Sweet 1999

Comparative Political Studies
http://cps.sagepub.com/
Judicialization and the Construction of Governance

ALEC STONE SWEET
Comparative Political Studies 1999 32: 147
DOI: 10.1177/0010414099032002001
The online version of this article can be found at:

http://cps.sagepub.com/content/32/2/147
Published by:
http://www.sagepublications.com
Additional services and information for Comparative Political Studies can be found at:
Email Alerts: http://cps.sagepub.com/cgi/alerts
Subscriptions: http://cps.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://cps.sagepub.com/content/32/2/147.refs.html
>> Version of Record - Apr 1, 1999
What is This?
Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Stone Sweet / JUDICIALIZATION
COMPARATIVE POLITICAL STUDIES
AND GOVERNANCE
/ April 1999
I present a theory of the emergence and evolution of governance, conceived as the process
through which the rules systems in place in any social setting are adapted to the needs of those
who live under them. The theory is composed of three elements: normative structure, dyadic
contracting, and triadic dispute resolution. I demonstrate that a move to triadic dispute resolution
leads the triadic dispute resolver to construct, and then to manage over time, specific causal rela-
tionships between exchange, conflict, and rules. In this way, political life is judicialized. Under
certain conditions, the triad will constitute a crucial mechanism of political change. I then ex-
plain judicialization and the dynamics of change in two very different polities: the international
trade regime and the French Fifth Republic. The conclusion draws out some of the implications
of the theory and data for our understanding of the complex relationship between strategic be-
havior and social structure.
JUDICIALIZATION AND
THE CONSTRUCTION
OF GOVERNANCE
ALEC STONE SWEET

Nuffield College, Oxford
T he triad—two contracting parties and a dispute resolver—constitutes

a primal social institution, a microcosm of governance. If this is so, in
uncovering the institutional dynamics of the triad, we uncover an essential
logic of government itself. Broadly stated, my objectives are twofold: to
defend the validity of these contentions and to demonstrate their centrality to
the discipline.
AUTHOR’S NOTE: For enormously helpful comments and criticisms, I am indebted to Christo-
pher Ansell, James Caporaso, Harry Eckstein, Henry Farrell, Neil Fligstein, Ron Jepperson, Pe-
ter Katzenstein, Robert Keohane, Nicholas Onuf, Paul Pierson, David Rowe, Wayne Sandholtz,
Martin Shapiro, Anne-Marie Slaughter, Rogers Smith, and Susan Sterrett. Earlier versions of
the article were presented at the Seminar on International Law and International Relations at
Harvard Law School (organized by Anne-Marie Slaughter and Andrew Moravcsik, November
1995); the Legal Theory Workshop at the Yale Law School (organized by Bruce Ackerman and
Owen Fiss, February 1996); two Workshops on International Law and International Relations
Theory at Yale University (organized by Harold Koh and Alexander Wendt, February 1996, and
COMPARATIVE POLITICAL STUDIES, Vol. 32 No. 2, April 1999 147-184
© 1999 Sage Publications, Inc.
147

148 COMPARATIVE POLITICAL STUDIES / April 1999
The article proceeds as follows. After introducing key concepts, I present

a model of a particular mode of governance. By mode of governance, I mean
the social mechanism by which the rules in place in any given community are
adapted to the experiences and exigencies of those who live under them. The
theory integrates, as tightly interdependent factors, the evolution of strategic
(utility-maximizing) behavior and normative (cultural or rule-based) struc-
ture. It captures dynamics of change observable at both the micro level, by
which I mean the behavior of individual actors, and the macro level, by which
I mean the institutional environment (or social structure) in which this behav-
ior takes place. In the discussion, the mechanisms of change that are endoge-
nous to the model are specified, and the conditions under which we would
expect to see these mechanisms operate, and fail to operate, are identified. I
then employ the model to explain two hard cases of systemic change: the
international trade regime, established by the 1947 General Agreement on
Tariffs and Trade, and the French Fifth Republic, founded in 1958. In the con-
clusion, I draw out some of the implications of the analysis for our under-
standing of the complex relationship between strategic behavior and social
structure.
DYADS, TRIADS, NORMATIVE STRUCTURE
The model comprises three core elements: the dyad, the triad, and norma-
tive structure.
The dyad, “the simplest sociological formation” (Simmel, 1950, p. 122),
is any “pattern of [direct] exchange” between two individuals or groups (see
Foster, 1977). The dyad alone defines, more or less comprehensively, a wide
range of basic human relationships. Examples exist wherever we look for
them. In marriage (the union of two people), in feudal polities (the tie between
serf and vassal), in parliamentary democracies (the dichotomy of party of
government and party of opposition), in industrial production (the interde-
pendence of capital and labor), and in international relations (the network of
allies and enemies), dyadic structures constitute core social identities of indi-
vidual entities. Because dyads bind single units together, they are primordial
by Harold Koh, October 1997); and the Research Seminar Series of the Center for Culture,
Organization, and Politics at the University of California–Berkeley (organized by Neil Flig-
stein, April 1998). The article began as an attempt to elaborate a theory of judicial politics capable
of synthesizing key insights of two rival approaches to the study of public law in political science: the
political jurisprudence of Martin Shapiro and other legal pluralists and behavioralists, and the nor-
mative jurisprudence of Rogers Smith and others who work at the intersection of case law and new
institutional approaches to politics. The article was transformed as a result of these discussions.

Stone Sweet / JUDICIALIZATION AND GOVERNANCE 149
social institutions. They are building blocks to society, that is, they can be
linked in chains and clusters to form larger social formations, and they
1
develop quite naturally within such constructs.
The normative basis of the dyadic form is reciprocity. Reciprocity is the
glue holding the dyad (society) together, which accounts for why it exists in
every human community about which we know anything (Gouldner, 1977).
Stripped to essentials, the norm holds that “people should help those who
help them” (Gouldner, 1977, p. 37). Promises made are to be kept; debts
incurred are to be repaid; kindnesses received are to be recognized and
returned. Reciprocity, according to Simmel (1950), gives to the dyad “a spe-
cial consecration,” by linking each party to “a common fate” (pp. 123, 135).
The notion that reciprocity is crucial to the maintenance of social systems
has been a staple of social science (e.g., Hobhouse, 1906, Malinowski, 1932,
Parsons & Shils, 1951). With the rise of neorationalism in contemporary
political science, analytical priorities have shifted from the normative and
social contexts of politics to strategic choice contexts within which individu-
als seek to maximize their utility. Although neorationalists also privilege the
dyad, they problematize reciprocity in particular and norms in general (e.g.,
Axelrod, 1986). Indeed, the paradigmatic metaphors of game theory—
prisoner’s dilemma, chicken, the assurance games—focus our attentions on
the fierce difficulties of establishing and maintaining dyadic cooperation.
Dyadic forms are inherently unstable, neorationalists tell us, because each
party faces powerful incentives to ignore normative obligations thereby
cheating on the other.
I will return to a discussion of neorationalism later. For now, it is enough to
recognize that dyadic forms can accommodate cooperation, which can be
socially enabling, and conflict, which can be socially debilitating.
The triad, two disputants and a dispute resolver, is a universal, if under-
theorized, phenomenon (but see Black, 1998, chap. 6). I understand the triad
to be a primal technique of organizing social authority and, therefore, of gov-
erning. The underlying reason for this is simple: The triadic entity is the guar-
antor of reciprocity. Quite literally rooted in the dyadic form, the triad brings
an external presence to the dyad, a presence whose interest is in the fate of “a
common fate,” that is, in the durability of social relationships across time.
Viewed functionally, triadic dispute resolution (TDR) serves to perpetuate
the dyad, given changes in the preferences or identities of the two parties, or
changes in the environment. As Simmel (1950) puts it: “The triad indicates
1. In noticing the ubiquity of relationships organized in twos, I do not mean to imply that
dyadic forms are all that matter. I focus on them, as a representation of “the social,” for theory-
building purposes, namely, to obtain advantages that come with reduction.

transition” (p. 145). The triadic entity responds to, and is a crucial agent of,
social change.
Two ideal types of TDR are relevant to the analysis. The first is consensual
TDR, triads constituted by the voluntary consent of both disputants, that is,
by an ad hoc act of delegation. The act recognizes but also confers social
authority, or legitimacy, on the third party. Siblings appeal to parents, class-
mates to one another or to a teacher, villagers to a shaman, a chief, or a sage.
The second type is compulsory TDR, triads that are permanently consti-
tuted by jurisdiction: dispute resolution processes are triggered by one
party to a dispute against the will of the other. In this type, office replaces
delegation (Shapiro, 1980, chap. 1), that is, an initial—constitutional—act
of delegation is frozen in place, for the life of the polity. Courts are the para-
digmatic form of compulsory TDR (but legislative bodies perform similar
social functions).
To move from the dyad to the triad is to construct a particular form of
governance—the triadic. In dyads, conflict can be debilitating, but conflict is
constitutive of the triad. Once activated, TDR performs governmental func-
tions: to generate normative guidance about how one ought to behave, to
stabilize one’s expectations about the behavior of others, and to impinge on
ex ante distributions of values and resources. Stated simply, the social func-
tion of TDR (governance) is to regulate behavior and to maintain social cohe-
sion as circumstances change.
The final element of the model is normative structure: the system of
rules—or socially constituted constraints on behavior—in place in any com-
munity. A great deal of controversy surrounds the subject of norms and rules,
their status and explanatory value. Although this article is partly a response to
this controversy, I do not attempt to resolve it here.
What I call normative structure is equivalent to what North (1990) calls
“institutions,” variously: “the rules of the game,” “customs and traditions,”
“conventions, codes of conduct, norms of behavior, statute law, common law,
and contracts” (pp. 3-6). It is congruent with how Eckstein (1988; see also
Wildavsky, 1987) conceptualizes culture: “mediating orientations,” those
“general dispositions of actors to act in certain ways in sets of situations”
(p. 790). It conforms to March and Olsen’s (1989) notion of rules: the
“beliefs, paradigms, codes, cultures, and knowledge” that permit us to “iden-
tif[y] the normatively appropriate behavior” (p. 22). It equates norms, as Tay-
lor (1989, p. 135) does, with ideologies and culture, and it conceives of institu-
tionalized rules, in Jepperson’s (1991, p. 145) terms, as “performance scripts.”2
Despite clear differences in how structure is understood, culturalists and
(at least a few) neorationalists agree on far more than we might expect. For
Eckstein (1988), culture allows people to “decode experience . . . to give it

meaning,” which “saves virtually all decision costs” (pp. 791-792). For North
(1990), institutions (a subset of which is “culture”—“a language-based con-
ceptual framework for encoding and interpreting . . . information”) “exist to
reduce the uncertainties involved in human interaction,” thus saving “trans-
action costs” (pp. 6, 17, 25; see also Johnson, 1997; Kreps, 1990). Normative
structures enable human interaction by simplifying the range of choices
available to individuals and by investing those choices with meaning.
Across the social sciences, change in normative structure has proved diffi-
cult to theorize (e.g., Eckstein, 1988; Powell & Dimaggio, 1991, chap. 1;
Taylor, 1989; Tsebelis, 1990, chap. 4). We better understand the logic of
institutional inertia. Rules facilitate exchange between individuals, creating
opportunities for collective action. Behavior that responds to these opportu-
nities, once locked in (e.g., in dyadic forms), reinforces normative structure.
In culturalist or constructivist terms, because normative structures constitute
individual and collective identities, and therefore give meaning to action,
they are difficult to change by way of action, without a concomitant change in
identities. In either case, it is clear that the reproduction of particular ways of
doing things inheres in the organization of human community.
In the next section, I model the transformation of the normative structure,
focusing on the dynamics of change that are endogenous to the logic of
dyads, triads, and rules.
CONSTRUCTING GOVERNANCE
Figure 1 depicts a simple model of the process by which systems of gov-

ernance emerge and evolve. An adequate theory of this process must account
for the following,
• strategic behavior, how individual actors conceive and pursue their interests
within any given community;
• policy making, how values and resources are distributed within any given com-
munity; and
2. I am aware that I have just assembled, in a very small tent, a disparate group of scholars
who traditionally do not agree on many first principles, least of all how we ought to think about
social structure. I have referenced them together for two reasons. First, I am seeking to build a
theory that strips governance down to its constituent elements, structure being one such element.
Although the scholars cited disagree for some good reasons, we can easily identify what each of
them means by structure; we can also see that, despite distinctly different approaches to research,
structure fulfills more or less equivalent functions. Second, if (as I am claiming) my theory is
relevant to the study of governance generally, my audience must be broad not narrow.

Figure 1. The construction of governance.
• systemic change, how the normative structure in place in any given community
is constituted, maintained, and revised.
The model breaks down this process into four stages, each a chronological
shift along a circular path, moving clockwise. Movement is generated by the
relative intensity of two relationships: of the dyad and triad and of normative
structure and strategic behavior. Although each shift is conditioned by what
has happened in prior stages, the discussion highlights distinct aspects of
these relationships.
Shift 1: Normative Structure to Dyadic Contract

The theory holds that we can move, by virtue of a self-sustaining process,
from a single dispute about the terms of a dyadic contract to an elaborate gov-
ernmental system.
To get to a dyadic contract, we need two individuals and at least a rudimen-
tary normative structure. By dyadic contract, I mean the rules of exchange—or,
those promises—voluntarily entered into between two persons. Contracts can
be implicit or explicit. The promises made in an “implicit dyadic contract” are
uncodified and “lack ritual or legal basis”; the “explicit dyadic contract”

codifies promises that are meant to be legally binding (Foster, 1977, p.16).
Both forms establish reciprocal rights and duties among two contractants.
In contracting, two individuals coordinate their self-interest in terms of
some shared view of the future. Such coordination is difficult, if not impossi-
ble, without at least a primitive cultural framework: in the form of language
(communication), and in the form of the norm of reciprocity, which is embed-
ded in notions of individual commitment, reputation, and responsibility.
Reciprocity, a crucial building block of society, enables the construction of
the dyadic form; as Gouldner (1977) has it, the norm constitutes a “starting
mechanism” that “helps to initiate social interaction” (p. 39).
Normative structure also serves to maintain dyadic contracts by facilitat-
ing dispute resolution. It does so in three ways, two of which are relevant at
this stage. First, at the level of the single actor, reciprocity—or a relevant rule
or established manner of doing things—can prevent disputes to the extent
that the norm provides individuals with behavioral guidance and an under-
standing of the consequences of reneging on a promise and that individuals
constrain their behavior accordingly. Second, once a dispute has erupted,
reciprocity and other relevant rules may provide the contracting parties with
the materials for settling the dispute on their own, dyadically. Such norms
furnish the bases for evaluating both the disputed behavior and potential solu-
tions to the conflict. The authority, or legitimacy, of these standards depends
heavily on their inherent neutrality with respect to the dispute, in the strict
sense that the relevant norms predate the dispute.
Shift 2: Dyad to Triad

The legitimacy of dyadic relationships is rooted in the self-interest of the
contracting parties. (At this point, I exclude from the analysis dyads consti-
tuted by coercion.3) For each party, the contract must be functional, in the
sense that its existence depends on the perception that the benefits of con-
structing and maintaining the dyad outweigh the costs and that the benefits of
a particular dyadic form outweigh the benefits of going it alone. The dyadic
contract coordinates egoistic motives, in the form of rules (reciprocal rights
and duties), for the life of the contract.
3. In excluding coercion from consideration, the model does some violence to reality. Surely
all dyadic relationships reflect or organize ongoing power relationships that contain elements of
(at least implied) coercion. I nevertheless theorize a consensual rather than a coercive model of
governance to focus attention on outcomes that result exclusively from the internal logic of rules,
dyads, and triads. Put differently, mine is a theory of ideational and normative—not physical or
material—power (influence).

Once formed, rules organize how conflict is identified and understood. As

a dyadic relationship proceeds or as circumstances change, the meanings
attached to the same set of rules by the contractants may diverge. Or the rela-
tive value of the dyadic relationship may decline for one or both of the con-
tracting parties. Or, having calculated “best strategies,” one or both of the par-
ties may succumb to incentives to renege on obligations (the prisoner’s
dilemma). In any case, the dyadic form generates a massive functional
demand for dispute resolution in the form of rule interpretation.
Dyads, aided by rules, often resolve disputes on their own. When they do
so, the dyad comprises, in the parlance of legal anthropology, a “legal level”
(Collier, 1973, chap. 1). If disputants fail to resolve their dispute dyadically,
they may choose to delegate the matter to a third party, thus constituting a
new legal level, the triadic.
This act of delegation can be understood as a simple—and universal—act
of “common sense,” as Shapiro (1980, p. 1) does. It can also be understood in
rationalist-utilitarian terms. Delegation is likely when, for each disputant,
going to a third party is less costly, or more likely to yield a desired outcome,
than either breaking the dyadic contract and going it alone or attempting to
impose a particular settlement against the wishes of the other disputant. For
each disputant, the short-run risk of delegation is the prospect of a settlement
in favor of the other. In the long run, however, the more two disputants inter-
act with one other—the more a relationship is perceived as beneficial by each
party—the less that risk matters. Other things being equal, each contracting
party may expect to win some disputes and to lose others, over time, against a
backdrop of absolute benefit. The social logic of delegation, like that of the
dyadic contract, is one of long-range utility: Each party must believe that it is
better off attempting to resolve a dispute than dissolving the relationship
altogether.
Of course, the calculation of costs and benefits need not favor the move
from dyad to triad. Both parties may possess a powerful commitment to
maintaining, rather than resolving, their conflicts. When the core identities of
the parties are constituted in opposition to one another, TDR will be anath-
ema. Examples might include certain conflicts between the United States and
the Soviet Union during the cold war, and between the Irish Nationalists and
the Ulster Unionists in Northern Ireland. Furthermore, in the absence of
minimal levels of trust, or lack of information about trustworthiness, the
more likely delegation itself will be viewed by one or both of the parties as
potentially more costly than beneficial. At a minimum, agreement on a dis-
pute resolver whose impartiality and wisdom is recognized may be a crucial
first step, but agreement may elude the parties. Thus, although the move to

Figure 2. The dispute resolver’s calculus.
triadic governance is a means of overcoming low levels of trust and weak

behavioral norms, that move itself is not automatic.
These points accepted, dyadic conflict and the delegation of these con-
flicts to a third party is the fuel that drives the model. If disputants choose not
to delegate or are unable to agree on procedures, if disputants are always able
to resolve their disputes dyadically, or if one of the disputants is always able to
impose a solution on the other, the theory implies, there would be neither
TDR nor evolution in governmental forms.
Shift 3: The Crisis of Triadic Legitimacy

Once constituted, the triadic dispute resolver faces a potentially intracta-
ble dilemma. On the one hand, the third party’s reputation for neutrality is
crucial to the social legitimacy of the triad itself (see Shapiro, 1980). Dispu-
tants would be loathe to delegate disputes if it were otherwise. Yet, in resolv-
ing disputes the third party may compromise her reputation for neutrality by
declaring one party the loser. That is, after all, what each of the disputants
hopes. We can express the dispute resolver’s dilemma as a fundamental inter-
est: Her interest is to resolve dyadic conflicts while maintaining the social
legitimacy of TDR. In pursuit of this objective, she deploys two main tactics.
First, the dispute resolver seeks to secure legitimacy by defending her
behavior normatively, as meaningfully enabled and constrained by rules
embedded in normative structure. Normative structure facilitates TDR, just
as it facilitates dyadic dispute resolution, by providing ready-made standards
of appropriate behavior and solutions to conflicts. Reciprocity—“promises
shall be kept”—animates the dyadic form; it also animates the exercise of
TDR, to the extent that the dispute resolver works to restore substantive fair-
ness and a sense of trust among the parties.

Second, the dispute resolver anticipates the disputants’ (or a communi-

ty’s) reactions to her behavior, especially if she decides or is asked to take a
decision. Compliance is a crucial test of the social legitimacy of consensual
TDR. Figure 2 depicts this calculus. Position A.1 represents the substantive
outcome desired by disputant A, and position B.1 represents the substantive
outcome preferred by disputant B. Outcomes situated between positions A.1
and A.2, (or substitute B.1-B.2 for B) represent outcomes that the dispute
resolver believes will not provoke A to refuse compliance. The space
between B.2 and A.2 constitutes the dispute resolver’s assessment of the
range of decision-making outcomes that will lead to the resolution of the dis-
pute, to compliance, and (much the same thing) to the reestablishment of a
disputed rule. The calculus also helps her to fashion settlements that avoid the
declaration of a clear winner or loser: In the area between B.2 and A.2, each
disputant achieves a partial victory. For some disputes, the positions of A and
B are more polarized, and no B.2-A.2 space exists; in such cases, the dispute
resolver is unable to deploy the tactic (she has an interest in creating such a
space by, for example, mediating between the parties). These are hard cases,
ones in which the dispute resolver can expect that any decision taken is likely
to result in public protests or even noncompliance. If she does attempt to
resolve the dispute by rendering a decision, her legitimacy will rest all the
more on the persuasiveness of normative justifications.
Shift 4: Triadic Dispute Resolution and Rule Making

Modes of TDR can be arrayed along a continuum constituted on one pole
by mediation and on the other pole by adjudication. In mediation, the dispute
resolver helps the disputants arrive at a mutually satisfactory settlement of
the conflict. In adjudication, the dispute resolver authoritatively resolves the
dispute on her own. In practice, nearly all TDR takes place on intermediate
points between these two poles. Dispute resolvers move back and forth along
the continuum—or threaten to do so—continuously, to enhance their flexi-
bility, limit their political exposure, and maximize their influence over the
disputants. In any move along the continuum toward adjudication, the dis-
pute resolver is led, with increasing precision and formality, to announce her
decision and to provide a rule-based justification for it.
In adjudicating, the dispute resolver simultaneously resolves a dyadic dis-
pute and enacts elements of the normative structure. Both are forms of rule
making. First, she makes rules that are concrete, particular, and retrospective,
that is, she resolves an existing dispute between two specific parties about the
terms of one dyadic contract. Second, in justifying her decision—in telling us
why, normatively, a given act should or should not be permitted—she makes

rules of an abstract, general, and prospective nature. This is so to the extent

that her decision clarifies or alters rules comprising the normative structure.
This latter form of triadic rule making constitutes a predictable response
to the crisis dispute resolvers face (Tactic 1 above). Yet, it raises a delicate
second-order legitimacy issue. From the point of view of the disputing par-
ties, it makes evident that the exact content of the rules governing the dispute
could not have been ascertained at the time the dispute erupted. The percep-
tion of the dispute resolver’s neutrality erodes as her capacity to make rules is
revealed. The dispute resolver can mitigate—but can never permanently
resolve—this problem. Most important, she can work to cast TDR as delib-
eration about the precise relationship of abstract rules to a concrete dispute,
portraying her decision as a record of these deliberations. In doing so, she
portrays triadic rule making as a by-product of TDR rather than an outcome
that she desires in and of itself.
In this way, TDR generates a discourse about how people ought to behave.
Because rules, reasoning about rules, and the adaptation of rules to specific
social needs constitute the core of this discourse (and, I would argue, of the
evolution of norms more generally4), precedent follows naturally. Precedent
helps to legitimize TDR by simultaneously acknowledging rule-making
behavior, while constraining that same behavior with a rule: that like cases
shall be settled likewise. In encapsulating this sequence—dyadic rules con-
flict deliberation triadic rule making precedent—TDR organizes discourse
about a community’s normative structure. In doing so, TDR performs a pro-
foundly governmental function to the extent that dyadic contractors are
drawn into this discourse and help to perpetuate it.
Shift 1: (Re)Constructing the Dyad

In moving through Shifts 1-4, we see how the dyad, the triad, and norma-
tive structure can be knotted together. And we see how a single dyadic con-
flict can generate a process of systemic change, the constitution or reconstitu-
tion of a mode of governance. Thus, a dyadic dispute erupts; the disputants
delegate the matter to a dispute resolver; the dispute resolver resolves the
4. My position is largely congruent with Robert Sugden’s. Sugden (1989, pp. 93-96) sug-
gests that norms (he focuses on “conventions”) develop in path-dependent, self-reinforcing
ways, one mechanism of which is the ubiquity, and naturalness, of normative reasoning itself.
Normative structures are inherently expansionary to the extent that they enable people to reason
from one situation to another, by way of analogy. The move to precedent (and, therefore, to
something akin to case law) is one result of analogous reasoning. If this is so, TDR is embedded
within, and further reinforces, the path-dependent nature of the greater process through which
rule systems evolve.

conflict in a process involving normative reasoning, revising (at least subtly)

normative structure. In returning to Shift 1, the impact of structure on strate-
gic behavior, we come full circle, to our initial starting point. But we find our-
selves in a rather different world this time: The individuals comprising the
dyad have learned something about the nature of their relationship (the rules
governing their exchange) and about the environment (the normative struc-
ture) that sustains it.
Put simply, TDR has reconsecrated the contract and reenacted the norma-
tive structure. The dispute resolver may have done so in a relatively conserva-
tive manner, fashioning a partial victory for each disputant and appealing to
rules whose prior existence is relatively unquestioned. In so doing, she has
reinforced existing structure while clarifying its domain of relevance and
application. The dispute resolver may have done so in a relatively progressive
manner, declaring a clear winner and loser while revising an existing rule or
crafting a new one. In so doing, she has reshaped normative structure,
expanding its domain.
Given two conditions, such rule making is likely to generate powerful
pedagogical—or positive feedback—effects, to be registered on subsequent
exchange and dispute resolution. First, contractants must perceive that they
are better off in a world with TDR than they are in a world without TDR. If
they perceive as much, and if they are rational in the sense of being utility
maximizers, they will evaluate the rulefulness of any potential action and
anticipate the probable outcome issuing from TDR. Second, the dispute
resolver must understand that her decisions have some authoritative—that is,
precedential—value.
If these conditions are met, TDR will inexorably become a powerful
mechanism of political change, and dyadic exchange will inevitably be
placed in the shadow of triadic rule making. As we move around the circle a
second time, and then again and again, this shadow will deepen and expand,
covering more and more forms of human interaction. A virtuous circle is
thereby constructed: To the extent that TDR is effective, it lowers the costs of
dyadic exchange; as dyadic exchanges increase in number and in scope, so
does the demand for the authoritative interpretation of rules; as TDR is exer-
cised, the body of rules that constitutes normative structure steadily expands,
becoming more elaborate and differentiated; these rules then will feed back
onto dyadic relationships, structuring future interactions, conflict, and dis-
pute resolution.
This dynamic, self-reinforcing process can be understood variously. Con-
ceived in economic terms, the process operates according to the logic of
increasing returns and path dependence (Arthur, 1994; North, 1990; Pierson,
1997). As it proceeds, dyadic exchange will be channeled down narrower and

narrower paths, that is, individuals will continuously adapt their behavior to
increasingly differentiated sets of rules. It is also institutionalization: A
process through which specific social contexts will be increasingly defined
by specific rules of behavior—those curated by the triadic entity. Individuals
will absorb and act on these rules, thus (re)making themselves and their
community.
Dyadic and Triadic Governance

Dyads and triads organize human community, constituting modes of gov-
ernance to the extent that they are institutionalized—constructed and main-
tained by rules. Dyadic forms flourish in hierarchy, often coexisting or sym-
biotic with triadic forms. Patron-client networks are an ubiquitous example
(Landé, 1977). When vertically stacked, as in Confucian and military sys-
tems, a chain of dyadic relationships links the rulers with the ruled and thus
establishes hierarchy. New institutionalist political economy captures such
hierarchies in “principal-agent” models of organization (Moe, 1987). Dyadic
contracts also give order to anarchy, as they do in international relations
(Stone, 1994). Examples include the “balance of terror” system of deterrence
(Hoffmann, 1991) and the elaborate world constructed by the bilateral treaty.
Triadic governance is institutionalized in one of two basic modes: consen-
sual and compulsory. My model demonstrates how a purely consensual form
of triadic governance can evolve. Movement around the circle is driven by the
complex mix of harmony and tension that inheres in the relationship between
rules and self-interest. Coercive sanctions and enforcement mechanisms are
conspicuously absent from the account (see Note 3). Instead, movement
depends on specific actors—the dyadic contractants and the triadic dispute
resolver—identifying their respective interests in some, rather than in other,
ways and behaving accordingly. To capture these requisite conditions, the
model is expressed in terms of statements that begin with “when,” “if,” and
“to the extent that,” statements that apply to specific contexts.
It follows that actors are always capable of blocking movement at crucial
points around the circle. Two contractants may choose to dissolve their con-
tract rather than delegate to a third party. The dispute resolver may render
capricious decisions without normative justification. After a dispute has been
adjudicated, and if a clear winner has been declared, the loser may refuse to
comply with the decision. If such behavior is, or becomes, the normal state
of affairs, triadic governance will be stillborn, and social entropy will
result. In such a state, reciprocity and other elements of normative structure
do not, on their own, sustain social exchange, and triadic dispute resolvers do

not function, on their own, to restore reciprocity given dyadic conflict and
change in exogenous circumstances.
The function of sanctions is to buttress dyadic and triadic dispute resolu-
tion. By sanctions, I mean social provisions that penalize noncompliance
with rules and triadic rule making. To the extent that they operate effectively,
sanctions counteract behavior that blocks the restoration of reciprocity
among disputants.
In consensual triadic governance, sanctions are informal but potentially
fully effective. Individuals will be led to abandon existing contracts, or avoid
entering into future contracts, with a chronic rule violator, an individual who
willfully disregards obligations imposed by a dyadic contract or by a triadic
dispute resolver. To the extent that this occurs, the violator forgoes the benefits
associated with social exchange and suffers stigmatization (the loss of reputa-
tion). If the violator’s behavior has led all other actors to refuse contractual
relations with the violator, social exclusion is the sanction. Banishment, virtual
death, is the ultimate penalty associated with consensual governance.
In compulsory triadic governance, explicit rules govern this sequence:
dyadic contract triadic dispute resolution decision compliance. Such rules
commonly enable the move from dyad to triad in the absence of the consent
of one of the parties, require the triadic dispute resolver to consider the com-
plaint,5 oblige the parties to comply with the terms of an eventual decision,
and organize enforcement measures in cases of noncompliance.
Like the move from dyadic to triadic governance, the transition from con-
sensual to compulsory TDR is inherently the stuff of political development.
The move is not automatic because the condition necessary for transition is
not sufficient. The condition is that, for any community or pool of potential
contractants, existing normative structure fails to provide an adequate frame-
work for social exchange, although the social demand for coordinative rules
and dispute resolution has increased. This may occur for a variety of reasons.
The potential contractants may be strangers, that is, they do not share a com-
mon normative structure. Or, changes in normative structure may not have
kept pace with changes in the nature of social exchange within a given com-
munity. A community of neighbors can become a group of strangers as a
result of migration, increased social differentiation, or the division of labor.
In any case, when existing rules cannot sustain social exchange at an optimal
level, people have an interest in developing new ones. The condition is not
sufficient because the construction of such rules is a potentially irresolvable
collective action problem. Thus, the model does not predict that TDR always
produces systemic change. On the contrary, when people (a) share a relatively
5. If only to decide not to decide.

comprehensive normative structure and (b) interact on an ongoing face-to-

face basis (i.e., where information relevant to exchange is virtually perfect
and transaction costs are virtually zero), TDR tends to reenact, rather than to
remake, social norms. In such contexts, existing normative structure is suffi-
6
cient, informal sanctions are highly effective, and mediation is preferred to
7
adjudication.
As political development has proceeded in the world, so has the ubiquity
of compulsory triadic governance. Generating and imposing new normative
structures that replace or supplement previously existing structures is per-
haps the only, or at least the most efficacious, means of providing a system of
governance for individuals who are otherwise strangers to each other. At the
extreme, organized coercion reinforces TDR processes, guaranteeing, with
force if necessary, social exchange, dispute resolution, and the enforcement
of rules. The modern state is the institutionalization of coercive TDR.
The Lawmaker and the Judge

With the development of the modern state, the authority to govern—the
power to resolve disputes and to make rules—tends to be divided among two
separate figures. Separation of powers doctrines notwithstanding, the law-
maker and the judge are not easily detached from one another.
The point can be made in terms of the model. The model demonstrates
how a full-blown system of governance can be constituted and maintained by
TDR processes alone, if Shifts 1-4 are iterated ad infinitum. The dispute
resolver governs by the pedagogical authority of her decisions. Triadic rule
making is legislative in nature: It adapts, over time, a given normative struc-
ture to the demands of dyadic exchange. But TDR is a relatively inefficient
means of rule making because it proceeds on a case-by-case basis. In delegat-
ing lawmaking powers to a legislator, a community establishes a far more
6. Game theorists make this point in the guise of the folk theorem, developed in the literature
on repeated games (e.g., Fudenberg & Maskin, 1986; Kandori, 1992).
7. Ellickson (1991) tells us that Shasta County ranchers refuse to use, or even educate them-
selves about, the laws meant to govern and resolve disputes concerning grazing rights, fencing,
and stray cattle. Invoking “good neighborliness,” they prefer to settle such disputes dyadically or
by quiet mediation, according to well-established norms. But when a Texas rancher moves into
the community and openly disregards these rules, litigation is the result. Collier (1973), in her
study of how the various legal levels operate among Mayan Indians in Zinacantan, found a com-
plex blending of dyadic dispute settlement and mediation, and a hostility toward formal Mexican
law and courts mitigated only in dealings with outsiders. At the time these books were published,
both communities shared relatively stable normative structures possessed of relatively high
social legitimacy. If in both communities TDR was ubiquitous, third parties were not used to pro-
voke, and dispute resolution was not expected to result in, normative change.

efficient means of revising normative structure. Nonetheless, in legislating,

the legislator performs a dispute resolution function. As part of normative
structure, laws help to prevent conflict from arising in the first place and to
facilitate the resolution of conflicts that do arise. Furthermore, because the
legislator fixes general rules for an entire community, it generates a crisis of
legitimacy no less intractable than that which afflicts the dispute resolver.
This crisis animates political life, in the form—again—of a quest to construct
8
rules to constrain rule making.
Even when judicial and legislative functions are separated, comparative
institutional advantage produces legislative-judicial interdependence. The
lawmaker makes rules for which the reach, among other things, is immedi-
ately general and prospective; the judge makes rules for which the reach,
among other things, is immediately particular and retrospective. If the judge is
expected to enforce the lawmaker’s law, and if this law is meant to be binding,
coercive TDR is required. If, for reasons elucidated, TDR results in rule making,
then compulsory TDR results in the authoritative reconstruction of the lawmak-
er’s law. The legislator therefore shares rule-making power with the judge.
TDR and Systemic Change

Triadic governance facilitates social exchange, and the adaptation of rule
systems to the exigencies of those who exchange, hence its social utility.9
Other things being equal, it must be that dyadic governance is inherently less
flexible, and more brittle, than triadic governance. Whereas conflict can
destroy dyadic contracts, conflict activates TDR and establishes the parame-
ters of a politics that can recast the normative basis of social exchange.
My theory holds that TDR, if exercised on an ongoing and effective basis, is
a crucial mechanism of social cohesion and change. To put it in constructivist
8. For example, the constant struggle to establish or to revise constitutions, electoral systems,
and the legislative and judicial processes.
9. I am not suggesting that the development of a stable mechanism of TDR is the only way to
achieve the virtuous circle depicted by the model. Cases in point are Avner Greif’s accounts
(1989, 1993, 1994) of how other (quasitriadic) mechanisms have performed similar functions. In
his analysis of trade relations in the Mediterranean region during the late medieval period, Greif
shows that the expansion of overseas commerce depended heavily on the activities of middlemen
(organized as the Maghribi Trader’s Coalition) operating within a relatively fixed rule system
(the Merchant’s Law). In Greif’s account, and in the theoretical and empirical materials I present
here, outcomes depend on the extent to which three factors—(a) social exchange, (b) organiza-
tional capacity to manage potential conflict associated with exchange, and (c) rule struc-
tures—develop together, thereby constituting a dynamic system of reciprocal influence. A
related theoretical framework has been developed to explain the dynamics of European integra-
tion (Stone Sweet & Caporaso, 1998a; Stone Sweet & Sandholtz, 1997).

terms, triadic governance coordinates the complex relationship between

structures and agents (Giddens, 1984), helping to constitute and reconstitute
both over time. In culturalist terms, it serves to counteract forces favoring
social “anomie” or “entropy” by adjusting general “guides to action” on one
hand to “the relentless particularity of experience” on the other (Eckstein,
1988, pp. 795-796) by, among other things, generating normative discourse.
In rationalist terms, the move from the dyad to the triad replaces games, such
as the prisoner’s dilemma or chicken, with an entirely different strategic con-
text. Although game theorists have begun to notice the challenge (e.g., Calvert,
1995), they have had difficulty modeling the kinds of triadic games implied
by this article (and by a good deal of judicial politics more generally), not
least because in these games the evolution of rule structures is endogenized
and normative reflection and argumentation are part of the game (see Stone
Sweet, 1998; Vanberg, 1998b).
In the next section, I demonstrate the power of the model to explain systemic
change, by which I mean a fundamental transformation in how normative
structure is constituted and sustained in any given human community. The
evolution from dyadic to triadic governance and the transition from consen-
sual to compulsory TDR are unambiguous examples. To the extent that such
evolution occurs, there will be a commensurate change in the social basis of
exchange, that is, in how individual actors understand and pursue their inter-
ests in coordination with other actors. Systemic change, then, implies the
transformation of collective and individual identity.
(RE)CONSTRUCTING THE POLITY
It would be a relatively simple task to demonstrate the general relevance of

the theoretical model to the field of judicial politics. One could, for example,
review the now burgeoning political science scholarship on the political
impact of judging around the world.10 For reasons that inhere in the theory,
we would learn that, in any given society, the judiciary’s share of total gov-
ernmental authority and influence varies with the degree to which it pos-
sesses and exercises the power to review the lawfulness of activity, public and
private. The task is simplified by the fact that political scientists generally
study courts, which are fully constituted mechanisms of coercive TDR.
I propose, instead, to examine the impact of TDR on two polities in which
judicial power had been initially, and by design, excluded. By judicial power,
10. After decades of neglect, the field of comparative judicial politics now thrives (e.g.,
Shapiro, 1980; Shapiro & Stone, 1994; Tate & Vallinder, 1995; Volcansek, 1992).

I mean the capacity of a triadic dispute resolver to authoritatively determine

the content of a community’s normative structure. In my two cases (the inter-
national trade regime and the French Fifth Republic), new normative struc-
tures (an international treaty, a national constitution) established rules gov-
erning relations between specific political actors (states in the GATT, elected
officials and state institutions in France). The regular use of TDR led to the
mutation of these relations, and new polities were thereby constituted.
I will use the term judicialization as shorthand for this mutation, for—the
same thing—the construction of judicial power. Judicialization is a process
sustained by the interdependence of dyads and triads, and of rules and strate-
gic behavior. It is observable, and therefore measurable, as modifications in
the conduct of dispute resolution and social exchange. The judicialization of
dispute resolution is the process through which a TDR mechanism appears,
stabilizes, and develops authority over the normative structure governing
exchange in a given community. The judicialization of politics is the process
by which triadic lawmaking progressively shapes the strategic behavior of
political actors engaged in interactions with one another.
A full treatment of the cases is beyond the scope of a single article.11 Of
necessity, my treatment is schematic and abbreviated, focusing on the rela-
tionship between specific theoretical predictions and empirical outcomes.
The theory asserts that TDR organizes political change so as to facilitate the
survival of societies in which individuals interact with each other on a con-
tinuous basis. The theory predicts that as the scope and intensity of these
interactions increase, so will the demand for the adaptation of normative
structure by way of dispute resolution. If and when dyadic dispute resolution
fails to satisfy this demand, there will be pressure to use TDR if a triadic
mechanism exists, or to invent such a mechanism if it does not exist. Once
individuals have moved to the triadic level, the internal dynamics of TDR will
drive processes of judicialization. The dispute resolver will seek to balance
the competing claims of disputants but will also generate precedent to legiti-
mize decisions. Triadic rule making will gradually reconfigure normative
structure and, in so doing, reconstruct social relations.
THE JUDICIALIZATION OF THE INTERNATIONAL TRADE REGIME
When GATT (1948) entered into force and was institutionalized as an

organization, “anti-legalism” reigned (Hudec, 1993, p. 137; Long, 1985,
11. I rely heavily on detailed studies of the judicialization of the GATT system (Hudec, 1992,
1993; Stone Sweet, 1997) and of the French Fifth Republic (Stone, 1992, 1996).

pp. 70-71). Diplomats excluded lawyers from GATT organs and opposed liti-
gating violations of the treaty. In the 1950s, TDR emerged in the form of the
Panel System. Panels, composed of 3-5 members, usually GATT diplomats,
acquired authority through the consent of two disputing states. In the 1970s and
1980s, the system underwent a process of judicialization. States began
aggressively litigating disputes; panels began treating the treaty as enforce-
able law, and their own interpretations of that law as authoritative; and jurists
and trade specialists replaced diplomats on panels. The process generated the
conditions necessary for the emergence of the compulsory system of adjudi-
cation now in place in the World Trade Organization (WTO).
Normative Structure and Dispute Resolution

GATT is the most comprehensive commercial treaty in history, today gov-
erning more than five sixths of world trade. From 1955 to 1974, membership
jumped from 34 to 100 states; 124 states signed the Final Act of the Uruguay
Round (establishing the WTO) in 1993. The treaty’s core provision is a gen-
eralized equal treatment rule, the Most Favored Nation (MFN) principle,
which rests on reciprocity: Each party to GATT must provide to every other
party all the advantages provided to other trading partners. The treaty further
prohibits, with some exceptions, import quotas. The organization also sup-
ports an interstate forum for legislating trade law: Eight rounds have reduced
most tariffs to the point of insignificance and, less successfully, restricted
nontariff barriers to trade.
The treaty exhorts members to settle their disputes dyadically, in accor-
dance with GATT rules. The potential for a trade conflict to move to a triadic
stage was implied: If State A could demonstrate that it had suffered damages
due to violations of GATT law committed by State B, State A could be
authorized by the GATT membership as a whole to withdraw advantages or
concessions that it would normally be required to accord State B. Almost
immediately, however, member states invented the Panel System to resolve
disputes.
As institutionalized in the 1950s, the system blended mediation and con-
sensual adjudication, against a backdrop of ongoing dyadic dispute resolu-
tion. Defendants could not be compelled to participate in TDR. By denying
consent, a state could block the construction of a panel, reject proposed pan-
elists, and refuse to allow a ruling to be reported. Relative to compulsory
forms of adjudication, the system appeared grossly inefficient. The original
function of panels, however, was to facilitate dyadic conflict resolution, not
to punish violators or to make trade law. Diplomats, trade generalists who
saw expedience in flexible rules and detriment in rigid ones, sat on panels.

Table 1
Dispute Settlement Activity in GATT: 1948-1989
1948- 1960- 1970- 1980-

1959 1969 1979 1989 Totals
Complaints
Complaints filed 53 7 32 115 207
Settled: Conceded by defendant 22 2 12 28 64
Settled: Withdrawn by plaintiff 10 0 5 40 55
Settled: Panel rulings 21 5 15 47 88
Panel activity
Panels convened 25 5 22 59 111
Rulings rendered 21 5 15 47 88
Published opinions 9 4 14 19 46
Rulings
No violation by defendant 6 0 7 7 20
Violation by defendant 15 5 8 40 68
Source: Adapted from data presented in Hudec, 1993, chap. 11.
When mediation failed, panels could, with the consent of the disputants,
resolve conflicts according to relevant treaty provisions.
Before 1970, states did not exploit the connection between TDR and rule
making. But, being both imprecise and rigid, the regime’s normative struc-
ture proved insufficient to sustain optimal levels of trade over time. The treaty
mixes a few hard obligations (the MFN norm and tariff schedules) with a
great many statements of principle and aspiration. Despite its flexibility,
important GATT provisions could be revised only by unanimous consent.
Although the success of GATT was partly due to normative imprecision—the
more vague a rule, the easier it was for states to sign on to it—textual impreci-
sion was often locked in by the unanimity requirement. The tension is obvi-
ous. Achieving optimal levels of exchange partly depends on the continuous
adaptation of abstract rules to concrete situations, but the GATT legislator
was ill-suited to perform this adaptation for the trade regime.
Dyad to Triad
Beginning in 1970, the largest trading states turned to the Panel System
not just to resolve their trade conflicts but also to make trade policy. Statistics
tell part of the story (Tables 1-3). After falling into desuetude in the 1960s
(only seven complaints filed), TDR exploded into prominence afterward. Of
the 207 complaints filed through 1989, 72% were filed after 1969 and 56% after
1979. The four largest trading states—Canada, the European Community,

Table 2
Dispute Settlement, 1980-88: Compliance Among Selected States
Complaints Adverse
Against Ruling Compliance* Noncompliance**
United States 36 9 6 4
European Community 30 9 5 3
Canada 9 6 4 2
Japan 14 4 4 0
All other states 14 7 7 0
Total 103 35 26 9
Source: Adapted from data presented in Hudec, 1992, pp. 34-35.
*Includes promises to comply made by defendant state.
**Includes 3 instances of noncompliance on the part of the EC; the disputes were settled to the
plaintiff state’s satisfaction after subsequent trade negotiations.
Japan, and the United States—dominated panel proceedings; in the 1980s,

more than 80% of all disputes registered involved two of these four states.
The expansion of global exchange and the domestic political conse-
quences of that expansion broadly explain the renaissance of TDR. Bilateral
exchange among the Big 4—Canada, the European Community, Japan, and
the United States—rose from $15 billion in 1959, to $44 billion in 1969, to
$234 billion in 1979, and to $592 billion in 1989. As trade redistributed
resources and employment across productive sectors within national econo-
mies, domestic actors mobilized to protect their interests. And as these
economies came to produce virtually the same products for export (e.g., elec-
tronics, automobiles, food products), trade relations were easily interpreted
in zero-sum terms.
By 1970, new forms of protectionism had proliferated, the gold standard
currency regime was rapidly disintegrating, and the American trade deficit
had become chronic. The need for clearer rules and better compliance was
acute. At the same time, the GATT legislator had failed to liberalize certain
crucial sectors (e.g., agriculture), to dismantle the mosaic of nontariff barri-
ers that had emerged in response to tariff reduction (e.g., restrictive licens-
ing policies and production standards), and to regulate other practices that
distorted trade (e.g., subsidies). Led by the United States, which was also
groping for ways to reduce its trade deficit, governments turned to the Panel
System.12
12. Disputants tend to litigate what diplomats failed to legislate. Conflicts over agriculture and
subsidies paralyzed trade negotiations, and they also dominated TDR processes after 1970. Of 115
complaints filed in the 1980s, 51 (44%) concerned trade in agricultural goods. Of the 44 disputes
filed citing one of the GATT codes, 21 (or 48%) relied on rules found in the subsidies code.

Table 3
The Review Activities of the French Constitutional Council: 1958-1993
1959-1973 1974-1980 1981-1987 1988-1993
Referrals 9 66 136 98
President 0 0 0 0
Prime minister 6 2 0 4
President of the Assembly 0 2 0 1
President of the Senate 3 0 2 3
60 deputies, or 60 senators — 62 134 90
Decisions* 9 46 92 70
Censuring text 7 14 49 38
Favorable to text 2 32 43 32
Source: Stone, 1996.
*Due to multiple referrals, the number of referrals since 1974 is larger than the number of decisions.
Three general motivations animated the move to TDR. In the vast majority
of instances, states initiated complaints in order to induce other states to mod-
ify their domestic trading rules. As we will see, the GATT panels proved to be
a relatively effective means of doing so. Second, states appealed to panels to
alter, clarify, or make more effective existing GATT rules. This motivation
overlaps the first, because virtually all trade disputes are translatable into a
general argument about the meaning and application of specific treaty provi-
sions. Disputants worked to convince panels to adopt their versions of GATT
rules to encourage the spread of practices they considered lawful and to dis-
courage practices they considered unlawful. Third, although difficult to ver-
ify, governments sometimes participated in TDR to delegitimize—and thus
facilitate the revision of—their own trade practices.13
To maximize their success, governments had a powerful interest in replac-
ing diplomats and generalists with lawyers and trade specialists. The Ameri-
cans understood this immediately; the Nixon administration turned GATT
litigation over to trade lawyers in 1970. By that year, the enormous complex-
ity of trade disputes—the resolution of which requires determining (a) the
13. In 1988, the United States instituted proceedings against the EC’s payment regime for
oilseed processing. A panel ruled that the program both discriminated against foreign processors
and functioned as an indirect subsidy for EC producers. France, invoking the consensus norm,
sought to suppress the decision but the EC adopted the ruling over France’s objection. The EC
then replaced the payment system with a new one. In effect, the EC had used TDR to delegitimize
an outmoded, costly program of which France had blocked revision within internal EC lawmak-
ing processes. Complaint No. 179, U.S. v. EC (22 April 1988). Complaints have been assembled
and numbered chronologically in Hudec (1993, Appendix). I use Hudec’s reference system to
refer to cases in this and subsequent notes.

extent to which a specific domestic law or administrative practice conforms

with treaty provisions and (b) the extent to which, in cases of nonconformity,
such a law or practice had caused, or might cause, trade distortions—was far
beyond the capacity of anyone but the lawyer and the expert. Once introduced
by the Americans, lawyerly discourse perpetuated itself. Lawyers filed
detailed legal briefs, attacking or defending particular national policies;
faced with detailed questions, panels gave detailed answers; lawyers then
understood the reasoning supporting such answers as guidelines for future
litigation strategies. The EC and Japan initially resisted the move to legalism,
but they became active participants after being bombarded with complaints
by the United States and Canada. By the early 1980s, all of the major trading
states had armed their Geneva staffs with permanent legal counsels.
Triadic Governance
In activating TDR, GATT members delegated to the Panel System an
authority that is inherently governmental. As panels exercised this authority,
they generated three sets of political outcomes; these outcomes can only be
explained by attending to the dynamics of TDR.
First, panels altered the terms of global exchange by provoking, with their
decisions, the modification of national trading rules. If complied with, every
decision declaring a national rule or practice inconsistent with GATT rules
concretely affects the lives of importers, exporters, consumers, and produc-
ers. Statistics (Tables 1-3) show that activating TDR worked in favor of plain-
tiff states: Plaintiffs enjoyed a success rate of 77% in the 1948-1989 period,
rising to 85% in the 1980s. The rate of compliance with adverse decisions
was 74% in the 1980-1989 period.
To resolve many of the most complex disputes, panels had no choice but to
reach far into national jurisdictions. Thus, a panel ruled that a U.S. law pro-
viding a special administrative remedy for patent infringement claims involv-
ing imported goods violated GATT because defendants stood a better chance
14
of winning in district courts. To arrive at this decision, panelists investigated
U.S. litigation rates and judicial outcomes, concluding that biases in the admin-
istrative procedure constituted a discriminatory bias affecting trade. In separate
cases, panels required Canada to force provincial governments to remove taxes
on foreign gold coins and to force provincial liquor boards to change regulatory
practices favoring domestic alcoholic beverages.15
14. Complaint No. 162, EC v. U.S. (29 April 1987).

15. Complaint No. 132, South Africa v. Canada (3 July 1984); Complaint No. 139, EC v.
Canada (12 February 1985).

Panels reinforced their influence over policy outcomes by elaborating

guidelines for state compliance. In explaining why a given national practice
was or was not inconsistent with GATT obligations, panels suggested GATT-
consistent versions of the practices in question. (Such behavior inheres in tri-
adic rule making.) In 1986, to take just one instance, the EC attacked the Japa-
16
nese system of taxation for alcoholic beverages. The system, which classi-
fied products into dozens of categories corresponding to different tax rates,
resulted in importers paying higher taxes than Japanese producers for similar
products. The panel declared the system to be inconsistent with the treaty and
announced a general rule: National tax schemes must treat all “directly com-
petitive” products equally. It then elaborated a hypothetical system based on
equal treatment, demonstrating precisely what a lawful system would look
like. The Japanese subsequently adopted a system similar to the panel’s.
Second, in response to the exploitation of TDR by states for their own
political purposes, panels reinvented themselves as judges, the authoritative
interpreters of the regime’s normative structure. This process, predicted by
the model, can be tracked and measured. As the number and complexity of
complaints grew, panels produced longer decisions and increasingly precise
interpretations of treaty provisions.17 In complicity with GATT litigators,
citations to past decisions became increasingly common and expected. Once
constructed as a precedent-based discourse about the meaning of GATT
rules, panel decisions became a fundamental source of those rules. (Such rule
making took place despite the absence of a doctrine of stare decisis in interna-
tional law and despite the refusal of the member states to formally recognize
the precedential value of decisions.) Certain treaty provisions (e.g., the MFN
norm, rules governing taxation and quotas) emerged as sophisticated, rela-
tively autonomous domains of legal discourse.18 In these domains, rules can
today only be understood in light of a dense and nuanced case law. Although
the substance of this law is far beyond the scope of this article, panels ratcheted
up national responsibility to justify any claimed exceptions to liberal trading
16. Complaint No. 154, EC v. Japan (6 November 1986).

17. In the 1948-1969 period, the average length of reported rulings was 7 pages; in the 1970-
1979 period, the average length rose to 15 pages; after 1985, the average reached 48 pages
(Hudec, 1992, p. 11).
18. Breaking down GATT complaints filed in the 1980s with reference to the article of the
Agreement in dispute provides some indication of the relative density of these areas. In 115 fil-
ings, disputants invoked specific parts of the Agreement 212 times. Four areas of the law account
for 71% of total claims: the MFN norm (Articles 1 and 2, 21%), nondiscrimination in taxation
and regulation (Article 3, 10%), elimination of quotas (Articles 11 and 13, 34%), and nullifica-
tion or impairment of benefits (Article 23, 6%). Of the 66 instances in which the special codes
were invoked, the codes on subsidies were involved 41 times (62%). See Stone Sweet (1997).

rules that, among other things, served to expand the grounds for future
complaints.
Panels also generated rules governing their own jurisdiction (Hudec,
1993, pp. 258-265). By the end of the 1980s, a stable case law asserted that,
among other things, panels could
• not only review the consistency of national acts with the treaty but also detail
what kinds of similar, if hypothetical, acts might violate GATT rules;
• announce answers to questions not raised by plaintiffs, but which were never-
theless relevant to other trade disputes; and
• report a ruling even if the dispute on which it was based had become moot (e.g.,
as a result of prior dyadic settlement), in order to clarify GATT rules and thus
facilitate future dyadic and triadic dispute resolution.
Third, judicialization processes reconstructed how states understood the

nature of their own regime. States reacted to the development of a rule-
oriented mode of governance not by suppressing it but by adjusting to it.
Their lawyers filed more and increasingly legalistic complaints, and their
diplomats ratified judicialization in official agreements. Thus, the 1979 “Un-
derstanding” on dispute settlement placed the GATT’s system on legal foot-
ing for the first time, codified dispute settlement procedures, and gave legal
force to panel reports. In 1981, citing the overwhelming complexity of litiga-
tion facing panelists, states permitted the establishment of a Legal Office
charged with rationalizing procedures and providing support for panel mem-
bers. And in the Uruguay Round (1986-1992), states asked an autonomous
group of experts to study how TDR could be strengthened. The fruit of their
efforts was the legal system of the WTO.
The Final Act of the Uruguay Round transformed GATT into the WTO.
The treaty, which is (implicitly) treated as a form of a constitutional law, pro-
vides for a system of compulsory adjudication of disputes. The new rules:
Automatically confer jurisdiction to panels on the reception of a complaint, no
longer permit unilateral vetoes of any stage by either party; and provide for a
broad range of measures to punish noncompliance. An independent appellate
body is charged with handling appeals from panels. The body is to be com-
posed of seven members who possess “demonstrated expertise in law.”
Undeniably, the move from consensual to compulsory TDR could not
have taken place without a convergence in the preferences of the most power-
ful trading states. The United States had advocated more efficient dispute set-
tlement since the 1970s. The Americans had even taken measures in domes-
tic law to unilaterally punish those who blocked or refused to comply with

GATT decisions, and the move provoked the EU to adopt similar measures.
Facing a trading world in which GATT rules might be enforced unilaterally
by the most powerful states, the rest of the world joined the United States and
Europe in working to strengthen multilateralism.
But, if converging state interests were crucial to the enhancement of TDR
in GATT, judicialization generated the context necessary for that conver-
gence. Judicialization is socialization. As states gained experience with dis-
pute settlement, as panels performed their dispute resolution functions, as a
stable case law enhanced legal certainty, GATT members could afford to
view triadic rule making as a useful, cost-effective guarantor of regime reci-
procity. In the 1980s, states did not consider abolishing the Panel System but
debated how best to enhance it. By the end of the decade, a collective future
without effective TDR was no longer a serious option.
THE JUDICIALIZATION OF THE FIFTH REPUBLIC
The 1958 Constitution was France’s 15th since the Great Revolution. Like
its predecessors, the constitution enshrined an official state ideology, the
“general will,” the twin corollaries of which are statutory sovereignty and the
prohibition of judicial review. In the 1970s, a process of judicialization
began. Legislators turned to the Constitutional Council to resolve their dis-
putes about the constitutionality of pending legislation; the Council
responded by developing the constitution as a system of substantive rules
governing policy making. As these interactions intensified, the Council
became an active participant in the legislative process; legislators became
active participants in the construction of constitutional law; and a new ideol-
ogy, that of constitutionalism, replaced the ideology of the general will.
Normative Structure and Dispute Resolution

Until the 1970s, the history of French constitutional law was barely more
than a chronicle of how state structures were successively remade by the
alternation in power of republicans, restored monarchs, emperors and gener-
als. As normative structures, French constitutions were brittle. They legiti-
mized the authority of those (temporarily) in control of the state but were
incapable of organizing enduring relationships between those who competed
for that control. The Fifth Republic—imposed in the midst of virtual civil
war—initially appeared just as brittle. A blueprint for Gaullist rule, the con-
stitution was not broadly welcomed or shared by the country’s political elites.
The political parties of the Left voted to reject the document, in a parliamen-
tary ballot they lost.

As ratified by popular referendum, the constitution departed from republi-

can traditions in two ways. First, it redistributed lawmaking power away from
parliament—the National Assembly and the Senate—and to the execu-
tive—the president of the Republic and the government (the prime minister
and ministers of state collectively). Statutes proposed by the government
were meant to be ratified by the legislature. Second, it established a new state
organ, the Constitutional Council, to police this redistribution. The founders
did not conceive of the Council as a judicial body. In contrast to all other
European constitutional courts, the Council does not hear appeals from the
judiciary or from individuals, and prior legal training or judicial experience is
not a requirement for membership.19 The founders rejected proposals to
include in the constitution a bill of rights over which the Council would exer-
cise jurisdiction, for fear of subverting statutory sovereignty.
In its original form, the constitution enabled but narrowly circumscribed
TDR. According to these rules, one of four officials—the president, the
prime minister, the president of the Assembly, or the president of the Sen-
ate—may ask the Council to review the constitutionality of a statute but only
after that statute has been definitively adopted by Parliament and before its
entry into force. If the Council determines that a statute’s provisions have
been adopted according to procedures that are inconsistent with constitu-
tional rules governing the legislative process, it annuls those provisions,
blocking their entry into force. Once a referral has been received, the Council
decides within a maximum delay of 30 days. In striking contrast to North
American constitutional judicial review, only statutes that have not yet been
promulgated are open to review in France; once in force, statutes are immune
from judicial scrutiny.
Two constitutional revisions modified the Council’s jurisdiction. In a
1971 decision, the Council—prompted by a referral of the president of the
Senate, and publicly encouraged by law professors and the media—annulled
a piece of government-sponsored legislation for the first time. In relying on a
rights text, it effectively incorporated a bill of rights, partly unwritten and ill-
defined, into the constitution. In 1974, a Constitutional Congress voted to
extend to parliamentary oppositions—formally, to any 60 deputies or sena-
tors—the power to refer statutes to the Council for review. Combined, these
changes radically expanded the rule-making capacities of TDR. Henceforth,
19. The Council is composed of nine members; the president of the Republic, the president of
the Assembly, and the president of the Senate each appoints three members who serve 9-year
terms.

any partisan dispute about legislation, once translated into a dispute about
constitutional rights, could be used to activate TDR.
Dyad to Triad
In the 1958-1970 period, the Council rendered only six decisions, siding
each time with the prime minister in disputes between the government and
Parliament over their respective legislative powers.20 Since 1974, the Coun-
cil’s caseload has been constituted, almost exclusively, by opposition refer-
rals alleging the unconstitutionality of legislation proposed by the govern-
ment and the parliamentary majority.21 When in power, parties of the Left and
Right have decried the Council’s growing authority over the lawmaking
processes, and both have threatened to abolish the organ to restore “the sover-
eignty of the general will.” In opposition, Left and Right have exploited,
without apology, the capacity of TDR to obstruct majority rule.
Oppositions are attracted to TDR for a simple reason: The Council is the
only state institution capable of altering legislative outcomes that is not con-
trolled by the government and the parliamentary majority. Referrals extend
the legislative process to include another stage: triadic rule making. The
move to TDR alters the strategic context of French policy making, redistrib-
uting political initiative in the opposition’s favor and reducing the influence
of the government and the majority over legislative outcomes. When the
opposition activates TDR, the government and the majority are placed on the
defensive, forced to participate in processes they do not control.
After 1974, referrals quickly became quasisystematic (Tables 1-3). Since
that year, opposition parties have referred every annual budget and, since
1981, virtually every major piece of legislation. In the 1974-1980 period, the
Giscard presidency, 46 laws were referred to the Council, or 6.6 per year; in
the 1981-1987 period, the first Mitterrand presidency, 92 laws, 13.1 per year,
were referred. The average number of references has remained above 10 per
year ever since. Expressed in different terms, since 1981 about 30% of all leg-
islation adopted has been referred,22 a huge ratio because most legislation is
politically noncontroversial and does not lead to a formal roll call vote at the
time of adoption. Substantively, the vast majority of referrals allege that the
referred law violates one or more constitutional rights capable of being
20. In 1962, the body also refused to rule on a complaint made by the president of the Senate.
21. By parliamentary majority, I mean those parties that support the government in parlia-
mentary votes.
22. Excluding the statutory ratification of international agreements.

recognized by the Council. Virtually costless, referrals work: Since 1981,

more than half (57%) of all referrals resulted in a Council annulment.
As predicted by the model, the regular use of TDR produced a self-
sustaining process of judicialization. Referrals provoked the Council to con-
struct the constitution, that is, to justify annulments in terms of an authoritative
interpretation of constitutional rules, and the construction of constitutional
law provoked more referrals. Also predicted (Shift 1: Reconstructing the
Dyad), triadic rule making exercised a powerful pedagogical influence on the
strategic behavior of lawmakers.
Most important, triadic rule making generated a stable politics of deter-
rence and anticipatory reaction. As the web of constitutional constraints on
lawmaking expanded and grew more intricate, the government became sus-
ceptible to a kind of constitutional blackmail. The opposition learned that it
could enhance its legislative influence by threatening to refer a bill under dis-
cussion to the Council, unless the government and majority agreed to amend
the legislation as the opposition saw fit. In the 1980s, parliament has adopted
hundreds of amendments, rewriting dozens of important bills, pursuant to
constitutional debates triggered by such threats (Stone, 1992, 1996; formal
model in Vanberg, 1998a).
As constitutional referrals, threats to refer, annulments, and constitutional
blackmail became banal elements of legislative process, lawmakers had
every reason to upgrade their legal expertise. The government, aided by its
legal advisor, the Council of State, began to review the constitutionality of all
draft bills, before submitting them to parliament.23 And the major political
parties turned to young law professors—specialists in “the new constitu-
tional law”—to help them draft referrals, respond to referrals, and to attack or
defend bills on the floor of parliament.
Triadic Governance
The move to triadic governance generated three sets of outcomes that
deserve our attention.
First, the Council evolved into a powerful policy maker, a kind of adjunct
legislative body with the capacity to veto, amend, and even propose legislative
provisions. Annulments have blocked or radically altered a score of major leg-
islative initiatives. In 1982, the Council vetoed the Left’s nationalization bill,
ruling that the legislation would not have provided sufficient compensation to
23. See Stone (1996) for details.

24. Respectively, Council decisions: 81-132 (1982); 84-181 (1984); 86-210 and 86-217
(1986); and 93-323, 93-325, and 93-326 (1993).

expropriated stockholders. In 1984, it vetoed key provisions of the press law,

thwarting the Left’s bid to counter the rapid concentration of the newspaper
industry. In 1986, it blocked the Right’s attempts to deregulate the press and
broadcast media. And in 1993, it gutted the Right’s attempts to restrict immi-
24
gration and to expand administrative authority to expel asylum seekers.
Furthermore, the Council, in clarifying exactly why a given legislative
provision is unconstitutional, provoked new legislative processes designed to
“correct” unconstitutional bills. These processes serve to implement the
Council’s rule making. Thus, in its decision on nationalizations, the Council
told the government exactly how stockholders must be compensated. The
government complained that its legislative authority had been preempted, but
dutifully incorporated the Council’s preferred compensation formula into a
new bill. The changes raised the cost of nationalizations by 25% but secured
promulgation. In more than 20 decisions on the penal codes (statutes specify-
ing crimes, penalties for committing crimes, and judicial procedures), the
Council has vetoed dozens of proposed legislative modifications, while lay-
ing down precise rules governing how these codes must or must not be
revised. The opposition has given agency to these rules by systematically
threatening to refer new reforms that do not sufficiently respect the dictates of
the Council’s case law.
Second, the Council reinvented itself as a court. Systematically implicated in
the legislature’s policy disputes, the Council worked to portray its decision
making as inherently judicial: a formal exercise in reasoning about rules. Coun-
cil decisions initially took the form of a series of terse syllogisms, containing
virtually no argumentation. In the late 1970s, the Council began producing
longer, more carefully crafted decisions, responding—point-by-point—to
25
arguments made in referrals. Predictably, referrals lengthened and became
more sophisticated. As decisions accumulated in areas of intensive legislative
activity, technical domains of case law inevitably emerged. The Council devel-
oped an array of linguistic formulas (one of several functional equivalents of
precedent in civil law systems) that it repeated again and again to clarify its
positions, and it began to reference, at first subtly, then more overtly, the com-
mentaries of leading legal scholars. In the late 1970s, a new autonomous field
of legal scholarship, devoted to the doctrinal analysis of the Council’s case
law, emerged. Constitutional law now flourishes in the academy.
25. Council decisions consist of numbered paragraphs. In the 1974-1979 period, the average
length of decisions (calculated annually) was 7 paragraphs, with a high of 8 in 1975. In the 1980-
1986 period, the average length of decisions (calculated annually) was 23, never falling below
13, with a high of 42 paragraphs in 1983 (Stone, 1992, p. 101).

Third, triadic rule making reconstructed the very nature of the French pol-
ity. Since the revolution, the constitution has been understood to be a law that
enabled state officials to govern but was unenforceable. Republican constitu-
tions proclaimed statutory supremacy and prohibited judicial review of statute
(perhaps a moot point because constitutions did not contain rights provisions,
and public liberties that were recognized in statute could be rescinded by
majority vote). In the absence of TDR, French constitutional law was static; it
developed no dynamic life of its own. Today, the constitution is a living law
that binds all public authorities in their interactions with each other and with
private individuals.
For the first time in history, French constitutional law is case law: The law
is what the Council interprets it to be (despite the formal absence of a doctrine
of stare decisis in the civil law tradition). In consequence, French legislative
politics, which operated on the basis of majority rule, have been reconstituted
as a constitutional politics, which operates as a continuously evolving, rule-
based discourse governing the exercise of legislative authority. In this poli-
tics, legislators continuously incorporate, into the language and practice of
policy making, a vocabulary and grammar of constitutional law. Lawmakers
do so to maximize their own political effectiveness. In their interactions with
each other, they debate and take authoritative decisions about the constitu-
tionality of statutes before them. This inherently judicial behavior, institu-
tionalized in the 1980s, is now a ubiquitous feature of the legislative
process.26 Referrals transfer these constitutional debates to the Council. In
this way, legislators participate in the construction of the constitution, pro-
viding the legal materials for constitutional adjudication, and legitimizing
the Council’s political authority.
Last but not least, judicialization also transformed the nature and function
of the French judiciary (Cour de Cassation, 1995). As constitutional review
steadily undermined legislative sovereignty, judges asserted their own
authority to interpret statutes and enforce the constitution. In the early 1980s,
the Supreme Court (Cassation) developed a rule that requires civil judges to
26. In parliament, formalized rituals—in the form of parliamentary “motions of unconstitu-

tionality”—organize deliberations. The motions require the chamber to debate and rule on the
bill’s constitutionality. The opposition, the government, and the majority support their respec-
tive positions by citing constitutional texts, legal scholarship, and the Council’s existing case
law. If the motion passes, the bill is declared unconstitutional, and it is killed. Because votes are
governed by party discipline, motions never pass. In the 1981-1987 period, the National Assem-
bly debated and voted on 94 motions of unconstitutionality; the Council rendered 93 decisions
over the same period (Stone, 1996).
27. In the presence of a law deemed unconstitutional, all judges can do is correct the law by
rewriting it, because a law once promulgated is immune to review.

do what traditional constitutional orthodoxy forbids, namely, to rewrite, by

interpretation, statutory provisions so that they conform to constitutional
27
law. Litigants now not only invoke constitutional rights in their arguments,
they sometimes win. Recently, France’s high administrative court, the Coun-
cil of State, has begun converting its own “general principles of law” (a body
of judge-made restrictions on administrative action that includes “individual
liberty,” “equality before the law,” and “freedom of conscience”) into equiva-
lent rights developed by the Council.
After two centuries, the French constitution is judicially enforceable law.
CONCLUSION
One virtue of the model is its inherent capacity to translate between micro-
level effects (or outcomes) and macro-level effects (or outcomes), simultane-
ously. The theory generates testable propositions about behavior and out-
comes at both levels. These propositions can be evaluated by focusing
empirical attention—at one point in time, at one level of analysis—on the
strategic interaction of individuals, the micro level, or the development of
normative structure, the macro level, while holding the other level constant.
The theory implies, however, that we will not be able to explain systemic
change adequately if we privilege, systematically and a priori, the causal
importance of one level vis-à-vis the other. The point deserves elaboration.
I have argued that how systems of governance emerge and evolve has every-
thing to do with the interdependence of rules and strategic behavior. If I am
right, Shift 2 is partly the province of neorationalism and game theory.
Rational individuals maximize utility by adjusting behavior—including how
they reason through, and talk about, norms—to changes in rules of the game.
In clarifying the scope and content of existing constraints, triadic rule making
shapes how players calculate the payoffs of available strategies. Stage 3—the
triadic dispute resolver’s response to the dilemma posed by the delegation of
political authority—can also be understood in neorationalist terms. Her inter-
est in her own survival leads the triadic figure to behave in predictable ways.
The analysis also suggests that neorationalism alone is inadequate to the
task of explaining systemic change. Game theorists rely heavily on struc-
ture—conceived as fixed rules—to provide the conditions necessary for pre-
dicting outcomes from strategic interaction. Although the macro level is an
integral part of any game theoretic analysis, all the action that matters actually
occurs at the micro level. Game theorists openly admit that they have barely
begun to theorize the dynamics of institutional change (see Tsebelis, 1990, chap. 4).
This article suggests that to the extent that neorationalism does not account for

crucial mechanisms of social change—including the relatively autonomous,

independent impact of normative discourse—its explanatory scope is lim-
ited. Put in terms of my theory, the value of neorationalism declines as we
move from the right to the left hemisphere of the circle (Figure 1).
The social world produces, consolidates, and stabilizes structure in vari-
ous ways. I have focused here only on the capacity of TDR to generate an iter-
ated, organized, and therefore social process of reasoning about rules. The
theory predicts that, once constituted, triadic governance will organize the
future by constructing and then managing the causal relationships between
social exchange, conflict, and normative development. Once these relation-
ships are established, TDR perpetuates a discourse about the pertinence of
rules to behavior, and this discourse gradually penetrates and is absorbed into
those repertoires of reasoning and action that constitute political agency.
Read this way, Stages 4 and 1 are partly the dominion of students of institu-
tionalization and structuration. Organizational theorists (Powell & Dimaggio,
1992), social psychologists (Rosenberg, 1995), constructivists (Giddens,
1984; Onuf, 1989, 1994; Wendt, 1992), culturalists (Eckstein, 1988), and
public law “new institutionalists” (Smith, 1988), reject the neorationalist
assumption that the essential properties of individual actors—or of rational-
ity itself—are exogenously fixed, or inherent. Instead, individual iden-
tity—how actors form, comprehend, express, and pursue their prefer-
ences—is understood to be socially constituted and therefore capable of
being socially reconstituted.
My theory, in effect, incorporates the constructivist point, without deny-
ing that self-conscious, strategic behavior at the individual level is a perma-
nent fixture of social life. Self-interested behavior, in fact, animates the
model. But the efficacy—or rationality—of this behavior is heavily condi-
tioned by macro-level structure, and components of this structure are signifi-
cantly autonomous from micro-level phenomena.
If I have correctly identified the causal linkages connecting dyadic
exchange, TDR, and normative structure, then those who initiate TDR cannot
meaningfully control the outcomes produced by triadic rule making. Viewed
over time, from the ex ante perspective of those who contract and delegate,
outcomes will not mechanically reflect the distribution of preferences and
capabilities among the actors within a given community. Rather, because tri-
adic rule making secures and enhances the relative autonomy of normative
structure—and of normative discourse—vis-à-vis actors, the world it builds
is only partly predictable by them and can therefore be only partly intended.
Put differently, the world of triadic governance evolves according to the logic
of path dependence, manifested by the increasing dominance of triadic rule
making (e.g., case law and precedent) over the content of normative structure,

and lock-in, manifested by the institutionalization of those forms of social

exchange that best respond to the evolution of this rule making (for applica-
tions to political analysis, see Pierson, 1996, 1997).
To return to my cases, the theory predicts that, given certain specified con-
ditions, a sustained move to TDR will reconstruct, gradually but inevitably,
the nature of governance. In the two polities examined here, and in more for-
28
mal tests of the theory, this prediction was borne out. States and French poli-
ticians began as jealous guardians of their own sovereignty, deeply hostile to
judges and to legalism; in pursuit of their own political objectives, they dele-
gated meaningful political authority to triadic dispute resolvers; and triadic
rule making reconfigured the Fifth Republic and the trade regime. The evi-
dence further suggests that the move to triadic governance stabilized and
made both polities more resilient in the face of potentially debilitating con-
flict. Propositions about the future are also suggested. There will be no
French Sixth Republic; rather, the French are already living in the new, more
supple republic institutionalized by triadic governance. And in the WTO, a
powerful supranational governmental authority will emerge, litigation will
steadily rise, and judges will generate an expansive legal discourse that will
gradually reshape the interstate discourse on trade.
To conclude, it bears repeating that systemic change is explained, but not
preordained, by the theory.29 Judicialization processes could have been
blocked or reversed: States and parliamentarians could have stopped
28. Stone Sweet and Brunell (1998) and Stone Sweet and Caporaso (1998a, 1998b) test a
series of specific hypotheses derived from the theory in their explanation of European legal inte-
gration, the process through which a transnational legal system for the EC has been constructed.
29. See Note 7 and corresponding text. Furthermore, I have left undertheorized, or ignored
altogether, certain patterns of behavior that are of obvious importance. First, I had little to say
about why and how actors negotiate the terms of their exchange in the first place, although the
more commitment-based are the rule structures they build, the more push for judicialization we
can expect (I thank Nicholas Onuf for reminding me of this point). Second, the kind of perpetual
motion machine theorized, one that produces ever higher levels of legal discourse and triadic
authority, has not functioned in most polities in the world, past or present. Two families of nega-
tive cases were nonetheless identified. Actors may be willing to tolerate, or even cultivate,
dyadic conflict without moving to TDR. In some cases, as when their respective identities are
constituted in opposition, disputants may have a higher interest in maintaining the conflict. In
other cases, for example, zero-sum situations, no joint gains issuing from dispute resolution are
possible, and in still others, neither party may be willing to budge from original, fixed, and radi-
cally opposed positions (there is no negotiating space for the triadic entity to exploit). A second
class of negative cases concerns instances in which the move to TDR does not lead to rule mak-
ing but to rule reinforcement and social control. Finally, I provide no theory of judicial rule mak-
ing, that is, how the dispute resolver interprets and makes rules. Nonetheless, core elements of
such a theory are implied in the Constructing Governance section (Shift 3, Note 4, and corre-
sponding text), and in my treatment of the two case studies.

activating TDR, they could have renegotiated new normative structures to

govern their relations, or they could have abolished judicial power.
But—unwilling to forego the benefits of TDR and unable to agree on alternative
arrangements— they did not.
Instead, they constructed triadic governance and triadic governance
reconstructed them.
REFERENCES
Arthur, W. B. (1994). Increasing returns and path dependence in the economy. Ann Arbor: Uni-
versity of Michigan Press.
Axelrod, R. (1986). An evolutionary approach to norms. American Political Science Review, 80,
1095-1111.
Black, D. (1998). The social structure of right and wrong. San Diego, CA: Academic Press.
Calvert, R. L. (1995). Rational actors, equilibrium, and social institutions. In J. Knight & I. Sened
(Eds.), Explaining social institutions. Ann Arbor: University of Michigan Press.
Collier, J. F. (1973). Law and social change in Zinacantan. Stanford, CA: Stanford University
Press.
Cour de Cassation. (1995). La Cour de Cassation et la Constitution de la République (The
Supreme Court and the Constitution of the Republic). Aix-en-Provence, France: Presses
Universitaires d’Aix-en-Provence.
Eckstein, H. (1988). A culturalist theory of political change. American Political Science Review,
82, 789-804.
Ellickson, R. C. (1991). Order without law: How neighbors settle disputes. Cambridge, MA:
Harvard University Press.
Foster, G. M. (1977). The dyadic contract: A model for the social structure of a Mexican peasant
village. In S. W. Schmidt, J. C. Scott, & L. G. Landé (Eds.), Friends, followers, and factions.
Berkeley: University of California Press.
Fudenberg, D., & Maskin. (1986). The folk theorem in repeated games with discounting or with
incomplete information. Econometrica, 50, 533-554.
Giddens, A. (1984). The constitution of society: Outline of the theory of structuration. Berkeley:
University of California Press.
Gouldner, A. W. (1977). The norm of reciprocity. In S. W. Schmidt, J. C. Scott, & L. G. Landé
(Eds.), Friends, followers, and factions. Berkeley: University of California Press.
Greif, A. (1989). Reputation and coalitions in medieval trade: Evidence on the Maghribi traders.
Journal of Economic History, 49, 857-882.
Greif, A. (1993). Contract enforceability and economic institutions in early trade: The Maghribi
trader’s coalition. American Economic Review, 83, 425-448.
Greif, A. (1994). Cultural beliefs and the organization of society: A historical and theoretical
reflection on collectivist and individualist societies. Journal of Political Economy, 102, 912-
950.
Hobhouse, L. T. (1906). Morals in evolution: A study in comparative ethics. New York: Holt.
(Reprint: 1968, New York: Johnson)
Hoffmann, S. (1991). Ethics and the rules of the game between the superpowers. In L. Henkin
(Contributors), Right v. might: International law and the use of force. New York: Council on
Foreign Relations.

Hudec, R. E. (1992). The judicialization of GATT dispute settlement. In M. H. Hart & D. B. Steger
(Eds.), In whose interest?: Due process and transparency in international trade. Ottawa,
Canada: Center for Trade Policy and Law.
Hudec, R. E. (1993). Enforcing international trade law: The evolution of the modern GATT legal
system. Salem, NH: Butterworth.
Jepperson, R. L. (1991). Institutions, institutional effects, and institutionalism. In W. M. Powell &
P. J. Dimaggio (Eds.), The new institutionalism in organizational analysis. Chicago: Univer-
sity of Chicago Press.
Johnson, J. (1997, Summer). Symbol and strategy in comparative political analysis. APSA-CP:
Newsletter of the APSA Organized Section in Comparative Politics.
Kandori, M. (1992). Social norms and community enforcement. Review of Economic Studies,
59, 63-80.
Kreps, D. M. (1990). Corporate culture and economic theory. In J. E. Alt & K. A. Shepsle (Eds.),
Perspectives on positive political economy. Cambridge, UK: Cambridge University Press.
Landé, C. H. (1977). The dyadic basis of clientelism. In S. W. Schmidt, J. C. Scott, & L. G. Landé
(Eds.), Friends, followers, and factions. Berkeley: University of California Press.
Long, O. (1985). Law and its limitations in the GATT multilateral trade system. Boston:
Martinus Nijhoff.
Malinowski, B. (1932). Crime and custom in savage society. London: Paul, Trench, and Trubner.
March, J. G., & Olsen, J. P. (1989). Rediscovering institutions. New York: Free Press.
Moe, T. (1987). An assessment of the positive theory of congressional dominance. Legislative
Studies Quarterly, 12, 475-520.
North, D. R. (1990). Institutions, institutional change, and economic performance. Cambridge,
UK: Cambridge University Press.
Onuf, N. (1989). World of our making: Rules and rule in social theory and international rela-
tions. Columbia: University of South Carolina Press.
Onuf, N. (1994). The constitution of international society. European Journal of International
Law, 5, 1-19.
Parsons, T., & Shils, E. A. (Eds.). (1951). A general theory of action. Cambridge, MA: Harvard
University Press.
Pierson, P. (1996). The path to European integration: A historical-institutionalist analysis. Com-
parative Political Studies, 29, 123-163.
Pierson, P. (1997). Path dependence, increasing returns and the study of politics. Study of Ger-
many and Europe Working Paper No. 7.7. Cambridge, MA: Harvard University Center for
European Studies.
Powell, W. W., & Dimaggio, P. J. (1992). Introduction. In W. W. Powell & P. J. Dimaggio (Eds.),
The new institutionalism in organizational analysis. Chicago: University of Chicago Press.
Rosenberg, S. W. (1995). Against neoclassical political economy: A political psychological cri-
tique. Political Psychology, 16, 95-136.
Shapiro, M. (1980). Courts: A comparative and political analysis. Chicago: University of Chi-
cago Press.
Shapiro, M., & Stone, A. (Eds.). (1994). Special Issue: The new constitutional politics in Europe.
Comparative Political Studies, 15.
Simmel, G. (1950). The sociology of Georg Simmel. New York: Free Press.
Smith, R. (1988). Political jurisprudence, the “new institutionalism,” and the future of public
law. American Political Science Review, 82, 89-109.

Stone, A. (1992). The birth of judicial politics in France: The Constitutional Council in com-
parative perspective. New York: Oxford University Press.
Stone, A. (1994). What is a supranational constitution?: An essay in international relations the-
ory. Review of Politics, 56, 441-474.
Stone, A. (1996). The Constitutional Council, constitutional politics, and malaise. In J.T.S. Keeler &
M. Schain (Eds.), New patterns of public policy in France: State responses to social change
and European integration. New York: St. Martin’s.
Stone Sweet, A. (1997). The new GATT: Dispute resolution and the judicialization of the trade
regime. In M. Volcansek (Ed.), Law above nations: Supranational courts and the legaliza-
tion of politics. Gainesville: University of Florida Press.
Stone Sweet, A. (1998). Rules, dispute resolution, and strategic behavior. Reply to Vanberg.
Journal of Theoretical Politics, 10, 327-338.
Stone Sweet, A., & Brunell, T. L. (1998). Constructing a supranational constitution: Dispute
resolution and governance in the European Community. American Political Science Review,
92, 63-81.
Stone Sweet, A., & Caporaso, James A. (1998a). From free trade to supranational polity: The
European Court and integration. In W. Sandholtz & A. Stone Sweet (Eds.), European inte-
gration and supranational governance. Oxford, UK: Oxford University Press.
Stone Sweet, A., & Caporaso, James A. (1998b). La Cour Européenne et l’intégration (The
European Court and integration). Revue Française de Science Politique, 48, 195-244.
Stone Sweet, A., & Sandholtz, W. (1997). European integration and supranational governance.
Journal of European Public Policy, 4, 297-317.
Stumph, H. P., Shapiro, M., Danelski, D. J., Sarat, A., & O’Brien, D. M. (1983). Whither politi-
cal jurisprudence: A symposium. Western Political Quarterly, 36, 533-570.
Sugden, R. (1989). Spontaneous order. Journal of Economic Perspectives, 3, 85-97.
Tate, C. N., & Vallinder, T. (Eds.). (1995). The global expansion of judicial power. New York:
New York University Press.
Taylor, M. (1989). Structure, culture, and action in the explanation of social change. Politics and
Society, 17, 115-188.
Tsebelis, G. (1990). Nested games: Rational choice in comparative politics. Berkeley: Univer-
sity of California Press.
Vanberg, G. (1998a). Abstract judicial review, legislative bargaining, and policy compromise.
Journal of Theoretical Politics, 3, 299-326.
Vanberg, G. (1998b). Reply to Stone Sweet. Journal of Theoretical Politics, 3, 339-346.
Volcansek, Mary L. (Ed.). (1992). Special issue on judicial politics in Western Europe. West
European Politics, 15.
Wendt, A. (1992). Anarchy is what states make of it: The social construction of power politics.
International Organization, 46, 391-426.
Wildavsky, A. (1987). Choosing preferences by constructing institutions: A cultural theory of
preference formation. American Political Science Review, 81, 3-21.

Alec Stone Sweet is official fellow, chair of comparative government, Nuffield College,
Oxford. His research interests include comparative politics, international relations, and
international law. Recent books include European Integration and Supranational Gov-
ernance (Oxford, 1998), coedited with Wayne Sandholtz; The European Court and the
National Courts—Doctrine and Jurisprudence: Legal Change in its Social Context
(Hart, 1998), coedited with Anne-Marie Slaughter and Joseph H.H. Weiler; and Govern-
ing with Judges: Constitutional Politics in Western Europe (Oxford, forthcoming).

Sweet 1999

Uploaded by

Copyright:

Available Formats

Sweet 1999

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

Sweet 1999

Uploaded by

Copyright:

Available Formats

Comparative Political Studies

Judicialization and the Construction of Governance

The online version of this article can be found at:

Email Alerts: http://cps.sagepub.com/cgi/alerts

>> Version of Record - Apr 1, 1999

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

ALEC STONE SWEET

T he triad—two contracting parties and a dispute resolver—constitutes

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

The article proceeds as follows. After introducing key concepts, I present

DYADS, TRIADS, NORMATIVE STRUCTURE

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Figure 1 depicts a simple model of the process by which systems of gov-

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Figure 1. The construction of governance.

Shift 1: Normative Structure to Dyadic Contract

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Shift 2: Dyad to Triad

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Once formed, rules organize how conflict is identified and understood. As

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Figure 2. The dispute resolver’s calculus.

triadic governance is a means of overcoming low levels of trust and weak

Shift 3: The Crisis of Triadic Legitimacy

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Second, the dispute resolver anticipates the disputants’ (or a communi-

Shift 4: Triadic Dispute Resolution and Rule Making

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

rules of an abstract, general, and prospective nature. This is so to the extent

Shift 1: (Re)Constructing the Dyad

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

conflict in a process involving normative reasoning, revising (at least subtly)

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Dyadic and Triadic Governance

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

5. If only to decide not to decide.

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

comprehensive normative structure and (b) interact on an ongoing face-to-

The Lawmaker and the Judge

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

efficient means of revising normative structure. Nonetheless, in legislating,

TDR and Systemic Change

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

terms, triadic governance coordinates the complex relationship between

(RE)CONSTRUCTING THE POLITY

It would be a relatively simple task to demonstrate the general relevance of

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

I mean the capacity of a triadic dispute resolver to authoritatively determine

THE JUDICIALIZATION OF THE INTERNATIONAL TRADE REGIME

When GATT (1948) entered into force and was institutionalized as an

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Normative Structure and Dispute Resolution

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

1948- 1960- 1970- 1980-

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014

Japan, and the United States—dominated panel proceedings; in the 1980s,

Downloaded from cps.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on April 8, 2014