to the
option you prefer: only by actual preference-frustration. Where Hobbes
thought the latter, Berlin argues persuasively for the former. In a free
choice, all the doors must be open, not just the one you push on;
otherwise you could make yourself free in a choice by adapting so as to
make sure that what you want you can get: what you want is an open
door, not a closed one.
3. Freedom in a choice requires consciously having the resources, personal,
natural and social, to be able to satisfy your will as between the options.
There are two ways in which those resources may be affected and the
choice hindered; one is by invasion, the other by vitiation. Invasion is a
hindrance that is triggered by your trying to satisfy your will, whereas
vitiation is a hindrance that comes about for independent reasons. In
effect, invasion comes about via the imposition of the will of another,
vitiation via unwilled forms of hindrance.
4. Both forms of hindrance are important but invasion has a particularly
inimical quality. In invading a choice, another agent or agency usurps
your control of what you do, whether totally or partially, and such
invasion naturally gives rise to resentment and outrage. Vitiation does
not give control in that sense to another agent; it may involve just the
limitation imposed by natural obstacles.
5. The difference between vitiation and invasion, and the complex charac-
ter of each form of hindrance, make the measurement of freedom a very
challenging task. This book generally avoids measurement problems,
relying instead on tests of intuitive sufficiency. Thus, as we shall see, it
invokes the eyeball test to determine a threshold at which people should
enjoy equal freedom as non-domination in relation to one another.
6. Freedom of choice, as it is discussed here, depends on how far you can act
according to the will or preference you form over the relevant options. As
this freedom is distinct from metaphysical free will, so it is distinct from
freedom of the will in the psychological sense in which this requires that
you form your will autonomously, however autonomy is understood.
7. The republican theory of freedom as non-domination argues for two
theses. The first is that there is no invasion of choice without domination,
the second that there is no domination of choice without invasion. B
dominates A in a choice to the extent that B has a power of interfering
or not interfering in that choice, in particular a power that is not cont-
rolled by A. And B interferes in A’s choice insofar as B removes, replaces or
misrepresents an option.
8. There is no invasion without domination, because interference may
be conducted under the control of the interferee. It may be like my
interference in denying you the key to the alcohol cupboard, except on
twenty-four hours’ notice, when this is an arrangement that you set up
and can suspend.
9. There is no domination without invasion, because my having a power
of uncontrolled interference in a choice, even a power I do not exercise,
means that whatever you choose, you choose in subjection to my will;
you depend on my will remaining a good will for being able to act
according to your wishes. Domination without interference may
involve me in invigilating what you do and/or, if you believe you are
invigilated, in intimidating you. And it may also be willed or unwilled.
10. Berlin argues that freedom can be reduced without actual frustration – by
interference with an unpreferred option – on the grounds that otherwise
you could make yourself free by adaptation: that is, by adjusting your
preferences. It is possible to argue that freedom can be reduced without
interference – as in the pure domination just discussed – on the parallel
grounds that otherwise you could make yourself free by ingratiation: that
is, by softening up the powerful. Freedom requires, not only that all
doors be open, but also that there be no door-keepers.
11. Assuming an absence of vitiation, freedom can be equated on this account
with non-domination. But consistently with holding by this equation, we
can allow that domination without interference or frustration is better
than domination with interference and without frustration; and that it is
better again than domination with both interference and frustration. We
need not rule on whether it is better in terms of freedom or in other terms.
12. How can we ensure your freedom in a choice? We must make sure that
you have the required resources, guarding against vitiation. And we must
protect you against invasion, guarding against anyone else’s domination.
Protecting you is distinct from making interference improbable, which
might be achieved by giving rewards for non-offences. It requires
obstructing interference, not probabilifying non-interference.
13. Interference may be more or less serious, whether it involves removing,
replacing or misrepresenting an option. And at whatever level of serious-
ness, interference may be available with greater or lesser ease to others: it
may be more or less eligible. Protection requires obstructing interference
so as to make it at least ineligible and to do this in a greater measure, the
more serious the interference is. The criterion of adequacy in such
protection is subjective and will be fixed later by the eyeball test.
Chapter 2. Social justice 297
14. Will a failure to resource and protect you in a certain choice count itself
as a form of interference? And will the power not to resource and protect
count as domination? Not necessarily. But often there is a fuzzy line in
this area between action and omission: a failure may count as interfer-
ence once the resourcing and protection has been established as a matter
of standard expectation.
chapter 2. social justice

1. Taking citizens to be all the more or less permanent, adult and able-
minded residents of a state’s territory, and taking the state to be an
agency that operates via different government branches, two questions
arise. What policies should the state be required in social justice to
impose? And what processes of decision-making should it be required
to follow if it is to constitute a politically legitimate decision-maker? Both
may get a hearing in standard theories of comprehensive justice, but it is
better to give them separate tags and keep them clearly apart.
2. Two assumptions guide this and almost all discussions of justice, social
and comprehensive. Justice requires the state to establish a proper balance
between the competing claims of citizens in the social order it establishes.
And in pursuing the goals to which people have claims, the state should
be expressively egalitarian: it should treat all its citizens as equals – and,
looking at this from the other side, all its citizens should be willing to live
on equal terms with others. The state is required to treat its citizens well,
advancing suitable goals, and in doing so to treat them as equals.
3. Theories of justice diverge on what goals the state should pursue in an
expressively egalitarian fashion – resources, utility, capability, or a mix of
these goods. And they differ on the issue of what expressive equalization
entails. Some strategies would involve a substantive form of equality,
whether in opportunity or outcome, others not.
4. A republican theory of justice would seek the expressive equalization
of freedom as non-domination: the promotion of freedom as non-
domination on the basis of an equal concern for each citizen. How should
we interpret this goal in more concrete terms? The ideal of the free citizen,
central to the tradition, provides a useful heuristic, at least on the assump-
tion that the society is operating within the circumstances of justice and
can enable each, in principle, to enjoy the status of a free citizen.
5. The ideal suggests that citizens should be guaranteed resources and
protections in the same range of choices – the basic liberties – on the
basis of public laws and norms and up to the limit dictated by the eyeball
test: absent timidity, each should be able by local standards to look
others in the eye without reason for fear or deference. Social justice, so
interpreted, would require each citizen to enjoy the same free status,
objective and subjective, as others. It would mandate a substantive
form of status equality.
6. This egalitarianism does not fall foul of the levelling-down objection,
for it does not imply that it might be desirable to reduce the level of the
well-off just for the sake of achieving equality. If everyone has, or can
access, an undominated status, there will be no call to level down. And
if it is impossible for all to achieve such a status then the circumstances
of justice fail and the expressive equalization of freedom as non-
domination will not argue for substantive status equality.
7. But the egalitarian ideal of equal status may seem to fall foul of another
objection. While requiring a free status for all, it allows inequalities in
people’s private wealth and power and in the private resources and
protections they can command. Are these inequalities likely to prove
objectionable? No, they are not. The ideal of equal status freedom is
social in character, requiring each to be able to command the respect of
others, and it puts significant constraints on how large material inequal-
ities can be.
8. How to identify the range of choices in which republican justice requires
people to be safeguarded? Assume that the range should be as large as
possible. The guiding idea, derived from the ideal of the free citizen, is
that only choices that are co-exercisable and co-satisfying should be
entrenched in this way: these are choices that all can exercise at once,
while each still enjoys the standard reward associated with the choice.
9. The co-exercisability constraint rules out choices that individuals
cannot perform individually, even with plausible state resourcing, or
cannot perform at one and the same time, whether for reasons of logic
or scarcity. On a similar basis, the co-satisfying constraint rules out
choices in which individuals cannot find satisfaction or choices –
harmful, over-empowering or counter-productive choices – in which
the satisfaction of some entails the dissatisfaction of others.
10. Some choices that are not co-exercisable or co-satisfying in themselves
can be reconstituted under social rules so that they do pass the two
constraints. Not everyone can take possession of things just as they wish
but all can take possession as they wish under suitable rules of owner-
ship. Not everyone can enjoy speaking to an audience if everyone is
speaking at the same time, but everyone can enjoy doing so under
Robert’s Rules of Order (Robert 2011).
Chapter 3. Political legitimacy 299
11. For this and other reasons the basic liberties are likely to assume a different
cast in different societies; they are not a natural kind. Amongst any set of
basic liberties, the only ones that need attention are the most distal and
the most general that meet the two constraints; look after these and you
will automatically look after others.
12. The republican assumption is that if people are publicly entrenched in their
enjoyment of the basic liberties, understood in this way, then they will be
positioned to live meaningful lives. It is possible to be as free as a full and
meaningful life requires and yet only to be as free as your fellow citizens.
13. Republican justice requires entrenching people in the exercise of the
basic liberties, whereas Rawlsian justice requires protecting them in the
basic liberties (first principle) and, in addition, guarding them against
certain material inequalities (second principle). This observation does
not tell against the republican theory, since the entrenchment of
freedom involves securing resources as well as protections and the free-
dom entrenched is freedom as non-domination, not freedom as non-
interference. Built on a strong counterpart of Rawls’s first principle, it
has no need of a counterpart of his second.
14. Republican justice would require a rich infrastructure for social life with
developmental, institutional and material aspects. It would argue for
insuring people on social, medical and legal fronts, rather than letting
them depend on the goodwill of philanthropists. And it would provide
for their insulation against others both in special, asymmetrical relation-
ships and on the generic front.
15. The insulation of people on the generic front requires a criminal justice
system. Republican theory provides a clear account of what criminaliza-
tion is, why criminalization is necessary, what acts should be criminal-
ized, and how criminalization should proceed.
16. This republican theory of justice operates with a modest, broadly
consequentialist principle that can serve in ideal or non-ideal theory.
It supports a rich and intuitive set of demands on the social order. And
it is quite distinctive in linking justice with a social status that people
enjoy in virtue of the laws they live under and the norms that come into
existence to support those laws.
chapter 3. political legitimacy

1. Where justice is a virtue of social institutions that imposes demands on the
horizontal relations between citizens, legitimacy is a virtue of political
institutions that imposes a demand on the vertical relations between the
citizenry and the state. Justice and legitimacy in this sense can come apart,
as illegitimate states may provide well for justice and legitimate states badly.
Empirically, the legitimacy of a state might require a measure of justice and
justice a measure of legitimacy, but they still make distinct demands.
2. The state is an agency, operating via government officials and bodies, that is
capable of making commitments to citizens and others and then living up
to them. Only an agency of this kind, and not for example an impersonal
rule of communal norms, can advance the cause of justice. And in order to
advance that cause it must be able to coerce its citizens, claiming a right
against any competitors. Hence the problem of legitimacy. What entitles
the state to impose a social order, just or unjust, on its people – particularly
on a people who may themselves differ in their views of justice? What gives
it a claim to the role of arbiter and decider in these matters?
3. The justice of a state’s laws would provide a pro tanto reason for
complying with them, the legitimacy of a state – judged by whatever
benchmark – would provide a pro tanto reason for accepting it. The
meaning of acceptance shows up in the fact that while the injustice of any
laws would allow citizens to try to change them, if not perhaps to ignore
them, the legitimacy of the state would oblige them to try to change the
laws only within the system.
4. The political obligation that goes with political legitimacy, then, is the
conditional obligation of citizens, if they oppose any laws on the grounds
of injustice, or indeed on other grounds, to oppose them only within the
system. Opposing some laws within the system is consistent, however,
with a campaign of civil disobedience in which campaigners do not
challenge the right of the courts to try them and penalize them.
5. Whether justice fails or not, there is a fall-back, content-independent reason
to comply with the law: that it provides a basis on which people form
mutual expectations about their behaviour. And even where legitimacy fails
there is a similar reason for opposing laws only within the system: that this
offers a promising way of inducing and maintaining legitimacy over the
longer term. This is an important observation, since many actual states fall
short of legitimacy, but it won’t play any role in the argument of this book.
6. What makes for legitimacy in the sense introduced? Hobbes, Locke and
Rousseau argued that a state can be legitimate only if it is set up with
the consent of citizens and continues over time to attract that consent.
More recent political philosophies, utilitarian, contractualist or what-
ever, have tended to ask after the performance of states in delivering
justice, whether social or more comprehensive justice, ignoring the issue
of whether they have a legitimate pedigree.
Chapter 3. Political legitimacy 301
7. The question of whether a state is legitimate is best taken as the
question of whether state coercion of citizens is consistent with their
continuing freedom. Whatever its failures in other respects, a regime
that preserves this freedom would be bound to count as more legitimate
than a rival that doesn’t.
8. If freedom means non-interference, then no state can preserve the
freedom of its citizens; every state imposes taxes and laws and penalties
and thereby interferes with its citizens. Adherents of this notion of
freedom must focus on the performance of states, neglecting their
pedigree, or argue that interference with consent is not interference.
The first approach would ignore the problem of legitimacy; the second
would make it insoluble.
9. If freedom means non-domination, however, another possibility opens
up. This is that if the citizenry control the interference of the state, as
you control my interference in the alcohol example mentioned in
Chapter 1, then that interference will not count as dominating and
the state will not deprive people of freedom.
10. Control in the sense envisaged here would require the people not just to
have an influence on government behaviour but to have an influence
that imposes a welcome direction on that behaviour. They must not
just make a difference to how government behaves, since the difference
made might be chaotic or wayward; it might be like my influence on the
traffic, if I tried to play traffic cop. People must make a difference that is
in some sense designed.
11. Such control is quite distinct from the consent that figures in other
theories of the legitimate state. You might have consented to an arrange-
ment without controlling what it allows to happen, as under a slave
contract. And you might have grown up under an arrangement without
your consent and still control what it allows: you might retain the right
of exit or the right to dictate the terms on which it operates in your case.
12. What form of control on the part of the citizenry could make the state
into an undominating force in their lives, ensuring its legitimacy? The
question bears first on the domain where people ought to control the
state and second on the nature of the control that they ought to be able
to impose in that domain.
13. The state you live under does not constrain you to live in political
society, since that constraint is imposed as a necessity of history. It does
not constrain you to live in this state rather than any other so long as it
allows emigration; that constraint derives from the political necessity
whereby other states deny a right of open immigration. And, finally, it
does not constrain you to live under coercively applied, as distinct from
voluntarily accepted, laws; the requirement on the state to treat all its
citizens as equals combines with the fact that some citizens have to be
coerced to support the functional necessity of universal coercion.
14. These observations show that the popular control that would give a state
republican legitimacy does not have to enable citizens to live out of
political society, or to live in a society of their choice, or to live under a
regime where they are not coerced to obey the law. It has to give them
control only in a domain where the state has discretion: that is, in the
matter of how it interferes in imposing taxes and laws and other measures.
15. But what is the nature of the control that the citizens would have to
exercise in this domain if the state is to count as legitimate? There are
three distinct and important requirements: the popular control exer-
cised has to be suitably individualized, unconditioned and efficacious.
16. Control will be individualized if the people enjoy an equally accessible
system of popular influence that imposes an equally acceptable direc-
tion on government. Control will be unconditioned if their directed
influence materializes independently of any other party’s goodwill: the
threat of popular resistance is enough to fasten it in place. And control
will be efficacious if their influence imposes that direction so unfailingly
that when decisions go against particular citizens, they can take this to
be just tough luck, not the sign of a malign will at work in their lives.
17. The requirements of republican legitimacy, then, are that the people
control their state in the domain where it has discretion and do so in an
individualized, unconditioned and efficacious way. This is a theory of
democracy insofar as it identifies an intuitive job specification for a
democratic system: a specification of what a system must achieve if it is
to give a distinctive form of kratos, or ‘power’, to the demos, or ‘people’.
18. This theory of democracy motivates a search for the institutions that
might implement that job specification. It pushes us to devise a repub-
lican model of the system of popular influence and direction that would
ensure legitimacy and establish a proper democracy.
19. The just and democratic state that is held out in prospect here is an
attractive ideal. It would provide for people’s enjoyment of the free status
of the citizen in the horizontal dimension of relations between people and
in the vertical dimension of their relations to the government.
20. Even the virtuous dispositions of the members of Kant’s kingdom of
ends would not ensure this status, since the less powerful members
would depend on the goodwill of the more powerful for escaping
interference; they would suffer domination. Thus the laws of a just
Chapter 4. Democratic influence 303
and democratic state can secure for people a public good that even the
highest level of individual morality cannot bring about.
21. It may be useful to distinguish, as we have done, between social justice
and political legitimacy, since it emphasizes the importance of democ-
racy. But even an approach that treats justice as a comprehensive, social-
cum-procedural ideal, as Rawls’s does, will be forced to recognize the
importance of democracy, provided that it endorses the core republican
ideal of freedom as non-domination.
chapter 4. democratic influence

1. In order to enjoy equally shared control over government, as republican
democracy requires, people must enjoy an equally accessible form of
suitably unconditioned and efficacious influence that imposes an equally
acceptable direction on the state. We look in this chapter at how such a
system of popular influence might materialize and ask in the next about
how it might impose a suitable direction on government.
2. A plenary legislative assembly might seem to promise a powerful system
of popular influence and a model of democracy. Ironically, it was first
introduced in that role by supporters of absolute monarchy, Jean Bodin
and Thomas Hobbes. They opposed the mixed constitution, arguing
that there had to be an absolute sovereign. While preferring a king, they
conceded that in principle the sovereign might be an elite committee or,
however implausibly, a committee of the whole.
3. But the idea of the empowered plenary assembly, endorsed most ful-
somely by Rousseau, runs into trouble with the discursive dilemma. Any
system of voting by individually consistent members is liable to generate
inconsistent policies and the most plausible way of remedying this is by
members considering case-by-case what the body as a whole should do to
restore consistency. A plenary assembly would be incapable of the
reflective exercise in deliberation that this would require.
4. The salient amendment is to opt for a representative assembly. But there
are two candidates for this role: an indicatively or statistically represen-
tative body, whose members have dispositions that mirror those of the
population, or a responsively representative body, whose members are
chosen, say electorally, to have dispositions that are sensitive to how the
people are disposed as a whole.
5. Mixing these models is not very plausible, since the electorally induced
motivations of responsive members are likely to undermine their claim to
indicative status. It may be an attractive idea to have a proportional
electoral assembly, but the grounds of the attraction cannot be the hope
of making the body representative of the people in both modes.
6. The responsive, electoral assembly is the more appealing alternative. It
should foster the use, and therefore the presence, of freedoms of speech,
association and travel amongst the people. Embodying a generate-and-
test methodology, it should eliminate both the false negatives and false
positives that can misdirect public policy. And the threat of electoral
sanctions should help to keep it under popular control.
7. An electoral assembly of the kind envisaged might be a representative
assembly, answering as a whole to the expectations of the people, or
an assembly of representatives, each answering primarily to their own
constituents. The divide corresponds broadly with that between a
Westminster and a Washington assembly, but while the issue of which
is preferable is bound to be important in more fine-grained institutional
design, we have to ignore it here.
8. Whatever form it takes, however, an electoral system of influence is not
guaranteed to provide for the individualized, unconditioned and effica-
cious system of directed popular influence that republican democracy
requires. Elections may be necessary for democracy, at least in practice,
but they are very unlikely to be sufficient.
9. Elections can give each citizen a vote with the same value, but that is
not sufficient to ensure that they each have equal access to influence, as
individualization requires. Equal access would require an equal chance
of being on the winning side on any randomly chosen issue and this
requirement is not going to be satisfied when there are sticky minorities
whose identity fixes how they are likely to vote on various questions.
This is the old problem of majority tyranny.
10. The only hope of remedying the problem is to allow individuals or
subgroups to contest various decisions on the grounds that the major-
itarian procedure does not give them an equal chance of influence in the
relevant area. Individual contestability has deep institutional implica-
tions, requiring transparency in public decision-making and impartial-
ity in the resolution of contestations. The resolutions of particular
issues might take any of a number of forms, including putting some
issues off the regular electoral agenda.
11. An electoral system will fail to give the people an unconditioned sort of
influence if the power of government itself, or of an agency like the army
or police or of any powerful elite, is such that electoral and contestatory
inputs are only allowed to have an impact when this suits that body. In
order to guard against such a danger the people must be ready to resist
Chapter 4. Democratic influence 305
measures that compromise their influence – they must be resistance-
prone – and the government must be disposed to back down: they must
be resistance-averse. But how to ensure the fulfilment of such conditions?
12. A good way of making government resistance-averse is to embrace an
arrangement like the traditional mixed constitution. This would ensure
the separating of the many powers of government, the sharing of each of
those powers by different authorities, and the recognition of popular
acquiescence as the ultimate guarantor of the constitution. It would
restrict the ability of those in power to close ranks against the people and
dispose government to avoid triggering popular resistance or doing
battle against it.
13. Under a mixed constitution, the voice of the people emerges from
coordination amongst distinct, interacting individuals and bodies, rather
than being identified with the voice of a single spokesperson, as sover-
eigntists like Hobbes, and indeed Rousseau, thought it had to be. They
were guilty of the fallacy of misplaced concreteness in thinking that there
could be no coherence or unity in the voice of the people, short of
establishing a single individual or corporate sovereign.
14. Turning to the second condition, the people will be resistance-prone,
ensuring that their influence is not conditional on any other party’s
goodwill, insofar as they are disposed to contest what government
proposes or decides on various fronts. This vigilance requires the virtue
displayed by those who identify with social movements that make their
arguments in public, on purportedly publicly acceptable grounds. This
is a motivated form of virtue, deriving from personal interest or sponta-
neous commitment, and ought not to be in short supply.
15. The ideal of the people’s sovereign assembly, as embraced by Rousseau,
implies that it is only this body that can speak for the people. Thus it
would challenge the right of individual contestation on the grounds,
derived from Hobbes by Rousseau, that if individuals were allowed this
right, then a further authority would be required to judge between those
individuals and the people as a whole, in which case that authority would
be sovereign. This shows how tight is the connection between the twin
republican ideas of the mixed constitution and the contestatory citizenry.
16. The final requirement that an electoral system of influence has to satisfy
is that it be suitably efficacious: whatever direction it serves to impose
on government – the topic of the next chapter – it does so with such an
effect that individuals who do not like what government does, either in
general or in its dealing with them, can see this as tough luck, not the
result of their subjection to an alien will.
17. The influence of the people under an electoral system is potentially
enormous, consisting in their actual electoral and contestatory inputs,
in their dispositions to make such inputs should government take a line
they do not like, or in their dispositions to make such inputs should
their own wishes shift in light of government policy. It may take an
active, a virtual or a reserve form.
18. Still, this influence is likely to be diluted if elected politicians are allowed
to dictate policy in domains where their electoral, party-political motives
are dominant or if private interest-groups are allowed to have a covert
or manipulative influence on government, forcing it to respond to their
special concerns.
19. The only hope of guarding against such influences requires setting up
unelected agencies that are appointed by elected representatives but do
not serve at their pleasure. These will include executive authorities like
electoral commissions or central banks, contestatory authorities like
ombudsman or auditor bodies, and, of course, the judicial authorities
represented in various courts and tribunals. These can help ensure
against the problems mentioned.
20. But will such authorities themselves dilute popular influence? Not if they
operate with publicly dictated briefs, under publicly imposed con-
straints, exposed to public challenge and review. In that case they can
count as suitably constrained indicative representatives of the people.
chapter 5. democratic control

1. No matter how powerful a system of popular influence, it will not support
republican democracy unless it serves to impose a popular direction on
government. Democracy requires more than popular influence, contrary
to the image promulgated in the work of Joseph Schumpeter and accepted
in most branches of political science.
2. There are two models of how popular influence might put a directive
stamp on government policy, thereby giving the people control. Under
one, the people would intentionally impose the required direction;
under the other, they would act in a way that has the unintended but
welcome effect of imposing a certain direction.
3. The intentional model suggests that people can form and then impose
their view of the public interest, which is best interpreted as the interest
that they each have in how things should be done, given that they are
required to live on equal terms with others. The trouble with the model,
however, is that it would require a great deal of unmotivated virtue in
Chapter 5. Democratic control 307
both the people and their representatives; it would be vulnerable to any
failure of virtue on their part.
4. The non-intentional model, classically illustrated in James Mill’s pro-
posal, suggests that if people just vote for their private preferences, and if
representatives have to answer for how they serve those interests in very
frequent elections, then, as by an invisible hand, those in power will tend
to maximize overall satisfaction: a utilitarian version of the public
interest. This model is equally problematic, because it would be vulner-
able to the efforts of special coalitions of voters or representatives to serve
their own particular ends.
5. But however problematic, these models have complementary attractions.
The first directs us to an appealing conception of the public interest,
the second to an appealing, invisible-hand mode, in which the public
interest might be promoted by popular influence. And it turns out that
there is a third model, to be developed here, in which those separate
attractions can be combined.
6. In order to introduce this model, we must distinguish between two
different procedures whereby people might try to identify and secure a
collective, universally beneficial arrangement. In the game of acceptance
they each seek to make the minimal concession required, perhaps in a
series of bids and counter-bids. In the game of acceptability they each
seek to move the group by appeal to considerations that all find accept-
able; they are all ready to recognize them as relevant in public decision-
making, even if some weight them more heavily than others.
7. In the acceptability game the considerations that make a plausible claim
to relevance, as participants abide by the norm of making an appeal to
all, are going to be supported by special norms. Each emerging consid-
eration will be such that, as a matter of common belief, everyone treats it
as relevant, everyone expects everyone to approve of this treatment and
to disapprove of the refusal to grant it, and everyone will be moved by
this approbative pattern to grant such treatment themselves.
8. The considerations that are supported in this way by norms may mark
out the convergent interests of the members of the group in some general
benefit or the concordant interests of individual members or subgroups
in receiving special treatment: like the interests subgroups have in being
treated as equals with others, these will be special interests that it is in the
convergent interest of all to recognize as important for each.
9. The currency of public policy-making norms that ought to emerge in
any group that operates under the acceptability game will put various
policies off the table, reducing the eligible candidates in any domain. But
they will also put various processes for selecting between eligible can-
didates off the table in any domain. The approved process in a particular
domain might involve election amongst members, a lottery device of
some kind, reference to another body, or whatever.
10. The acceptability game is clearly present in the operation of private
associations like a condominium of apartment owners. But it ought also
to have a presence under the sort of electoral-cum-contestatory regime
described in the last chapter. It will make room for the emergence of
policy-making norms at the various, mutually accessible, sites of debate,
formal and informal, that the system sets up.
11. The sorts of norms that we might expect to emerge will license the
importance of equal influence and associated aspects of equal status,
register the demands of personal space and freedom, establish desiderata
on how public decision-making is to be conducted, and help to fix the
range of duties that the state is expected to promote. And they will also
identify historically inherited features of the system as fixtures that,
fortunately or unfortunately, are not up for revision.
12. Such emerging norms of policy-making are bound to serve a constrain-
ing and regulatory function, ensuring that while different policies are
identified and decided in ways that suit different factions, they are all
fixed in a way that conforms to the requirements of the commonly
endorsed norms. Not everything will be decided in a deliberative way
but everything will be done under a form of deliberative regulation.
13. This regulation may not be very salient when it is working well. Its effect
will be to put off the table an indefinite range of policies and processes,
making them more or less unthinkable for members of the relevant
community. That effect may be quite important and yet not be much
remarked within the community.
14. If a political system operates under the sort of popular influence described
in the last chapter, imposing norms of the kind described in this, then it
will have a dual aspect, each with its own temporal register. In the short
haul, the system will engage people in electoral and contestatory exercises
of a familiar kind. But over the long haul it will engage them in the
exercise of ensuring the compliance of public policy-making with the
norms that get established amongst citizens, or at least amongst citizens
who are willing to live on equal terms with others.
15. This dual aspect of a functioning democracy can be illustrated in actual-
world politics. Oliver MacDonagh’s studies of the changes in government
policy as Britain democratized over fifty years or so in the nineteenth
century serve this purpose nicely. They reveal a pattern whereby publicity
Chapter 5. Democratic control 309
given to what are scandals by contemporary norms, now in this domain,
now in that, generates popular outrage and then forces government to
respond with remedial measures. These measures become progressively
more substantive as renewed publicity reveals the failure of existing
interventions to rectify the problem.
16. The dual-aspect model of democracy combines the attractions of the
intentional and non-intentional pictures. In this model, government is
forced to promote the public interest, conforming to public norms in
the policies adopted and the processes used to decide them. And in this
model it does so via an invisible hand akin to that which is held to
operate in the open market, not by the formation and imposition of a
general will or intention. Where the market harnesses people’s eco-
nomic influence so as to support competitive pricing, this model would
harness their political influence – ideally, their individualized, uncon-
ditioned and efficacious influence – so as to ensure compliance with
policy-making norms.
17. The model makes sense of how government can satisfy Lincoln’s three
demands. Government will be for the people insofar as it is forced to
satisfy accepted norms. It will be by the people insofar as the influence
that forces it to do this originates, directly or indirectly, with the people.
And it will be of the people insofar as this influence, being uncondi-
tioned, is independent of the goodwill of any other party; it prevails
regardless of the wishes of those in government, a local powerful elite, or
indeed a foreign country.
18. The dual-aspect model allows the emergence of a state that has a single
voice, operates to a coherent set of ends, and can reliably incur and
discharge commitments to its citizens and other national or international
bodies. It allows the state to take on the conversable form that, as we saw,
any state must assume if it is to be able to discharge standard duties.
19. The people under the dual-aspect model appear in two roles. The
people as a plurality establish and maintain the state insofar as it is
their influence, directly or indirectly applied, that shapes policy-making
and imposes a norm-complying pattern on it. But in establishing the
state, those people incorporate as a group entity, taking the form of a
singular agency rather than a plurality of many agents; the people in this
sense just is the state.
20. In the first role they may be described, in Sieyes’s term, as the constitut-
ing people; in the second as the constituted people. Both as constituting
and constituted, the people rule themselves, satisfying Lincoln’s triad.
This ambiguity in talk of popular self-rule has an ancient pedigree,
appearing in the medieval use of civitas to refer either to the citizenry or
to the state.
21. Under the dual-aspect model, as distinct from the sovereignist picture,
the constituting people have a priority over the constituted. While the
constituted people counts as a law-making sovereign that is not itself
subject to the law, as in Bodin and Hobbes’s ideal, it exists only by
virtue of the constituting people; being formed by the constituting
people, it cannot dominate them; and it can be replaced by the con-
stituting people as they set aside the constitution under which it comes
into existence.
22. Under the dual-aspect model, finally, there is no paradox in the claim
that the people are causally responsible for the constitution, yet exist
only in virtue of the constitution. It is the constituting people who are
responsible historically for the constitution and they do not require the
constitution in order to exist. And both the constituting and consti-
tuted people are responsible in the sense of being able to alter the
constitution at any time. The constituted people can alter it by acting
within the rules for amendment and the constituting people can alter it
even without conforming to those rules.
References
Abizadeh, A. (2008). ‘Democratic Theory and Border Coercion: No Right to

Unilaterally Control Your Own Borders’. Political Theory 36: 37–65.
Abramson, J. (1994). We, the Jury: The Jury System and the Ideal of Democracy. New
York, Basic Books.
Ackerman, B. (2000). ‘The New Separation of Powers’. Harvard Law Review 113:
633–729.
Adams, J. (1776). Thoughts on Government Applicable to the Present State of the
American Colonies. Philadelphia, John Dunlap.
Agamben, G. (2005). State of Exception. Chicago University Press.
Alexander, J. M. (2008). Capabilities and Social Justice: The Political Philosophy of
Amartya Sen and Martha Nussbaum. Aldershot, Ashgate.
Amar, A. R. (1988). ‘Philadelphia Revisited: Amending the Constitution outside
Article V’. University of Chicago Law Review 55: 1043–104.
Anderson, E. (1999). ‘What is the Point of Equality?’ Ethics 109: 287–337.
Anscombe, G. E. M. (1957). Intention. Oxford, Blackwell.
Appiah, K. A. (2010). The Honor Code: How Moral Revolutions Happen. New York,
Notrons.
Arendt, H. (1958). The Human Condition. University of Chicago Press.
(1973). On Revolution. Harmondsworth, Pelican Books.
Aristotle (1996). The Politics, ed. S. Everson. Cambridge University Press.
Arneson, R. (1989). ‘Equality and Equal Opportunity for Welfare’. Philosophical
Studies 56: 77–93.
Arrow, K. (1963). Social Choice and Individual Values. New York, Wiley.
Atiyah, P. S. (1979). The Rise and Fall of Freedom of Contract. Oxford University Press.
Bailyn, B. (1967). The Ideological Origins of the American Revolution. Cambridge,
MA, Harvard University Press.
Bakan, J. (2004). The Corporation: The Pathological Pursuit of Profit and Power.
New York, Free Press.
Baldwin, T. (1984). ‘MacCallum and the Two Concepts of Freedom’. Ratio 26: 125–42.
Barry, B. (1995). Justice as Impartiality. Oxford University Press.
Bartels, L. M. (2008). Unequal Democracy. Princeton University Press.
Beitz, C. (1989). Political Equality: An Essay in Democratic Theory. Princeton
University Press.
(2009). The Idea of Human Rights. Oxford University Press.
311
312 References
Bell, D. (2010). China’s New Confucianism: Politics and Everyday Life in a Changing
Society. Princeton University Press.
Bellamy, R. (1999). Liberalism and Pluralism: Towards a Politics of Compromise.
London, Routledge.
(2007). Political Constitutionalism: A Republican Defense of the Constitutionality
of Democracy. Cambridge University Press.
Benhabib, S. (1996). ‘Towards a Deliberative Model of Democratic Legitimacy’ in
Democracy and Difference: Contesting the Boundaries of the Political, ed.
S. Benhabib. Princeton University Press.
Bentham, J. (1843). ‘Anarchical Fallacies’ in The Works of Jeremy Bentham, ed.
J. Bowring. Edinburgh, W. Tait, vol. ii.
Berlin, I. (1969). Four Essays on Liberty. Oxford University Press.
Besold, C. (1618). Politicorum Libri duo. Frankfurt, J. A. Cellii.
Besson, S. and J. L. Marti (2008). Law and Republicanism. Oxford University
Press.
Bodin, J. (1967). Six Books of the Commonwealth, ed. M. J. Tooley. Oxford, Blackwell.
Bohman, J. (2007). Democracy Across Borders: From Demos to Demoi. Cambridge,
MA, MIT Press.
Braithwaite, J. (1997). ‘On Speaking Softly and Carrying Big Sticks: Neglected
Dimensions of a Republican Separation of Powers’. University of Toronto Law
Journal 47: 305–61.
(2002). Restorative Justice and Responsive Regulation. New York, Oxford University
Press.
Braithwaite, J., H. Charlesworth and A. Soares (2012). Networked Governance of
Freedom and Tyranny: Peace in East Timor. Canberra, ANU Press.
Braithwaite, J. and P. Pettit (1990). Not Just Deserts: A Republican Theory of
Criminal Justice. Oxford University Press.
Bramhall, J. (1658). ‘The Catching of Leviathan or the Great Whale’ in Castigations
of Mr Hobbes . . . Concerning Liberty and Universal Necessity, J. Bramhall.
London, John Crooke.
Bratman, M. (1987). Intention, Plans, and Practical Reason. Cambridge, MA, Harvard
University Press.
(2004). ‘Shared Valuing and Frameworks for Practical Reasoning’ in Reason and
Value: Themes from the Moral Philosophy of Joseph Raz, ed. J. Wallace, P. Pettit,
S. Scheffler and M. Smith. Oxford University Press, pp. 1–27.
(2007). Structures of Agency. New York, Oxford University Press.
Brennan, G. and A. Hamlin (1995). ‘Economizing on Virtue’. Constitutional
Political Economy 6: 35–6.
Brennan, G. and L. Lomasky (1993). Democracy and Decision: The Pure Theory of
Electoral Preference. Oxford University Press.
(2006). ‘Against Reviving Republicanism’. Politics, Philosophy and Economics 5:
221–52.
Brennan, G. and P. Pettit (2004). The Economy of Esteem: An Essay on Civil and
Political Society. Oxford University Press.
Brennan, J. (2011). The Ethics of Voting. Princeton University Press.
References 313
Brett, A. S. (2011). Changes of State: Nature and the Limits of the City in Early
Modern Natural Law. Princeton University Press.
Brettschneider, C. (2007). Democratic Rights: The Substance of Self-Government.
Princeton University Press.
Brugger, W. (1999). Republican Theory in Political Thought: Virtuous or Virtual?
New York, Macmillan.
Buchanan, A. (2002). ‘Political Legitimacy and Democracy’. Ethics 112: 689–719.
(2004). Justice, Legitimacy and Self-Determination: Moral Foundations for
International Law. Oxford University Press.
Burke, E. (1999). ‘Speech to the Electors of Bristol’ in Select Works of Edmund
Burke. Indianapolis, Liberty Fund.
Burnheim, J. (1985). Is Democracy Possible? Cambridge, Polity Press.
Cabot, V. (1751). ‘Quod Non Omnis Status Reipublicae Simplex Est’ in Novus
Thesaurus Juirs Civilis et Canonici. G. Meerman. Amsterdam, Pieter de Hondt.
2: 622–3.
Canning, J. P. (1983). ‘Ideas of the State in Thirteenth and Fourteenth Century
Commentators on the Roman Law’. Transactions of the Royal Historical Society
33: 1–27.
Carroll, L. (1895). ‘What the Tortoise said to Achilles’. Mind 4: 278–80.
Carter, I. (1999). A Measure of Freedom. Oxford University Press.
(2008). ‘How are Power and Unfreedom Related?’ in Republicanism and Political
Theory, ed. C. Laborde and J. Maynor. Oxford, Blackwell.
Christiano, T. (2008). The Constitution of Equality: Democratic Authority and its
Limits. Oxford University Press.
Christman, J. (2009). The Politics of Persons: Individual Autonomy and Socio-
historical Selves. Cambridge University Press.
Cicero, M. T. (1998). The Republic and the Laws. Oxford University Press.
Clark, A. (1997). Being There: Putting Brain, Body and World Together Again.
Cambridge, MA, MIT Press.
Cohen, G. A. (1979). ‘Capitalism, Freedom and the Proletariat’ in The Idea of
Freedom, ed. A. Ryan. Oxford University Press.
(1993). ‘Equality of What? On Welfare, Goods, and Capabilities’ in The Quality
of Life, ed. M. C. Nussbaum and A. Sen. Oxford University Press, pp. 9–29.
(2008). Rescuing Justice and Equality. Cambridge, MA, Harvard University Press.
Cohen, J. (1989). ‘Deliberation and Democratic Legitimacy’ in The Good Polity, ed.
A. Hamlin and P. Pettit. Oxford, Blackwell, pp. 17–34.
(2004). ‘Minimalism About Human Rights: The Most We Can Hope For’.
Journal of Political Philosophy 12: 190–213.
(2009). Philosophy, Politics, Democracy. Cambridge, MA, Harvard University Press.
(2010). Rousseau: A Free Community of Equals. Oxford University Press.
Cohen, M. (1933). ‘The Basis of Contract’. Harvard Law Review 4: 553–92.
Coleman, J. (1974). Power and the Structure of Society. New York, Norton.
(1990). ‘The Emergence of Norms’ in Social Institutions: Their Emergence,
Maintenance, and Effects, ed. M. Hechter, K.-D. Opp and R. Wippler,
Berlin, de Gruyter, pp. 35–59.
314 References
Connolly, W. E. (1969). The Bias of Pluralism. New York, Atherton.
(1993). The Terms of Political Discourse. Oxford, Blackwell.
Constant, B. (1988). Constant: Political Writings. Cambridge University Press.
(2003). Principles of Politics Applicable to All Governments. Indianapolis, Liberty
Fund.
Cornish, W. R. and G. d. N. Clark (1989). Law and Society in England 1750–1950.
London, Sweet and Maxwell.
Costa, M. V. (2007). ‘Freedom as Non-Domination, Normativity and
Indeterminacy’. Journal of Value Inquiry 41: 291–307.
Dagger, R. (1997). Civic Virtues: Rights, Citizenship, and Republican Liberalism.
Oxford University Press.
Demosthenes (1939). Against Meidias, trans. A. T. Murray. London, Heinemann.
Dennett, D. C. (1992). Consciousness Explained. New York, Penguin.
(1996). Darwin’s Dangerous Idea: Evolution and the Meanings of Life. New York,
Simon and Schuster.
Dietrich, F. and C. List (2007). ‘Arrow’s Theorem in Judgment Aggregation’.
Social Choice and Welfare 29: 19–33.
(2008). ‘A Liberal Paradox for Judgment Aggregation’. Social Choice and Welfare
31: 59–78.
Dowlen, O. (2008). The Political Potential of Sortition. Exeter, Imprint Academic.
Dryzek, J. (2003). Deliberative Democracy and Beyond: Liberals, Critics, Contestations.
Duff, R. A. (2001). Punishment, Communication, and Community. Oxford University
Press.
Dunn, J. (2005). Democracy: A History. New York, Atlantic Monthly Press.
Dworkin, G. (1988). The Theory and Practice of Autonomy. Cambridge University
Press.
Dworkin, R. (1978). Taking Rights Seriously. London, Duckworth.
(1986). Law’s Empire. Cambridge, MA, Harvard University Press.
(2000). Sovereign Virtue: The Theory and Practice of Equality. Cambridge, MA,
Harvard University Press.
(2006). Is Democracy Possible Here? Principles for a New Political Debate.
(2011). Justice for Hedgehogs. Cambridge, MA, Harvard University Press.
Eisgruber, C. L. (2001). Constitutional Self-Government. Cambridge, MA, Harvard
University Press.
(2002). ‘Constitutional Self-Government and Judicial Review: A Reply to Five
Critics’. University of San Francisco Law Review 37: 115–90.
Elster, J. (1979). Ulysses and the Sirens. Cambridge University Press.
(1983). Sour Grapes. Cambridge University Press.
(1986). ‘The Market and the Forum: Three Varieties of Political Theory’ in
Foundations of Social Choice Theory, ed. J. Elser and A. Hillard. Cambridge
University Press.
(1999). Alchemies of the Mind: Rationality and the Emotions. Cambridge University
Press.
References 315
Ely, J. H. (1981). Democracy and Distrust: A Theory of Judicial Review. Cambridge,
Eskridge, W. N., Jr and J. Ferejohn (2010). A Republic of Statutes: The New
American Constitution. New Haven, CT, Yale University Press.
Espejo, P. O. (2011). The Time of Popular Sovereignty: Process and the Democratic
State. University Park, Pennsylvania State University Press.
Estlund, D. (2007). Democratic Authority: A Philosophical Framework. Princeton
University Press.
Ferguson, A. (1767). An Essay on the History of Civil Society. Edinburgh, Millar and
Caddel (reprinted New York, Garland, 1971).
Fishkin, J. (1991). Democracy and Deliberation: New Directions for Democratic
Reform. New Haven, CT, Yale University Press.
(1997). The Voice of the People: Public Opinion and Democracy. New Haven, CT,
Yale University Press.
Fleurbaey, M. (2008). Fairness, Responsibility and Welfare. Oxford University Press.
Forst, R. (2002). Contexts of Justice: Political Philosophy beyond Liberalism and
Communitarianism. Berkeley, University of California Press.
Frankfurt, H. (1969). ‘Alternate Possibilities and Moral Responsibility’. Journal of
Philosophy 66: 829–39.
(1987). ‘Equality as a Moral Ideal’. Ethics 98: 21–43.
(1988). The Importance of What We Care About. Cambridge University Press.
Franklin, J. (1991). ‘Sovereignty and the Mixed Constitution: Bodin and his Critics’
in The Cambridge History of Political Thought 1450–1700, ed. J. H. Burns and
M. Goldie. Cambridge University Press.
Fukuyama, F. (2011). The Origins of Political Order: From Prehuman Times to the
French Revolution. New York, Farrar, Straus and Giroux.
Fuller, L. L. (1971). The Morality of Law. New Haven, CT, Yale University Press.
Garsten, B. (2006). Saving Persuasion: A Defense of Rhetoric and Judgment.
Cambridge, MA, Harvard University Press.
Gaus, G. (1983). The Modern Liberal Theory of Man. London, Croom Helm.
(2011). The Order of Public Reason: A Theory of Freedom and Morality in a Diverse
and Bounded World. Cambridge University Press.
Gauthier, D. (1986). Morals by Agreement. Oxford University Press.
Geuss, R. (1981). The Idea of Critical Theory. Cambridge University Press.
Gilbert, M. (2001). ‘Collective Preferences, Obligations, and Rational Choice’.
Economics and Philosophy 17: 109–20.
(2006). A Theory of Political Obligation. Oxford University Press.
Gilovich, T., D. Griffin and D. Kahneman, eds. (2002). Heuristics and Biases: The
Psychology of Intuitive Judgment. Cambridge University Press.
Goldberg, J. (2005–6). ‘The Constitutional Status of Tort Law: Due Process
and the Right to a Law for the Redress of Wrongs’. Yale Law Journal 115:
524–627.
Goldman, A. (1999). ‘Why Citizens Should Vote: A Causal Responsibility
Approach’. Social Philosophy and Policy 16: 201–17.
Goldsworthy, J. (1999). The Sovereignty of Parliament. Oxford University Press.
316 References
Goodin, R. E. (1996). ‘Institutionalizing the Public Interest: The Defense of
Deadlock and Beyond’. American Political Science Review 90: 331–43.
Goodin, R. E. and F. Jackson (2007). ‘Freedom from Fear’. Philosophy and Public
Affairs 35: 249–65.
Green, J. E. (2004). ‘Apathy: The Democratic Disease’. Philosophy and Social Criticism
30: 745–68.
Guerrero, A. (2010). ‘The Paradox of Voting and the Ethics of Political
Representation’. Philosophy & Public Affairs 38: 272–306.
Guinier, L. (1994). Tyranny of the Majority: Fundamental Fairness in Representative
Democracy. New York, The Free Press.
Gutmann, A. and D. Thompson (1996). Democracy and Disagreement. Cambridge,
Gwyn, W. B. (1965). The Meaning of the Separation of Powers. Nijhoff, The Hague.
Habermas, J. (1984–9). A Theory of Communicative Action. Cambridge, Polity
Press, vols. i and ii.
(1994). ‘Three Normative Models of Democracy’. Constellations 1: 1–10.
(1995). Between Facts and Norms: Contributions to a Discourse Theory of Law and
Democracy. Cambridge, MA, MIT Press.
Hansen, M. H. (1991). The Athenian Democracy in the Age of Demosthenes. Oxford,
Blackwell.
Hardt, M. and A. Negri (2000). Empire. Cambridge, MA, Harvard University Press.
Harman, G. (1986). Change in View. Cambridge, MA, MIT Press.
Harrington, J. (1992). The Commonwealth of Oceana and a System of Politics.
Cambridge University Press.
Hart, H. L. A. (1955). ‘Are There Any Natural Rights?’ Philosophical Review 64: 175–91.
(1961). The Concept of Law. Oxford University Press.
(1973). ‘Rawls on Liberty and its Priority’. University of Chicago Law Review 40:
534–55.
Hayek, F. A. (1988). The Fatal Conceit: The Errors of Socialism. University of
Chicago Press.
Hayward, C. (2011). ‘What Can Political Freedom Mean in a Multicultural
Democracy? On Deliberation, Difference and Democratic Government’.
Political Theory 39: 468–97.
Hegel, G. W. F. (1991). Elements of the Philosophy of Right, ed. A. W. Wood.
Held, D. (2006). Models of Democracy, 3rd edn. Cambridge, Polity.
Herreros, F. (2006). ‘Screening Before Sanctioning: Elections and the Republican
Tradition’. European Journal of Political Theory 5: 415–35.
Hill, L. (2000). ‘Compulsory Voting, Political Shyness and Welfare Outcomes’.
Journal of Sociology 36: 30–49.
Hobbes, T. (1994a). Human Nature and De Corpore Politico: The Elements of Law,
Natural and Politic. Oxford University Press.
(1994b). Leviathan, ed. E. Curley. Indianapolis, Hackett.
(1998). On the Citizen, ed. and trans. R. Tuck and M. Silverthorne. Cambridge
University Press.
References 317
Hobbes, T. and J. Bramhall (1999). Hobbes and Bramhall on Freedom and Necessity,
ed. Vere Chappell. Cambridge University Press.
Hoekstra, K. (2001). ‘Tyrannus Rex vs. Leviathan’. Pacific Philosophical Quarterly 82:
420–46.
(2006). ‘A Lion in the House: Hobbes and Democracy’ in Rethinking the
Foundations of Modern Political Thought, ed. A. S. Brett and J. Tully.
Holmes, S. (1995). Passions and Constraint: On the Theory of Liberal Democracy.
University of Chicago Press.
Honig, B. (2001). Democracy and the Foreigner. Princeton University Press.
Honneth, A. (1996). The Struggle for Recognition. Cambridge, MA, MIT Press.
Honohan, I. (2002). Civic Republicanism. London, Routledge.
Honohan, I. and J. Jennings, eds. (2006). Republicanism in Theory and Practice.
London, Routledge.
Hont, I. (1995). ‘The Permanent Crisis of a Divided Mankind: “Contemporary
Crisis of the National State” in Historical Perspective’ in The Contemporary
Crisis of the Nation State?, ed. J. Dunn. Oxford, Blackwell, pp. 166–231.
Hume, D. (1875). Of the Independence of Parliament. Hume’s Philosophical Works,
ed. T. H. Green and T. H. Grose. London, vol. iii.
(1994). Political Essays. Cambridge University Press.
Husak, D. (2008). Overcriminalization. Oxford University Press.
Jackson, F. (1987). ‘Group Morality’ in Metaphysics and Morality: Essays in Honour
of J. J. C. Smart, ed. P. Pettit, R. Sylvan and J. Norman. Oxford, Blackwell,
pp. 91–110.
James, S. (1997). Passion and Action: The Emotions in Seventeenth-Century
Philosophy. Oxford University Press.
Joyce, J. M. (1999). The Foundations of Causal Decision Theory. Cambridge
University Press.
Kalyvas, A. (2005). ‘Popular Sovereignty, Democracy and the Constituent Power’.
Constellations 12: 223–44.
Kalyvas, A. and I. Katznelson (2008). Liberal Beginnings: Making a Republic for the
Moderns. Cambridge University Press.
Kant, I. (1996). Practical Philosophy, trans. M. J. Gregor. Cambridge University Press.
(2005). Notes and Fragments, ed. Paul Guyer. Cambridge University Press.
Kaufman, A. (2006). Capabilities Equality: Basic Issues and Problems. London,
Routledge.
Kelly, E. (2009). ‘Criminal Justice without Retribution’. Journal of Philosophy 106:
440–62.
Kelsen, H. (1961). General Theory of Law and State. New York, Russell and Russell.
(1970). The Pure Theory of Law. Berkeley, CA, University of California Press.
Keohane, N. O. (1980). Philosophy and the State in France: The Renaissance to the
Enlightenment. Princeton University Press.
Ketcham, R., ed. (2003). The Anti-Federalist Papers. New York, Signet Classic.
Kharkhordin, O. and R. Alipuro, eds. (2011). Political Theory and Community
Building in Post-Soviet Russia. London, Routledge.
318 References
Knights, M. (2005). Representation and Misrepresentation in Later Stuart Britain:
Partisanship and Political Culture. Oxford University Press.
Kornhauser, L. A. (1992a). ‘Modelling Collegial Courts. I. Path-Dependence’.
International Review of Law and Economics 12: 169–85.
(1992b). ‘Modelling Collegial Courts. ii. Legal Doctrine’. Journal of Law,
Economics and Organization 8: 441–70.
Kornhauser, L. A. and L. G. Sager (1993). ‘The One and the Many: Adjudication in
Collegial Courts’. California Law Review 81: 1–59.
Kramer, M. (2003). The Quality of Freedom. Oxford University Press.
(2008). ‘Liberty and Domination’ in Republicanism and Political Theory, ed.
C. Laborde and J. Maynor. Oxford, Blackwell, pp. 31–57.
Kukathas, C. (1989). Hayek and Modern Liberalism. Oxford University Press.
Kymlicka, W. (1995). Multicultural Citizenship. Oxford University Press.
Laborde, C. (2008). Critical Republicanism: The Hijab Controversy and Political
Philosophy. Oxford University Press.
Laborde, C. and J. Maynor, eds. (2007). Republicanism and Political Theory.
Oxford, Blackwell.
Langton, R. (1997). ‘Disenfranchised Silence’ in Common Minds: Themes from the
Philosophy of Philip Pettit, ed. H. G. Brennan, R. E. Goodin, F. C. Jackson
and M. Smith. Oxford University Press.
Languet, H. (1994). Vindiciae, Contra Tyrannos. Cambridge University Press.
Larmore, C. (2003). ‘Public Reason’ in The Cambridge Companion to Rawls, ed.
S. Freeman. Cambridge University Press, pp. 368–93.
(2012). ‘What is Political Philosophy?’ Journal of Moral Philosophy 9.
Lessig, L. (2011). Republic, Lost: How Monday Corrupts Congress – And a Plan to Stop
it. New York, Hachette.
Levin, M. (1984). ‘Negative Liberty’. Social Philosophy and Policy 2: 84–100.
Lewis, D. (1969). Convention. Cambridge, MA, Harvard University Press.
Libourne, J. (1646). The Legal Fundamental Liberties of the People of England,
Asserted, Revived, and Vindicated. London.
Lieberman, D. (2006). ‘The Mixed Constitution and the Common Law’ in The
Cambridge History of Eighteenth-Century Political Thought, ed. M. Goldie and
R. Wokler. Cambridge University Press, pp. 317–46.
Lind, J. (1776). Three Letters to Dr Price. London, T. Payne.
Lindahl, H. (2007). ‘The Paradox of Constituent Power: The Ambiguous Self-
Constitution of the European Union’. Ratio Juris 20: 485–505.
Lipset, S. M. (1960). Political Man: The Social Bases of Politics. New York,
Doubleday.
List, C. (2004). ‘The Impossibility of a Paretian Republican? Some Comments on
Pettit and Sen’. Economics and Philosophy 20: 1–23.
(2006a). ‘The Discursive Dilemma and Public Reason’. Ethics 116: 362–402.
(2006b). ‘Republican Freedom and the Rule of Law’. Politics, Philosophy and
Economics 5: 201–20.
List, C. and P. Pettit (2002). ‘Aggregating Sets of Judgments: An Impossibility
Result’. Economics and Philosophy 18: 89–110.
References 319
(2004). ‘Aggregating Sets of Judgments: Two Impossibility Results Compared’.
Synthese 140: 207–35.
(2011). Group Agency: The Possibility, Design and Status of Corporate Agents.
List, C. and B. Polak (2010). ‘Symposium on Judgment Aggregation’. Journal of
Economic Theory 145 (2): 441–66.
Lively, J. (1975). Democracy. Oxford, Blackwell.
Lively, J. and J. Rees, eds. (1978). Utilitarian Logic and Politics: James Mill’s ‘Essay on
Government’, Macaulay’s Critique, and the Ensuing Debate. Oxford University
Press.
Locke, J. (1960). Two Treatises of Government. Cambridge University Press.
Long, D. C. (1977). Bentham on Liberty. University of Toronto Press.
Lovett, F. (2001). ‘Domination: A Preliminary Analysis’. Monist 84: 98–112.
(2010). Justice as Non-domination. Oxford University Press.
Lovett, F. and P. Pettit (2009). ‘Neo-Republicanism: A Normative and
Institutional Research Program’. Annual Review of Political Science 12: 18–29.
Luhmann, N. (1990). Essays on Self-Reference. New York, Columbia University Press.
MacCallum, G. C. (1967). ‘Negative and Positive Freedom’. Philosophical Review
74: 312–34.
McCormick, J. P. (2011). Machiavellian Democracy. Cambridge University Press.
MacDonagh, O. (1958). ‘The 19th Century Revolution in Government: A
Reappraisal’. Historical Journal 1: 52–67.
(1961). A Pattern of Government Growth 1800–60. London, MacGibbon and Kee.
(1977). Early Victorian Government. London, Weidenfeld and Nicolson.
McGeer, V. and P. Pettit (2009). ‘Sticky Judgment and the Role of Rhetoric’ in
Political Judgment: Essays in Honour of John Dunn, ed. R. Bourke and
R. Geuss. Cambridge University Press, pp. 48–73.
McGilvray, E. (2011). The Invention of Market Freedom. Cambridge University
Press.
Machiavelli, N. (1965). The Complete Works and Others. Durham, NC, Duke
University Press.
Mackie, G. (2003). Democracy Defended. Cambridge University Press.
McLean, J. (2004). ‘Government to State: Globalization, Regulation, and
Governments as Legal Persons’. Indiana Journal of Global Legal Studies 10:
173–97.
McMahon, C. (2005). ‘The Indeterminacy of Republican Policy’. Philosophy and
Public Affairs 33: 67–93.
Madison, J., A. Hamilton and J. Jay (1987). The Federalist Papers. Harmondsworth,
Penguin.
Mandeville, B. (1731). Free Thoughts on Religion, the Church and National
Happiness. London.
Manin, B. (1997). The Principles of Representative Government. Cambridge
University Press.
Mansbridge, J. (2009). ‘A “Selection Model” of Political Representation’. Journal of
Political Philosophy 17: 369–98.
320 References
Margalit, A. (2010). On Compromise and Rotten Compromises. Princeton University
Press.
Markell, P. (2008). ‘The Insufficiency of Non-Domination’. Political Theory 36: 19–36.
Markovits, D. (2005). ‘Democratic Disobedience’. Yale Law Journal 114: 1897–952.
Marti, J. L. and P. Pettit (2010). A Political Philosophy in Public Life: Civic
Republicanism in Zapatero’s Spain. Princeton University Press.
Maynor, J. (2003). Republicanism in the Modern World. Cambridge, Polity Press.
Mercier, H. and D. Sperber (2011). ‘Why Do Humans Reason? Arguments for an
Argumentative Theory’. Brain and Behavioral Sciences 34: 57–111.
Michelman, F. I. (1999). Brennan on Democracy. Princeton University Press.
Mill, J. S. (1964). Considerations on Representative Government. London, Everyman
Books.
(1978). On Liberty. Indianapolis, Hackett.
Millar, F. (1998). The Crowd in Rome in the Late Republic. Ann Arbor, University of
Michigan Press.
Miller, D. (1984). ‘Constraints on Freedom’. Ethics 94: 66–86.
Milton, J. (1953–82). Complete Prose Works of John Milton. New Haven, CT, Yale
University Press, vols. i–viii.
Montesquieu, C. d. S. (1989). The Spirit of the Laws. Cambridge University Press.
Morris, I. (2010). Why the West Rules – For Now: The Patterns of History, and What
they Reveal about the Future. New York, Farrar, Straus and Giroux.
Mueller, J. W. (2007). Constitutional Patriotism. Princeton University Press.
Murphy, L. and T. Nagel (2004). The Myth of Ownership. New York, Oxford
University Press.
Nagel, T. (1987). ‘Moral Conflict and Political Legitimacy’. Philosophy and Public
Affairs 16: 215–40.
(1991). Equality and Partiality. Oxford University Press.
Nathan, G. (2010). Social Freedom in a Multicultural State: Towards a Theory for
Intercultural Justice. New York, Palgrave Macmillan.
Negri, A. (1999). Insurgencies: Constituent Power and the Modern State. Minneapolis,
MN, University of Minnesota Press.
Nelson, E. (2004). The Greek Tradition in Republican Thought. Cambridge University
Press.
Niederberger, A. and P. Schink, eds. (2012). Republican Democracy: Liberty, Law
and Politics. Edinburgh University Press.
Nozick, R. (1974). Anarchy, State, and Utopia. Oxford, Blackwell.
Nussbaum, M. (1992). ‘Human Functioning and Social Justice’. Political Theory
20: 202–46.
(2006). Frontiers of Justice. Cambridge, MA, Harvard University Press.
O’Donnell, G. (2004). ‘Why the Rule of Law Matters’. Journal of Democracy 15:
32–46.
O’Neill, M. (2008). ‘What Should Egalitarians Believe?’ Philosophy and Public
Affairs 36: 119–56.
O’Neill, O. (1979–80). ‘The Most Extensive Liberty’. Proceedings of the Aristotelian
Society 80: 45–59.
References 321
Ober, J. (1996). The Athenian Revolution. Princeton University Press.
Olsaretti, S. (2004). Liberty, Desert and the Market. Cambridge University Press.
Otsuka, M. (2003). Libertarianism without Inequality. Oxford University Press.
Paley, W. (2002). The Principles of Moral and Political Philosophy. Indianapolis,
Liberty Fund.
Parfit, D. (1984). Reasons and Persons. Oxford University Press.
(2000). ‘Equality or Priority?’ in The Ideal of Equality, ed. M. Clayton and
A. Williams. New York, St Martin’s Press, pp. 81–125.
Patten, A. (2002). Hegel’s Idea of Freedom. Oxford University Press.
Pettit, P. (1986). ‘Free Riding and Foul Dealing’. Journal of Philosophy 83: 361–79.
(1990). ‘Virtus Normativa: A Rational Choice Perspective’. Ethics 100: 725–55;
reprinted in P. Pettit (2002) Rules, Reasons, and Norms. Oxford University Press.
(1991). ‘Decision Theory and Folk Psychology’ in Essays in the Foundations of
Decision Theory, ed. M. Bacharach and S. Hurley. Oxford, Blackwell; reprin-
ted in P. Pettit (2002) Rules, Reasons, and Norms. Oxford University Press.
(1993). The Common Mind: An Essay on Psychology, Society and Politics, paperback
edition 1996. New York, Oxford University Press.
(1994). ‘Enfranchising Silence: An Argument for Freedom of Speech’ in Freedom
of Communication, ed. T. Campbell and W. Sadurksi. Aldershot, Dartmouth,
pp. 45–56.
(1996a). ‘Freedom and Antipower’. Ethics 106: 576–604.
(1996b). ‘Institutional Design and Rational Choice’ in The Theory of Institutional
Design, ed. R. E. Goodin. Cambridge University Press.
(1997a). ‘A Consequentialist Perspective on Ethics’ in Three Methods of Ethics:
A Debate. M. Baron, M. Slote and P. Pettit. Oxford, Blackwell.
(1997b). ‘Republican Theory and Criminal Punishment’. Utilitas 9: 59–79.
(1997c). Republicanism: A Theory of Freedom and Government. Oxford University
Press.
(1998). ‘Reworking Sandel’s Republicanism’. Journal of Philosophy 95: 73–96.
(1999). ‘Republican Liberty, Contestatory Democracy’ in Democracy’s Value, ed.
C. Hacker-Cordon and I. Shapiro. Cambridge University Press.
(2000a). ‘Democracy, Electoral and Contestatory’. Nomos 42: 105–44.
(2000b). ‘Minority Claims under Two Conceptions of Democracy’ in Political
Theory and the Rights of Indigenous Peoples, ed. D. Ivison, P. Patton and
W. Sanders. Cambridge University Press, pp. 199–215.
(2001a). ‘Capability and Freedom: A Defence of Sen’, Economics and Philosophy
17: 1–20.
(2001b). ‘The Capacity to Have Done Otherwise’ in Relating to Responsibility:
Essays in Honour of Tony Honore on his 80th Birthday, ed. P. Cane and
J. Gardner. Oxford, Hart, pp. 21–35; reprinted in P. Pettit (2002) Rules,
Reasons, and Norms. Oxford University Press.
(2001c). ‘Deliberative Democracy and the Discursive Dilemma’. Philosophical
Issues (supp. to Nous) 11: 268–99.
(2001d). ‘Non-Consequentialism and Political Philosophy’ in Robert Nozick,
ed. D. Schidmtz. Cambridge University Press.
322 References
(2001e). A Theory of Freedom: From the Psychology to the Politics of Agency.
Cambridge and New York, Polity and Oxford University Press.
(2002a). ‘Is Criminal Justice Politically Feasible?’ Buffalo Criminal Law Review,
Special Issue ed. Pablo de Greiff, 5(2): 427–50.
(2002b). ‘Keeping Republican Freedom Simple: On a Difference with Quentin
Skinner’. Political Theory 30: 339–56.
(2003a). ‘Akrasia, Collective and Individual’ in Weakness of Will and Practical
Irrationality, ed. S. Stroud and C. Tappolet. Oxford University Press.
(2003b). ‘Deliberative Democracy, the Discursive Dilemma, and Republican
Theory’ in Philosophy, Politics and Society Vol. 7: Debating Deliberative
Democracy, ed. J. Fishkin and P. Laslett. Cambridge University Press, pp. 138–62.
(2004a). ‘The Common Good’ in Justice and Democracy: Essays for Brian Barry,
ed. K. Dowding, R. E. Goodin and C. Pateman. Cambridge University Press,
pp. 150–69.
(2004b). ‘Depoliticizing Democracy’. Ratio Juris 17: 52–65.
(2005a). ‘The Domination Complaint’. Nomos 86: 87–117.
(2005b). ‘The Elements of Responsibility’. Philosophical Books 46: 210–19.
(2005c). ‘On Rule-Following, Folk Psychology, and the Economy of Esteem:
Reply to Boghossian, Dreier and Smith’: contribution to Symposium on
P. Pettit’s Rules, Reasons, and Norms. Philosophical Studies 124: 233–59.
(2006a). ‘The Determinacy of Republican Policy’. Philosophy and Public Affairs
34: 275–83.
(2006b). ‘Rawls’s Peoples’ in Rawls’s Law of Peoples: A Realistic Utopia, ed.
R. Martin and D. Reidy. Oxford, Blackwell.
(2007a). ‘Joining the Dots’ in Common Minds: Themes from the Philosophy of
Philip Pettit, ed. H. G. Brennan, R. E. Goodin, F. C. Jackson and M. Smith.
Oxford University Press, pp. 215–344.
(2007b). ‘Rationality, Reasoning and Group Agency’. Dialectica 61: 495–519.
(2007c). ‘Resilience as an Explanandum of Social Theory’ in Contingency, ed.
I. Shapiro and S. Bedi. New York University Press.
(2007d). ‘Responsibility Incorporated’. Ethics 117: 171–201.
(2007e). ‘Free Persons and Free Choices’. History of Political Thought, Special
Issue on ‘Liberty and Sovereignty’, 28: 709–18.
(2008a). ‘The Basic Liberties’ in Essays on H. L. A. Hart, ed. M. Kramer. Oxford
University Press, pp. 201–24.
(2008b). ‘Freedom and Probability: A Comment on Goodin and Jackson’.
Philosophy and Public Affairs 36: 206–20.
(2008c). Made with Words: Hobbes on Language, Mind and Politics. Princeton
University Press.
(2008d). ‘Republican Liberty: Three Axioms, Four Theorems’ in Republicanism
and Political Theory, ed. C. Laborde and J. Manor. Oxford, Blackwell.
(2008e). ‘Value-mistaken and Virtue-mistaken Norms’ in Political Legitimization
without Morality?, ed. J. Kuehnelt. New York, Springer, pp. 139–56.
(2009a). ‘Corporate Responsibility Revisited’. Rechtsfilosofie & Rechtstheorie, Special
Issue on ‘Philip Pettit and the Incorporation of Responsibility’, 38: 159–76.
References 323
(2009b). ‘Varieties of Public Representation’ in Representation and Popular Rule,
ed. I. Shapiro, S. Stokes and E. J. Wood. Cambridge University Press.
(2010a). ‘Legitimate International Institutions: A Neorepublican Perspective’ in
The Philosophy of International Law, ed. J. Tasioulas and S. Besson. Oxford
University Press.
(2010b). ‘Representation, Responsive and Indicative’. Constellations 3: 426–34.
(2010c). ‘A Republican Law of Peoples’. European Journal of Political Theory,
Special Issue on ‘Republicanism and International Relations’, 9: 70–94.
(2011a). ‘The Hedgehog’s Fantasies’: review of Ronald Dworkin, ‘Justice for
Hedgehogs’. Times Literary Supplement. London, News International.
(2011b). ‘The Instability of Freedom as Non-Interference: The Case of Isaiah
Berlin’. Ethics 121: 693–716.
(2012a). ‘The Inescapability of Consequentialism’ in Luck, Value and
Commitment: Themes from the Ethics of Bernard Williams, ed. U. Heuer and
G. Lang. Oxford University Press.
(2012b). ‘Two Republican Traditions’ in Republican Democracy: Liberty, Law
and Politics, ed. A. Niederberger and P. Schink. Edinburgh University Press.
(in press). Just Freedom. New York, W.W. Norton and Co.
Pettit, P. and D. Schweikard (2006). ‘Joint Action and Group Agency’. Philosophy
of the Social Sciences 36: 18–39.
Pettit, P. and M. Smith (1996). ‘Freedom in Belief and Desire’. Journal of
Philosophy 93: 429–49; reprinted in F. Jackson, P. Pettit and M. Smith
(2004) Mind, Morality and Explanation. Oxford University Press.
Pinker, S. (2011). The Better Angels of our Nature: Why Violence has Declined. New
York, Viking Penguin.
Pitkin, H. (1972). The Concept of Representation. Berkeley, University of California
Press.
Pitkin, H. F., ed. (1969). Representation. New York, Atherton Press.
Pocock, J. (1975). The Machiavellian Moment: Florentine Political Theory and the
Atlantic Republican Tradition. Princeton University Press.
Pogge, T. (1990). Realizing Rawls. Ithaca, NY, Cornell University Press.
Polybius (1954). The Histories. Cambridge, MA, Harvard University Press.
Prentice, D. A. and D. T. Miller (1993). ‘Pluralistic Ignorance and Alcohol Use on
Campus’. Journal of Personality and Social Psychology 64: 243–56.
Price, R. (1991). Political Writings. Cambridge University Press.
Priestley, J. (1993). Political Writings. Cambridge University Press.
Przeworksi, A. (1999). ‘A Minimalist Conception of Democracy: A Defense’ in
Democracy’s Value, ed. C. Hacker-Cordon and I. Shapiro. Cambridge
University Press.
Quiggin, J. (2011). Zombie Economics: How Dead Ideas Still Walk Among Us.
Raab, F. (1965). The English Face of Machiavelli: A Changing Interpretation
1500–1700. London, Routledge.
Raventos, D. (2007). Basic Income: The Material Conditions of Freedom. London,
Pluto Press.
324 References
Rawls, J. (1955). ‘Two Concepts of Rules’. Philosophical Review 64: 3–32.
(1971). A Theory of Justice. Oxford University Press.
(1993). Political Liberalism. New York, Columbia University Press.
(1995). ‘Political Liberalism: Reply to Habermas’. Journal of Philosophy 92:
132–80.
(1999). The Law of Peoples. Cambridge, MA, Harvard University Press.
(2001). Justice as Fairness: A Restatement. Cambridge, MA, Harvard University
Press.
Raz, J. (1986). The Morality of Freedom. Oxford University Press.
Rehfield, A. (2005). The Concept of Constituency: Political Representation,
Democratic Legitimacy, and Institutional Design. Cambridge University Press.
(2006). ‘Towards a General Theory of Political Representation’. Journal of
Politics 68: 1–21.
Reid, J. P. (1988). The Concept of Liberty in the Age of the American Revolution.
Chicago University Press.
Reidy, D. (2007). ‘Reciprocity and Reasonable Disagreement: From Liberal to
Democratic Legitimacy’. Philosophical Studies 132: 243–91.
Richardson, H. (2002). Democratic Autonomy. New York, Oxford University Press.
Richter, M. (1977). The Political Theory of Montesquieu. Cambridge University Press.
Riker, W. (1982). Liberalism against Populism. San Francisco, W. H. Freeman and Co.
Ripstein, A. (2009). Force and Freedom: Kant’s Legal and Political Philosophy.
Risse, M. (2004). ‘Arguing for Majority Rule’. Journal of Political Philosophy 12:
41–64.
Robbins, C. (1959). The Eighteenth Century Commonwealthman. Cambridge, MA,
Harvard University Press.
Robert, H. M., H. M. Robert, III, D. H. Honemann, T. J. Balch, D. E. Seabold and
S. Gerber (2011). Robert’s Rules of Order Newly Revised, 11th edn. Cambridge,
MA, Da Capo Press.
Roemer, J. (1998). Equality of Opportunity. Cambridge, MA, Harvard University
Press.
Rosanvallon, P. (2006). Democracy, Past and Future. New York, Columbia
University Press.
Rousseau, J. J. (1997). Rousseau: ‘The Social Contract’ and Other Later Political
Writings, trans. Victor Gourevitch. Cambridge University Press.
Rubenfeld, J. (2001). Freedom and Time: A Theory of Constitutional Self-
Government. New Haven, CT, Yale University Press.
Ryan, M. (1999). ‘Bartolus of Sassoferrato and Free Cities’. Transactions of the Royal
Historical Society 6: 65–89.
Ryle, G. (1949). The Concept of Mind. University of Chicago Press.
Sandel, M. (1996). Democracy’s Discontent: America in Search of a Public Philosophy.
Sanyal, S. (forthcoming). ‘A Defence of Democratic Egalitarianism’.
Scanlon, T. M. (1998). What We Owe to Each Other. Cambridge, MA, Harvard
University Press.
References 325
Scheffler, S. (2005). ‘Choice, Circumstance and the Value of Equality’, Politics,
Philosophy and Economics 4: 5–28.
Schmidtz, D. and J. Brennan (2010). A Brief History of Liberty. Oxford, Wiley-
Blackwell.
Schmitt, C. (2005). Political Theology: Four Chapters on the Concept of Sovereignty.
Chicago University Press.
Schumpeter, J. A. (1984). Capitalism, Socialism and Democracy. New York, Harper
Torchbooks.
Schwartzberg, M. (2007). Democracy and Legal Change. Cambridge University
Press.
Searle, J. (1995). The Construction of Social Reality. New York, Free Press.
Sellers, M. N. S. (1995). American Republicanism: Roman Ideology in the United
States Constitution. New York University Press.
Sen, A. (1970). ‘The Impossibility of a Paretian Liberal’. Journal of Political
Economy 78: 152–7.
(1983a). ‘Liberty and Social Choice’. Journal of Philosophy 80: 18–20.
(1983b). ‘Poor, Relatively Speaking’. Oxford Economic Papers 35: 153–68.
(1985). Commodities and Capabilities. Amsterdam, North-Holland.
(2002). Rationality and Freedom. Cambridge, MA, Harvard University Press.
(2009). The Idea of Justice. Cambridge, MA, Harvard University Press.
Shapiro, I. (2003). The State of Democratic Theory. Princeton University Press.
Shapiro, S. (2011). Legality. Cambridge, MA, Harvard University Press.
Sharp, A., ed. (1998). The English Levellers. Cambridge University Press.
Shiffrin, S. (2000). ‘Paternalism, Unconscionability Doctrine, and
Accommodation’. Philosophy and Public Affairs 29: 205–50.
Sidney, A. (1990). Discourses Concerning Government. Indianapolis, Liberty Classics.
Sieyes, E. J. (2003). Political Writings. Indianapolis, Hackett.
Simmons, A. J. (1976). ‘Tacit Consent and Political Obligation’. Philosophy and
Public Affairs 5(3): 274–91.
(1979). Moral Principles and Political Obligations. Princeton University Press.
(1999). ‘Justification and Legitimacy’. Ethics 109: 739–71.
Sintomer, Y. (2007). Le Pouvoir au Peuple: Jurys citoyens, tirage au sort et democratie
participative. Paris, Edition la Decouverte.
Skinner, Q. (1974). ‘The Principles and Practice of Opposition: The Case of
Bolingbroke versus Walpole’ in Historical Perspectives: Studies in English
Thought and Society in Honour of J.H. Plumb, ed. N. McKendrick. London,
Europa Publications.
(1978). The Foundations of Modern Political Thought. Cambridge University Press.
(1996). Reason and Rhetoric in the Philosophy of Hobbes. Cambridge University
Press.
(1998). Liberty Before Liberalism. Cambridge University Press.
(2005). ‘Hobbes on Representation’. European Journal of Philosophy 13: 155–84.
(2008a). ‘Freedom as the Absence of Arbitrary Power’ in Republicanism and
Political Theory, ed. J. Maynor and C. Laborde. Oxford, Blackwell.
(2008b). Hobbes and Republican Liberty. Cambridge University Press.
326 References
(2009). ‘A Genealogy of the Modern State’. Proceedings of the British Academy
162: 325–70.
Slaughter, S. (2005). Liberty beyond Neo-Liberalism: A Republican Critique of
Liberal Government in a Globalising Age. London, Macmillan Palgrave.
Smith, A. (1976). An Inquiry into the Nature and Causes of the Wealth of Nations.
(1982). The Theory of the Moral Sentiments. Indianapolis, Liberty Classics.
Smith, M. (1994). The Moral Problem. Oxford, Blackwell.
Sober, E. and D. S. Wilson (1998). Unto Others: The Evolution and Psychology of
Unselfish Behavior. Cambridge, MA, Harvard University Press.
Sosa, E. (2007). A Virtue Epistemology. Oxford University Press.
Spitz, J.-F. (1995). La Liberte Politique. Paris, Presses Universitaires de France.
Steiner, H. (1994). An Essay on Rights. Oxford, Blackwell.
Stewart, A. T. Q. (1993). A Deeper Silence: The Hidden Roots of the United Irish
Movement. London, Faber.
Stilz, A. (2009). Liberal Loyalty: Freedom, Obligation, and the State. Princeton
University Press.
Stout, J. (2004). Democracy and Tradition. Princeton University Press.
Strauss, L. (2000). On Tyranny. Chicago University Press.
Strawson, P. (1962). Freedom and Resentment and Other Essays. London, Methuen.
Sugden, R. (1998). ‘The Metric of Opportunity’. Economics and Philosophy 14:
307–37.
Sunstein, C. R. (2009). Republic.com 2.0. Princeton University Press.
Talisse, R. B. (2007). A Pragmatist Philosophy of Democracy. London, Routledge.
(2009). Democracy and Moral Conflict. Princeton University Press.
Tan, K.-C. (2008). ‘A Defense of Luck Egalitarianism’. Journal of Philosophy 105:
665–90.
Taylor, C. (1985a). Philosophy and the Human Sciences. Cambridge University Press.
(1985b). Philosophy and the Human Sciences: Philosophical Papers 2. Cambridge
University Press.
Temkin, L. (1996). Inequality. Oxford University Press.
Thaler, R. and C. Sunstein (2008). Nudge: Improving Decisions about Health,
Wealth and Happiness. London, Penguin Books.
Tilly, C. (1975). ‘Reflections on the History of European State-Making’ in The
Formation of National States in Western Europe, ed. C. Tilly. Princeton
University Press.
(2007). Democracy. Cambridge University Press.
Tomasi, J. (2012). Free Market Fairness. Princeton University Press.
Trenchard, J. and T. Gordon (1971). Cato’s Letters. New York, Da Capo.
Tuck, R. (2006). ‘Hobbes and Democracy’ in Rethinking the Foundations of
Modern Political Thought, ed. A. S. Brett and J. Tully. Cambridge University
Press.
(2008). Free Riding. Cambridge, MA, Harvard University Press.
Tully, J. (2009). Public Philosophy in a New Key. Cambridge University Press, vols.
i and ii.
References 327
Tuomela, R. (1995). The Importance of Us. Stanford University Press.
Tyler, T. R. (1990). Why People Obey the Law. New Haven, Yale University Press.
Urbinati, N. (2006). Representative Democracy: Principles and Genealogy. University
of Chicago Press.
Vallentyne, P. and H. Steiner, eds. (2000a). Left-Libertarianism and its Critics. New
York, Palgrave.
(2000b). The Origins of Left-Libertarianism. New York, Palgrave.
Van der Rijt, J.-W. (2012). The Importance of Assent: A Theory of Coercion and
Dignity. New York, Springer.
Van Gelderen, M. and Q. Skinner (2002). Republicanism: A Shared European
Heritage, 2 vols., Cambridge University Press.
Van Parijs, P. (1995). Real Freedom for All. Oxford University Press.
Vatter, M. (2011). ‘Natural Right and States of Exception in Leo Strauss’ in
Crediting God: Sovereignty and Religion in the Age of Global Capitalism, ed.
M. Vatter. New York, Fordham University Press.
Vieira, M. B. and D. Runciman (2008). Representation. Cambridge, Polity Press.
Vile, M. J. C. (1967). Constitutionalism and the Separation of Powers. Oxford
University Press.
Viroli, M. (1995). For Love of Country. Oxford University Press.
(2002). Republicanism. New York, Hill and Wang.
Waldron, J. (1999a). The Dignity of Legislation. Cambridge University Press.
(1999b). Law and Disagreement. Oxford University Press.
(2006). ‘The Core of the Case Against Judicial Review’. Yale Law Journal 115:
1346–1406.
(2007). ‘Pettit’s Molecule’ in Common Minds: Themes from the Philosophy of
Philip Pettit, ed. G. Brennan, R. E. Goodin, F. Jackson and M. Smith. Oxford
University Press, pp. 143–60.
Waley, D. (1988). The Italian City-Republics, 3rd edn. London, Longman.
Walzer, M. (1981). ‘Philosophy and Democracy’. Political Theory 9: 379–99.
(1983). Spheres of Justice. Oxford, Martin Robertson.
Warren, M. E. (2011). ‘Voting with Your Feet: Exit-based Empowerment in
Democratic Theory’. American Political Science Review 105: 683–701.
Warren, M. E. and H. Pearse, eds. (2008). Designing Deliberative Democracy.
Watson, A. (1985). The Digest of Justinian, Four Volumes. Philadelphia, University
of Pennsylvania Press.
Watson, G. (2003). ‘Free Agency’ in Free Will, 2nd edn, ed. G. Watson. Oxford
University Press.
(2005). Agency and Answerability: Selected Essays. Oxford University Press.
Weber, M. (1947). The Theory of Social and Economic Organization. London,
William Hodge.
Weingast, B. (1997). ‘The Political Foundations of Democracy and the Rule of
Law’. American Political Science Review 91: 245–63.
Weinstock, D. and C. Nadeau, eds. (2004). Republicanism: History, Theory and
Practice. London, Frank Cass.
328 References
White, S. and D. Leighton, eds. (2008). Building a Citizen Society: The Emerging
Politics of Republican Democracy. London, Lawrence and Wishart.
Whitehead, A. N. (1997). Science and the Modern World. New York, Simon and
Schuster.
Williams, B. (2005). In the Beginning was the Deed: Realism and Moralism in
Political Argument. Princeton University Press.
Williamson, T. (2000). Knowledge and its Limits. Oxford University Press.
Winch, D. (1978). Adam Smith’s Politics: An Essay in Historiographic Revision.
Winch, P. (1963). The Idea of a Social Science and Its Relation to Philosophy. London,
Routledge.
Wirszubski, C. (1968). Libertas as a Political Ideal at Rome. Oxford University Press.
Wolf, S. (2010). Meaning in Life and Why it Matters. Princeton University Press.
Woolf, C. N. S. (1913). Bartolus of Sassoferrato. Cambridge University Press.
Young, I. (1990). Justice and the Politics of Difference. Princeton University Press.
(2000). Inclusion and Democracy. Oxford University Press.
Zucca, L. (2007). Constitutional Dilemmas: Conflicts of Fundamental Legal Rights in
Europe and the USA. Oxford University Press.
Name index
Abizadeh, A. 162 Brennan, Geoffrey xii, 12, 15, 20, 117, 121, 128, 169,
Abramson, J. 197 182, 200, 236, 247
Ackerman, B. 285 Brett, A. S. xii, 287
Adams, John 134 Brettschneider, C. 265
Agamben, Georgio 173 Brugger, W. 3
Alexander, J. M. 79 Buchak, Lara 32
Alipuro, R. 252 Buchanan, A. 146, 148, 151
Amar, Akhil 292 Burke, Edmund 206
Anderson, E. 91, 93 Burnheim, J. 204
Anscombe, G. E. M. 53
Appiah, K. A. xii, 128 Cabot, V. 222
Arendt, Hannah 12, 18, 289 Canning, J. P. 288
Aristotle 6, 189–90, 291 Carroll, Lewis 278
Arneson, R. 79 Carter, Ian xii, 44, 46, 51
Arrow, Kenneth 194 Charlesworth, H. 222
Atiyah, P. S. 151 Christiano, Thomas 145
Christman, J. 48
Bailyn, B. 6–7 Churchill, Winston 242
Bakan, J. 116, 235 Cicero, M. T. 6, 88, 221
Baldwin, T. 48 Clark, A. 115, 225
Barry, Brian 143 Cohen, G. A. 20, 79, 96, 183–4
Bartels, L. M. 210 Cohen, Joshua 13, 14, 15, 23–5, 79, 123, 126,
Bartolus of Sassoferrato 287–8 144–5, 267
Beitz, Charles xii, 143, 146, 214, 262 Cohen, M. 111
Bell, D. 148, 151 Coleman, J. 84, 115
Bellamy, Richard 3, 207, 216 Connolly, W. 180, 251
Benhabib, S. 215 Constant, Benjamin 16, 200, 232–3
Bentham, Jeremy 8, 9, 10, 123, 149, Cook, Maeve 44
249, 293 Cornish, W. R. 115
Berlin, Isaiah 9, 17, 22, 29, 30–5, 39, 41–3, 48, Costa, M. V. 58
64–7, 149, 150, 294, 295
Berntson, Daniel 42 Dagger, Richard 11
Besold, C. 222 Demosthenes 135
Besson, S. 3 Dennett, D. C. 203, 225
Bodin, Jean 12–13, 14–15, 189–90, 220–1, 223–5, Dietrich, F. 106, 194
290, 293–4, 303 Dowlen, O. 12, 196
Bohman, J. 3, 128 Dryzek, John 145
Bolingbroke, Viscount 6–7 Duff, R. A. 118
Braithwaite, John xii, 3, 117, 119, 121, 222 Dunn, J. 180
Bramhall, Bishop J. 29, 142 Dworkin, Ronald 11, 20, 48, 78, 79, 148,
Bratman, M. 48, 277 180, 186
329
330 Name index
Eisgruber, C. L. 265 Hayward, Clarissa 44, 63, 126
Elazar, Yiftah 16 Hegel, G. W. F. 126, 222, 224, 290
Elster, Jon 15, 84, 253, 267, 278 Held, David 181
Ely, John Hart 201, 217 Herreros, F. 244
Eskridge, William 257 Hill, L. 210
Espejo, P. O. 238, 289 Hobbes, Thomas 9, 12–13, 14–15, 28–31, 33–5,
Estlund, David 144–5, 148 41–3, 61, 64–7, 141, 142, 151, 189–90, 220–1,
Ewing, Ben 120, 174 223–5, 228, 229, 289, 293–4, 295, 303, 305
Hoekstra, Kinch xii, 221, 189
Faini, Matteo 196 Holmes, Stephen 23–5, 201
Ferejohn, John xii, 274 Honig, B. 227, 274
Ferguson, Adam 173–4, 227 Honneth, A. 184
Fisher, Richard xii Honohan, I. 3
Fishkin, James 197 Hont, Istvan xii, 286
Fleurbaey, Marc xii, 79 Hume, David 151, 246–7
Forst, Rainer xii, 143 Husak, D. 119
Fowler, A. 210
Frankfurt, H. 29, 48, 51, 81 Ivison, Duncan xii
Franklin, J. 222
Fukuyama, F. 6, 135, 164 Jackson, Frank xii, 34, 35, 164
Fuller, L. L. 221 James, Susan xii, 48
Jay, J. 6–7, 76, 211, 244
Gadeke, Dorothy 62 Jennings, J. 3
Garber, Dan xii Joshi, Hrishikesh 53
Gardner, John xii Joyce, J. M. 34
Garsten, B. 259
Gatens, Moira xii Kahneman, D. 246
Gaus, G. 48, 255, 261 Kalyvas, A. 8, 289, 292
Gauthier, David 145 Kant, Immanuel 12, 17, 44, 134, 142–3, 144, 164,
Geuss, R. 59 191, 223–4, 229, 302
Gilbert, M. 138, 277 Katznelson, I. 8
Gilovich, T. 246 Kaufman, A. 81
Goldberg, J. 111 Kelly, E. 121
Goldman, A. 169 Kelsen, H. 287
Goldsworthy, J. 278 Keohane, Nan xii, 12
Goodin, R. E. xii, 34, 35, 245 Ketcham, R. 6–7, 199
Gordon, T. 7–8 Kharkhordin, O. 252
Green, J. E. 227 Knights, M. 270
Griffin, D. 246 Kolodny, Niko xi, 23–5, 204, 211
Guerrero, A. 169, 211 Kornhauser, Lewis 192
Gutmann, Amy 144–5 Kramer, Matthew xii, 44, 46, 51
Gwyn, W. B. 221 Krygier, Martin xii
Kukathas, Chandran xii, 135
Habermas, Juergen 12, 15–16, 23–5, 144–5, 201, Kymlicka, Will 178
253, 267
Hamilton, A. 6–7, 76, 211, 244 Laborde, Cecile xii, 3, 178
Hamlin, A. 247 Lacey, Niki xii
Hand, Learned 101 Lane, Melissa xii
Hansen, M. H. 12, 189, 196, 222 Langton, R. 202
Hardt, M. 289 Languet, H. 150
Harman, G. 194 Larmore, C. 20, 254
Harman, Liz xii Leighton, D. 3
Harrington, James 6–7, 17, 134, 228 Lessig, L. 235
Hart, Herbert 84, 96, 100, 221 Letsas, George xii
Hayek, F. A. xii, 134 Levin, M. 115
Name index 331
Lewis, D. 255 Nozick, Robert 11, 79, 145
Libourne, J. 8, 83 Nussbaum, M. 8, 79, 81, 83, 104–5, 112, 126
Lieberman, D. 222
Lind, John 150–8 Oastler, Richard 272
Lindahl, Hans 289, 292 Ober, Josh xii, 188–9, 196, 222
Lipset, S. M. 227 O’Donnell, G. 221
List, Christian xii, 32, 49, 67, 76, 106, 116, 120, 133, Olsaretti, S. 29, 40, 54, 72, 148, 151, 166
168, 192, 193, 194, 195, 198, 206, 223, 277, 283, 289 O’Neill, M. 91
Lively, J. 191, 249, 251 Otsuka, Michael xii, 29, 158
Livy 6
Locke, John 93, 141, 173, 204, 292 Paley, William 8, 9, 10, 11, 22, 123, 150, 188–94
Lomasky, L. 12, 200 Parfit, Derek 81, 89, 91, 100, 184
Long, D. C. 9 Patten, Alan xii, 184
Lovett, Frank xii, 3, 7, 18, 58, 90, 116, 180 Patton, Paul xii
Luhmann, Niklas 193, 284 Pearse, H. 197
Pettit, P. 3, 8, 9, 11, 12, 15, 17–20, 26, 27, 28–9, 30, 32,
MacCallum, G. C. 48 34, 38, 41, 42, 48, 49, 51, 52, 58, 59, 60, 64, 67, 74,
McCormick, J. P. 6, 213, 217, 222, 227, 231 76, 84, 90, 93, 100, 116, 117, 119, 120, 121, 124, 127,
MacDonagh, Oliver 270–2, 274, 308 128, 129, 133, 140, 150, 157, 161, 168, 182, 189, 192,
Macedo, Stephen xii, 255 193, 194, 195, 198, 202, 204, 206, 218, 221, 223, 231,
McGeer, Victoria xii, 246, 259 234, 236, 245, 246, 247, 259, 262, 277, 281, 284,
McGilvray, Eric 3, 45, 111 287, 289
Machiavelli, Nicolo 6, 84, 174, 221, 222, 227 Pitkin, Hannah 198, 199, 207, 244
Mackie, G. 194 Plato 6
McLean, J. 133 Plunkett, David xi
McMahon, C. 58 Pocock, John 3, 8
Madison, J. 6–7, 76, 211, 244 Pogge, Thomas 143
Mandeville, Bernard 246–7 Polak, B. 193
Manin, B. 195 Polybius 6, 12, 188–9, 221
Mansbridge, J. xii, 195, 195 Prentice, D. A. 84
Margalit, A. 216 Price, Richard 6–8, 16, 169, 173, 218, 247
Markell, P. 127 Priestley, Joseph 6–7, 204
Markovits, D. 138 Przeworksi, A. 241
Marti, Jose xii, 3, 124, 234
Maynor, J. 3 Quiggin, J. 112
Mercier, H. 254
Michelman, F. I. 147 Raab, F. 6–7
Mill, James 191, 249–51, 307 Rakove, J. 219
Mill, John Stuart 98, 119, 135, 199, 210 Raventos, D. 112
Millar, F. 222 Rawls, John 11, 20, 41–3, 76, 78, 79–80, 82, 90, 94,
Miller, D. 39, 49, 84 107–10, 122, 123, 124, 125, 143, 144–5, 146, 150–8,
Milton, John 6–8, 158 184–6, 254, 287, 294, 299
Mirabeau, Honoré 199 Raz, Joseph xii, 78, 136, 137
Montesquieu, Baron de 6–7, 221, 226, 243 Rees, J. 191, 249
Morris, I. 96 Rehfield, A. 195, 201
Mueller, J. W. xii, 140, 238 Reid, J. P. 6–7, 168–9
Murphy, L. 99 Reidy, D. 143
Richardson, H. 3, 58, 128
Nadeau, C. 3 Richter, M. 222
Nagel, Thomas 99, 143 Riker, William 23, 194, 294
Nathan, G. 178 Ripstein, A. 168, 229
Nebel, Jake 184 Risse, M. 214
Negri, A. 289 Robbins, C. 6–7
Nelson, Eric 6 Roemer, J. 79
Niederberger, A. 3 Rosanvallon, Pierre 235
332 Name index
Rosen, Gideon xii Sugden, R. 46
Rousseau, Jean-Jacques 5, 12–18, 130, 141, 142, Sunstein, C. R. 56, 234, 262
147, 189, 191, 223–5, 228, 229, 243, 289, 293–4, 305
Rubenfeld, J. 291 Talisse, R. B. 148, 178
Runciman, D. 195 Tan, K. C. 126
Rutten, A. R. 219 Tasioulas, John xii
Ryan, Magnus xii, 288 Taylor, Charles 45, 146
Ryle, Gilbert 225 Temkin, L. xii, 81
Thaler, R. 56
Sadurski, Wojciech xii Thompson, Dennis 144–5
Sager, Larry 192 Tilly, C. 164, 242
Sandel, Michael 12 Tomasi, J. 111
Sanyal, S. 126, 132 Trenchard, J. 7–8
Scanlon, T. M. xii, 143 Tuck, R. xii, 169, 189
Scheffler, S. 91, 114 Tully, J. 23–5, 75, 217, 238
Scheppele, Kim xii Tuomela, R. 277
Schink, P. 3 Tyler, T. R. 84, 118
Schmidtz, D. 20
Schmitt, Carl 173 Urbinati, Nadia 195, 290
Schumpeter, Joseph 22–3, 201, 241–2, 243, 294
Valentini, Laura xii
Schwartzberg, M. 23–5, 217
Vallentyne, P. 79
Schweikard, D. 277
Van Gelderen, M. 3
Scott, Sir Walter 271
Van Parijs, P. 42
Searle, J. 277
Vatter, M. xii, 131
Sellers, M. N. S. 6–7
Vieira, M. B. 195
Sen, Amartya xii, 8, 69, 79, 81, 83, 87, 104–5, 106,
Vile, M. J. C. 221
112, 124, 126, 145, 155, 158, 171, 248
Viroli, M. 3, 228
Shapiro, Ian xii, 201
Shapiro, S. 84, 292 Waldron, J. xii, 23–5, 34, 79, 145, 213, 214
Sharp, A. 81, 83 Waley, D. 196
Shiffrin, S. 59, 98 Walzer, Michael 23–5, 79
Sidney, Algernon 6–8, 204 Warren, M. E. 168, 197
Sieyes, E. J. 285, 286 Watson, A. 131
Simmons, A. J. 141, 144, 146 Watson, G. 48
Singer, Peter xii Weale, Albert xii, 30
Sintomer, Y. 197 Weber, Max 145–6
Skinner, Quentin xii, 3, 5, 6–7, 9, 17–20, 26, 28–9, Weingast, Barry xii, 219
58, 195, 254, 259, 287 Weinstock, D. 3
Slaughter, S. 3 White, S. 3
Smith, Adam 87, 104–5, 128, 249 Whitehead, A. N. 224–5
Smith, Melanchton 199 Williams, B. 145, 148
Smith, Michael xii, 27, 48, 59, 60, 281 Williamson, T. 32
Soares, A. 222 Wilson, D. S. 84
Sober, E. 84 Winch, P. 84, 87
Sosa, E. 198 Wirszubski, C. 2
Soskice, David xii Wolf, S. 102
Southwood, N. xii, 171 Wolff, Jonathan xii, 30
Sperber, D. 254 Wollheim, Richard 30
Spitz, J. F. 16 Woolf, C. N. S. 189, 190, 288
Steiner, H. 44, 51, 79, 95
Stewart, A. T. Q. 219 Yong, Caleb 136, 214
Stilz, A. xi, 94, 143 Young, I. xii, 227, 257
Stout, J. xii, 227
Strauss, Leo 131 Zucca, L. 106
Strawson, Peter 27, 43, 60 Zuehl, Jake 267
Subject index
absolutism 12–13 political obligation of 138

acceptability game 252–64, 307–8 role of 14–15
acceptance game 252–9 civic justifiability 143, 146
adaptation 31, 64–7 civic vigilance 226–7
admission-cost regulation 117–18 coercion
appeal 216 consent to 151
arbitrary interference 58–9 and freedom 148, 149, 301
assemblies and law enforcement 160
elected 197, 303–4 necessity for 163, 165, 183–4, 301–2
indicative assemblies 195–7, 198, 203–5 and social justice 136
indicative-cum-responsive assemblies 199–200 collective exercisability 94–7
plenary 188–94, 303 collective satisfaction 99
representative 195–208, 303–4 colonial rule 281
of representatives 205–7 committee-of-the-whole as sovereign 13
responsive 303–4 communal disapproval 84
autonomy 48 communal norms 134, 258
communitarian opposition 11–18
basic liberties, see liberty community competition 164
benevolent despotism 10, 60, 185–6 consequentionalism 124
British Columbia Citizens’ Assembly 197, 200, constituting people, in dual-aspect democracy
201, 205, 232–3, 235 285–8
brute luck 79 constitution, mixed 5–6, 12–13, 220–5, 305
constraints
capability theory 79 adaptation to 31
Capitalism, Socialism and Democracy 241–2 breach by sovereign 220–1
choice 26–7, 69–74, 92 constitutional 5, 221
and benefits to all 100 deliberative 15–16
co-exercisable 93, 94–7, 105, 106, 299–300 egalitarian 78
co-satisfying 93, 94–7, 98–101, 299–300 equality 79
counter-productive 100–1 on interference 63
distality of 102–3 of law 216, 223
and domination 161–2 of rationality by state 282–3
entrenchment of 102–3 under shared control 175–6
and external costs 100–1 consultation 216
facilitating 69–74 consumer behaviour 248–9
institutional environments for 111 contestability, individualized 213–14,
and meaningful life 101–4 215–18, 304
over-empowering 99 contestation 14–15, 215, 220, 225–9, 260
see also constraints; freedom contestatory vigilance 227–8
citizens contract, freedom of 150
contestatory role of 14–15, 220, 225–9 contractualism 144
333
334 Subject index
control mixed constitution in 283–4
and consent 157–60, 301 role of the people 285–8, 292
idea of 153–6 role of the state 282–5
and influence, see influence see also democracy, Washington system of;
intentional 243–7 democracy, Westminster system of
inter-agential 171
nature of 153–60 egalitarianism, expressive 78–9, 89
non-intentional 248–51 elected politicians, see politicians
popular, see popular control electoral commissions, independent 232–3
unintentional 248–51 electoral influence 209–11, 237, 305–6
courts, in democracy 285 electoral institutions 207–8
criminalization 117–22, 299 enfranchisement of citizens 168–9
English republic 6–7
decision-making, collective 258 equality
deliberative regulation 264–9 commitment to 255
democracy of influence 262
Athenian model 188–9 material 125, 126
Churchill on 242 and social justice 77–92
and constitutionalism 23 in society 2–3, 169
contestatory 227–8, 238 equalization, expressive 79, 89–90
deliberative 267–8, 269–74 An Essay on Government 249–51
dual-aspect model of 252–79, 308–10 exercisability, individual 94–7
electoral 238 eyeball test 84–5, 86–7, 98, 105, 124, 176
emerging view of 21–4
and freedom 22 fallacy of misplaced concreteness 224–5
and justice 24–5 federalist/anti-federalist debate 6–7
model of 4 Florence 6
non-directive influence 22–3 free choice, resources required 36–7
and non-interference 150 free-man 17–18
Schumpeter on 241–2, 306 free-person heuristic 82–7, 93
slow 270–2 freedom
theory of 4, 179–81, 302 in accord with common discourse 9
Washington system of 206–7, 246, 284–5, 304 of association 201–2
Westminster system of 206–7, 246, 284–5, 304 basic 92–107, 201–2, 297–8, 299
democratic control, see control identifying 106
democratic state 179–86 variations in 104–7
appeal of 181–4 of choice 69–74, 83, 103, 295–6
dependent sovereignty of constituted people co-exercisable 93, 94–7, 105, 106, 299–300
289–90 co-satisfying 93, 94–7, 98–101, 299–300
derived returns, equal 80–1 of contract 150
despotism, of custom 135 coordination rules 97
direct returns, equal 80–1 criteria for 92–4, 173–4
Discourses on Livy 6 culturally variable rules 105, 106
discrimination 78 dependency on goodwill 169
domination and exercisability 94–7
by corporate bodies 115–17 and general protection insulation programmes
by employers 115 117–22
and interference 50, 165, 296 hindrance probability 33–5, 295
and invasion 28, 50, 295–6 individual satisfaction 98
necessity for 163, 165 institutional environments for 111
protection against 109–10 and insurance programmes 112–14
types of 63 and law 93
see also control as legal liberties 42
dual process 23 and material environment 110–12
dual-aspect model of democracy 252–79, 280–1 measurement of 44–7
Subject index 335
minority groups 211–13 incorporated people 289
and moral personality 94 independence of the people 218–19
as non-domination 4, 5, 6, 7–8, 9, 18, 22, 25, indeterminacy 282
67–9, 82–7, 88–92, 123, 126–7, 152–3, 166, individual rights 217
183, 293, 294, 296, 297, 301 individual satisfaction 98
factors reducing 27–8 individualized contestability 213–14, 215–18, 304
hindrances to, see hindrances industrialization, of nineteenth-century England
summary 294–7 270–2
as non-frustration 64–9, 296 influence
as non-interference 9–10, 64–9, 123, 149–52, accessibility to 187, 303, 304
293, 301 campaign supporters 234–5
opportunity-freedom 46–7 directed 167, 187, 240–52
physical restrictions on 96 electoral 209–11, 229–31
and political legitimacy 146–9 equally accessible 187, 213–15, 262
procedural 185 failure of 203
and public entrenchment 86–8, 93 individualized 209–18, 262
rule-dependent 97 media 234, 235
and self-determination 16 minority groups 211–13
and special protection insulation programmes plenary assembly 188–94
114–17 popular 168–70, 229–31, 242, 260, 280, 301, 306
of speech 201–2 private lobbies 231, 233–5
and status 83–4, 90–2, 181, 298 reserve 230–1
three theories of 64–7 unconditioned system of 218–29
to interfere 71 unelected authorities 231, 235–8
of travel 201–2 varieties of 156–7
see also choice; liberty virtual 230
freedom-related words 20–1 without control 240–3
French Constituent Assembly 199 see also control
ingratiation 64–7, 72
general protection insulation programmes 117–22 insulation programmes
generic hindrances 37–40 general protection 117–22
global sovereignty 18–19 special protection 114–17
government insurance programmes 112–14
intentional direction of 243–7 intentional-control model 275–7, 306–7
non-intentional direction of 248–51 intentional-direction model 247
oversight of 226 interest-group pluralism 250
and public interest promotion 249–51 interference
Great Reform Bill (1832) 270 arbitrary 58–9
with choice 49, 69–74, 165
Hand test 101 controlled 153
harmful choice 100–1 and eyeball test 84–5, 86–7, 98, 105, 124, 176
hindrances impacts of 229
generic 37–40 and legitimacy 149–52
invasive 35–49, 69, 295 modes of 56
probability of 33–5, 295 moral constraints on 63
resources for free choice 36–7 nature of 50–6
specific 37–40 non-arbitrary 152
to freedom of choice 27–35 paternalistic 58–9
variations in 35 and subjection 56–9
vitiating 35–49, 295 uncontrolled 58–9
will-imposed 44 intimidation 60–4
humanitarianism 271–2, 273 intolerability, of employment conditions 272
invasion, and domination 28, 50, 295–6
impartiality 13–14, 215 invasive hindrances 35–49, 69, 295
impersonal norm 134 investments, equal 80–1
336 Subject index
invigilation 60–4 non-domination 4, 5, 6, 7–8, 9, 18, 22, 25, 67–9,
invisible-hand mechanism 248–9, 250, 307 82–7, 88–92, 123, 126–7, 152–3, 166, 183,
Italian–Atlantic republicanism 5, 8, 12, 14, 15, 293, 294, 296, 297, 301
16–18, 218–19, 220, 293–4 non-intentional model 306, 307
Italy non-interference 9–10, 64–9, 123, 149–52, 293,
medieval 2–3, 6 301
Renaissance 2–3, 6 norms 84, 128, 134, 135, 255–6, 257, 258, 261–9,
275–7
jointly intentional action 277 policy-relevant 261–9, 276, 307–8
justice
and coercion 136 ombudsmen 216
criminalization 117–22 opportunity-cost regulation 117–18
implementation by state 133–4, 299–300 opportunity-freedom 46–7
infrastructural programmes 110–12 option-luck 79
and material equality 125, 126 options
model of 107–22 and distality of choice 102–3
and non-interference 124 removal of 50–6
procedural provisions 76 orthonomy 27, 48, 281–2
Rawls’s theory of 107–10 over-empowering choices 99
and social order 76
see also republican justice; social justice participatory engagement 227
paternalistic interference 58–9
kingdom of ends 181–2, 183 patriotism 140, 261–2
knave’s principle 246–7 people, in dual-aspect democracy 285–8, 289
philanthropy 113
laissez-faire doctrine 271–2 plenary assembly 188–94, 303
law policy-condition 265
coercive 84, see also coercion political campaign supporters 234–5
and liberties 106 political corruption 247
legitimacy political legitimacy 3, 4, 18–19, 20, 76, 77, 294
alternative uses of terminology 145–6 acceptance of regime 137, 300
and civic justifiability 146 and civil disobedience 138
and control 177, 301 content-independent conformity to law 139
international 146 as distinct from social justice 130–1
political, see political legitimacy eclipsed in later political thought 142–5
levelling-down 88, 298 and freedom 146–9
Leviathan 142 in freedom-denying regime 148
liberalism 10–11 illegitimate institutions 139
liberty, see freedom illegitimate laws 139
lobbies, private 231, 233–5 imposition of rules 141
lottery-based system, of assembly membership just illegitimacy 140
196–7, 198 and justice system 133–4
luck-egalitarian theory 79 lawful opposition to 137–8, 139–40
legitimacy threshold 139
manipulation 55–6 as non-domination 152–3
material equality 125, 126 as non-interference 149–52
media influence 234, 235 and norms 134, 135
minority/majority divides 211–13 as on–off 139
mixed theories 79 and patriotism 140
morals 84 political obligation of citizens 138, 300
multinationals, and domination 115–17 as popular control 146–53
summary 299–303
natural resources for free choice 36 traditional 141–2
need, and eyeball test 86 and utilitarianism 142–5
non-directive influence 22–3 political liberty 41–2
Subject index 337
political lobbies 231, 233–5 republicanism
political ontology, principles of 288–92 and communitarian doctrine 11–18
political party policy 285 and liberalism 10–11
politicians, elected and political legitimacy 147–8, 294
dependence on private lobbies 234 resistance (civil disobedience) 138, 173, 219
electoral influence 209–11, 237 resourcing choice 69–74
self-interest of 231, 232–3 responsive assemblies 197–8, 201–5
popular control 160–79 restraint, absence of 9
domain of 160–6 retributivism 121–2
efficacious system of 167, 175–9 return, equal rates of 80–1
individualized system of 168–70, 302 revealed will 47–9
and influence 240–3 rewards 55
nature of 166–7 Robert’s Rules of Order 101, 298
unconditioned system of 170–4 Rome
veto over government 168 institutions 6
see also control subjection in 2
popular direction 245 rule-dependent choices 99
poverty 104–5 rule-dependent liberty 97
power rules, culturally variable 105, 106
asymmetrical 62
individualization of 226 Social Contract 142
reputation of 61–2 social justice 4, 18–19, 20
power sharing 222–3 and basic liberties 77
powers, separation of 143, 221–3 and coercion 136
preference-satisfaction 30–1, 34 and democracy 24–5, 303
priority of constituting people 289 and equality 77–92, 297
private interest 244–5 connection between 77–81
procedural liberties 185 republican justice equality strategy
process-condition 265 87–92
proportionality, in electoral system 199 republican justice freedom goal 82–7
protection and freedom 77
against domination 109–10 parental strategy analogy 80–1
of choice 69–74 and political legitimacy, see political legitimacy
public debate 215 principles (Rawls) 79
public interest 244–5 republican theory of 77, 88–92
public person 14 summary 297–9
see also justice
re-election, of politicians 232, 241, 247 social legitimacy, see political legitimacy
real will 47–9 social resources for free choice 36–7
reciprocal exchange 78 sovereignty
reflective equilibrium 20–1, 294 of constituted people 289–90, 293–4, 305
regularities of behaviour 128 international 18–19, 20
replaceability of constituted people 291–2 special insulation protection 114–17
representation, mechanisms of 199 specific hindrances 37–40
republican justice The Spirit of the Laws 6–7
character of 122–9 spontaneous norms 135
and criminalization 117–22 state
demands of 126–7 as agency 133, 300
equality strategy 87–92 and criminalization 117–22
freedom goal 82–7 and dual-aspect democracy 282–5
minimalism of 123, 126 and justice implementation 133–4
principle of 123–5 types of 189–90
role of law and norm in 127–9 state egalitarianism 78–81
see also justice; social justice status, and freedom 83–4, 90–2, 181, 298
republican rhetoric 254 straw voting 193, 284
338 Subject index
subjection unfree 162
evil of 1–2 utilitarianism 10, 142–5
and interference 56–9
supervision, by public 5 Venice 6
vitiation 28
threshold benefits, equal 80–1 of choice 63, 295
tough-luck test 177–9, 229 vitiating hindrances 35–49, 295
transparency, in decisions 215 voting 168, 192–4, 209–15, 216–17, 260, 283
uncontrolled interference 58–9 The Wealth of Nations 249

undominating will of constituted people 290–1 wealth regulation 127–8
unelected authorities 231, 235–8 Whigs 6–7

Pettit (2013) On The People's Terms - A Republican Theory and Model of Democracy

Uploaded by

Copyright:

Available Formats

Pettit (2013) On The People's Terms - A Republican Theory and Model of Democracy

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pettit (2013) On The People's Terms - A Republican Theory and Model of Democracy

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What is Philip Pettit's argument?

What is Philip Pettit's argument?

more information – www.cambridge.

According to republican political theory, people’s freedom as persons

Philip Pettit is L. S. Rockefeller University Professor of Politics and

The Seeley Lectures include

(1) Strange Multiplicity: Constitutionalism in an Age of Diversity

Cambridge University Press

© Cambridge University Press 2012

This publication is in copyright. Subject to statutory exception

First published 2012

Printed and bound in Great Britain by the MPG Books Group

Library of Congress Cataloguing in Publication data

isbn 978-1-107-00511-2 Hardback

Cambridge University Press has no responsibility for the persistence or

List of tables page x

Introduction. The republic, old and new 1

1.1 Variations in hindrance page 35

three core ideas

the liberal opposition

the communitarian opposition

building philosophically on republican history

the emerging view of democracy

justice and democracy

In order to gain a good understanding of the concept of freedom as non-

1. what has to be hindered to reduce

Freedom, robustness and probability

Table 1.1 Variations in hindrance

Hindrance is more probable If you choose X If you choose Y

You actually 1. choose X 2. choose X

P(X)(H if X) + P(Y)(H if Y) will reduce to P(Y)P(H if Y). That probability

2. vitiating and invasive hindrances

The resources required for free choice

Two kinds of hindrance

The signiﬁcance of invasive hindrances

The measurement of freedom

Revealed will and real will

3. no invasion without domination

The nature of interference

Table 1.2 Modes of interference

Do I interfere in your choice a. objectively (whether signalling b. cognitively (but not

1. by removing an option? Yes. Real obstacle. Yes. Pretend obstacle.

Interference is not sufﬁcient for subjection

4. no domination without invasion

Invigilation and intimidation

Three theories of freedom

Freedom, robustness and probability, again

5. how can we ensure your freedom of choice?

Resourcing and protecting

Actions and omissions

1. social justice and equality

proportion to the children’s different capacities to gain educational

The equality strategy of republican justice

2. the basic liberties

Towards a meaningful life

* The freedom to express what you think.

* The freedom to practise the religion of your choice.