Warlords and Coalition Politics in Post-Soviet States

The breakup of the USSR was unexpected and unexpectedly peaceful. Though
a third of the new states fell prey to violent civil conflict, anarchy on the
post-Soviet periphery, when it occurred, was quickly cauterized. This book
argues that this outcome had nothing to do with security guarantees by Russia
or the United Nations and everything to do with local innovation by ruthless
warlords, who competed and colluded in a high-risk coalition formation
game. Drawing on a structured comparison of Georgian and Tajik militia
members, the book combines rich comparative data with formal modeling,
treating the post-Soviet space as an extraordinary laboratory to observe the
limits of great powers’ efforts to shape domestic institutions in weak states.

Jesse Driscoll is Assistant Professor of Political Science at the School of

International Relations and Pacific Studies, University of California, San
Diego. He was an OCV (Order, Conflict, and Violence) Fellow at Yale
University and a GAGE (Governing America in a Global Era) Fellow at the
University of Virginia, and most recently a member of the Harvard Academy
for International and Area Studies from 2009 to 2013. His work has been
published in the Journal of Conflict Resolution, Research and Politics, and
the Journal of Survey Statistics and Methodology.
Cambridge Studies in Comparative Politics

General Editors
Kathleen Thelen Massachusetts Institute of Technology
Erik Wibbels Duke University

Associate Editors
Robert H. Bates Harvard University
Gary Cox Stanford University
Stephen Hanson The College of William and Mary
Torben Iversen Harvard University
Stathis Kalyvas Yale University
Peter Lange Duke University
Margaret Levi Center for the Advanced Study of the Behavioral Sciences,
Stanford University
Helen Milner Princeton University
Frances Rosenbluth Yale University
Susan Stokes Yale University
Sidney Tarrow Cornell University

Other Books in the Series

Ben W. Ansell, From the Ballot to the Blackboard: The Redistributive
Political Economy of Education
Leonardo R. Arriola, Multi-Ethnic Coalitions in Africa: Business Financing
of Opposition Election Campaigns
David Austen-Smith, Jeffry A. Frieden, Miriam A. Golden, Karl Ove Moene,
and Adam Przeworski, eds., Selected Works of Michael Wallerstein:
The Political Economy of Inequality, Unions, and Social Democracy
Andy Baker, The Market and the Masses in Latin America: Policy Reform
and Consumption in Liberalizing Economies
Lisa Baldez, Why Women Protest: Women’s Movements in Chile
Stefano Bartolini, The Political Mobilization of the European Left,
1860–1980: The Class Cleavage
Robert Bates, When Things Fell Apart: State Failure in Late-Century Africa
Mark Beissinger, Nationalist Mobilization and the Collapse of the Soviet
State
Nancy Bermeo, ed., Unemployment in the New Europe
Carles Boix, Democracy and Redistribution
(continued after the Index)
Warlords and Coalition Politics in
Post-Soviet States

JESSE DRISCOLL
University of California, San Diego
32 Avenue of the Americas, New York, ny 10013-2473, usa

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781107063358
© Jesse Driscoll 2015
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2015
Printed in the United States of America
A catalog record for this publication is available from the British Library.
Library of Congress Cataloging in Publication data
Driscoll, Jesse, 1978–
Warlords and coalition politics in post-Soviet states / Jesse Driscoll.
pages cm. – (Cambridge studies in comparative politics)
isbn 978-1-107-06335-8 (hardback)
1. Warlordism–Georgia (Republic) 2. Warlordism–Tajikistan. 3. Coalition
governments–Georgia (Republic) 4. Coalition governments–Tajikistan. 5. Georgia
(Republic)–Politics and government–1991– 6. Tajikistan–Politics and
government–1991– I. Title.
jz1317.2.d75 2015
320.94758–dc23 2014047366
isbn 978-1-107-06335-8 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
urls for external or third-party Internet Web sites referred to in this publication
and does not guarantee that any content on such Web sites is, or will remain,
accurate or appropriate.
I get it. I know what you want. I understand. You want to make up
a list, on your computer, of all our bad men. ‘Terrorists.’ You want to
cross names off the list when they were killed or jailed. To see that we
Tajiks can take care of our own. But we can. We did. You’ll see.
Yuri, Dushanbe, 2007
Contents

Acknowledgments page xi

1 Revisionist History 1
2 Predator Collusion: A High-Stakes Game 30
3 Kto Kogo? 46
4 Warlord Coalitions and Militia Politics 85
5 Coup-Proofing 123
6 Implications 173

Appendix A: Case Selection and External Validity 197

Appendix B: Mathematical Proofs 207
Appendix C: Ninety-Seven Anonymous Warlords 213
References 221
Index 237

ix
Acknowledgments

None of this would have been conceivable – or be meaningful to me – without

the love and support of my wife Emma; my parents, Mary and Wally; my sister,
Amanda; and a number of friends that are increasingly indistinguishable from
family. (“…We’re all on one road … and we’re only passing through …”)
This book builds upon an ongoing research program that matured while I
pursued graduate studies at Stanford’s Department of Political Science. I owe
a great deal to the patient guidance of David Laitin, James Fearon, and Jeremy
Weinstein, and I owe far more to the friendly assistance of my exceptional
graduate student cohort. But the book also exists because of a full decade
of itinerant labor. During this time I gratefully accepted institutional and
financial support from diverse sources: The Center for Russian and Eastern
European Studies at Stanford University; the Russian Language program at
Middlebury College; the Caucasus Research Resource Centers (CRRC); the
Program on Order, Conflict, and Violence (OCV) at Yale University; the
American University of Central Asia in Bishkek; the International Research
and Exchanges Board (IREX); the Governing America in a Global Age (GAGE)
program at the University of Virginia; and probably a few more that I am
forgetting. Jorge Dominguez and the Harvard Academy for International and
Area Studies at Harvard University deserve special thanks, both for connecting
me with the Cambridge University Press team and giving me time to share my
work with a fantastic network of young interdisciplinary scholars. I am grateful
to all of these institutions for help getting my career started.
Along the way, many scholars shared critical thoughts on primitive drafts of
the manuscript. The argument evolved in response to critiques from Dominique
Arel, Robert Bates, Mark Beissinger, Timothy Blauvelt, Keith Darden, Georgi
Derluguian, Paula Garb, Hans Gutbrod, Steph Haggard, Stephen Hanson,
Miles Kahler, Stathis Kalyvas, Charles King, Bethany Lacina, David Lake,
Jay Lyall, Ghia Nodia, Roger Petersen, Robert Powell, Phil Roeder, John

xi
xii Acknowledgments

Schoeberlein, Yuhki Tajima, Barbara Walter, and Christoph Zuercher. Special

thanks must be extended to James Fearon, who continued to give me advice on
how to streamline and simplify the argument long after professional obligations
expired. The blunt advice of the anonymous reviewers and series editors
at Cambridge University Press is largely responsible for the structure and
intellectual coherence of the final book manuscript.
For most of the data in this book I relied on the goodwill and honesty of
former combatants in Tajikistan and Georgia. The work must be dedicated, in
large part, to Yuri, Vasha, Samojon, Sergei, and others who made introductions,
vouched for my character, and shared difficult memories. None of their real
names appear in this book for human subjects reasons that should be quickly
apparent. Key respondents divulged their stories with the expectation that I
would write a fair and accurate history. This book represents my best effort to
honor their trust. I reserve a different sort of gratitude for the various families
that opened their doors to me along the way – people who were not my human
subjects, but whose humanity inspired me to try to write a good book about
real people. Memories from the field fade unevenly, but I am always going
to wonder what happened to Lira and Jamiliya and Siyl, to Anthony and
Nino, to Elizabeth and Katherine, to Bahron and Mahmuba and Ikron and
Shodmon. Most of my professional colleagues will recognize that the methods
employed in this sort of fieldwork are not well-incentivized by the discipline
of political science. As such, I am in an unusual sort of debt to David Laitin,
whose scholarship and example convinced me that it really was all right, even
in security studies, to lean heavily on the kinds of insights that only emerge
through years of patient ethnography. That unusual debt extends to Stanford
University’s Department of Anthropological Sciences, personified by “Down
Town” Melissa Brown and Kylea Liese. And to Azriel, who was one of the first
to voice her concerns that this book was a thing of evil. (Mine!)
I have an enviable first job. The University of California at San Diego is
a wonderful employer. Steph Haggard, Peter Gourevitch, Miles Kahler, and
Barbara Walter have all been tremendous professional mentors in different
ways. All empowered me, with their scholarship and their gentle advice, to
write exactly the book that I did. I have also had the opportunity to teach
some of the world’s brightest students – some of whom I employed as the book
moved toward the final stages of completion. Elaine Denny provided useful
feedback on an early draft of the material related to human subjects and issues
of anonymity. Karen Simpson and Megan Becker assisted in the preparation
of this manuscript with careful copyediting. Daniel Maliniak created original
maps, helped with a few of the figures, and continues to keep me honest with
the data. John Porten organized the index.
I wouldn’t have made it to Stanford at all if I hadn’t met Charles King and
Michael Brown just exactly when I did as an undergraduate at Georgetown
University. If I wind the clock back further, I find that it is important that I
use this space to thank Penny Johnston and Mark “Mad Dog” Bowman for
Acknowledgments xiii

different kinds of virtuous labor as underpaid public high school teachers at

Mead High School in Spokane, Washington.
But past is past, and I’m working to be here now. My wife Emma was the
first person thanked, but she insisted that she be thanked again at the end, as
well. It isn’t very much fun to live with someone whose work involves writing
self-serious paragraphs about the representation of distant people in war zones.
Emma, and our wonderful dog Sally, help to remind me that even when I’m
most proud of my work, and even when I convince myself that my work is
important, the truth is my work isn’t my life. It’s just work.
1

Revisionist History

As many as 500,000 people lost their lives in the wake of the Soviet
experiment. Civil wars were fought in Azerbaijan, Bosnia, Croatia, Georgia,
Kosovo, Moldova, the North Caucasus, Romania, and Tajikistan. Though
the thought experiment requires a grisly kind of arithmetic, social scientists
can assert with confidence that longer civil wars likely would have resulted in
many more deaths. How did order consolidate so quickly in the post-Soviet
space?
This book presents a host of new data and original game theory to revisit
the basic intuition of Thomas Hobbes (1651): anarchy creates strong incentives
for people to build states. I demonstrate that political order arose out of violent
anarchy because violence entrepreneurs – warlords hereafter – realized that the
great powers would pay handsomely for local order. Order facilitates efficient
markets (for foreign investors) and local-language intelligence collection (for
foreign militaries). Warlords understood that they were in a position to extort
certain rents of sovereignty from the international system and wanted to be
bought out in the scramble that followed the collapse of the USSR. The ancient
truism that “war is bad for business” was quickly grasped by certain individuals
who realized that they were in a rare position to extort civilian governments
directly – and the international community indirectly – with anarchy. Foot
soldiers were recruited from the sub-proletarian underclass through promises
of future state spoils. Some warlords initially colluded to provide order, access
international wealth, and allocate themselves monopoly rents from the state
apparatus that fell under their control. A local puppet president served as
a placeholder for opaque coalition politics. Many warlords became violence
subcontractors for the regime. Some did not. Complicated bargaining followed.
Back-room deals were struck. A great deal of property changed hands. Peace
emerged as local criminals developed techniques to hold civilians hostage and

1
2 Jesse Driscoll

rewrite local history to their advantage. In other words, the warlords became
the state.

post-soviet settlement patterns: successful

third-party interventions?
It is clear that helping people to build states in the wake of civil disorder will
be a core foreign policy challenge for the United States and the United Nations
for the foreseeable future. The threat of catastrophic terrorism has changed the
terms of the debate about national security, increasing the emphasis on threats
that can emerge from very weak actors in the international system. Events that
transpire in the global periphery can directly threaten the safety of voters living
in states shielded by strong professional militaries and oceans.
This book is motivated by a question rarely given voice by Western diplomats
or academics: What did the Russians get right in their relations with their
new periphery? After all, wars that broke out during the Soviet collapse were
shorter – and thus far less bloody – than similar civil wars emerging from
decolonization. The average length of post-Communist wars is only 3.9 years,
compared to an average length of 9.8 years for all civil wars since 1945.1
Violence in the wake of the Communist collapse was brutal, but drawn-out
insurgencies would have produced many more deaths. Appendix A at the end
of this volume shows that even with numerous statistical controls, the subset
of civil wars resulting from the breakup of the USSR was a group of unusual
outliers in terms of overall duration.
Is this unusual regional trend attributable to local politics and
path-dependent institutional history? Or is it attributable to the successful
actions of a third-party intervention force, with Russia acting as the lead state?
Reasonable people disagree.2 Each cell of Figure 1.1 is meant to represent
a different internally consistent narrative of how civil wars remain settled,
including different assumptions about the possible role of third-party foreign
assistance in shaping peace processes.3 The horizontal dimension is the assumed

1 Nearly a dozen individual states housed civil wars that each surpassed a half-million deaths –
Afghanistan, Angola, Cambodia, China, the Democratic Republic of the Congo, Ethiopia,
Greece, Lebanon, Mozambique, Uganda, and Vietnam. These conflicts lasted 16.5 years on
average. Lacina (2006) demonstrates that the length of a civil war is a robust predictor of the
overall battle deaths.
2 This disagreement may be no one’s fault. Military professionals, diplomats, missionaries,
and development assistance professionals often hold different root assumptions about what
it takes for a war to stay resolved. Theoretical assumptions inevitably leak into descriptions. By
selectively omitting deviant facts, different narratives can be fit to the same observations.
3 Much of the data produced from humanitarian disaster zones chronicle the critical role
being played by the intervention force, confirming the need for ongoing foreign assistance.
Heathershaw (2007) argues that in authoritarian regimes recovering from civil conflict, social
scientists often become complicit in this interpretative exercise. See also Heathershaw (2008,
2009).
Revisionist History 3

Decisive Military Stalemate Between

Victory Factions
Possibility
of Third-Party
[Postmodern] Liberal
Monitoring/
Imperialism Interventionism
Enforcement

No Possibility
of Third-Party Realism Militia Coalition Politics
Monitoring/
Enforcement

figure 1.1. Disputed Narratives: What mechanisms keep civil war settled?

military balance between the incumbent and insurgent armies at the time of
settlement. The vertical dimension is the assumed ability of foreign powers to
monitor and enforce outcomes relevant to the settlement.
In the lower left corner of Figure 1.1 one finds most self-styled “realists.”
They maintain that the central mechanism that keeps civil wars settled
is military hegemony by a sovereign authority within recognized interstate
borders. This would have been called “the king’s peace” in prior eras. Probably
the most famous account of how states emerge from civil war comes from
seventeenth century philosopher Thomas Hobbes (1651), who articulates a
straightforward case for peace through military conquest by the agents of a
sovereign. The social contract, for Hobbes, is imposed. Citizens are made
subordinate to the ruler violently, opponents are disarmed, and order emerges.4
A strong state apparatus is the best inoculation against civil war.5
A number of independent research programs – most prominently those of
Licklider (1995) and Fearon (2004) – have confirmed that, since 1945, the most
stable civil war settlements are those that end with military victory. The military
contest often takes a long time – approximately a decade on average.6 Many
“negotiated settlements” are face-saving arrangements that codify the de facto

4 This interpretation of Hobbes (1651) draws heavily on the synthesis of the realist canon in
Wagner (2007). On pages 126–127, Wagner references the central argument in Fearon and
Laitin (1996) to suggest the evolution of the current boundaries of nation-states as “natural”
responses to differential comparative advantages in counterinsurgency by different language
speakers.
5 For compelling evidence that state weakness is statistically correlated with the outbreak of
large-scale civil violence, see Fearon and Laitin (2003).
6 Military victory, when it comes, rarely requires comprehensively and decisively defeating a
conventional rebel army on the battlefield. Much more common is selective co-optation of
insurgent field commanders during the closing phases of asymmetric irregular war. As such,
the coalition of social forces that constitutes “the state” changes from the beginning of the
conflict to its end.
4 Jesse Driscoll

military balance.7 Toft (2010) forcefully argues that deterrence – fear of a

technologically enabled security sector – is the mechanism that is most likely
to be responsible for post–civil war peace.8 National veterans are the heroic
actors, and the story of their decisive victory is passed on from one generation
to the next in monuments, museums, and military academies.
Though the collapse of the Soviet Union has been described by economist
Douglass North as “perhaps the most striking case of internally induced
rapid demise in all of human history,” even the most disadvantaged states
on the Soviet periphery were “born strong” in important respects, inheriting
huge institutional advantages compared to the postcolonial states of Africa
that achieved independence in the 1960s and 1970s.9 Although the Eurasian
states that joined the United Nations in the early 1990s lagged behind in
terms of per capita gross domestic product (GDP) – difficult to measure in
states transitioning to market systems – decades of institutional development
translated into huge advantages when it came time for post-Soviet populations
to improvise the construction of war machines. There can be no doubt that this
is part of the explanation for the quick resolution of the post-Soviet wars. The
Soviet experience bequeathed to the first generation of post-independence elites
a well-organized party network, borders and administrative units, a centralized
media distribution system, a secret police apparatus (with transnational
linkages to other republics-turned-states), a national language, an official
history (institutionalized with maps, censuses, and museums), as well as an
educated and largely literate population that anticipated that these institutions
would endure.10
Completely different assumptions and mechanisms support a more benign
“liberal intervention” narrative of civil war termination, located in the top right
quadrant of Figure 1.1. Especially since the end of the Cold War, policymakers
have sought solutions to military stalemates that do not rely on grinding
military attrition. An empirical research agenda demonstrating the efficacy of

7 King (1997), Fearon and Laitin (2007), and McCormick, Horton, Harrison (2007) propose
moderate policies based on this insight. Luttwak (1999) is also consistent with this line of
reasoning.
8 I have located Toft’s scholarship in the “realist” camp for the purpose of this chapter because
it is clear that she sees herself in opposition to liberal voices (e.g., those in the upper-right
quadrant of Figure 1.1). With that said, in my reading Toft is equivocal on the role of
foreign governments; she does admit a limited role for foreign governments in promoting
“security sector reforms” during the implementation phase of postwar peace processes. She
is vexed that U.S. threats to intervene militarily to facilitate decisive victories lack credibility
(160–162), so perhaps she would prefer to be identified with the “postmodern imperialists” in
the upper-left-hand corner.
9 North (2006), 4. A broader exposition of his views on the dissolution of the Soviet experiment
can be found on pages 146–154 of the same volume.
10 For a good introductory overview of the nature and sources of Soviet institutional advantages
in producing compliant behaviors in the rural periphery, see Roeder (1993); Jones-Luong
(2002), chapter 1; and Brown (2007).
Revisionist History 5

third-party interventions to end conflict has coevolved with the expansion of

United Nations Peacekeeping Operations (UNPKOs) in the last two decades.
Remarkable scientific progress has been made on the question of whether and
how outsiders can assist in ending civil wars. They can, and we know quite a bit
about how they can. Peacekeepers can provide security guarantees that allow
for disarmament (e.g., Walter 1997, 2002), provide neutral monitoring of the
terms of agreement (e.g., Fortna 2008), marginalize holdouts (e.g., Stedman
1997), and gradually establish trust between warring parties via multifaceted
mediation programs (e.g., Doyle 2006). In the quarter-century since the end
of the Cold War, there has been an audience for arguments explaining how
a benign international gendarmerie might help establish order. The United
Nations – assisted by a plethora of international organizations, social scientists,
and private actors – has established many peacekeeping missions. Much of the
foreign aid that reaches post–civil war societies is directed towards paying the
salaries of, and meeting program goals drawn up by, liberal interventionists.
Most of the rotating class of Americans and Western Europeans who staff
embassies and the offices of aid organizations housed in the capital cities of the
post-Soviet republics imagine themselves to be day-by-day peace builders. Their
reports state plainly why they believe their programs are vital to the persistence
of peace.
It is common for realists to caricature the arguments of liberal intervention-
ists as utopian.11 A few contemporary liberal interventionists have met this
critique head-on, acknowledging that identities and interests do not need to be
fundamentally transformed for a civil war to end. War ends because actors
with the capacity to undermine order-providing institutions with violence
come to believe that it is not in their best interest to sabotage order. But
when locals are stuck in a costly stalemate, neutral foreigners can sometimes
save lives by helping the warring sides extract themselves from pointless
attrition. An influential deductive approach to civil war settlement over the
last decade argues that foreign interveners can shape the postwar institutions
by altering players’ strategies without altering their underlying preferences. The
transformation of conflict identities may be important in the long run, but it
is often the work of many generations and is not necessarily relevant to the
contours of war termination.
A more urgent task, as Barbara Walter (1997) has argued, is the creation
of a secure framework to ensure rebel disarmament. Negotiators attempting
to end a civil war grapple with different challenges than diplomats negotiating

11 To the extent that liberals’ optimism relies on the gradual transformation of identities or
the alleviation of deep grievances, these charges are deserved. Much of the programming
of humanitarian relief agencies has a striking resemblance to missionary work. Most of the
professional bureaucrats who serve as a rotating middle class, drifting across the world’s war
zones, are motivated by a desire to assist in transnational and transhistorical processes of social
transformation.
6 Jesse Driscoll

an armistice after interstate war. At the end of an interstate war, both armies
remain intact and can retreat behind internationally reified boundaries. Ending
a civil war, it is argued, requires that one side or the other formally lay down its
arms. The winners – who will then control all the guns – have a very difficult
time making their commitments to honor the terms of the ceasefire credible.12
This approach to the problem of civil war suggests that credible third-party
security guarantees, and subsequent monitoring, can help sculpt peace accords
that would otherwise crumble under the weight of the security dilemma.
But once the possibility of third-party intervention to sculpt war outcomes
is considered, it also becomes necessary to consider the upper left corner of
Figure 1.1. The threat of transnational mass-casualty terrorism changed the
conversation about involvement in other people’s civil wars. Certain weak
states, once peripheral to American interests, are now treated by great powers as
potential security concerns, rather than just troubling manmade humanitarian
disasters. The situation is new, and its implications poorly understood, but
Western governments grasp that stabilizing weak states is not simply about the
humanitarian mission of saving lives – it is also about self-protection. And in
this new world, the same constituencies who would balk at their tax dollars
ending up in unsavory pockets can be blackmailed into tolerating autocratic
corruption. Violence against human dignity is weighed against the risk of
ideologically hostile regimes emerging from pockets of anarchy in the Middle
East and Central Asia. For certain autocrats, the claim to be “too weak” to
control one’s territory can, perversely, bring more foreign aid in the service
of decisively defeating terrorists. Much of this extortion dynamic depends
on variables that are imagined or kinds of intelligence that are intrinsically
suspect.13

12 This commitment problem complicates the diplomatic resolution of civil wars through many
mechanisms. It is thought to render stable postwar power sharing extremely difficult. Fearon
and Laitin summarize the core of this asymmetric commitment problem: “Rebel groups aim
at regime change because they could not trust the government to implement the policies they
desire even if the government formally agreed to do so. After the rebel group disbands, or after
the central government regains strength, or because of monitoring problems arising from the
nature of the policy aims (for example, redistribution), the central government would renege
on policy concessions it made to end a war. Thus rebel groups must often fight for ‘all or
nothing.’” Fearon and Laitin (2007), 2. For a review of theoretical and formal literature on
commitment problems in civil war, see Blattman and Miguel (2009) and Walter (2009).
13 The word “imagined” is perhaps too provocative, giving the impression that national
interests are completely constructed. States are constructed as strategic allies partially as a
product of their geographic location vis-à-vis perceived enemies, partially based on objective
characteristics of a country (e.g., the presence of oil, democratic institutions, nuclear weapons,
military bases, diaspora linkages, or density of ideologically radical subpopulations), and
partially a figment of political practice. See Gourevitch (1978). During the Cold War,
post-revolutionary leaders could install Communist Party Structures (Single Party Regimes)
and count on some aid from the USSR. Today there is little doubt that democracies in
strategically important neighborhoods – Israel, the Philippines, Taiwan, and most recently,
Georgia – have been able to attract bilateral aid from the United States by a similar logic.
Revisionist History 7

Direct military interventions into other states’ civil wars to shape the
contours of settlement, facilitating decisive victory for one faction or another,
will be familiar to students of imperial history. The mechanisms tend to
emphasize sinister kinds of meddling: sharing signals intelligence and military
satellite information, providing sophisticated weapons, liquidating potential
spoilers, picking winners, picking losers. The end of the Cold War; the
demonstrated ability of weak actors to cause great damage to the interests of
strong states; the spread of new technology; and the growing consensus by elites
in Russia, China, the United States, France, and Great Britain – the permanent
five members of the UN Security Council – that their security interests are tied
up in the outcomes of civil wars fought in weak or failed states are combining to
facilitate the emergence of a new kind of “Post-modern Imperialism,” according
to Fearon and Laitin (2004).14 These behaviors are distinguished from classical
imperialism in that the intervener acts on behalf of the entire state system,
and does not want to stay in the territory – the intervention force wants to
go home as soon as possible, but to do so it must leave a stable partner
government in charge of the territory. The kinds of policies that result are not
always compatible with the idealized prescriptions championed by the liberal
interventionists. But great powers do, if only rarely, find it is in their national
interest to guarantee decisive victory for one side or the other in someone else’s
civil war.
Consider the two maps in Figures 1.2 and 1.3. If one doubts that Russian
military power was decisive in shaping the contours of military settlements in
the early 1990s, one has only to notice the persistence of breakaway regions
in Georgia, in clear contrast to the territorially intact map of Tajikistan.
The “frozen conflicts” inside territory claimed by Georgia pit the national
government in Tbilisi against Russian-backed secessionist statelets in Abkhazia
and South Ossetia. To the question “Why was the map redrawn in the South
Caucasus and left intact in Central Asia?” one can do worse than answer with
the crude observation that “Political elites in Russia just wanted it that way.”
As discussed at length in the chapters that follow, Russian peacekeeping – or
“peacemaking” as the word mirotvorchestvo is more accurately translated –
was never meant to facilitate general disarmament. Russian troops – sometimes
still in familiar Soviet uniforms, and sometimes wearing black ski masks –
and paramilitary units from neighboring republics (the North Caucasus) and
states (Uzbekistan) rallied across new interstate borders.15 In Georgia, borders

14 The authors identify four general challenges for peacekeeping missions sent after humanitarian
disasters break out in badly governed parts of the world: 1) recruitment (“who sends troops?”),
2) coordination (“who acts as the ‘lead state,’ taking responsibility for critical tasks of
coordination?”), 3) accountability (“what happens if peacekeepers are not neutral?”), and
4) exit (“at what point can the intervention terminate?”). As we shall see, these questions had
unusually clear answers in the post-Soviet wars: 1) the CIS 2) Russia 3) nothing 4) maybe
never. Locals were not tempted to try to “wait out” the Russian military force.
15 King (2000) and Derluguian (2005), 262–273.
8 Jesse Driscoll

Legend
Kazakhstan Mountains
Tashkent Country Border
Andizjan
Regional Border
(Khojand)
Kilometers
Khojand
Uzbekistan 0 50 100 200
Uroteppa
Samarkand
Kyrgyzstan
China
Panjakent Leninobod
Gharm
(Gharm)
Central Territories
Dushanbe
Kofarnihon
Kalikhum
Gorno-Badakhshon (GBAO)
Khatlon Merghob
Kalininobod Kulob Alichur
Kurgon-Tubbe (Kulob)
Khorog

Afghanistan
Pakistan

figure 1.2. Tajikistan.

were essentially redrawn by Russia to coerce elites in Tbilisi to join the

new Commonwealth of Independent States (CIS). In Tajikistan, after initial
lukewarm support for Khojandi secessionists, it was decided in Moscow that
Tajikistan would stay intact as a buffer state to shield the rest of inner Asia
from chaos in Afghanistan. It was difficult to misinterpret Boris Yeltsin’s 1993
statement that the Tajik–Afghan border was “in effect, Russia’s.”16
These mechanisms hint at another partial explanation for the unusually
short length of the post-Soviet wars. These wars did not develop into “proxy
wars,” with different regional powers backing different clients. The violence
of the post-Soviet wars occurred in the periphery of the USSR, and against
a backdrop of total state failure that was no less revolutionary for being
largely nonviolent.17 Other great powers calibrated their foreign policies to
give Russia, a stumbling nuclear superpower, a wide berth. Potential external

16 On Tajikistan, Fearon and Laitin report: “Russian peacekeepers were able and willing to (in
the words of several informants) ‘liquidate’ spoilers. They were able, as in Tajikistan, to pick
a warlord favorable to them and provide him the military support necessary to compel other
pretenders into negotiations.” Fearon and Laitin (2004), 27. Footnote 56 is informative as
well.
17 Our social science theories are simply not up to the task of task of explaining the contingencies
of revolutionary politics, even when the stakes are very high. No one can state with confidence
why it was that Boris Yeltsin emerged standing on top of the tank instead of an aggressive
military populist.
Revisionist History 9

Russian Federation
Gagra

Gudauta Abkhazia
Sukhumi Svanetia
Tkvarcheli

Ochamchira
Gali
Legend Mingrelia Gori
Zugdidi
Mountains South Ossetia
Country Borders Kutaisi Tskhinvali
Poti
Regional Border Samtredia Dushet'i
Breakaway Regions Ozurgeti
Telavi
Mtskheta Lagodekhi
Kilometers Adjara
0 25 50 100 Batumi
Tbilisi Tsnori
Akhaltsikh
Rustavi

Turkey Armenia Azerbaijan

figure 1.3. Georgia.

interveners – Iran, Turkey, the United States, or China – might have been able
to find security, economic, or humanitarian justifications to expand their role
in the South Caucasus or Central Asia, setting off a competitive scramble to
affect outcomes. Proxy war dynamics might have ensued. This did not happen,
however, in large part because possible gains were weighed against the risk that
perceived interference into Russia’s traditional zone of control would empower
hard-liners in Moscow. Nothing in Georgia, Tajikistan, or Chechnya could be
worth trading Boris Yeltsin for Alexander Lebed. Georgians and Tajiks under-
stood that foreign governments would be second-guessing Moscow’s prefer-
ences while improvising policy in Russia’s traditional sphere of influence.18
By the late 1990s, when Russia was more of a “normal country,” the
impulse for Western governments to tinker with institutions under the aegis of
democracy promotion and meddle in the security affairs of post-Soviet states

18 MacFarlane (1999) provides a sympathetic summary of how Russia’s geopolitical predicament

was viewed at the time. The message that the United States’ central foreign policy priority
vis-à-vis Russia was stable control of nuclear materials by a friendly government in Moscow –
and that every other interest would be subordinated to that goal – was signaled clearly,
early, and often. See Allison and Blackwill (1991), 90–91; Nunn (1991); Kennan (1997);
and Kubicek (1999–2000), 548–549, and the retrospective summary of Cornell (2001), 367.
Despite some pressure from Congress, the only thing that pushed the Clinton Administration
to even entertain discussion of military intervention in Russia’s war in Chechnya was the (false)
claim that Dzhokhar Dudayev had acquired a nuclear bomb. See Goldgeier and McFaul (2003),
139–149.
10 Jesse Driscoll

reasserted itself.19 The results are particularly visible in contemporary Georgia.

But one can safely generalize that it was not simply a strong outside intervener
tipping the military balance that cauterized the wars that broke out as the USSR
collapsed – it was the absence of multiple competing interveners. So if one is
tempted to fall back on the crude realist observation that “Russia just wanted
it that way,” for completeness one should add “and it matters quite a bit that
just Russia wanted it that way.”
All three of the perspectives sketched earlier – the realist, the liberal interven-
tionist, and the postmodern imperialist perspective – contain important insights
into the stable settlement patterns of the post-Soviet wars. A realist approach
emphasizes the Soviet institutional legacy – clearly visible in the de facto and
de jure borders of the recognized and unrecognized states of the region. New
leaders inherited unusually strong coercive machinery, facilitating the rapid
creation of sovereign domestic authority structures.20 A liberal, apologetic
for Russia’s CIS interventions into its periphery, would emphasize Russia’s
role in protecting embattled minorities and trying to assist as a third-party
peace mediator in a humanitarian capacity. It is a political necessity that any
document that passes through the Security Council of the United Nations be
calibrated to reflect that narrative.21 Many harbor conspiracy theories – very
difficult to falsify – that Russia’s true role was more sinister: strengthening
security forces in some places and sabotaging them in others, sharing signals
intelligence and military satellite information selectively with clients, and
targeting certain individuals for assassination. The provocative phrasing
“postmodern imperialism” is meant to emphasize that Russia was not driven
by desire for imperial glory, but was executing the will of many foreign actors,
all colluding to ensure a decisive outcome in the name of international security.
But my book sustains a different narrative, emphasizing processes in the
lower right corner of Figure 1.1. I provide evidence that in Georgia and Tajik-
istan local militias never disarmed, and that local elites never really followed
through on the liberal development script. They instead improvised creative
solutions that were self-enforcing. “The state” described here is analagous to
a semipermeable membrane for violence entrepreneurs – “warlords” – who
weigh their life opportunities as social bandits against their life opportunities
as agents of an internationally recognized sovereign. The utility of such an
approach would be measured by its ability to generate predictions about when
warlords are likely to yoke themselves to the governing coalition and when
they are likely to resist consolidation. But if it is the case that incumbent and
insurgent militia commanders were potentially interchangeable candidates for

19 I borrow this phrasing from Shleifer (2005).

20 Roeder (2007) suggests that some aspects of this machinery generate momentum of their own,
creating demands for sovereignty that ended the federal experiment of the USSR.
21 For an outstanding overview to the arguments made by the Russian state to justify its
interventions, see O’Prey (1996).
Revisionist History 11

postwar government spoils, then from the point of view of field commanders,
civil war settlement was a coalition formation process.
This study defines “the state” somewhat pragmatically, as a bundle of formal
and informal contracts regulating and incentivizing the behavior of citizens
within a particular geographic space (ideally an internationally recognized
territory). Shared norms exist that govern enforcement power for contracts,
but I believe it is appropriate to consider the point of view of the violence
entrepreneurs who compete for the right to selectively enforce those contracts
to their benefit.22 Great power interests and institutional inheritance set the
chessboard, but, viewed from the point of view of individual warlords, the
nature of the spoils being contested made “robust third-party monitoring”
impossible. The micro-contours of post-independence redistribution politics
hinged on the particulars of land reform, privatization in opaque and
closed-bid environments, complex credit swaps cemented via marriages, money
laundering across borders, and asset price-fixing schemes designed to pass debt
on to the next generation. Profiting from these kinds of dealings required
informal understandings between warlords – and the argument in this book
is that those informal understandings, combined with certain inherited state
institutions, ultimately provided the orderly structure that foreigners later tried
to “shore up” with aid and assistance. Warlords understood perfectly that great
powers could not send much money or guarantee regime support until after
strong local clients had emerged, via a violent process of sorting and attrition
on the streets. They bargained among each other and ultimately backed the
ascendency of a civilian figurehead. The president rehabilitated warlords and
granted them ministry positions in exchange for security. The consolidation
process that followed had characteristics of a high-stakes lottery. Unlucky
warlords eventually found themselves isolated and liquidated when the winds
of political fortune turned. Lucky warlords were allowed to disappear into the
police and military bureaucracy, where many still reside.

coalition formation by interchangeable predators

The violent conflicts that emerged out of the collapse of the Soviet Union were
often extensions of nonviolent property disputes. These property disputes were
often between members of the same ethnic group.23 All players understood that

22 It is plausible to me that the norm of territorial sovereignty – now enshrined in international

law and constantly reified in contemporary discourse – evolved as a technology for strong states
to efficiently enforce a “division of responsibility that diminishes the number of occasions for
violent conflict between or among states.” Wagner (2007), 126. See also Chapters 3, 4, and
6 generally. On the closely related question of why the “nation state” (as opposed to simply
“the state”) has evolved as the primary unit of political order in the modern world system, see
Anderson (1991) and Fearon and Laitin (1996). A useful genealogy the word “sovereignty” is
found in Krasner (2001), especially 184–219.
23 See Handelman (1995), Varese (2001), and Reno (2002).
12 Jesse Driscoll

local solutions were necessary. The story of how warlords arrived at these local
solutions involved highly contingent local brokering, bargaining, and coalition
formation, with a great deal of side-switching and ambiguity of alliances.
Russia’s geopolitical strength loomed in the background of local strategizing,
as a structural constraint to be maneuvered around and the engineer of critical
model parameters. But in this account Russia was not a “peace enforcer” in the
sense that it did not monitor or enforce the bargains that were necessary to keep
peace. This book’s account emphasizes the agency of locals, who ultimately
selected and maintained order-producing equilibria. It also emphasizes the
fundamental interchangeability of warlords as providers of local order.
The nonviolent equilibrium that ultimately emerged in Georgia and Tajik-
istan did not require disarmament. Indeed, as the far right column in Table 1.1
summarizes, the processes that are described in this book did not require many
of the steps that are often presented by liberals as prerequisites to lasting peace.
The risk of war decreased even while the postwar states developed in a manner
that is thought to exacerbate the long-term risk of renewed violence: autocratic
rule centered on a strong president, statelets dependent on external protection,
and the normalization of corrupt and patronage-ridden governance. Political
order was produced by a process that had little to do with the cognitively
appealing script of democratization, legitimacy, or the rule of law. I anticipate
that the account of militia politics sketched in the third column will be utterly
uncontroversial to Georgian and Tajik citizens who lived through the war and
its aftermath.24 Yet the account will run counter to much of the official history
of these two states’ civil war settlements. There will be little discussion of party
reform or rehabilitation of professional security services. I do not describe
attempts to alleviate grievances or heal the rifts between social forces, nor do I
chronicle attempts by third-party mediators to facilitate roundtables for patient
negotiations between elites. The kinds of things that would count as success
stories for most Western conflict resolution professionals are almost completely
absent from this book’s account.
This book explores the possibility that international mediation and multilat-
eral aid play a critical role in the settlement – but not the role that development
professionals believe they are playing. In this book’s account, the central
mechanism of civil war settlement is bribery. Some warlords collude together to
create a state, hoping to secure foreign aid and investment. A clear prediction
emerges: as the pool of potential bribes increases, the number of warlords who
opt into the state will increase. Warlords keep their armies intact so that they
can, in principle, extract rents through the credible threat of a coup. Foreign
aid professionals can have a positive effect on war settlement not despite but
because of the fact that these funds can be stolen by violence entrepreneurs.
What is important is the liquidity provided by the influx of the donor economy,

24 If there is anything controversial, I anticipate it will be framing the post-Soviet wars in a

way that emphasizes actors’ material interests at the expense of ethnic identities or ideological
beliefs.
Revisionist History 13

table 1.1. Contested Mechanisms of State-Building after Civil War

Proposed Process Alternative Process

(State-Building Canon) (Militia Politics)
PROBLEM #1: Disarming and Militias consolidate
“Militia disbanding militias themselves in response
Demobilization” through the to strong incentives to
If militias have more rehabilitation of eliminate imitators and
guns than the police or security forces; control entry into the
army, how does Liquidating “spoilers” violence market.
post–civil war with foreign assistance. Process gets more
disarmament work? violent as foreign
assistance is imminent,
as foot soldiers
understand that the
stakes are very high.
PROBLEM #2: Institutionalization “Trust” relationships
“Coup-Proofing” and formalization of between presidents
How does a civilian robust political and armed groups,
president get power arrangements – ideally based on promises of
away from the armed “legitimized” by future aid. Viability of
gangsters who widespread democratic bargains depends on
installed him? participation. international
conditions and ability
of president to
successfully manage
divide-and-rule.
PROBLEM #3 Conditional foreign A red herring. If
“Positive Sovereignty” aid; building civil bilateral or multilateral
How does a weak society, standard aid is sufficient to pay
national government “Development bribes and secure
commit to rehabilitate Economics Toolkit” regime without taxing
institutions and from international citizenry, there is no
provide public goods? development agencies, reason to expect
technical help with institutional reform.
counterinsurgency. Politics are about
coup-proofing and
managing geopolitical
narrative to ensure aid
flows – not providing
public goods.
14 Jesse Driscoll

which is then diverted. The content of third-party aid programs may not be
relevant, even though the presence of third-party actors is.
This account emphasizes the ability of domestic actors in a civil war zone
to anticipate and frustrate the desires of foreign development professionals
to make politics open and transparent. If the main axis of politics hinges
on warlords’ ability to coordinate and credibly threaten to restart the civil
war and the president’s use of divide-and-rule tactics to prevent this outcome,
it is not obviously in anyone’s interest for promises to be codified in
writing, or channeled through transparent formal institutions. An autocratic,
patronage-based favor economy is central to the provision of domestic order,
where wealth is secured through selective enforcement of property rights and
certain warlords are permitted to loot the state from within.
Though the paired cases in this study are outliers in certain respects, the
kinds of arrangements observed in Georgia and Tajikistan are representative
of civil war settlements generally. It is inconvenient for the humanitarian aid
complex to openly acknowledge the fact that most states that fought civil wars
since 1945 emerged from the violence as dictatorships. Only a quarter of the
settled civil war observations in the Doyle and Sambanis (2006) dataset are
characterizeable as democratic (combined Polity Score of 6 or higher) five years
after civil war.25 Postwar authoritarian institutions matter – and in Appendix
A, statistical evidence will be presented that shows that building parties after
civil wars seems to be a relatively popular strategy for keeping the war from
breaking back out again.26 In the post–Cold War era, the modal form of
governance after civil war is a “hybrid regime” – a semi-authoritarian regime
that holds elections.27 This book emphasizes that governance reform is partially
theatrical: regimes make cosmetic efforts towards democracy and reforms to
receive multilateral aid, and then use this aid to pay off criminal factions who
keep violence off-camera.
An approach that views the state from the bottom up, as an emergent
equilibrium of joint warlord strategies, illuminates aspects of war termination
that are often obscured. In this book’s account, foreign governments can use
their financial, diplomatic, and military tools to shape civil war settlements

25 On the standard combined −10 to 10 scale, modal and median polity scores five years
after the end of the civil war are −7 and −2, respectively. Given that governments that can
insulate themselves from coups are more likely to sustain the transition to nonviolent politics,
scholars and practitioners alike will be drawn into the business of assisting with coup-proofing
weak states after civil war. Fully functioning democracies are obviously the best long-term
inoculation against the threat of a coup – and Wantchekon (2004) provides good reasons for
this to be so – but these are relatively rare outcomes in the wake of civil violence.
26 Girod (2012) provides evidence that elites in the world’s weakest and most vulnerable states
are fully capable of anticipating the kinds of institutions that foreigners expect to see, and, since
the end of the Cold War, can be incentivized by multilateral aid and low-interest international
loans to build efficacious political institutions.
27 For variations on this argument, see Levitsky and Way (2005), Pevehouse (2005), Carothers
(2006), and Goldsmith (2008).
Revisionist History 15

indirectly by altering certain parameters – but the strategic choices that create
the equilibrium are made by people who will continue to live in the war zone
after the foreigners depart. The coalition-formation approach has a number
of observable implications that differ from those of traditional two-player
bargaining models. Instead of treating the armed strength of a rebel challenger
as an exogenous model parameter, this approach seeks to explain variation in
the relative size and strength of the incumbent and insurgent coalitions. The
central strategic choice in the game presented in this book is whether a warlord
will choose to contribute his armed forces to the emerging warlord coalition
(the state) or remain outside this coalition (as a rebel).
A coalition-formation approach can shed light on contemporary counterin-
surgency practices. Though in principle a large number of rebel “spoilers” may
complicate negotiations to end a civil war, in practice much of what militaries
actually do in the name of counterinsurgency is obviously aimed at sowing
rebel fragmentation so that the state can “pick up the pieces.”28 Embattled
governments often create institutions to facilitate side-switching by insurgents
as irregular wars grind to a slow finish. Incorporating enemy combatants is
a triple-win for the incumbent regime, as it simultaneously reduces the ranks
of the enemy, provides an “exit option” for demoralized rebels who do not
want to keep fighting, and brings hardened soldiers with local knowledge onto
their side. It also mitigates the commitment problem that is so vexing to liberal
interventionists, as individual rebels simply do not disarm – they keep their
weapons, hedge their bets, and gradually transmogrify into state agents. These
processes of defection, desertion, and incorporation are rarely placed at the
center of strategic analysis by counterinsurgency professionals. This book is
meant as a partial corrective to that trend, emphasizing that extortion and
bribery were critical mechanisms in settling the post-Soviet wars.

empirical methodology
This volume represents my best attempt to synthesize the narratives of many
people who experienced the wars of independence in Georgia and Tajikistan

28 On spoilers, see Stedman (1997) and Cunningham (2006). On ruthless counterinsurgency

tactics aimed at sowing fragmentation, see Kalyvas (2006) and Davenport (2007). Examples
of side-switching between incumbents and insurgents abound. More than 5,000 former rebel
fighters from the Moro National Liberation Front in the Philippines were trained, armed, and
assigned to active units in the Armed Forces of the Philippines in the late 1990s, according
to Felter (2005). The 2006 “Anbar Awakening” in Iraq would have been impossible if former
insurgents were not permitted to reinvent themselves and find employment in the Iraqi Security
Forces, as Anderson (2007) and Ucko (2009) report. Tarzi (2005), Jalali (2006), Rashid
(2008), and Giustozzi (2009), all make it clear that the United States’ strategy for post-Taliban
Afghanistan involved incorporating prominent warlords, many of whom have found their way
into the army, police, private security companies, or into opaque relationships with the Karzai
government.
16 Jesse Driscoll

firsthand as party activists, militia members, financiers, or leaders, but also

as civilians, refugees, and family support networks. I traveled to Central Asia
for the first time in 2005. I spent nearly twenty-five uninterrupted months
in Tajikistan and Georgia between 2006 and 2008. I employed ethnographic
methods – including living in a remote village on the Tajiki–Afghan border –
to gain rapport with interview subjects. I lectured at local universities to
cement my bona-fides as a credentialed academic with ties to the Academies of
Sciences in Tajikistan, Georgia, Kyrgyzstan, and Abkhazia. I oversaw multiple
data collection projects to meet ambitious young students. I gradually forged
relationships with academics, journalists, expatriate professionals, current and
former government employees, and eventually many former militia members
in both countries. I conducted more than 300 interviews, 173 of which were
with former combatants. These life histories form the locus of what I believe
and understand about the consolidation process.
As social scientists, we want to know, from someone else’s perspective, why
what they did “made sense at the time.” If no counterfactuals can be identified
that would have changed their strategies, it is likely that the researcher is asking
the wrong questions. If one course-corrects in the field, and tries to find the
right questions, then one is theory-building (and no longer hypothesis testing).
But the alternative – pressing forward and imposing causal explanations and
constraints that are inconsistent with subjects’ imagined counterfactuals –
squanders the goodwill of busy people. Mathematical modeling is a technology
that strives to creates precise, unambiguous statements about the psychological
states of these subjects. This is an interpretative task, the result of “guessing at
meanings, assessing the guesses, and drawing explanatory conclusions from the
better guesses.”29
As I began to break the process of civil war settlement down into tractable
parts, it became possible to refine analogies and construct models that captured
strategic dilemmas from the perspective of participants. More importantly,
locals helped me discard models and analogies that were inappropriate. I was
well positioned to ask follow-up questions, and the answers often led me in
directions that would never have occurred to me if I had been building the
models from the comfort and safety of my home institution. I often had the
opportunity to probe counterfactuals with the relevant local actors: to get them
to state plainly what they believed would have happened if things had gone
differently. Different individuals often identified the same “critical junctures”

29 Geertz (1977), 20. Rational choice frameworks put the burden on the scientist to be explicit
about what “truth” she imagines she has distilled from vapor of cultural nuance. Distillation is
inevitable, even for the postpositivist skeptic. More than 99 percent of one’s daily ethnographic
observations are discarded immediately, inevitably, with or without the filter of theory. The
words cannot be transmitted to paper fast enough. But note that if the interpretive process
occurs across a linguistic or cultural barrier, then at some point the scientist is required to
speak on behalf of “The Other.” That simply comes with the territory. Satz and Ferejohn
(1994) chart a useful path through the minefield of representational debates.
Revisionist History 17

in the consolidation process – suggesting shared understandings of the choke

points through which contingent histories could not pass.
Inevitably I met respondents based on a cascading web of contacts. Given
the obvious selection problems implicit in snowball sampling, I was initially
reticent to display the quantitative data at all or run regressions. I have
since come to the conclusion that the sample is sufficiently representative
for the intrasample variation to reveal interesting patterns.30 Approximately
two thirds of the interviews took place in the capital cities of Dushanbe and
Tbilisi. The rest took place in regional capitals or rural townships. In both
Tajikistan and Georgia the sample is weighted somewhat toward pro-regime
militias, which was appropriate for my research question, but which limits the
generalizability of valid inferences.
I put time and creative energy into strategies to correct for urban bias and
network selection effects. I gradually realized that the same recycled stories
were the result of selection effects: I was too often collecting the life histories
of the “winners,” who had managed to stick around the capital, and missing
the perspective of the “losers,” who had been forced to retreat to rural areas
or migrate. Starting in the fall of 2006 and throughout 2007, I made inroads
with failed milita groups to see if this was the case. In Georgia, this meant
actively seeking out groups that were purged from the state relatively early (the
Zviadists, the National Guard, the White Eagles, etc., instead of just former
Mkhedrioni), and groups that operated outside of the capital city (Mkhedrioni
branches in Gori, Kvimo-Khartli, etc.).
In Tajikistan, it meant venturing outside of Dushanbe for months at a time. I
lived in the village of Kalikhum, on the Tajik–Afghan border, in the winter and
spring of 2007. This village was in former opposition-controlled territory, and
was one of the primary crossing points for the United Tajik Opposition (UTO)
insurgency. I met few combatants during those cold months. I did not penetrate
rural social networks in any meaningful way. I kept my eyes open, however, and,
through pathways that I could not understand, twenty-nine representatives of

30 It is worth emphasizing that these are extremely small countries. Key respondents in both
countries were critical in providing multiple introductions in networks that branched out
unpredictably in different directions. In Georgia, I had two different retired members of the
Ministry of the Interior providing my first set of introductions, as well as a number of random
connections that turned into profitable networks (a store clerk with whom I struck up a chance
conversation, the bouncer at the nightclub beneath my apartment building, etc.). The Special
Analytic Division of the Ministry of the Interior was also an unexpected resource, generous
with time and access. In Tajikistan introductions came far slower. A number of embassy and
NGO drivers knew my project but did not introduce themselves until I was quite far along in
the process, both because of a general cultural aversion to discussing the war and a fear that
my project would inevitably attract the wrong kind of attention. Still, my reputation gradually
spread, and eventually people came out of the woodwork. As I became confident that I was
asking the right questions, I suddenly had more respondents than I could schedule. As word
spread that I was a serious person writing a serious book, people wanted to make sure their
stories were included.
18 Jesse Driscoll

various Pamiri/Gharmi militias – including two former field commanders –

sought me out after I returned to the capital city. Much of my perspective on
the civil war can be traced to my long, cold months in those mountains.31
All of the militia members I interviewed were men. Because initial intro-
ductions were often made in social settings, about a dozen interviews, mostly
in Georgia, were conducted with groups of men from the same fighting unit.
There is no doubt that this biased these responses somewhat. Particularly when
conducting interviews with large groups of Geogians at the same time, there
was often a hidden threshold in the interview process beyond which people
started contradicting each other, arguing among themselves, and revising each
others’ statements. The main data that went missing in group interviews – due
to strong norms of masculinity that proscribe certain kinds of conversations –
is the role of parental pressure, particularly from mothers and mothers-in-law,
which emerged only in responses with key interviewees. The model that emerges
is therefore biased toward a set of microfoundations that treats the individual
(man) as the unit of analysis rather than the family network. I suspect that
the shame associated with “letting the family down” motivated many of my
respondents to keep fighting, even when things got very bad, but my research
design makes it impossible to test this hypothesis.
Many of my respondents used to be employed in the police or the army. At
the time of the interview, many were marginally employed.32 Still, respondents
spoke clearly from their social position – often including self-awareness of
their own biases. When I asked questions about their behaviors, most could
reconstruct plausible explanations. They would try to make me understand
that they weren’t stupid, crazy, or deserving of pity. They believed they had
acted reasonably, even deterministically, under the circumstances. Many held
bleak and ironic attitudes, and could angrily recall the jarring transition from
being “the law” to being “on the wrong side of the law” in a few arbitrary
weeks. The random-seeming nature of coalition formation had a cheapening
effect on their views of state legitimacy. The process was reflective. The model
anchoring this book reflects shared understandings.33

31 Kalikhum was effectively cut off from the capital for nearly nineteen months during the war.
This led to a great deal of starvation and many deaths from exposure. But when this period was
described spontaneously by schoolteachers, government workers, NGO workers, and medical
professionals – the most educated citizens of the village, who would be the ones to know better,
if anyone would – the suffering was consistently blamed on “the Russians” who “bombed the
roads.”
32 The most productive members of the Tajik male labor force reside in Russia and my data reflect
this – more than a quarter of my respondents were only “passing through” the capital city on
their way back to construction jobs in Moscow, St. Petersburg, or Siberia.
33 Respondents were not passive conduits to the past. A few asked me what I was doing so far
from home, what incentivized me to chase ghosts or get the names right, what my family
thought of my spending nights having conversations about someone else’s civil war. I tried to
answer honestly, but the changing answers that I gave to these questions were constitutive in
ways I did not fully appreciate at the time. I acquired a personal stake in getting the story right.
Revisionist History 19

Respondents tended to be generous with their time and candid with their
opinions. Many were flattered that I was so interested in finding the right
analogies to capture the essence of their experiences. I gradually learned to
ask the right questions and articulate persuasive analogies. How did militia
leaders initially emerge during the chaos of the Soviet collapse? What promises
did they use to initially recruit foot soldiers? Why did this recruitment spiral
out of control in some places and not others? Why did some recruits switch
captains or hold steady in the face of vicious street violence while others
deserted? How did militias adapt their organizational structures after the
war? Why did militia leaders empower a president, if they knew he would
eventually be in a position to isolate and liquidate them?34 For many veterans,
the war was a coming-of-age ritual, and they spoke about their militia with
the nostalgia that American men from small towns use when they describe
the exploits of their high school football team. The initial difficulty of finding
respondents in Tajikistan and the nature of the interview process often lent
itself to long and relatively unstructured conversations. Georgian veterans tend
to enjoy discussing their militia experiences in large groups. Tajiks do not.
As any veteran can attest, it is difficult to assess the truth of first-person war
stories. I was surely exposed to many tall tales. Data analysis and summary
statistics in Chapter 4 are drawn from the subset of questions that were asked
universally and for which I believe I received honest responses from most
interview subjects.35

It is difficult to be specific about how the process changed my orientation toward the research
problem, but I did change. As I learned more about how the Soviet gray market functioned
to empower certain family networks above others and how familial finance networks could
be laid directly atop late Soviet mobilization dynamics, the decision for a certain kind of
young male to join a militia began to feel overdetermined. Archived emails show that –
for months – the nonmobilizations in Kyrgyzstan, Turkmenistan, and Uzbekistan were the
puzzling outcome. For a time I became entranced by factors widely considered to be beyond the
ken of rational choice: charismatic leadership, self-reinforcing emotional and crowd dynamics,
religious beliefs, and conspiracy theories. Subsequent interpretation is further complicated by
the fact that ethnographers do not retain an identity wholly distinct from their subjects. I will
probably always think of the Tajik village of Kalikhum as “my” village, though of course it is
nothing of the sort. I will probably always carry around a bit of anger on behalf of my subjects.
This anger is fully irrational, based on grievances that were never my own. The information
transmission process goes both ways. Although few of my respondents will ever read this book,
it is likely that as a result of answering my questions and reverse-engineering my reasons for
asking them, some of these men understand the wars in the same way that I do.
34 These questions may not be “the right questions” in any objective sense, but they do have
the advantage of being questions with locally verifiable answers. The longer you ask these
kinds of questions to serious people, in a serious way, the harder it becomes for them to
lie as they answer. This is the sort of small-n research design advocated by a colloquium of
anthropologists and political scientists in Bowen and Petersen (1999). The accumulated life
histories of combatants, elite interviews with commanders, and ethnographic experiences were
invaluable in the formation of theory. See Aunger (1995) and Wood (2007).
35 This leaves the social scientist with the difficult problem of what to do with data collected from
interview respondents who lied outright. Not everyone trusted me at first. Some respondents
20 Jesse Driscoll

At first, I naively tried to avoid implicating my respondents in descriptions of

abjectly criminal activities. It quickly became obvious that this was impossible.
I settled for a policy of total anonymity for all respondents.36 Certain subject
matter was carefully filtered by the respondents themselves. It was very rare
for anyone to discuss money in specific denominations. I learned a great
deal about the banalities of racketeering, price-fixing, money laundering, and
other mainstays of the post-Soviet underworld, but specific information about
the dispensation of family assets was carefully guarded. Details were not
volunteered and I never asked. Resentment toward the current generation of
“big men” was tempered by the realities of the embedded favor economy. There
are strong norms against taking the side of outsiders against the nation, tribe,
or family. I never stopped being an outsider. My field interviews improved
substantially when I accepted the limits of my role. I acquired a reputation
for being primarily interested in political stories and having no particular local
agenda. I also took conscious steps to control the subject matter, intentionally
shying away from asking respondents about their own violent activities. The
main way I did this was to make it clear that I was studying institutions –
militia structures – and not people. I overcame my skepticism of questions
like “in your group did you observe …” when I realized that respondents

left the interview not trusting me at all. (Who could blame them?) Strange as it sounds, many
of the respondents who assumed I was a spy, or affiliated with the U.S. military, were often
more willing to engage with me. I can only speculate why this was the case. Perhaps they saw
me as being part of the same fraternity of warrior-defenders, or that we shared the experience
of making bad youthful decisions and getting in over our heads. But some respondents were
certainly sociopathic liars. Some respondents misrepresented basic facts about events, their
beliefs, and their roles. There is little to be said here beyond the obvious: I listened hard, asked
follow-up questions when I could, and usually stayed until the person wished to stop talking.
I can only promise readers what I promised my respondents: that I have tried in this text to
represent the spirit of their comments accurately, to never print anything I did not believe to
be true, and to present quotes in the spirit in which they were spoken. All subsequent errors
in the modeling and interpretation of these soldiers’ narratives are my own.
36 I recorded only the subject’s first name, or a pseudonym, and interview date to preserve
anonymity. In a few cases, the subject insisted that I record his full name. For my own safety,
and that of my respondents, I never complied with these requests. The protections that I deemed
necessary for myself and my respondents were substantially more serious than anything that
anyone in the Human Subjects Committee of my home institution thought to impose. I initially
hoped it would be useful to have respondents sketch pictures of the security structures at
different periods of time. These exercises quickly devolved into incoherence – pages filled with
circles, lines, and scribbled names that I realized would never be coded or systematized. I
began to brainstorm strategies to collect individual-level characteristics to predict which militia
members joined the state, which factions ended up with which jobs, and which factions were
“weeded out.” Meandering conversation threads then gave way to more structured discussions
about who was getting what, how and when side-switching between militia commanders was
considered, etc. Though I eventually became discouraged with the effort to map individual
militia trajectories, the exercise provided the impetus for the collaborative efforts with the
Small Arms Project to code warlord biographies and the impetus for removing individual
identifiers from analyzed data.
Revisionist History 21

could describe their own actions in the third person, with plausible deniability
and emotional distance from actions that were regretted in retrospect.37 But
interviews often sprawled, and what I ultimately ended up collecting were life
histories. With time, behind the veil of anonymity my methods provided, I
received rich anecdotes.
It is reasonable to second-guess the motives of networked strangers who
arrive from far away states to ask questions about a war. Having extended
conversations about post-Soviet security structures occasionally led to uncom-
fortable and conspiracy-laden conversations about whether old Russian phone
taps had been replaced with American post-9/11 security assistance. Following
Derluguian (2005), I took notes but did not record interviews – and am sure
that if I had started taping voices I would have received less access and very
different kinds of data.38
I was living in Kyrgyzstan in 2005 when the Andizjan events unfolded across
the border in Uzbekistan.39 I recall distinctly the feeling of vulnerability when
the Internet stopped working correctly. I decided that if I was going to continue

37 There is obviously some risk that the leading structure of the interview questions biased
respondents toward remembering the worst in their counterparts. This means that the degree
of theft and indiscipline may be overstated.
38 In keeping with Whyte (1982) and Emerson, Fretz, and Shaw (1995), sometimes I judged
that taking notes would disturb the intimacy of the conversation, though I would, if necessary,
pause the respondent to scribble a particularly juicy quote, translated to English in real time. In
these cases, field notes were recorded immediately after an interview session. Most interviews
were either conducted in respondents’ homes or in mutually chosen public spaces. I often
compensated respondents for their time with a culturally appropriate token gift (often baked
goods or alcohol). In public and social settings, I quickly learned that it was important for
me to insist I “host” the interview to maintain control of the environment. This allowed me
some control over the timing of the questions, when to stop, and some ability to slow the
speed of toasts. On a few occasions, my embarassingly low tolerance for alcohol affected
my ability to complete the interview, but in many cases these social settings produced key
informants and a range of introductions. My inferences are inevitably “weighted” toward
a small number of key informants with whom I established rapport, conducted multiple
follow-up interviews, and who supplied introductions with others who corroborated their
claims. The majority of interviews were conducted in Russian in Georgia, and in either Tajik
or Russian in Tajikistan. Approximately a quarter of the interviews in Georgia and a third of
the interviews in Tajikistan were conducted in the presence of a translator (always a young
man of university age). Eventually I found that navigating the contours of the language barrier
provided many opportunities for follow-up questions, and the intimacy allowed me to guide
the interview. My knowledge of the civil wars expanded organically along with my language
abilities. The quality of my data also improved as I mastered enough local detail to ask
face-saving follow-up questions and signal that I had recognized a half-truth.
39 What is not disputed is that there was an attempted prison break and many, many people died.
Then the narratives diverge. Akiner (2005) went to Andizjan two weeks after the uprising
to conduct interviews and concludes that the demonstrations were a “carefully prepared”
(10) attack on the Uzbek government, organized by armed militants (part of the “Akromiya
movement”), who were multinational in composition (30–31, 27–29). Her version of events
is contradicted by Bukharbaeva (2005), Daly (2005), the OSCE (2005), Ilkhamov (2006), and
Kendzior (2006).
22 Jesse Driscoll

operating in an unfriendly authoritarian environment, I needed to adapt. I

quickly educated myself about how Internet servers work, and I stopped
assuming my email communications were private. For important topics I began
to rely on pen and paper. But these realizations gradually transformed my
relationship with “the field.” Especially in Tajikistan, it meant proceeding very
slowly over months that turned into years. After a few frightening encounters,
I was warned by a trusted advisor that my research design was making me look
quite a bit like a spy – and that if I ran afoul of the wrong character in the wrong
security bureaucracy, it could easily result in my permanent disappearance.40
After returning to the United States, to limit my own liability, I made a point of
destroying all of my field notes that could be used to reconstruct contemporary
networks.
As a personal matter, I have opted for a rather distant and antiseptic tone
in this volume with respect to acts of violence. After a few uncomfortable
conversations, I tried to enforce a general policy of stopping interviews when
subjects divulged personal memories of violence or war crimes. This was not
always possible. Many interviews had a confessional aspect to them. Some men
tried to shock me with graphic descriptions of tactics – inflicted or received –
to test, embarrass, or entice me. As an author, I have learned that there is no
catharsis to be gained by seeing these descriptions in print. Lurid descriptions
of mutilations, written from the safety of the ivory tower, cheapen the horror
of the experiences for participants. The deepest scars – posttraumatic stress
and survivor guilt – are invisible. Ethnographers working in conflict zones
eventually learn what police detectives and competent military professionals
have understood for time immemorial: there are serious mental and emotional
costs associated with treating sociopaths as subjects.41
Sometimes I got scared. The interview subjects always noticed. The interview
ended shortly afterwards. That I conducted so many interviews is evidence
that, from my perspective, the participants in these wars were not all thugs,
gangsters, and sociopaths. I could not have created the necessary intimacy
if they were. Obviously there were strategic reasons to present distorted and
self-serving versions of certain facts, but most respondents seem to remember

40 A Tajik doctoral student and personal friend was recently placed under house arrest in
Tajikistan, pending a trial for charges of treason and espionage, for qualitative observational
research (conducting interviews in Badakhshon) on behalf of a non-Tajik Principal Investigator,
who works at a well respected European research institute. He has since been released but it
is not clear if he will ever be permitted to return. The risks to my local contact network are
just as serious. Though there is technically an amnesty law on the books, many respondents
became visibly uncomfortable when the conversation turned to Rakhmonov, contemporary
politics, or anything that might pique the interest of eavesdroppers. Journalists are routinely
intimidated in Dushanbe. Independent of each other, both of my regular translators in
Tajikistan (both aspiring journalists) requested that I omit their names from this volume and
all future publications.
41 I occasionally wonder how my life would be different if I had read E. Valentine Daniel (1996)
before going into the field, rather than afterwards.
Revisionist History 23

doing what they thought was appropriate at the time. Respondents were
reasonably clear, once I asked the right kinds of questions, about what games
they understood themselves to be playing, why they adopted the strategies that
they did, and why they succeeded or failed.
I decided early in the research process to build my theory on the perspective
of actual people as much as possible. People do not, as a rule, like to be studied.
And as Jarvis Cocker observed, “everybody hates a tourist.” People do, as a rule,
like the idea that their stories will be recorded for posterity. Ethnographers – by
their invasive presence – force these decision heuristics into conflict. As the line
between researcher and subject blurs, what occasionally emerges is a crucible
for creative, cooperative theory-building. People can tell when their words are
being received with empathy (i.e., when they are being treated as subjects) and
when their words are being clinically recorded for some other purpose (i.e.,
when they are being treated as objects). I found that once a subject decides that
the researcher is actually listening – and taking anonymity promises seriously –
the researcher gets much better data.

structure of the book

My book compares the experiences of two countries that disintegrated
violently, but then reconstituted relatively quickly, after the collapse of the
Soviet Union. In the cases I selected for this study, the third-party intervention
force – a Russian force – was not benign or neutral, and it did not make a con-
vincing attempt to implement liberal policies to facilitate general disarmament.
Nonetheless, local actors managed to sculpt self-enforcing arrangements by
cannibalizing inherited Soviet institutions. But contemporary order-providing
bargains in the two countries are very different from each other. I will argue
that the differences are better explained as geopolitical adaptations by local
innovators than as structural inheritances from Soviet times.
Because no one anticipated the timing of the Soviet Union’s collapse, the
disintegration and reconstitution of the successor states provides the closest
analogue to a laboratory experiment that students of state failure and recovery
are likely to find. A shared institutional and historical legacy guarantees that
many unobserved contextual variables were held constant in the aftermath
of independence. This book traces the experiences of two Eurasian countries,
Georgia and Tajikistan, through the first decade of their independence. Both
states collapsed into anarchy as the result of dynamics unleashed in their
inaugural elections. The hostile and impassable mountains, the absence of
a national army, the easy access to illicit funds from drug corridors, the
weighty history of bad governance, ethnic cleavages, opaque clan networks,
and foreign-backed insurgencies all combined to make them poor candidates
for quick resolution or negotiated peace. At the beginning of 1995, Georgia was
in a state of near-total anarchy, with the state apparatus delegitimized and its
territorial integrity splintered by multiple successful secessionist movements.
24 Jesse Driscoll

Tajikistan was vying with Chechnya for the dubious distinction of being
the most brutal war in the post-Soviet space. To the surprise of everyone –
conflict scholars and area specialists alike – stable regimes in both countries
consolidated during the next few years. Over the course of the next decade,
rulers who were originally installed at gunpoint by paramilitary warlords
managed to wrest control of the state apparatus from the armed groups that
installed them. Faith in the permanence of the state gradually returned as
ceasefires calcified into armistice across Eurasia. Today the settlements appear
quite stable.
The divergent experiences of Georgia and Tajikistan provide a rare oppor-
tunity to examine “the state” from the ground up: sets of local understandings
between armed groups congealed eventually into self-enforcing and predictable
strategies. It is possible for social scientists to catch a rare glimpse of the
process by which order emerged in the post-Soviet periphery. In both countries,
clan-based militias and criminal warlords installed a civilian regime in the
capital city that was capable of appealing directly to international donors. The
regime started as a cosmetic legitimizing device for violent militias. Armed
groups fought each other for the right to extort presidents, who doled out
privatization rights and ministry positions to buy loyalty. Over time, post-Soviet
leaders learned how to pit their enemies against one another in the coalition for-
mation process. A critical component of this strategy was incorporating former
political enemies into the state apparatus, allowing rebels to reinvent themselves
as important regime allies. Nongovernmental organizations (NGOs) managed
the delivery of humanitarian aid, partial privatization enriched local actors,
and warlords became tax collectors. European and American aid for targeted
institution building and state reconstruction was captured by local agents, who
kept violence mostly out of sight and deep structural reforms off the table.
State-building entailed the emergence of unaccountable patronage networks
inside regime ministries – the “systematic corruption” decried by intellectuals
and NGOs – which gardually harmonized informal wartime institutions with
formal state structures.
The post-Soviet wars are often explained as exemplar cases of ethnic war:
where historically inherited categories determined, and could thus be used to
predict, wartime affiliations.42 This book takes as its starting point the easily

42 I have in mind Posen (1993) and Van Evera (1994). Kaufmann (1996) argues that once ethnic
violence reached a certain threshold, the sides are fixed, the stakes are zero-sum, and negotiated
settlements short of secession and population separation are practically impossible. Petersen
(2002) emphasizes that deep-seated historical grievances are often articulated by fighters as
causal explanations for their behaviors. He argues that, in the aftermath of the Soviet federal
experiment, it was clear that the inversion of long-established ethnic hierarchies was inevitable
in many of the new states. It was not difficult, in the absence of strong state institutions, to use
the powerful emotions that accompanied status reversals to motivate young men to commit
acts of symbolic violence against their neighbors. Laitin (1999a) presents a “tipping” model
for vigilante violence in a multi-linguistic context, designed to create local status-hegemony
Revisionist History 25

visible counter-trend toward interethnic cooperation – by far the most common

outcome between the myriad ethnic groups that did not begin killing each other
when the Soviet Union collapsed.43 The idea that the violent homogenization
of Abkhazia is somehow more of a “natural” or “inevitable” outcome than
the stubbornly enduring peaceful coexistence of ethnic groups in the Ferghana
Valley ought to be anathema to modern thinking liberals. When all of the “dogs
that did not bark” are taken into account, explanations for violence based on
ethnic difference and historical grievance are suspect. Ethnicity was certainly
not a very reliable predictor of loyalty in either of the violent state consolidation
projects that this book describes.44
Much of the data that are presented in this book focus on events in Tbilisi
and Dushanbe, the respective capital cities of Georgia and Tajikistan. This
is intentional. When Georgian and Tajik soldiers talked about the politics of
war termination, the politics that they wanted to talk about was primarily the
politics of dividing up the best turf. The capital is where one is most likely to find
airports, high-end restaurants and casinos to cater to the wealthy transnational
investor class, banks, embassies, physical symbols of state authority (such as
the statehouse and parliament building), television transmitters, public utilities’
junctions, and centers of cultural production (such as universities and opera
houses). Calls for nationalist mobilization that come from the capital city are
more likely to be taken seriously by rural youth. Whoever controls the capital
city controls the image of that country to the world. A relatively orderly capital
city is a keyhole that foreigners are permitted to look through.45 In this account,
the visible signs of well-functioning state institutions – shiny buttons on pressed
uniforms, television programming to commemorate national and religious
holidays, schools where nationalist myths are standardized as curricula,
military parades to display soldiers marching in well-ordered ranks, clerks
at desks empowered with stamps, clean parks with well-maintained statues,
well-run elections and parliaments – are epiphenomenal of understandings
between domestic armed groups. The majority of citizens observe public rituals
and are held in awe.

for a language group to maximize the life opportunities of the vigilantes’ children. For Laitin,
violence can be rationalized as facilitating a societal “tip” in the status of one’s native tongue,
with the benefit being passed on to the next generation in the form of high-paying jobs
as government clerks. Fearon (1998) explicitly builds the assumption of zero-sum tradeoffs
between the values of different ethnic groups into his model of violent secession.
43 Fearon and Laitin (1996), 716.
44 Though it is beyond the scope of this study, the same is also true in Chechnya. The
Russian government enlisted rehabilitated Chechen rebels in the last decade and Lyall (2010)
demonstrates that they are unusually effective at the sordid tasks of counterinsurgency. Balta
(2007) confirms this account. Enloe (1980) documents military co-optation across ethnic lines
around the globe.
45 Bates (1984) demonstrates that whoever controls the capital of an African state can extract
wealth from rural poor by leveraging import–export bottlenecks. This is also clearly the case
in many post-Soviet republics.
26 Jesse Driscoll

Social scientists often rely on models to understand observations and

experiences. Formal models derive their power from their claim of being able
to clearly capture the essence of a situation. Models are aesthetically attractive
to readers who appreciate clear and explicit presentation of theory. Combining
formal modeling with narrative – the “analytic narrative” approach – ensures
that the argument is defined precisely.46 The narrative establishes who the
actors are, their preferences, the available set of strategies, and possible
outcomes. The formal model specifies how the interplay of strategies worked
to create a particular stable outcome (equilibrium). In dynamic settings, the
concept of subgame perfection is a powerful explanation for why certain
behaviors and ideas “stick” in some conditions and not others. The analytic
narrative approach also has advantages for social science presentation because
the formal model guarantees that analytical rigor is maintained, but readers are
not forced to break the narrative prose for hypothesis testing.47 Filtering a set
of noisy facts about the world through a mathematical prism allows for clear
articulation of causal processes and washes out irrelevant details.
In a highly critical essay, Jon Elster (2000) raised a number of programmatic
objections to the analytic narrative approach. Eschewing real people in favor
of bloodless rational agents has pitfalls, namely, the assumption of rational
actors, high levels of aggregation, little concern for intentions and beliefs,
and no concern for uncertainty. I am sympathetic to Elster’s concern that, as
a community, political scientists have become a bit too comfortable taking
the logic of the firm and applying it to states, ethnic groups, rebel factions,
political parties, and terrorist cells. Everyone knows that none of these
organizations really have interests or beliefs: only people do. With that in
mind, this book’s empirical strategy and presentation are calibrated to be
directly responsive to Elster’s concerns. The individual human being is the
unit of analysis in all of the chapters that follow. I describe a violence market
in weak states between warlords and recruits, coalition formation processes
between strategizing warlords, and divide-and-rule politics between a weak
president and his warlords-turned-ministers. In all of these examples, strategic
calculations are made by human beings under conditions of great stress and
uncertainty. Neither perfect rationality nor perfect information is taken for
granted.
Once analysts abandon institutions as the unit of analysis and replace them
with thinking agents, Ferejohn (2004) hints that persuasive accounts of any
political phenomenon worth studying – such as the quick resolution of the
post-Soviet wars – should include both external and internal explanations.

46 See Bates et al. (2000, 2000).

47 A reader who dismisses the assembled quotes as an unscientific assemblage of accumulated
conspiracy theories, acquired haphazardly from a snowball/convenience sample, can at least
have faith that all of the tall tales and street stories congealed into something internally
consistent.
Revisionist History 27

External explanations relate to intersubjectively realized constraints that are

“out there” in the world. Internal explanations come from within an individual,
such as psychological biases, emotional responses, and culturally contingent
belief structures, and are much harder to measure.48 Collecting data on the
psychology of civil war actors is difficult and dangerous, and I now appreciate
better why studies that provide compelling “internal” accounts of civil war
behaviors are so rare.49 As I investigated the militia structures that formed the
fighting units of the post-Soviet wars in Tajikistan and Georgia, I was granted an
unusually clear view into the worldviews that informed civil war participants.
If social science is judged by its ability to integrate subaltern perspectives into
hegemonic discourses – and I see no reason it should not be – then adopting the
vantage point of excluded actors, to the extent that these actors let their voices
be heard, seems to be a decent place to start. I have reproduced a number of
my interview subjects’ quotes faithfully, and scattered them through the text.
Chapter 2 presents a theoretical account of civil war settlement in which
no one actually disarms. A ruler in the capital city (identified generically
as a figurehead president) is responsible for acting as the public face of the
nation to the international community. The president remains in power by
establishing informal patronage networks and distributing wealth to local
strongmen (identified generically as warlords) who, in turn, use the threat
of violence to keep citizens from organizing to challenge the state. Bargains
are self-enforcing because warlords who install a president can extort him
post-installation with the threat of a coup. As payment for providing order,
the president is forced to give the warlords relatively unfettered access to state
ministries. Parameters required for an orderly situation to stabilize from the
model are clear: shared expectations of foreign wealth and low “reservation”
options for warlords who opt to retain criminal empires outside of the regime’s

48 Consider the now-classic rationalist account of revolutionary social mobilization presented

by Timur Kuran (1991). Costly social mobilization is a coordination problem. Risky protests
against the government might be worthwhile if they are likely to induce “tipping” behavior –
creating a situation where the regime collapses and there is no authority to retaliate against
activists. An individual’s choice to participate in a revolution will depend on the tradeoff
between an external payoff (i.e., personal rewards and punishments, which will vary depending
on whether they are acting alone or in a group) and an internal payoff (i.e., the psychological
costs of preference falsification). See Petersen (2001), especially chapter 2, for a discussion of
how different community thresholds yield different equilibrium strategies.
49 Participant observation has been the research design that most consistently yields persuasive
“internal explanations” for violence. Seminal ethnographic (and often selfnographic) investi-
gations into the social structures and psychological dispositions that make violence thinkable
include Orwell (1952), Capote (1965), Fanon (1968), Herr (1968), Scott (1976), Popkin
(1979), Pileggi (1985), Katz (1988), Vigil (1988), Jankowski (1992), Buford (1993), Espiritu
(1996), Grossman (1996), Brass (1997), Sacco (2003), and Jackall (2005). For an excellent
introduction to the anthropology of violence, see the assembled essays in Scheper-Hughes and
Bourgois (2004).
28 Jesse Driscoll

distributive politics framework. These are both parameters that great powers
can manipulate without any particular local knowledge.
With the structure of strategic interaction defined and the theoretical
framework established, the book unfolds in a roughly chronological manner.
Rather than present historical snapshots, each chapter explains a process of
equilibrium selection. Each chapter explains how Georgian and Tajik actors
understood their choice set at the time, and why they chose particular strategies,
intentionally emphasizing processes at the expense of variables.
Chapter 3 defines the primary strategic actors in this study –
warlords – and explains how they dragged Georgia and Tajikistan into violent
state failure. Central to the story in both cases were escalating fears of radical
programs for social redistribution and geopolitical re-alignment. With the
authority of old Soviet hierarchies contested, and established favor economies
disrupted by emigration and economic collapse, there was deep uncertainty
about who would be the terminal enforcer of contracts and provider of order.
The paramilitarization of politics fueled cycles of fear, and anarchy washed over
the countryside. Russian artillery and air power made certain non-negotiable
realities of the post-Soviet international system clear by backing the regime in
Tajikistan and backing ethnic minority secessionists in Georgia. But despite
military cauterization, Russian soldiers could not credibly commit to policing
the particulars of the deals that ended the war.
Chapter 4 describes the process of coalition formation in Georgia and
Tajikistan. Certain warlords in both states colluded together to invite a “puppet
president” to speak on behalf of the state and shake hands with foreign
donors. There were not even superficial attempts at disarmament – militias
in the original winning coalition self-consciously became the state. Warlords
in Georgia and Tajikistan understood that if they could form a coalition, lay
claim to the capital city, and act like a unified force – reappropriating symbols
and rituals from defunct Soviet institutions long enough to convincingly
masquerade as a government – they would gain access to foreign aid and
recognition. This would, in turn, allow them to defeat rival factions. Warlords
understood themselves to be participating in something analogous to a costly
lottery or a game of musical chairs. Some warlords ended the game inside
the state and others did not. There was a period of jockeying for position
in the postwar government. During this period, militia members – many of
whom were recruited based on promises of future state spoils – competed
with newcomers in the violence game, emerging from urban slums or refugee
camps. The post-war police forces could not cartel the local violence market
as militias proliferated. Violence during the “Time of Troubles” was primarily
instrumental in this account, allowing militia members to gather information
on the strength of various warlords, the resolve of various militia recruits,
and, ultimately, the final composition of the ruling coalition. Once it became
clear which warlords were winners and which were losers, many foot soldiers
abandoned their commanders and returned home.
Revisionist History 29

Chapter 5 examines the process by which the civilian “puppet” presidents

ultimately managed to free themselves from the threat of a coup. The story
of how divide-and-rule tactics were used to transform this symbolic power
into real bargaining leverage is ultimately a story of different militia captains
being played against each other. Once merged into the state, armed groups
provided contract enforcement and laundered foreign assistance into the local
economy, and by keeping the risk of coup constant, ultimately found themselves
filling a vital institutional role. Warlords in both states understood the necessity
of presenting a united front to extort the president, but faced a severe
collective action problem against crafty civilian rulers who took advantage
of splits and cycling in shifting warlord coalitions. Detailed case studies
trace the particulars of the divide-and-rule strategies used by Shevardnadze in
Georgia and Rakhmonov in Tajikistan as their respective super-presidencies
consolidated over the 1990s. Divergent outcomes in the first decade of the
twenty-first century are briefly sketched. Though Georgia and Tajikistan both
gained independence with unconsolidated territory, scrap-heap armies cobbled
together from identical weapon systems, and similar values on virtually every
structural variable that one would use to predict persistent state weakness, they
have embarked on strikingly different geopolitical and institutional trajectories.
Chapter 6 concludes the study with a brief discussion of policy implications,
including contemporary observations relevant to the civil war in Ukraine. If
the reader concludes that the post-Soviet wars – long considered to be about
indivisible issues of identity – were actually resolved through brokering and
buy-out, this study has political implications. Normatively, the news is mixed.
On one hand, short civil wars save lives. On the other hand, the processes of
war termination described in this volume implicate interveners in structural
violence, extortion, and bribery.
2

Predator Collusion
A High-Stakes Game

A persuasive account of state recovery must answer two questions. First: How
did civilian executives become strong once formal institutions collapsed? The
answer, as already forecast in the first chapter, is that they were figurehead
placeholders for coalitions of warlords who “ran the streets” out of sight. But
because this arrangement ultimately benefitted the president at the expense of
the warlords who installed him, it is reasonable to ask a second question: Why
did warlords agree to install a president if they knew that a possible result was
that he would use divide-and-rule tactics to cut them out of the spoils? The
answer is that although some high-profile warlords were jailed or killed in the
consolidation lottery, others slid out of view, reinventing themselves as state
agents. A few became quite wealthy.
In this simplified account, state-building is a constantly renewing process
of contracting and bargaining between violence entrepreneurs. Warlords are
locked in competition, and their rivalry can easily turn violent. This violence
can end only with military victory or through a process of coalition building.
Though they have the option of working together, cooperation is risky.
If a group of warlords can assemble a coalition with sufficient military
power to seize the capital city and achieve international recognition, it can
install a civilian regime. This civilian regime will gain access to foreign
aid, military assistance, and low-interest capital investment. The new regime
will become the warlords’ hostage, and will immediately transfer most of
these new rents to them to stay in power. If the gains associated with
seizing the capital city and extorting the rents of sovereignty are greater
than the expected utility of outright war, all warlords may rationally abjure
violence.
This chapter presents an account of civil war settlement under conditions
of state failure. Though it is presented in the form of a two-stage, n-player
coalition game, one does not have to be a student of game theory to understand
the argument in this chapter (proofs and formal propositions can be found

30
Predator Collusion: A High-Stakes Game 31

in Appendix B). In the first stage, warlords choose either to fight or join a
coalition and back the ascendency of a president. If no president is installed,
the game ends with continued warfare. If a president is installed, a second stage
takes place in which the president distributes the wealth of the state – newly
increased, as the president is able to get more aid and other benefits from
foreign actors – among various warlords. Warlords observe the distribution
and choose either to accept the transfer or attempt to remove the president
in a coup. All players are assumed to understand this basic game in the
same way, second guess each other’s strategies, and maneuver strategically.
Analysis of the model reveals a few analytically distinct equilibria. One of
particular interest is an equilibrium in which all warlords merge into a single
coalition – monopolizing the production of violence. Other equilibria describe
stalled negotiations or persistent state failure. Order-providing institutions and
understandings are sustainable despite the inability of foreigners to monitor
or enforce local arrangements. Peace is self-enforcing without the need for an
external guarantor.
In the “liberal interventionism” framework referenced in Chapter 1, the
central problem of civil war settlement is convincing rebels to disarm. Because
warlords’ bargaining power is assumed to be inextricable from their capacity
as violence entrepreneurs, in this model no one disarms. One might observe
certain kinds of cosmetic disarmament – warlords may don suits, affix a party
lapel pin, and reinvent themselves as party officials or vote brokers – but they
maintain control of men and weapons. Order is contracted through a process
of incorporation and buy-out, with payments taking the form of graft: state
offices, black-market monopolies, or rigged privatization schemes. “The state,”
in this account, is a legitimizing device for warlords who have reinvented and
redefined themselves.

a game
The strategic contest takes place in a small, internationally recognized sovereign
state. This state contains lootable resources and government positions, and
the conflict is over the right to appropriate these spoils through selective
enforcement of property rights. The actors in this contest are warlords –
violence entrepreneurs with private armies, locked in a struggle for power.
Assume the state contains n > 2 warlords indexed by i, W = {1, 2 . . . n}. At the
beginning of the game, every warlord i ∈ W simultaneously chooses either to
fight or join in a coalition to install a president. Each warlord i has the option to
“Fight” to capture the capital city, exclude rivals from power, and expropriate
state wealth v for himself.
Fighting imposes costs c on each warlord, as sustaining a militia cannibalizes
productive assets and exposes his family to some risk of violence. Civil war
is costly and unpredictable from the perspective of a warlord. A charismatic
leader’s ability to sustain a militia depends on a host of military, social, and
32 Jesse Driscoll

psychological variables that cannot be easily predicted (see Chapters 3 and 4).
A prominent warlord can be killed by a ricocheting bullet or replaced by a crafty
lieutenant (see Chapters 4 and 5). Warlords are forced to choose strategies
without full knowledge of how their own relative capabilities will compare
to those of their opponents over the course of the war.
A simple way to capture the contingent character of this process is to treat
warlords as symmetric and interchangeable. If all n warlords play “Fight,” all
will receive payoffs of nv − c. This could represent, with an equal probability,
either that the territory of the state fragments into proportional warlord
fiefdoms, or that one warlord wins and will control all the spoils. Fighting
imposes costs either way.
As an alternative to going alone, warlords can work together to “Install”
a president. If they succeed, they will form a group with sufficient domestic
armed power to provide order in the capital and minimally secure the borders.
The coalition will then temporarily abjure violence and back the ascension of a
civilian government, headed by a figurehead president P.1 If enough warlords
collude together, a government emerges capable of appealing to international
donors directly. The “stability threshold” s represents the number of warlords
necessary to control the capital against rival warlords outside the coalition,
making it safe for foreign governments to open embassies and diplomatically
recognize P’s regime. If s or more warlords work together, then a government
emerges with sufficient domestic power to acquire foreign aid, claim the
country’s seat at the United Nations, and secure foreign investment. A second
stage of the game begins (which will be discussed a bit later in this chapter).
If fewer than s warlords opt to collude together, then the government that is
installed will be incapable of controlling the countryside, and warlords revert
to fighting. The miscalculating warlords receive a sucker’s payoff of nv − c − w
for any warlord i who played “Install” while others played “Fight.”2
The “stability threshold” s is a fixed parameter. It is a benchmark for how
many warlords are necessary to install a president in the first stage of the
game. It also represents the minimum number of warlords necessary to keep
a president in power in the second stage of the game. In Rousseau’s classic stag

1 It would be more realistic to complicate warlords’ coalition formation problem, imagining the
universe of national intellectuals, statesmen, heads of prominent tribes, or bureaucrats forming
a pool of potential presidents P = {f1 , f2 , . . . fz }. In this more complicated, realistic, and dynamic
version of the game, when warlords opt into a coalition, they would have to choose not only
to “Fight” or “Install,” but also to “Install” while declaring loyalty to a potential figurehead. I
assume that the process of coordinating on an acceptable figurehead is resolved extra-model.
2 The size of the parameter w is independent of the number of warlords who played Install, but
w >0 by assumption. A failed attempt at coalition formation ought to be costly in some way,
if only because resources invested in diplomacy trade off with investments in military strategy.
One might imagine that a warlord who incorrectly anticipates a coalition forming suffers a
reputation cost in the eyes of potential recruits, having endorsed a vision of an emergent stable
coalition that does not emerge (i.e., after miscalculating, warlord i would appear less wise, or
less informed, than competitors whose predictions of total war proved correct.).
Predator Collusion: A High-Stakes Game 33

hunt, all of the hunters have to work together to bring down a stag. But it is
not a logical or realistic requirement for all of the warlords to have to work
together to install a figurehead president. And unlike many coalition formation
games analyzed in institutionalized settings, there is nothing particularly special
about the 50 percent threshold for a simple majority. The number of warlords
who have to work together to stabilize a state sufficiently in the eyes of the
international community varies by context. A low stability threshold means
that a government can access foreign aid and investment even if there are
large pockets of territory controlled by unaffiliated warlords. A high stability
threshold means that a government needs to incorporate most of the warlords
before it gets access to these rents of the international system.
Warlords who chose to fight in the first stage rather than support the
president will be cut out of these spoils, and they have no realistic chance of
displacing the entire coalition of warlords that now claims the capital city. For
a warlord i who remains outside of a consolidating state, the next best thing
he can do with his private army will yield a payoff of r, with 0 ≤ r ≤ nv − c. The
game ends for these warlords.
Based on what I heard during interviews, r can vary a great deal depending
on a warlord’s economic and social endowments, regional geopolitics, and the
ability of the regime to get foreign support to coerce recalcitrant warlords.
Some warlords may be able to flee to the mountains or across international
borders and keep the fight alive for years. They may be able to start new lives
as narcotics traffickers or soldiers of fortune. They may set up “shadow state”
institutions in territories beyond the reach of state authorities. They may be
allowed to walk away from their armies and simply disappear. Or they may
not. They may end up on no-fly lists, tagged and tracked for the rest of their
lives by the installed regime. They may be quickly and efficiently executed. They
may die slow deaths from starvation and exposure.3
The number of warlords who play “Install” in the first round can be
called k. Call the subset of k warlords who play “Install” W P ⊆ W, such that
W P = {i, j, . . . q}. So long as k ≥ s, a coalition government forms. Power-sharing
follows, and a figurehead president P is installed to shake hands with foreign
heads of state. Warlords do not disarm.4 They keep access to men and weapons
through a variety of invisible channels, and are well positioned to extort the

3 What individual warlord characteristics might change the reservation value r? The obvious
starting point is expectations of great power support to the consolidating regime, but students
of counterinsurgency might quickly add to the list, suggesting ethno-linguistic differences,
diaspora networks stretching to foreign capitals, transnational criminal linkages, a homeland
in impenetrable mountains, access to particularly lucrative drug trafficking routes, ideological
or religious predispositions, or managerial genius suitable to managing clandestine networks.
Appendix B has an extension showing that heterogeneity in r can yield unique equilibrium
predictions.
4 Disarmament of armed rebels has been described as the “fundamental barrier to civil war
settlement” in Walter (1997, 2002).
34 Jesse Driscoll

president by threatening a coup. For the president P and the k warlords in W P ,

the second stage of the game begins.
The first thing that happens in this second stage is that the influx of
foreign aid, foreign civilians, and illicit rent-seeking opportunities increase
the lootable wealth in the state from v to v∗ . This new wealth v∗ comes
into existence in a form that is controlled by P, as he is the one who
doles out cabinet appointments. If foreign investors want to extract mineral
wealth from the country’s interior, build an oil pipeline across the territory,
or sell liquor or cigarettes in the capital, they will have to broker with the
president’s agents. The same is true for foreign militaries that want to silence
transnational dissidents or liquidate terrorists living in the territory. A president
is a focal symbol of order and stability. He is the face of the winning warlord
coalition.
Though v∗ is nominally controlled by the figurehead president, power still
comes from the barrel of a gun. The president does not need all of the
k warlords to rule, but will pick l ≥ s warlords to form his inner circle.
Call the subset of l warlords selected by the president W L ⊆ W P such that
W L = {i, j, . . . q}. Warlords in W L merge their memberships with the army
and police forces. But the game is not over. The threat of a coup still hangs
over all distributional politics that follow. Warlords in W L are well positioned
to replace one civilian president with another if their demands are not met.
P must publicly distribute v∗ among the l warlords in W L and himself, with
his decision represented by x = (xi , xj , . . . xq , xP ). Wealth transfers take the form
of ministry positions, nonenforcement of tax laws, closed-bid contracts, and
rigged privatization schemes, all of which are arrangements designed to be
opaque to foreign observers. Each of these warlords observes his transfer and
chooses simultaneously whether he wants to “Coup” or “Accept” the president.
It is necessary for s or more warlords to choose “Accept” after receiving their
share of x, or the president P will be deposed. In this case, P ends the game
with a payoff of zero.
Choosing “Coup” invites violence, which imposes costs on the warlord. Call
these costs c, invariant from the first stage, for simplicity in notation. He has
some positive probability p of successfully deposing the president. If he keeps
the president’s distribution of x, a warlord would replace P and seize whatever
share of the rents the president was planning to appropriate to his office (xP ).
But why stop there? Perhaps a warlord could completely exclude rivals after
installing himself as head of state, claiming the entirety of v∗ for himself.5 Every

5 Embedded in these payoffs is the assumption that the coup winner can maintain the favor of the
international community and investors (or extort them with the threat of domestic chaos) so the
country will continue to receive v∗ . At the risk of jumping ahead, payoffs in a coup should reflect
the most optimistic possible post-coup situation from the perspective of a warlord, as after
backwards induction, the “ceiling” of a coup payoff will represent the “floor” of a risk-adverse
president P’s transfer to a warlord. It is plausible that in the event of a poorly executed palace
coup, v∗ would shrink back down to v. In this case expected coup payoffs are only pv − c. If this
is were known to be true by the warlords, these losses would ultimately benefit the president P,
Predator Collusion: A High-Stakes Game 35

warlord is a capable observer of the political environment, so for simplicity let

us assume that each warlord i can correctly assess his subjective probability of
successfully carrying out a palace coup at a given time.6 To simplify matters
and highlight essentials, let us also assume p is invariant to l, and assume a
successful palace coup would net the warlord all of the assets in the state.7
The coup payoff can be described as pv∗ − c. At the risk of repetition, if fewer
than s warlords play “Accept,” then the coalition breaks down and all of the
warlords fight among themselves, each hoping to personally ascend to the role
of president. In this case, the president receives zero and each warlord in W L
receives his coup payoff pv∗ − c. If the number of warlords playing “Accept”
is equal to or greater than s, P stays in power and all of the warlords in W L
receive their transfer according to x.
What happens if more than one warlord “Coups” at the same time? To
maintain the assumption of warlord symmetry, the couping warlords should
have equal odds of ending up in power. If there are m warlords attempting a
palace coup, they each have an expected payoff of mp v∗ − c.8 But to preview
analysis in Appendix B, all that is going to matter is the payoff based on the
probability of pulling off a successful coup when no one else is trying to do
the same, which has been defined previously as pv∗ − c. If a warlord i in W L
attempts a coup alone and does not succeed (which will occur with probability
1 − p), it does not affect other warlords’ payoffs so long as s or more support
the president.
But what happens to warlords who played “Install” in the first round, but
were then not selected to be part of the inner circle? In every case where k > l,
there will be at least one warlord i ∈ W P ∈ / W L . This unlucky warlord i was
not transferred anything as a part of x, and is incapable of challenging the
combined force of the l warlords who now back the president. The game ends
for this warlord with a payoff of zero. This warlord played the lottery and

who now only has to pay v∗ − l(pv − c). One can get approximately the same results by allowing
c to be different in the first and second stages of the game (c1 and c2 ), incorporating certain losses
in investment as higher costs of fighting c2 in the second stage.
6 In this stylized setting, incorporated into that probabalistic judgment is not only the
simultaneous defection of k − s additional warlords, but also defeating all of them in the
scramble that will follow.
7 I have no particular theoretical priors on the question of whether the probability of a successful
coup should vary with the number of warlords in W L . It could be that a large number of
warlords in the coalition implies a smaller probability of any one of them successfully seizing
power. It seems equally plausible that a larger number of warlords competing behind the throne
makes coup prospects for any warlord i higher because of collective action problems associated
with organizing a counter-coup. General results still hold if we let p(l) be the probability of
ending up in power after playing “Coup,” as a function of number in W L , so long as p(l) is
decreasing, but not decreasing as fast as 1/l. The derivative of l(p(l)v∗ − c) with respect to l
must be positive.
8 At first this probably seems counterintuitive, as it implies that a warlord playing “Coup” alone
has a higher payoff than playing p in a group ( mp v∗ − c < pv∗ − c). This confusion is an artifact
of the notation – p should change between the two settings. If all k warlords play “Coup,” the
∗
coup succeeds with certainty (p = 1) and m = k, so the expected payoff would be vk − c.
36 Jesse Driscoll
Stage I Stage II
Exiting Anarchy Divide-and-Rule

If k ≥ s, call these k
n warlords in W warlords W P W.
all choose to They go on to
“Fight” or “Install”. Stage 2.
Call the number of v increases to v*, figurehead P selects l ≥ s warlords W L W P
warlords playing President P is installed. and pays them each what they
would receive in a coup, pv* – c
“Install” k.

Any warlord
W p W L gets zero.

If k < s, war continues. If

The game ends for any
war continues, “Fight”
warlord W P, who gets r.
yields v/n − c, and “Install”
yields v/n – c – w.

figure 2.1. The Game (off-the-equilibrium-path outcomes not shown).

lost, and will be marginalized by being shut out of state jobs and racketeering
rents. His political rehabilitation will be revoked. It will not be taboo to notice
his prior criminal behaviors. His army will probably disband itself as soldiers
realize their patron cannot pay them.

analyzing the game: three different stable outcomes

Despite the fact that warlords have the first and last move in the game,
regardless of parameter values, the figurehead president, once installed, ought
to be able to remain in power by buying off a coup. He should be able to
appropriate substantial rents for himself from his ability to play warlords
against one another. He can build a version of a minimum winning coalition
for domestic stability and create a “shadow state,” which is exclusively loyal
to him and exists in parallel to regular state institutions. A president can –
and therefore will – pay s warlords exactly what they would receive in a
coup (pv∗ − c), pay the rest zero, and keep everything else for himself. Figure
2.1 displays the basic choice sets for warlords who expect that a figurehead
president will play the game in this way.9
Knowing this, what is a warlord to do in the first stage? It depends on what
other warlords are doing. If most are fighting, he should fight as well. But even if
enough warlords are working together for a figurehead president to be installed,

9 For those readers who are not acquainted with the jargon of game theory, this simplified
representation of the game is meant as a schematic of what would happen if every player
anticipated that every other player were going to act strategically.
Predator Collusion: A High-Stakes Game 37

each warlord i must compare his opportunities in W P to his reservation value r.

Because s is static, every warlord i’s utility is decreasing in k. Every warlord who
plays “Install” worsens the W L “insider lottery” odds for a warlord in W P .
Because warlords are symmetrical and interchangeable to the president, and
the president does not need all of them, in the first stage warlords are gambling
when they install a president. Still, if v∗ and s are high and the reservation
value r is low, a warlord might opt into a high-stakes “liquidation lottery” – a
probabalistic game of installing a president with the full knowledge that some
installers will be killed as the ruling coalition is trimmed, but others will be in
a position to extort the president and get away with it.
In Appendix B, three different classes of subgame perfect Nash equilibria
are identified, none of which involve the play of weakly dominated strategies:
the state failure equilibrum, various full incorporation equilibria, and various
partial incorporation equilibria. These equilibria have different distributional
consequences.

• State Failure: All warlords fight. The state failure equilibrium is inefficient
but robust. If all warlords are planning to wage war and seize the capital,
an individual warlord i can only make himself worse by not taking part in
the scramble.
• Partial Incorporation: Some install a president, some refuse. In a partial
incorporation equilibrium, some warlords form a coalition to install a
president and some warlords remain outside the consolidation process.
All warlords expect the president will use divide-and-rule tactics to
play one against another, assembling the “cheapest” governing coalition
of warlords possible, and keeping the rest for himself. Installing the
president is a gamble – but a gamble that pays off only for some of the
warlords. The odds of being selected as part of the winning coalition W L
diminish as k grows with every additional warlord who opts to enter
the state in the first period. For some parameter values, it makes sense
for some, but not all, warlords to collude in the extortion game, and
warlords in the state coalition prefer that holdouts remain outside state
structures.
• Full Incorporation: All warlords install a president. In a full incorporation
equilibrium, all warlords act together to install a president. The president
will only distribute the rents of sovereignty v∗ among only s warlords,
so installing a president is a gamble. Even so, for certain parameters, the
potential benefits of being selected to be a member of W L and extort the
president for a share of v∗ are sufficiently attractive to outweigh the certain
reservation value r one would receive staying outside the state, which
induces every warlord to join the state.
What might be observed in a state failure equilibrium? Joint strategies
by warlords create the Hobbesian nightmare of all against all, as various
factions fight each other for survival. Tactical bargains between warlords break
38 Jesse Driscoll

down and rampant side-switching and coalition politics make it impossible

to identify the “master cleavage” of the conflict.10 Violent and unpredictable
anarchy, chaotic looting, and social disintegration follow. Groups may ravage
the countryside with no desire to create order and no political project beyond
day-to-day enrichment and survival.
What might be observed in a partial incorporation equilibrium? Some
warlords have been incorporated into the state apparatus and some have not.
Some warlords inside the state do very well, but some warlords who reject
state authority do just as well. Civil war may simmer at a low intensity. The
sovereignty of the state may be challenged by territorial “shadow states,”
strong organized criminals who operate in defiance of regime preferences or
foreign-backed insurgencies. One might observe stalled peace processes, persis-
tent low-intensity conflict between a loose coalition of urbanized gangsters who
control the capital city and rural gangsters who reject the regime’s authority,
or just a generically “weak state.” What all these situations have in common
is that some “insurgent” warlords reason that there is more to be gained
at the fringes of state control than there is to be gained serving as regime
agents.
What might be observed in a full incorporation equilibrium? The vast
majority of warlords will have been incorporated into the state apparatus
and reinvented themselves as regime allies, policemen, or organized criminals
with strong ties to the regime. Different organizational and institutional forms
comprise the “rules of the game” at this stage, but the underlying arrangements
are not considered to be open for modification. Although the threat of a
coup hangs over the distribution of spoils, these threats can take the place of
actual changes in government. The figurehead president manages a patronage
network that allows warlords to reinvent themselves as state agents in the army,
ministry of interior, tax police, or local government. Violence entrepreneurs are
sated with rents received, and unaccountable patronage networks emerge inside

10 By “master cleavage,” I mean the division between the protagonist and the antagonist in
the war’s “master narrative” – e.g., the incumbents vs. the insurgents, the Islamists vs. the
secularists, the red team vs. the blue team, etc. As Kalyvas (2003) forcefully articulated,
the reductionist impulse to define a war as being reducible to a two-player game – two
“sides” divided by a “master cleavage” – is inherently political, often misleading, and has
real implications for how a conflict is rendered in collective memory. One explanation for the
systematic bias in favor of narratives that emphasize the “master cleavage,” also from Kalyvas
(2004), is that journalists and researchers tend to conduct their research in urban areas – where
they have access to English-speaking academics, NGO workers, and state representatives –
rather than in rural areas, where most of the violence takes place in most civil wars. Another
explanation is that civil war actors themselves are often quite sophisticated propagandists, and
to attract financial and military from great powers, local allies, or far-flung diaspora groups
they misrepresent the content of their motivations or ideology, their unity of command, and
any other “off message” detail that might confuse a potential donor with a short attention span.
Prominent pleas for disaggregation of the “country-war” as the unit of analysis are Laitin and
Brubaker (1998) and Kalyvas (2007).
Predator Collusion: A High-Stakes Game 39

regime ministries. Relationships with the president and with other warlords –
not formal institutions – are the mortar that hold the arrangement together.
Taken sequentially, the three classes of equilibria provide an informal
account of how failed states rehabilitate themselves after extended periods
of violent anarchy. All warlords recognize that a state failure equilibrium
is inefficient and to be avoided if possible. Some warlords initially collude
to provide order, gain access to international wealth, and gain monopoly
rents from the state apparatus that falls under their control. A local “puppet
president” is selected as a placeholder for opaque coalition politics. Compli-
cated bargaining follows and backroom deals are struck. As foreign aid and
investment increase and the president acquires a reputation for fair dealing,
parameter shifts gradually facilitate a switch from a partial incorporation
equilibrium toward a full incorporation equilibrium. In this scenario, peace and
order are supported by warlords’ ability to extort presidents directly and the
international community indirectly. Many warlords transmogrify into violence
subcontractors for the regime.

the stability threshold and the reservation value

To preview the analytic narrative that follows, two parameters critical to the
comparison of Georgia and Tajikistan are the stability threshold (s) and the
reservation value (r). The story of consolidation in the two decades since
independence is the story of declining reservation values. The obvious exception
to the general trend of reservation value decline is in the South Caucasus,
where certain Abkhaz and South Ossetian warlords remain shielded by Russian
military power. The general decline in fortunes for warlords remaining outside
the states can be either endogenous to the processes modeled (i.e., as a stable
governing coalition emerges, it is increasingly capable of coercing atomized
criminal competitors) or exogenous (i.e., outsiders, especially great powers,
provide the coercion). Exogenous shifts in the reservation value are of obvious
interest for students of international affairs. The obvious exception to the
general trend of reservation value decline is in the South Caucasus, where
certain Abkhaz and South Ossetian warlords remain shielded by Russian
military power. The timing of the 1997 peace accords in Tajikistan conforms
with shifts in Russian military policy toward Shah Massoud, the Tajik warlord
in Afghanistan, who stopped providing safe havens for warlords across the
border. The U.S. invasion of Afghanistan dramatically reduced the expected
utility of running a nonstate armed group in Central Asia in a way that
Tajik warlords simply could not have anticipated on September 10, 2001. As
reservation values fall, the attractiveness of taking part in the v∗ lottery – even
if it is acknowledged to be a rigged lottery – increases.
The stability threshold s for a particular state is determined by international
actors, notably foreign governments and offshore capital markets. It captures
how a regime is regarded by important international actors that are in a position
to provide recognition, aid, and military assistance. These actors attempt to
40 Jesse Driscoll

include in their assessment of local conditions (military technology available to

warlords, terrain, distributions of popular support for radicals in the popula-
tion at large), but their access to information about these conditions is limited.
Neighborhood effects matter a great deal. In some settings, foreign powers are
desperate to recognize an agent – any agent – to “keep a lid” on revolutionary
activism in a state and will send assistance to keep their selected beneficiary
in power. If a regime has strong external backers, then the stability threshold s
may be n3 , 4n , or lower. A low stability threshold is more likely when the threat of
anarchy from certain out-of-coalition gangsters threatens the collective security
of the international community. At the opposite extreme, if foreign powers
dislike the political orientation of the de facto government and respond with
sanctions or shunning (as often observed in postrevolutionary settings), the
stability threshold s, then, may be close to n. A high stability threshold is justi-
fiable when a regime faces strong international pressure and the out-of-coalition
alternatives are seen as viable alternatives by the international community.
Analyzing the consolidation game with different stability thresholds is meant
to capture differences between post–civil war regimes capable of attracting
foreign patrons easily and post–civil war governments that are under constant
pressure from outside forces. How does the game play out differently? When s
is very high, most of the warlords in a country are necessary to provide order.
This increases the “lottery odds” for a single warlord, making him more likely
to be in a position to extort the president and get away with it. When s is
very low, a small coalition of warlords is sufficient to keep the president in
power. Equilibrium predictions are equivocal. If s is low, it is relatively easy to
form a government but relatively difficult for the president to credibly commit
to distribute wealth widely. This is a situation that has lower “lottery odds”
from the perspective of any single warlord, as the president can easily play one
out-of-coalition warlord against another and drive expected rents toward zero.
By contrast, divide-and-rule is more difficult if s is high, but it is also more
difficult to install a president in the first place.
When the stability threshold is extremely high, a partial incorporation
equilibrium closely resembles a full incorporation equilibrium. Nearly all the
warlords have to be bought off to achieve a minimum of stability. When
the stability threshold is very low, a partial incorporation equilibrium has
a very different feel – more of a high-stakes standoff between factions. Full
incorporation, if it emerges at all, does so only when warlords’ reservation
value r drops precipitously toward zero. Chapter 5 frames Tajikistan and
Georgia as exemplar cases of “low stability threshold” and “high stability
threshold” consolidation projects.

observable implications
This chapter was motivated by the question: If warlords were the ones best
positioned to profit from the breakdown of social order, and they knew that
Predator Collusion: A High-Stakes Game 41

they were eventually going to be divided against each other by the president,
why would they ever install a president in the first place? The answer is that
they anticipate the total amount of wealth in the country to extort (v) will
increase once the floodgates of foreign aid (v∗ ) are opened. The decision to
install a puppet president is akin to buying a lottery ticket on inclusion in a
winning coalition. Winners get to extort the president for offices, privatization
rights, and de facto monopolies. How many opt in to “play the lottery” depends
on their outside options, the stability threshold (higher implies better lottery
odds), their odds of successful coup once president is installed (more go in
if odds are better), and the value of increased revenues from international
support (higher is obviously better). Equilibrium selection is fundamentally a
matter of local politics, and very difficult to predict in advance. What coalitions
emerge is a result of politics, persuasion, personalities, and path-of-play as
much as parameters. Informal patronage structures are assumed to run parallel
to formal institutions. Warlords opt to civilianize themselves and loot the
state from within because they can predict a vast quantity of potential wealth
available to the “shadow state” coalition.
Social order after civil war emerges out of collusion by predatory violence
entrepreneurs in this account. Instead of treating the armed strength of a rebel
challenger as an exogenous model parameter, or giving causal weight to the
policies of the third-party intervener, this approach seeks to explain variation
in the relative size and strength of the incumbent and insurgent coalitions. The
comparative statics of the model are straightforward and intuitive. Conditional
on having achieved a partial incorporation equilibrium, the likelihood of a
full incorporation equilibrium should increase with v∗ . A higher ratio of v∗
to v means more wealth for the warlords to steal. The reservation value
for staying outside the consolidation process r is also a critical parameter,
as discussed earlier. For easy exposition, in Figure 2.2, I assume the most
difficult case for consolidation: one in which r never declines. Regardless of
how many warlords install a president, in both worlds I hold r identical to
the “war economy” payoff, nv − c. This means that by choosing to “Fight”
in the first stage, a warlord i can guarantee himself a “war economy” flow
payoff. In this simplified setting only positive inducements – higher v∗ –
can induce a change in strategy. More generally, in this stylization higher
costs of war and violence c translate directly into low reservation wage “war
economy” payoffs, making it easier for the president to convince recalcitrant
warlords to join the state. One might think about of v∗ as carrots and c as
sticks.
It is clear in the top graph in Figure 2.2, which shows the payoffs in a
partial incorporation equilibrium, that the payoffs to warlords who remain
outside the state are identical to those inside the state. Because warlords
are symmetric in power by assumption and indifferent between strategies in
equilibrium, the model suggests that it is very difficult for anyone – foreign
analysts or the warlords themselves – to predict which warlords will accept
the figurehead president and which ones will continue to oppose the state’s
42 Jesse Driscoll

Spoiler/Fighting Payoffs
Installing President Payoffs
Payoffs
per
Warlord “State Failure” “Partial Incorporation”
(by strategy) Equilibrium Equilibrium

0 25 s 50 75 100
% of Total Warlords

Spoiler/Fighting Payoffs
Installing President Payoffs
Payoffs
per “Full Incoporation”
Warlord “State Failure” Equilibrium
(by strategy) Equilibrium

0 25 s 50 75 100
% of Total Warlords

figure 2.2. Comparative Statics: Outsiders assist consolidation by increasing v∗ .

consolidation project. Because the president can tailor v∗ rewards to field

commanders individually, the model implies that in a partial incorporation
equilibrium no measurable characteristics of a warlord should predict loyalty
or defection because they would already be priced into the payoff offer that the
president used to buy their loyalty. Homogeneous warlords may, in equilibrium,
rationally adopt heterogeneous strategies.
The notion that partial incorporation equilibrium can be sustained even with
completely identical warlords is a subtle but important departure in this book’s
treatment of the post-Soviet wars from many other accounts. That certain
armed actors attempt to stay out of the consolidation process, calculating
that they can do better at the fringes of state control, may have nothing to
do with linguistic or religious difference, regional economic grievances, or
anything else. Many analysts have succumbed to the temptation to label the
politicized cleavages that emerged from the post-Soviet wars as “primordial” or
“natural.” A perspective based on coalition formation, by contrast, is designed
Predator Collusion: A High-Stakes Game 43

to accommodate side-switching political reversals. When v∗ gets very high or

the reservation value r drops, the model predicts that cultural distance will
be paved over, as warlords opt to quietly join the state. Differences can be
deconstructed based on cold calculations of interest. The Muslim territory of
Adjara can, at the whim of a few people, rejoin the Georgian polity.11
These comparative statics are not unqualified. Strategies are determined
jointly, and the model explicitly admits the possibility of a violent state
failure equilibrium regardless of model parameters. This has troubling policy
implications for outsiders attempting to shape incentives in failing states. If
armed actors have decided to seize the statehouse or extort the president,
eleventh-hour wealth transfers may simply increase the value of the prize
being contested. Once Georgian and Tajik criminal actors believed that others
planned to break from the social contract and attempt to seize the capital
with armed force, no amount of aid was likely to halt the slide toward violent
anarchy. Put differently, there are two fully sustainable types of “war equilibria”
possible in this model, and only one of them, the partial incorporation
equilibrium, is even in principle susceptible to manipulation by outsiders.
Returning to Figure 1.1 in Chapter 1: the assumption of symmetric warlords
anchors this book in the lower half of the table, where foreigners are assumed
to be blind to local nuance. Against the anarchic backdrop of state failure,
I assume it is too costly for foreigners to shape coalition politics in real
time.12 All of this suggests a relatively blunt set of financial, diplomatic,
and military tools.13 These tools, and their inherent limitations, are clearly
familiar to contemporary American practitioners of counterinsurgency, and
they were equally well understood by Russian military professionals in the early
1990s. Subsequent chapters will tackle the politics of equilibrium selection,
and the movement from one type of equilibrium to another (“state failure,”
“partial incorporation” or “full incorporation”) over time. But the account will
emphasize that Russians attempted to shape the contours of war termination
indirectly – primarily by inducing warlords to cast their lot with the government
(in Tajikistan) or stay outside the consolidating state (in Georgia).
Russia’s role in both Georgia and Tajikistan continues to be disputed by local
civil war participants and will be the source of debate by future generations of
Georgian and Tajik historians. To the extent that there was variation between
Russia’s policy toward Georgia and its policy toward Tajikistan (as most
observers would agree that there was), the model cannot explain it – only
assume it. The model cannot account for adaptive changes to postwar social
structures in response to foreign aid programs, as v∗ is assumed to be divided up
in the form of graft. The model cannot explain why some separatists succeeded

11 Marten (2012), 80–84.

12 See, for instance, Filkins and Gall (2010).
13 See Fearon and Laitin (2004), 33, FN 70.
44 Jesse Driscoll

in making good on their demands and others did not.14 Nor can it explain why
violence escalated to war in some parts of the Soviet periphery and not others.
What exactly does the model do? The model frames the politics of regime
consolidation as being primarily about the strategic relationships between
warlords as they second guessed each other’s strategies. If the president is a
placeholder for a warlord coalition, the easy distinction between “formal”
(desirable) and “informal” (corrupt) institutions is destroyed. Presidents are
installed, or not, because of joint warlord strategies. Once installed, presidential
reneging is not constrained by the rule of law, but by the demonstrated ability
of warlords to cooperate and collude together – to threaten a coordinated
coup if the president breaks his promises, or to demand that some warlords be
included in (or excluded from) the inner circle, W L . Inherited beliefs and social
structures constrained the path of play.15 The model also sheds light on the
question of why the period just before and after presidents were installed was
so violent. The presidents had not yet selected W L , and warlords in W P were
jockeying for position in a high-stakes lottery, with a share of v∗ as the prize.
The model does not attempt to explain the expansion and contraction of militia
memberships – it explicitly treats warlords, and the militias they command, as
fully interchangeable, identical, and unitary. Viewed from the streets, of course,
warlords are not at all interchangeable. Popular and capable warlords could
recruit larger armies and demand to be “bought out” by the state. Wartime
coalition politics were tremendously uncertain and contingent, and there was
great uncertainty about who the warlords were and which ones were strongest.
Many foot soldiers came to imagine that their individual contributions of labor
to a militia could change the course of the consolidation process. All of this is
detailed in Chapter 4.
The model also emphasizes that the president is in an agency relationship
with warlords, whose threat of a coup hangs over distributional politics.
Incrementally stripping bargaining power away from the men with guns is a
local process, with the threat of anarchic violence ever present. The model
describes an oligopoly of violence emerging based on structures of personalist
rule. These structures are described in Chapter 5.

14 In the language of the model, some warlords have higher reservation wages than others because
of historical inheritance, or unusually good luck at attracting foreign assistance. Appendix B
has an extension showing that heterogeneity in r can yield unique equilibrium predictions.
15 Analysis of beliefs generally pushes the scientist toward ephemeral matters: trust, shared
values, focal points for identifying when to coordinate for a coup, and other gritty particulars
of human political action. These kinds of politics are virtually impossible to analyze in the
absence of formal institutions, however, because doing so requires assumptions about what
other warlords would do off-the-equilibrium-path. Data on counterfactuals are rarely reliable.
Subgame perfection cuts through this knotty problem by assuming that all of the warlords hold
the correct beliefs, informed by history and culture, and are at least in part endogenous to the
path of play, as described in Ferejohn (1991), 285. See also Bates et al. (2000), 699–700 for a
useful discussion of the disadvantages of incorporating models of incomplete information into
analytic narratives, and Thelen (1999), generally.
Predator Collusion: A High-Stakes Game 45

Finally, this account gives tremendous agency to warlords, as the political

actors capable of overturning local order. Who were these warlords? How
did they come to seize the reins of politics in post-independence Georgia and
Tajikistan? Why did violence escalate so quickly? Before we can exit from
anarchy, we must first enter it. The narrative journey from independence to
state failure begins in the next chapter.
3

Kto Kogo?

The collapse of the USSR was unexpected and unexpectedly peaceful.1 The
ideological superstructure disintegrated, the leviathan ceased to exist, yet across
Eastern Europe, Russia, and Central Asia, violence rarely escalated. A few
social actors seized security structures and dragged their states into chaos, but
this sort of thing was not even attempted in the Baltic states or in most of
the new states of Central Asia.2 In the Kyrgyz city of Osh, in the ethnically
mixed Ferghana Valley, there were violent pogroms in 1990 – followed by
a court-led investigation by the new Kyrgyz government in 1991, where
forty-six of the forty-eight participants in the pogroms charged were found
guilty.3 Georgi Derluguian (2005) recounts the story of the tiny Caucasian
republic of Kabardino-Balkaria, where escalation tactics by rowdy warlords
were cauterized by local innovation.4 Are there patterned regularities to the

1 For seminal contributions to social scientists’ understanding of the Soviet collapse, see Roeder
(1993), Solnick (1998), Bunce (1999), Treisman (1999), McFaul (2001), Beissinger (2002),
Brown (2007), Roeder (2007), and Kotkin (2008). For students seeking a shorter summary, see
North (2006), 146–154.
2 Bunce (1999) makes a convincing case that the variation in post-communist violence can be
predicted by institutional characteristics, particularly by the composition of the armed forces
(102–126).
3 This event is documented extensively in Tishkov (1995).
4 Derluguian’s tells the tale of the dog that did not bark like this: Kabardino-Balkaria had
emerged from an impenetrable mountain backwater to a relatively wealthy ski resort town
during the 1980s. The Balkars, 10 percent of the new republic’s population, found themselves
underrepresented when inaugural elections removed Soviet ethnic quotas. Deadly ethnic war
was looming. A local hero, fellow sociologist Yuri Shanibov, stepped in. Transforming himself
into the pious Muslim “Musa Shanib,” and, backed by a rowdy crew (“athletes – wrestlers,
boxers, martial artists – veterans of the Afghan war and simple hooligans [ready] for the fight”
266), he brokered an elite compromise that satisfied the Balkars and staved off a repeat of
Chechnya. Or so the story goes. The rarity of English-language accounts like Derluguian’s,
however, draws attention to an observational bias at work on studies of violence and order:

46
Kto Kogo? 47

divergent outcomes – violent or nonviolent, democratic or nondemocratic –

in the post-Soviet space? David Laitin (2006), in a playful summary of
Derluguian’s class-based analysis, speculates the following answer to this
important question:

First, there is the nomenklatura, the high officials of the Soviet state. … Provincial Soviet
life involved families buying state or party appointments, in order to then distribute
bribe-friendly posts to relatives. All this was quite comfortable for the nomenklatura
until the state began to unravel. They then had to make a historic choice: they
could steal what they could of state assets and run; they could seek support from
the newly reconstituted centre in Moscow to help them regain power; or they could
transmogrify into nationalist elites and seek to lead independent states. … Second, there
are the national intellectuals, a sub-class of the industrial proletariat. … Universities,
Palaces of Culture and local soviets assured positions for this new class of national
intellectuals. As Lenin hoped they would, these regional official intelligentsias defused
nationalist mobilizations; but since the holders of the positions had no prospects for
mobility outside their titular republics – ‘their credentials did not travel beyond the
republic’s borders’ – a core of bored intellectuals developed in each republic with close
ties both to one another, and to national intellectuals from other republics. … Third,
there are the ‘sub-proletarians’ … [a] ‘residual’ class [that] includes those who remained
outside the state hierarchy, surviving on subsistence agriculture and migration to
perform seasonal work, or carry out petty trade. … [T]heir alliances and relationships
[determined] … likely outcomes: formal democracy, restored autocracy or chaos. The
route to chaos in the Caucasus tended to begin with the retreat of the nomenklatura, now
facing those they had oppressed and without a Soviet big brother to protect them. They
either escaped with the money acquired through the sale of state assets or remained at
home and re-identified themselves as national intellectuals. An alliance between national
intellectuals and sub-proletarians tended to follow. This alliance, in large part because
the national intellectuals could not rein in the rowdy habitus of the sub-proletarians,
drove all the regions of the Caucasus where it occurred into near chaos, and several over
the brink. … The chaos in the Caucasus was but one of three patterns that emerged from
the relationship between nomenklatura, national intellectuals and sub-proletarians. In
the post-Soviet republics as a whole, the most prevalent pattern was the reclamation
of power by a powerful rump of the nomenklatura, this time capable of controlling
the sub-proletarians, with the relatively weak national intellectuals brought into the
ruling coalition. These allies feared the rise of sub-proletarians as mujahedin, and in the
name of secularism and modernity re-imposed authoritarian rule. The Central Asian
republics are the key instances of this pattern. In the third pattern, of which the Baltic
States are the prime examples, the sub-proletarians were weak. An alliance of the
national intellectuals, whose considerable social power derived from their elite positions
in pre-revolutionary civil society, and a reformed nomenklatura, which re-identified with
the national intellectuals – and responded to the lure of membership in the EU – paved
the way for a peaceful transition to representative democracy.5

Our community devotes orders of magnitude more energy to writing books about a very rare
outcome, the breakdown of order (in, for instance, Chechnya) than we do to writing books
about the modal outcome, which is the maintenance of order (in, for instance, Adygeya).
5 Laitin (2006).
48 Jesse Driscoll

Across post-Communist Eurasia, the slide toward violent state breakdown –

where it occurred – began when important social actors lost faith in the ability
of the new regimes to secure their interests. This loss of faith was not inevitable,
even when the weighty structural trends identified by Laitin are taken into
account. Though the shock to state capacity that accompanies decolonization
often leads to civil war, peaceful decolonization is the modal outcome. Though
war occurred in a third of the newly independent states that emerged from
the breakup of the Soviet state, in two thirds of the new states violence did
not escalate beyond street demonstrations. Roeder (2007) emphasizes that
the Soviet legacy bequeathed its federal units (its Soviet Socialist Republics
[SSRs]) all of the raw materials for the provision of order: a strong Party
system for the distribution of spoils, police and transportation infrastructure,
material for a national army, and the institutions necessary to maintain
self-renewing self-referential nationalist histories. McFaul (2002) shows that
the broad trend across post-Soviet Eurasia was pacts between ex-Communist
Party elites to effectively restrict access to the political space and freeze politics
in an authoritarian mode. Many of these arrangements can, at some risk of
controversy, be said to persist to this day.
What unfolded in Georgia and Tajikistan to make them exceptions to this
rule? In these states, elites could not quickly cauterize violence, and the descent
into anarchy was horrifyingly rapid. This chapter makes the case that war, as
it emerged in Georgia and Tajikistan, can be best understood in the context
of a more general scramble for post-Soviet spoils. Though these civil wars
are typically discussed in the context of “ethnic conflict,” or the preferences
of leaders in Moscow, this chapter suggests that the core political failure in
both cases was a violent breakdown of intra-ethnic bargaining within the
post-independence national elite, after certain members of the elite succumbed
to the temptation of allying with warlords.

post-soviet predators: who were the warlords?

The breakdown of the Soviet state involved a large-scale defection of the Soviet
bureaucracy. The result was a gradual collapse, and in some cases a complete
collapse, of the public’s faith in the state to provide that most basic public good:
security. Three classes of actors began second-guessing each other’s strategies.
The first were national elites. There was an obvious symbiosis between
the interests of the nomenklatura and the various criminal networks that
could access state transit networks to move resources within the Soviet Union.
Rehabilitating these transit networks, and securing new wealth attained in the
post-Soviet fire sale, required forging local alliances with individuals capable
of guaranteeing property rights against a backdrop of anarchy.
The nomenklatura began crafting a complex set of bargains with a second set
of actors, which I have already defined as warlords. These men proved capable
of recruiting private armies and “running the streets” when suddenly there was
Kto Kogo? 49

no one to make arrests. Many of these warlords affiliated with new radical
parties.
A third set of strategic actors were the street recruits who served at the behest
of the warlords. For a number of years, the primary social actors in Georgia
and Tajikistan were well-armed militias – with rotating leaderships and flexible
memberships – recruited and loosely organized by family, clan, and kinship
(avlod, in Tajik) ties. Foot soldiers were recruited primarily, but, as we shall
see, not exclusively, from the criminal or rural underclass.
As the political order began to fray, many elites in the national republics
realized that the end of communism and the transition to independence would
bring economic contraction. In the short and medium term, there would be
a smaller “pie” to divide, as they anticipated the dual loss of fiscal transfers
from Moscow and various efficiencies associated with membership in the Soviet
Union. The struggle for control over resources began in the late Soviet period
and quickly extended to every facet of social organization. Whether elites were
squabbling about how to divide up property rights over natural resources,
ownership of factories servicing obsolete industrial production quotas, or even
battles over protected institutional niches (e.g., museums, university structures,
churches and mosques, sports complexes), similar battles unfolded. Everyone
understood that there would be no social safety net to manage the resulting
dislocations.6
Solnick (1998) chronicles this process analytically, creatively deploying
the principal–agent framework. Well-connected party bureaucrats (“agents”)
strategically timed their “defection” (liquidating state assets, putting the profits
into offshore accounts, and moving their families to secure enclaves) against a
“principal” (the “Soviet State” or the “median Soviet voter” – both of which
were increasingly imaginary). Because this self-serving behavior was possible
only if the probability of monitoring and subsequent punishment was small,
the dynamics of the resulting state failure took on characteristics of a bank
run: once it became obvious that some people were defecting, everyone else
wanted to defect as well. Those in power “rushed to claim … assets before
the bureaucratic doors shut for good.”7 Watchful eyes in Moscow could not
fix their gaze everywhere at once, and first-movers in peripheral bureaucracies
were able to use official letterheads and unofficial influence networks to legalize
practices that had been tacitly accepted for decades. Mid-level bureaucrats
were well positioned to simply take things they wanted – dachas, town
cars, downtown apartment buildings, factories, mineral deposits, commodities
futures, and, ultimately, even phones and carpets from the buildings in which
they worked.

6 Olcott (1994), 45.

7 Solnick (1998), 6–7, 251.
50 Jesse Driscoll

Many accounts of the early 1990s tend to emphasize the continuity of

political cadres from the late Soviet period in the Caucasus and Central Asia.8
Overemphasizing continuity at the top of the power pyramid led many analysts
to ignore a parallel trend: traditional elites were inviting new actors into
the political arena to secure their holdings. It was a time of unprecedented
social opportunity for men who could organize guns. This environment, where
money had recently become worthless, rewarded individuals with a talent for
activating latent social capital and articulating a political program that could
justify using violence against unarmed civilians. As a result, during this period,
any journalist or social scientist who went to these collapsing states looking for
evidence of “ethnic hatred” or “ingrained patterns of resentment” was sure to
find exactly what he or she was looking for. As a general rule, these men who
would be warlords had no trouble at all justifying their actions and explaining
to anyone who would listen why stakes were indivisible.
The emergent class of power brokers – the warlords – were men who could
command respect among a social milieu composed of alcoholics, criminals, day
laborers, hooligans, and the serially unemployed. The slow disintegration of
order in the late Soviet period provided these sub-proletarians a unique window
of opportunity to convert their violent social capital into a better future.9 Many
had settled into pedestrian lives at the time when social order began to collapse.
The men who would emerge to lead them were not always from the same
social class. The Soviet implosion was a rare second chance for a certain type
of person to remake his life. Local drug dealers began to imagine themselves
running countrywide smuggling operations. Rural bazaar protection racketeers
began to imagine themselves managing urban hotel chains. Dzhaba Ioseliani,
the eventual leader of the Mkhedrioni in Georgia, had been a playwright
and artist in the USSR. In communist times, “General” Dumbadze worked
in a sewing factory, and “Chief” Kobalia managed supplies and inventory
at a bakery.10 Prominent field commanders in Tajikistan included a former
construction engineer, a cement factory floor manager, a cotton gin operator,
a number of schoolteachers, a visa expediter, a popular singer, a bartender,
and a bus driver. The security services, police, and military produced their
share of charismatic leaders, and had a comparative advantage in acquiring
weapons and recruiting shooters. All of these men had come of age in a system
in which familial and kin networks were necessary for everyday life, and they
recruited their friends and allies with the understanding that getting a foothold
within the new state bureaucracy represented their best chance to increase
their life opportunities. The most powerful Tajik and Georgian paramilitary

8 For generalizations based on elite continuity and institutional continuity, see Suny (1995);
Jones-Luong (2002); Collins (2006); and McFaul (2002).
9 I gratefully borrow the phrase “sub-proletarian” from Georgi Derluguian, who borrowed it
from Pierre Bourdieu’s observations of social structures in Algeria.
10 Darchiashvili (1997a).
Kto Kogo? 51

commanders – Tengiz Kitovani and Sangak Safarov – had both served time in
the Soviet prison system.
Not everyone was equally talented at this sort of social organization.
It required not only day-to-day charisma and a gift for improvisational
management, but also the rarer ability (much prized in CEOs) to project and
impose a strategic vision. It also rewarded natural talent for oratory – or at least
for a particular kind of calibrated hate speech. Amidst the collapse of Soviet
social order, many learned for the first time that they had a gift for “power
language” and guiding the mob, and could deploy symbols and phrases to
generate authority out of thin air.11 With a certain natural leadership, aided
by a few resources such as weapons, ammunition, and some guarantee of
immunity from prosecution, certain men found themselves transforming their
prison gangs, martial arts dojos, or even (in the case of Dzhaba Ioseliani) their
theater clubs into potent political and economic actors. One could rise from
obscurity to local celebrity by demonstrating a willingness to make good on
violent threats.
Many warlords reappropriated religious symbols and iconography. It is not
fully inaccurate to say that the Mkhedrioni (“The Knights of Georgia”) were
a paramilitary wing of the Georgian Orthodox Church. Many recruiters and
enforcers displayed ornate crosses. Many opposition leaders in Tajikistan were
holy men, and in the euphoric aftermath of the USSR some of them competed
with each other to articulate the most extreme possible positions. It is easy
to exaggerate the importance of religious ideological content by simply taking
these extremists at their word. But as the quotes in this chapter make clear, most
of these ecclesiastical arguments were weaponized primarily in interaction with
locally relevant gendered claims. The ancient, emotionally salient claim that the
war’s winners would lay claim to the bodies of the losers’ females was made
early and often.
There were three analytically distinguishable pathways to wealth for these
warlords. The first was to encourage their followers to go out, take things, hurt
people, and then to sell protection against the anarchy that they could unleash.
The second, requiring more subtlety, was to serve as a broker, contract
enforcer, and “silent partner” for political or economic elites. At least for a time,
neither the Soviet state nor the post-independence governments could serve
these functions. Some warlords took advantage of new economic opportunities
and became businessmen themselves. Once an import–export bottleneck was
controlled by the warlord, there was a temptation to cut politically connected
civilians out of the loop.

11 A classic observational study of crowd dynamics used the following phrasing: “When studying
the imagination of crowds we saw that it is particularly open to the impressions produced by
images. These images do not always lie ready to hand, but it is possible to evoke them by
the judicious employment of words and formulas. … Reason and arguments are incapable of
combatting certain words and formulas.” LeBon (1895), 96–97.
52 Jesse Driscoll

The third was aggrandized extortion, such as threatening the president with
a coup and the entire state with a return to general civil war. As the following
chapters show, in the 1990s, Georgian and Tajik warlords realized they had the
ability to replace presidents, trigger investor pullouts, and drag their countries
into anarchy. This extortion dynamic is easy to miss in two-player bargaining
models, or in approaches that define “warlords” as criminal actors distinct from
the state.12 Warlords had only to make these threats credible to be bought off –
not to actually make good on them.
The kinds of informal arrangements that emerged in parallel to state
institutions are well understood. The most important relationship for any
would-be criminal businessman was with “a roof” – a patron with ties to
the shadowy political superstructure, capable of structuring the formal “rules
of the game” to the advantage of insiders. This arrangement is described by
North, Wallis, and Weingast (2009) as “The Natural State” or a “Closed Access
Society”:

The natural state reduces the problem of endemic violence through the formation of a
dominant coalition whose members possess special privileges. The logic of the natural
state follows from how it solves the problem of violence. Elites – members of the
dominant coalition – agree to respect each other’s privileges, including property rights
and access to resources and activities. By limiting access to these privileges to members
of the dominant coalition, elites create credible incentives to cooperate rather than fight
among themselves. Because elites know that violence will reduce their own rents, they
have incentives not to fight.13

Warlords were ideally suited to serve as pivot players in this relationship.14

Some members of the emerging political class began to use militias as a source
of political insurance against being cut out of the spoils of privatization. For
warlords, these arrangements offered a rare opportunity to improve their
lives. Their challenge was to convert their ability to organize armed social
capital into wealth, power, and possibly even job security and respect over the
long term. What followed was essentially the high-stakes coalition formation
process modeled in the previous chapter: Winners would receive a position in

12 Kimberly Marten (2012) defines warlords as always and already distinct from the state in her
theoretical framework: “[T]he principal actor (the state) relies on an agent (the warlord) to
fulfill assigned tasks. … A warlord bargain is logically akin to the problem faced by states that
employ private security contractors as a cost-saving measure, and risk having their overall
strategic plans undermined by undisciplined tactical behavior that they cannot control.” 30.
13 North, Wallis, and Weingast (2009), 18.
14 For extensive details of the kinds of bargains that emerged in Georgia and Tajikistan, see
Aves (1996), Jones (1997), Akhmedov (1998), Nourzhanov (2005), Pirseyedi (2000), ICG
(2004), de Waal (2005), Rubin (1998), Torjensen (2005), and Slade (2007). In Georgia the
most profitable smuggling industries were either drugs, black market currencies, or citrus fruits
from Abkhazia, which could be resold on the Siberian black markets at a huge markup. Cotton
mono-cropping dominates the Tajik economy, as documented in Van Atta (2008).
Kto Kogo? 53

a government security ministry, gain control over a populated fiefdom (often

a privileged position in a wealthy kolkhoz, or de facto mayoral rights over a
small town or city), receive monopoly taxation rights for a bazaar or a lucrative
gray-market bottleneck, or obtain the ability to dictate property rights over cash
crops and mineral deposits. Losers faced exile or worse.
Violent social mobilization was an outgrowth of predatory politics that were
ubiquitous during the breakup of the Soviet Union. The details of how these
processes spiraled out of control to total state failure require more local flavor,
which will be added presently. For a time, predatory criminal militias were in
the best position to take advantage of this uncertainty. Subsequent chapters of
this book will provide evidence that the nationalist warlords understood that
they were temporarily vital but functionally interchangeable with each other as
local order-providers, and thus ultimately disposable when the political winds
changed. But certain lucky warlords managed to play the violent high-stakes
consolidation game successfully, emerging from the lottery as very rich men.

why did the followers follow?

What leads someone to join a violent militia? There are three broad answers to
this question. The first suggests that “the costs of collective action” are simply
exaggerated. To the extent that post-Soviet militia formations conflated the
rhetorical power of nationalist politics with the intoxicating allure of criminal
fraternities, members may have simply calculated that they could have a pretty
good time as violent extortionists when there was suddenly no one to arrest
them.15 For certain young men ritualized kinds of violence are a great deal
of fun.16 So long as this violence was committed by agents of local majorities
against unprotected minority targets, these actions did not come with a risk
of retaliation or escalation.17 At the extremes of this line of argument one
can notice that the breakdown of social order allowed certain sociopaths
to do what they wanted to do anyway: rape and kill their neighbors and
afterwards seize their neighbors’ apartments. Certain men found that they

15 See Keen (1998), Kaldor (1999), King (2001a), and Zuercher (2005). A hauntingly memorable
expression of this argument as applied to Bosnia’s war can be found in Sacco (2003).
16 See Buford (1993), Mueller (2000), and especially Petersen (2002). Various musings on human
nature consistent with this insight are scattered throughout Keegan (1994).
17 When mobilization was not perceived as being risky, violence against helpless minorities
was sometimes just a pick-up game; (see Mueller (2000), Fearon and Laitin (2000), and
King (2004a) on this point). But there is increasing consensus that practically nothing that
happens in a civil war zone can be described adequately without some account of how armed
groups organize themselves, and how they relate to the civilians who give them moral and
economic encouragement. Prominent scholarly contributions have documented how social
belief structures sustain militia solidarity in the face of disciplined state resistance (e.g., Petersen
[2001], Wood [2003]); emphasized social stigma against violating taboos (e.g., Brass [1997],
Ellis [1999]); or shown how the structural constraints that shaped militia recruitment shaped
subsequent tactics and strategy (e.g., Fairbanks [2002], Weinstein [2006]).
54 Jesse Driscoll

enjoyed the psychotic rush of breaking taboos against violence and the power
that came from the reversal of long-held status hierarchies. As one of my Tajik
respondents admitted:

“If they hit one us they knew the rest of us … we would send more people after them,
after their women, their children. And they knew that we would remember. We would
keep coming and coming and never stop, never! Not until they are dead or they run –
run to Moscow or Tashkent, to drive a cab and live like a slave with no visa. If they
wanted to live here [in Dushanbe], they knew they would need to deal with us. … If you
have a gun, and I just have this [bottle], but I’m willing to use it on you, and you’re
afraid, I will win. That’s what war is!”18

Sociologists specializing in gang behavior tend to take umbrage at this

caricature of the criminal mind. Most police officers and military professionals
would take umbrage as well. Not every soldier, gang member, or criminal
who commits violence suffers from psychopathology. Many participants in
the post-Soviet wars were clear-headed and relatively well-educated products
of the Soviet lower middle class. This suggests a second set of answers, one
emphasizing the sociology of institutions. The psychological, moral, and social
character of militia solidarity groups anchors identities for disaffected young
men while inculcating members in a culture of violence.19 Militia organizers in
weak states free ride on masculine norms of honor, potent religious and national
scripts, and other engines of social solidarity.
Students of economics and management science can be counted on to
invoke a third class of answers: Militia recruits clearly responded to material
incentives. From the perspective of a rural Georgian or Tajik, or a high school
dropout in the capital city, part of the appeal of participation in these popular
movements was that they lowered the barriers for entry into a respectable
life. They knew that, if lucky, in return for their sacrifices they would be
allowed to don uniforms and join the ranks of the state security services. Social
rewards would be passed on to families of the winners at the expense of the
losers. Treating militia formations as organizational structures allows them to
be modeled as if they were aberrant firms: consuming resources, organizing
labor, and ultimately producing violence.20
None of these three approaches is obviously wrong. Coding each of my
173 interviews by rhetorical content reveals systematic patterns in respondents’

18 Interview conducted May 21, 2006 in Dushanbe.

19 Useful contributions informed by this tradition include Albini, Rogers, and Shabalin (1997);
Goldstone (1994); Blasi, Kroumova, and Kruse (1997); Darchiashvili (1997b); Tishkov (1997);
Derluguian (2005); Whitlock (2005).
20 Useful contributions informed by this tradition include Popkin (1979); Mason and Krane
(1989); Gambetta (1996); Dixit (2004), and more recently, Weinstein (2006), Berman (2009),
and Shapiro (2013). Post-Soviet case data consistent with this school of thought can be found
in Volkov (1999, 2002) and Varese (2001).
Kto Kogo? 55

table 3.1. Cheap Talk: Some of the Same Scripts Were Well Rehearsed

Variables All Georgia Tajikistan

(%) (%) (%)
Agreed that “You Can Only Trust Your Family” 82 81 83
Stayed Because Life Was Better Inside Group 94 94 94
Militia Sometimes Engaged in Theft or Rackteering 90 84 96
Stayed Because Friends Were in the Militia Too 74 78 70
Joined to “Help Win the Civil War” 76 68 84
Most Militia Members Talked about Politics 36 30 42
Joined to Protect Their Families 83 83 84
Joined to Influence Politics in the Capital City 83 90 77
Stayed for Promise of a Job 71 68 73
Observations 173 84 89

accounts of the sustaining mechanisms for militia membership.21 Note in Table

3.1 that certain reasons for joining a militia were common. Local concerns
dominated recruitment narratives. The awareness that these militias were
players in a larger game for political dominance in the capital cities of the newly
independent states penetrated deep into the ranks of the militias. Most joined
with friends.22 Most could articulate the cause for which they fought. Most
were emphatic that they joined to protect their families, but this was easily
reconciled psychologically with heading to the capital city when the time came
as it was obviously where the high-stakes political questions would be resolved
and where the crowds would demonstrate. Groups were organized with an eye
toward potential personal gains from war: securing their family property first,

21 I borrow the distinction between “triggering mechanisms” and “sustaining mechanisms” from
Petersen (2001), 13. The influence of Elster (1989) on Petersen’s argument is clearest on pages
82–88 and 113–123.
22 Militia members in both states often used the plural (“we”) when discussing decisions to
join a militia, switch commanders, or return home during the violent purges of the “Time
of Troubles”. When militia members had to lobby together to demand payment from a
political or warlord patron, these ties proved quite important. I speculate that the psychological
mechanisms that kept soldiers working together through difficult times – honor and shame –
were channeled primarily through these initial solidarity networks, as these men knew they
would eventually return to the same families, village, and neighborhood networks where there
were few secrets and no privacy. It is also notable that dense social networks connecting militia
members to each other did not always stretch upward to the commander, though this was more
likely to have been the case in Tajikistan. Slightly more than half of the Tajik respondents
reported having a family member in the same militia, as opposed to only seven Georgians.
This sheds light on the higher likelihood of Georgian militia members to switch commanders
during the course of the fighting than their Tajik counterparts, constrained by avlod
obligation ties.
56 Jesse Driscoll

but also taking things from their neighbors and anointing new political patrons
who would turn a blind eye to their theft.
Most militias were not sites for ideological indoctrination. Only 40 percent
of Tajik respondents and 30 percent of Georgian respondents characterized
their armed groups as being social environments where they remember most
people being motivated by politics. Banditry, roadblocks, racketeering, and
other predatory behaviors were commonly reported in both states. Many
of the men who were best positioned to take advantage of the breakdown
of the economic, political, and social order were those with little to lose:
criminals, manual laborers, sub-proletarian petty entrepreneurs, and the
serially unemployed. Others were simply young enough to have no real
responsibilities. Excitement, social recognition, and the opportunity to shoot
guns with friends were all mentioned as motivations in the early phases of
independence.
Social mobilization was dangerous in the late Soviet period because it
meant challenging ruthless and well-tested Soviet security structures, but
these structures vanished in 1991. Most participants invoked “arming for
community defense” dynamics, suggesting instability was stoked by the kinds
of local security dilemmas envisioned by Posen (1993).23 After the easy answer
was coded, however, follow-up questions often revealed exactly what Kalyvas
(2006) would have predicted: A complex local milieu of local jealousies, very
old family and neighborhood feuds, and no shortage of class- and clan-based
resentment. Many recruits were being mobilized to go take things (“take things
back”) from rival groups. Many reported encouragement by girlfriends or
female family members in doing this. Because recruits expected to keep what
they looted, up to and including homes and property, many began to think
of themselves as soldiers of fortune, serving as armed representatives of family
interests.24 There was consistent agreement that the family was the relevant unit
of practical politics, and the “selective incentives” offered were often framed as
side payments to mothers, mothers-in-law, or extended family.
But when pressed, more than two thirds of interview respondents admitted
that they stuck around because they wanted a job. A general focus on social
solidarity and criminal entrepreneurship in militia groups should not obscure

23 Posen (1993) argues that the offense–defense indistinguishability problem in irregular war can
be a self-fueling engine of conflict. Self-defense militias can easily go on the offense, and it is
impossible to misunderstand or ignore this fact. He also suggests an incentive to strike first
in this setting, as whoever holds the territory being fought over at the time the international
observers arrive to begin peacekeeping and mediation efforts tends to hold it indefinitely. See
Petersen (2011) for an extensive treatment on the implications of this second hypothesis.
24 For consistent interpretations, see Rubin (1993), Goldenberg (1994), Fairbanks (1995), Aves
(1996), Bornet (1998), Brenninkmeijer (1998), Zviagelskaya (1998), Anchabadze (1999),
Fairbanks (2002), Darchiashvili (2005), de Waal (2005), Torjensen (2005), and Zuercher
(2005).
Kto Kogo? 57

the main currency that was used to recruit militia members during this period:
an emerging “futures market” in political favoritism, contracted by warlords
at the time of recruitment. In the post-Soviet wars, the decision to join a militia
amounted to risking one’s life in exchange for some short-term mix of security
and loot, and the long-term promise of better life opportunities if a recruit
aligned himself with a worthy patron. In interview after interview, I learned that
militia recruits were often explicitly offered the opportunity to rise along with
their field commanders, who (by this point) had convinced their men that they
were going to be architects in the new national future. In practical terms, this
meant the potential for the marginally employed leftovers of the Soviet system
to reimagine themselves in jobs with pensions, and even some possibility of
social respect and family prestige.
A problem with this recruitment technique was that it was all based on
what economists call “cheap talk.” No one knew which warlords would
emerge as real political players over the long haul. National currencies had
not been established and the ruble was in a free-fall, making spot payments
for labor impossible. When militia members began to use violence to kill and
maim each other, joining up was tantamount to placing a bet on their patron
emerging in the faction that controlled ministries within the capital city after the
sorting was through. In other words, joining was analagous to playing a costly
lottery. To make the matter even worse, there was a problem of intertemporal
commitment. There was no way to write a contract that would be enforced
at a later time. Should a warlord beat the odds and successfully ascend to
prominence, he would no longer need any individual who had risked his life to
contribute to the warlord’s rise. He might be tempted to keep the spoils solely
for his family network. So although it was true that aligning oneself with the
correct militia captain might secure a recruit access to the wealth associated
with control of the state in theory, in practice this wealth would be realized
only if the warlord who recruited him both emerged as a powerful member of
the new governing apparatus and was willing to dole out patronage to loyal
subordinates when the time came.
How was this constraint overcome? Many respondents emphasized that the
transactions in the militia labor market were not about money. They described
organizational patterns more often associated with faith-based enterprises
or a religious revival. Warlord success in sustaining recruitment was often
reducible to the intangibles of charismatic authority: Who had the power
to make young men believe that their contribution and sacrifice could tip
the odds in their militia’s favor? Many men described enduring terrible costs
because they had strong convictions that their commander’s understanding of
conflict dynamics warranted sacrifice. One of the hallmarks of a successful
paramilitary commander was an ability to predict politics and rationalize
wartime events. The key to keeping morale high was constructing reasonably
coherent narratives about why they would win – or why they could not lose.
Certain symbolic events (e.g., displays of Russian air power, elections, foiled
58 Jesse Driscoll

table 3.2. Beyond Cheap Talk: Evidence of Complex Recruit Motivations

Variables All (%) Georgia (%) Tajikistan (%)

Joined a Militia in an Urban Area 67 92 43
Most Militia Operations “Close to Home” 22 18 27
Had Family Members in the Same Militia 31 9 52
Really Trusted His Commander 47 30 62
Joined or Stayed in a Militia for “Good Times” 29 46 13
Observations 173 84 89

assassination attempts) had dramatic effects on militia memberships, shattering

the credibility of certain commanders (whose predictions had been falsified) and
increasing the cache of others (whose predictions had been verified).25
In anticipation of the case studies that follow, Table 3.2 displays a summary
of the differences in recruitment patterns between Tajik and Georgian militias.
Most Georgians reported that they joined a militia in an urban setting, recalling
their initiation conversations at bars and cafes, at rallies, or in underground
political clubs. Most Tajiks in the sample report being recruited in their
rural homesteads. This should not be interpreted as evidence that Georgian
paramilitaries were a uniquely urban phenomenon. A few of my respondents
hailed from rural areas, and I am certain that if I had spent more time in rural
areas of Georgia, I could have discovered many more. But they described a
pilgrimage of friends borrowing a car and heading to Tbilisi or a nearby town
to be “introduced” to a local boss. By contrast, even the modest 43 percent
of Tajiks who claim to have been recruited in cities is probably a bit of a
misleading figure because rural regional capitals, such as Kurgon-Tubbe and
Kulob, would probably not be considered “real cities” by most Georgians in the
sample.
Another striking difference, revisited in the text that follows, is that
Georgians were far less likely than Tajiks to report really trusting their
commander throughout the war. This is likely both a cause and an effect of the
common practice of militia members defecting from one militia captain and
joining another. This practice, which is discussed at length in Chapter 4, was
pervasive in Georgia and comparatively rare in Tajikistan. About 40 percent of
the Georgians and 30 percent of the Tajiks who took part in the urban warfare
in my sample were “opportunistic joiners,” which is to say they were men who

25 A theory based on what recruits believed to be true at the time is difficult to test. Convincing
recruits to risk their lives in war is difficult, as most of the important benefits of a victory will be
realized independent of individual participation and the risk of dying usually outweighs any
individual benefit. See Olson (1965), Tullock (1971), Popkin (1979), Lichbach (1995), and
Grossman (1999). There was no effort made to systematically interview non-joiners or gather
information that could be used to test theories of recruitment in a rigorous fashion.
Kto Kogo? 59

joined their militia only after the end of major fighting (coded by the scourging
of the lowlands in Tajikistan, or the end of the Abkhaz campaign in Georgia).
Georgians were more likely to switch factional alliances during this period than
Tajiks. Only 13 percent of the members of the Popular Front reported switching
commanders, which is approximately the same fraction of Georgians in my
sample who did not switch factions.
Finally, more than three times as many Georgians as Tajiks admitted that one
of their primary motivations for joining or staying with a militia was “fun,”
“excitement,” “a good alternative to boredom,” or related concepts. Social
mobilization in Georgia escalated over many years, amidst the euphoria of a
nationalist movement that saw itself as the harbinger of a brighter future. In
Tajikistan, mobilization took place primarily after the unexpected dissolution
of the USSR in 1991, and in an environment of fear that descended, rapidly,
into terrible violence.

escalation processes in two post-soviet states

At least since the Brezhnev era, in peripheral republics like Georgia and
Tajikistan, true power lay in a coalition composed of the Party, which policed
entry into the political arena and managed relations with Moscow, and the
“shadow economic elite,” which provided luxury goods and allowed access
to real wealth. There was often a visible class structure in the late Soviet
periphery. For example, one set of families would occupy the state house and
manage external relations, while another set would run the streets and tax
the bazaars.26 In the final throes of the Soviet experiment, it was extremely
demoralizing to be a member of the first group, voicing the slogans of scientific
socialism, while watching criminal capitalist entrepreneurs secure offshore
holdings. Georgia’s preeminent political scientist, Ghia Nodia, summarizes the
social understanding of the nomenklatura on the eve of independence in the
following way:
The delegitimization of the public sphere implies that decent people should avoid
politics, or enter them only grudgingly, on the grounds that ‘someone has to do the
dirty work.’ And since entering politics means demeaning yourself, perhaps you should
reward yourself by being dishonest for your own personal benefit. The term ‘public
good’ does not exist and is hard to translate.27

26 For an primer on social and economic relations in the Soviet periphery see Derluguian (2005),
especially chapters 3 and 4. See also Simis (1982), especially 65–95, 126–143, 157–179, and
Remnick (1994), 180–215.
27 Nodia (1996). This resonates with the more general summary of North, Wallis, and Weingast
(2009) in their discussion of the “natural state”: “Personal relationships, who one is and who
one knows, form the basis for social organization and constitute the arena for individual
interaction, particularly personal relationships among powerful individuals. Natural states
limit the ability of individuals to form organizations. … Identity … is inherently political.” 18.
60 Jesse Driscoll

Against this backdrop, the rise of politically connected militias in the

post-independence capitals of Tbilisi and Dushanbe raised the stakes of
politics quickly. When new political parties were flanked by young men
with Kalashnikov rifles, it was reasonable for civilians to panic. As these
new political parties began to flirt with extreme arguments, such as
advocating radical redistribution of wealth and geopolitical realignment,
counterrevolutionary militias formed. Although it is beyond the scope of
this book to address the question of why the problem of post-independence
resource redistribution led to violence in these two states and not others, the
scholarly literature suggests three kinds of salient factors.
First, groups that anticipated becoming relative losers in the consolidation
process were able to resurrect their chances by enlisting help from beached
Russian military units. It is very difficult to account for the mobilization
strategy of the Abkhaz secessionists or the Popular Front for Tajikistan (PFT),
or either of their rapid escalations to violent tactics, while omitting expected
Russian military assistance as a causal variable. Lukewarm support from
the Kremlin for the fledgling Nabiev and Gamsakhurdia governments was
interpreted in the periphery as a signal that players in Moscow preferred
regime change. One of the main determinants of which countries fell victim
to serious civil violence was a belief by latent insurgent groups that if they
raised the specter of violence and disorder, they could draw Russian military
and political aid to their cause, in a humanitarian capacity, adding men and
material to aid in their struggle. In Abkhazia, Adjara, South Ossetia, Tajikistan,
Transdenestria, and Karabakh, expectations of Russian acquiescence or
support were instrumental in crisis decision making.28
Second, constructing ethnic and regional difference as a zero-sum game was
unusually easy in Georgia and Tajikistan for reasons that are presented at
length in the text that follows.29 The various warring groups could draw on
Soviet institutional resources, which were mapped onto ethno-federal hierar-
chies, to recruit armies quickly and to infuse certain questions of solidarity

28 For a fuller explication of this argument, see Laitin (2001).

29 I tend to be skeptical of these kinds of arguments on their merits. Although it is certainly
the case that different ethnic and regional groups in Georgia and Tajikistan saw themselves
as distinct, cultural distance is a poor predictor of violent conflict. Cultural frictions
over appropriate gender relations, abstract fear of “The Other,” and profound aesthetic
disagreements are likely inevitable. Laitin (1977), 139–161, presents a set of arguments
suggesting that differences in worldviews caused by language differences may be a permanent
fixture of the human condition. Nonetheless, as long as representatives of different groups can
calculate costs and benefits, there are almost always bargains or arrangements that can resolve
these differences nonviolently. Even in the absence of deterrence – such as when the Soviet
state apparatus simply disappeared – it is empirically possible for people who do not much
care for each other to coexist in amicable silence. Politics between ethnic groups only become
zero-sum as a result of political construction. De Figueiredo and Weingast (1999) present a
formal model of how violence can construct ethnic identity. For a review of this literature’s
findings and testable hypotheses, see Fearon and Laitin (2000).
Kto Kogo? 61

or distributional politics with a “natural” sense of identity.30 Preexisting

autonomous republic boundaries and kolkhoz-level hierarchical authority
structures were a huge asset in mobilizing the masses. People knew where they
fit on the map. Area specialists were not surprised to see Georgia break down
along ethnic lines and Tajiks mobilize based on kolkhoz and avlod structures.
Third, and perhaps most importantly, these states achieved independence
with no national military. The Transcaucasus Military District remained under
Russian control. The 201st Motorized Division in Tajikistan continued to take
orders from Moscow. Local police forces could keep order for a time, but in a
situation with no leviathan of last resort, local gangsters were tempted to esca-
late tactics and dare retaliations that they knew would probably not come.31
In Georgia, as nationalist militia groups and extremists rose to political
prominence in Tbilisi, minority ethnic enclaves began to fear for their safety
in the new Georgian state. Some communities organized militias to defend
themselves. Many more accommodated themselves to the new politics, waiting
for the violent nationalism to blow over. A few others explored the possibility
of Russian military protection, and improvised efforts by rump Soviet security
forces to protect Ossetian minorities were decried in Tbilisi as Russian meddling
in Georgian sovereign affairs. Georgian nationalists in the capital began to
fight among themselves over how to interpret the new state of affairs. Some
factions tried to deescalate the tensions and serve as moderate peace brokers,
but they could not restrain aggressive warlords. What emerged was political
gridlock and an angry echo chamber among the new elites in Tbilisi, with ethnic
minorities listening fearfully across a language barrier.32 As the Soviet Union
unraveled, these escalations eventually led to full-blown wars of secession in
South Ossetia and Abkhazia, and the general disintegration of state authority.
A similar drama played out in Tajikistan. Islamic activists entered into
tactical political alliances with democratic nationalists. Agendas for radical
redistribution were discussed for the first time in generations. Elites from the
wealthiest and most politically connected regions – who had monopolized state
positions and reported directly to Moscow for decades – feared the entry of new
groups into the power structure. As Tajikistan was an impoverished cotton
colony with some of the worst inequality in the former Soviet Union, stakes
were tremendously high. Most of the Russian middle class fled, leaving a civil
society void. Proposals for redrawing the internal boundaries of the republic,
essentially proposals to replace old gerrymandering with new gerrymandering,
threatened established regionally based patronage networks that kept rural

30 See Toft (2005) and Roeder (2007). For a carefully reasoned response to Roeder, see Lacina
(2014).
31 For an analogous argument tested in the context of Indonesia’s transition to democracy in the
mid-1990s, see Tajima (2014).
32 This is a distillation of many complex events summarized in Darchiashvili (1997a), Cornell
(2001, 2002), Darchiashvili and Nodia (2003), and Wheatley (2005).
62 Jesse Driscoll

populations dependent on urban elites in Dushanbe and Khojand.33 Once these

proposals were on the table, they could not be easily removed. These embattled
elites joined forces with rural criminal networks and local strongmen, who
cannibalized local military garrisons to create voluntary village defense units.34
The result was political fragmentation and brutal war.

georgia: 1989–1994
In its first three years of independence, the Georgian state confronted four
different ethno-territorial secessionist challenges, a military coup, economic
collapse, and rural warlordism. Cumulatively, the wars had a relatively modest
human toll compared to wars in Tajikistan or Chechnya. They culminated in
full-scale ethnic cleansing and the displacement of as many as 200,000 ethnic
Georgians, but probably no more than 20,000 total deaths. The legacy of
this violence is substantial, however, casting a long shadow over the current
Georgian political landscape.
As Tbilisi evolved in the Russian cultural imagination as an island outpost of
civilized luxury on the unruly mountain frontier, wealth and social opportunity
accrued to the urban ruling class in Georgia over many generations.35
Georgians who attended the right schools and managed their careers carefully
could access all that Moscow had to offer, fully enjoying the social, cultural, and
economic benefits afforded to the Communist Party elite. Some have suggested
that in Georgia the Communist Party had more in common with an aristocratic
social stratum than an actual political party.36
Outside of Tbilisi, the situation was very complex. Soviet ethno-federal
structures that protected minorities were particularly important to social order
in Georgia, given the well-understood political advantage that the majority
group held in the union. A “separate-but-equal” system had allowed dozens of
demographically small – but linguistically distinct – groups to live peacefully
as neighbors in the South Caucasus for a half-century.37 The dominance of

33 See Dudoignon (1998), Rubin (1998), Roy (2000), Whitlock (2005), and Collins (2006). For
some, references to political Islam cloud what would otherwise be a relatively straightforward
story of postcolonial distributive politics. For others, it is naive to separate the issue of political
Islam from neighborhood geopolitics given the close geographic proximity to Afghanistan and
linguistic proximity to Iran.
34 These groups began to carry out offensive operations outside their home territories almost
immediately. Bushkov and Mikulskii (1995), Kuzmin (1997), and Markowitz (2012), 60–65.
35 In the Soviet ethnic status hierarchy urban Georgians enjoyed most of the important social
protections associated with “favored lords”, in the typology presented by Laitin (1998), 59–82.
36 Aves (1993), 230. The perceived social status advantages that Party affiliation conferred on
Georgians is reflected in their party membership. With 8.5 percent of the Georgian population
registered as official members of the Communist Party, Georgians had one of the highest party
membership rates in the twilight of the USSR, besting Russians (7.9 percent) and Azeris (5.4%).
37 See Martin (2001). Blauvelt (2013) reports that Georgian is the most widely spoken of four
languages in the Karvelian group of South Caucasian languages (the other being Mingrelian,
Kto Kogo? 63

ethnic Georgians increased in the republic over the seventy years of Soviet
rule, but Georgia remained ethnically diverse. According to the 1989 census,
Georgians made up 70 percent of the population, followed by the Armenians
(12.7 percent), Russians (6.3 percent) and Azerbaijanis (5.7 percent). Georgia
had a number of enclaves where Georgian was barely spoken, and a politically
estranged class of elites who mediated between their Abkhaz, Adjar, Armenian,
Azeri, Laz, Mingrelian, Ossetian, or Svan-speaking constituencies and the
powers-that-be in Moscow and Tbilisi.38
Georgia’s descent into civil war is difficult to disentangle from the dele-
gitimization of communist ideology, which left a gaping political vacuum.
Independence meant that Moscow would no longer be the site of future elite pil-
grimages, or the guarantor of minority rights.39 This meant that ethno-federal
preferences, job quotas, and protected slots for minority-language primary
education were all suddenly negotiable. As the inaugural election empowered a
generation of self-assured Georgian elites, the threat of permanent second-class
citizenship began to crystallize in the minds of ethnic minorities. Under the
best of circumstances, it would have been difficult to convince Abkhaz, Adjars,
Armenians, Azeris, and Ossets that they would fare as well as minority groups
within a rapidly nationalizing Georgian republic as they had fared under the
“affirmative action empire” of the Soviet Union.40

Laz and Svan). Though Russian historically played an important role in administration and
in inter-ethnic communication, Georgian was made the official language during Georgia’s
brief period of independence in 1918–1921, and following the Soviet military conquest, the
Georgian SSR was one of the few union republics that successfully implemented paperwork
in the titular language under the Soviet “indigenization” policy. Publishing in Georgian and
Georgian-language educational institutions, libraries, and cultural and literary organizations
were supported during the Soviet period, and local Georgian leadership made fluency in
Georgian a criterion for cadre advancement within the Republic. Russian was still necessary
for educational and professional advancement opportunities on the all-Union level and access
to literature and popular culture, and served as the language of inter-ethnic mobility for
ethnic minorities in Georgia. Instruction in Russian was mandatory, though primary and
secondary education was also available in “official minority languages”: Abkhaz, Armenian,
and Azerbaijani.
38 At this point, some readers will be tempted to ask why the population did not all “just
become Georgian” or “just learn Georgian.” The short answer is that Georgian is difficult
to learn, they resented the forced choice, and they calculated that their children would “never
be Georgian enough” to ascend to a position of power in Tbilisi. For experimental evidence
that they were correct in their estimation, see Driscoll and Blauvelt (2012). Moreover, deeply
ingrained legacies of ethnically defined identity were at the heart of the Soviet experiment.
For those who wished to become bilinguals, Laitin (1998) provides myriad rationales for the
language of social mobility to be Russian, not Georgian. Laitin (1994) notes that multilingual
settings often empower a small set of linguistically gifted elites, who can become trilingual with
lower-than-average investment and “charge for translation services.” 622.
39 By “educational and administrative pilgrimages” (142), see Anderson (1991), 142. See also
137–143 generally.
40 For expansions on this point, see Martin (2001) and Laitin (1998), chapters 3, 4, and 12. For
similar arguments applied to South Ossetia and Abkhazia in particular, see Derluguian (2005)
64 Jesse Driscoll

From the perspective of ethnic minorities, what made things worse was
that the emergent elites were often relatively well-known criminals. Criminal
entrepreneurship in crass collusion with security services was nothing new in
the South Caucasus. Organized crime had flourished for decades with a wink
from Tbilisi, and since the 1970s, these networks had gained notoriety as vori v
zkone (“Thieves in Law”).41 Individuals with ties to wineries, citrus orchards,
hazelnut farms, and other luxury goods could profit by exploiting the shortages
and bottlenecks inevitable in a command economy. As latent tensions between
Russians and Georgians grew heated via electoral rhetoric, it was increasingly
“impossible to separate political violence from struggle for economic power”
among this elite.42 By the late 1980s, as a natural transit point to the Black
Sea, Turkey, and the European markets beyond, Georgia had an off-the-books
“shadow economy” that was estimated to be one of the largest anywhere in the
Union.
The slow collapse of authority sent uneven ripples through the Soviet
military and security hierarchy. In the South Caucasus, lucrative rents
associated with guaranteeing road, border, and bazaar security passed to
actors with the ambition and entrepreneurial spirit to monetize transnational
security networks. As the de facto sources of order inside the boundaries of
a new state, military field commanders found themselves guaranteeing vital
transit lines. These men evolved somewhat naturally into brokers – first for
bullets and ammunition to militias, and later for gasoline, drugs, cigarettes,
alcohol, consumer staples, and everything else. Alliances began to form between
uniformed state agents and criminals, as one respondent recalled:
“It was May of 1989 … I remember because I was just home from prison … An old
school friend, who was then in the police, was the first to talk to me about real politics.
I remember he said ‘This can’t end except with blood. Everyone is watching Berlin,
watching the Baltics, watching Budapest. There are not enough eyes and ears to pay
attention to us down here. If things go the way they are going, this will all fall in our
laps.’ At first I thought he was trying to trick me, to get me to say something that would
send me back to jail … but all the plans were his. He had friends in The [Communist]
party that could get us guns, ammunition … my job was to talk to people I knew who
could shoot [straight] and be trusted.”43

Independence came to Georgia on the wave of nationalist challenges on the

Baltic model. Mass protests against Soviet power were held, culminating
in demonstrations with many thousands of participants in the capital.44

233–237, Laitin (1999a) 48–56. In particular, residents of Abkhazia, where the Abkhaz had
slowly become a demographic minority in their own Soviet autonomous republic, believed
they were on the brink of being stripped of their linguistic, social, and economic autonomy.
See Cornell (2002).
41 See Mars and Altman (1983).
42 Slider (1997), 168.
43 Interview conducted November 1, 2006 in Tbilisi.
44 Slider (1997), 159–162.
Kto Kogo? 65

Communist authorities were utterly discredited by their decision to violently

disperse a demonstration on April 9, 1989, leaving nineteen dead and hundreds
injured. The martyrs empowered radical voices at the expense of moderates,
creating a set of self-sustaining anarchist attitudes that spread from educated
youth throughout the intelligentsia. This political culture festered during the
long window of political paralyzation between 1989 and 1991 – a period in
which all state institutions began to be seen as illegitimate.45 Koba, one of my
first key informants, presented the following account of his decision to join the
Mkhedrioni (“The Knights of Georgia”):

“I had never considered breaking the law … My mother always told me to be good, to
stay out of trouble, and that I should try to keep my grades good so I could join The
Party like my older brother Levan … But then when the Russians killed the women on
April 9 [1989], everything changed for me. … They murdered women for demonstrating
peacefully against their fascist state. They used gas, and they cut up their bodies with
shovels. How could we stand by while they did this to us? Our sisters? Our mothers?
What kind of men would we be? My father and I discussed this many nights … [A] friend
from school said he could introduce me to Dzhaba [Ioseliani], so I went to a meeting.
After hearing him speak, I knew that he was the man who would save us. … He was our
Martin Luther King.”46

There are elements of this story that were common across Georgian recruitment
narratives. The first is the reference to the April 9, 1989 massacre of female
activists as a focal event evoking an emotional response.47 It delegitimized
the authority of the Soviet police apparatus, and served as a gendered
rallying cry for young Georgian males to redeem themselves with reciprocal
violence. Second, the presence of family dynamics and the subtle importance
of parental approval surrounding the decision to join a violent militia group
emerged during many of my extended life interviews with key informants.48

45 Nodia (2002), 424 and Areshidze (2007), 19.

46 Interview conducted February 28, 2006 in Tbilisi.
47 Stereotypes of Georgians as an “emotional” people were regularly invoked in descriptive
accounts of the rise of the militias. This sort of facile cultural stereotyping cannot explain why
violent mobilization succeeded in Georgia and not in other republics. (A primary stereotypes of
Tajiks is of a quiet and subservient people). Yet cultural memory is real. The visceral emotional
response to the April 9 events still resonated in the imaginations of my respondents more than
fifteen years after the fact – manifesting in raised voices, angry gestures, and (in one case) the
suspension of an interview, when the respondent flew into a rage when I seemed unsuitably
unmoved.
48 Although family dynamics are impossible to code and not the focus of this research design, it
would not be inconsistent with my data to hypothesize that joining a militia was a life path
that would be discouraged for married sons with obligations or who were seen as obvious
social risers, and subtly encouraged for sons who had not demonstrated an aptitude for school
or a strong work ethic. There are a number of anecdotes from my data of families intervening
when a “promising” son allowed himself to be drawn into the violent milieu of militia politics.
One particularly prominent case – a former Minister for Culture and Sport for the Saakashvili
government – served with his Mkhedrioni unit in Abkhazia before his parents bought him
66 Jesse Driscoll

Third, the transcendent, charismatic authority of the personage of Dzhaba

Ioseliani featured in many accounts of Mkhedrioni recruits. Informal networks
that connected recruits to groups of university art students or other activist
communities with anarchist ties were reflected in stories. Fourth and finally,
my interview data reveal a ubiquitous willingness on the part of Georgian
nationalists to blame “the Russians” for everything that went wrong during
this period.
Anti-communist, anti-political, anti-state, and anti-Russian rhetoric began
to blur together. It is difficult to overstate the visceral anti-Russian character
of Georgian nationalism. Xenophobic rhetoric migrated into the mainstream
of political speech acts. It became commonplace to label political opponents
and ethnic minorities as nefarious foreign agents. The blunt stick of language
laws escalated the stakes of politics sharply, as the Georgian leadership gave
official status to the Georgian language. This curtailed job opportunities and
social mobility for minorities, and made it clear that the wave of nationalism
would not simply “blow over.”49 Many Russian-speaking ethnic minorities felt
they were being asked to choose between cultural assimilation, ghettoization,
or permanent resettlement outside of Georgia. Much of the Russian population
left Georgia and did not return.50
Many former Mkhedrioni members described its founder and spiritual
leader, Dzhaba Ioseliani, as an anarchist at heart – someone more interested
in tearing down the old order than in presiding over new institutions. As a
teacher of theater, he also had superb instincts for what kinds of symbolic
actions would allow youth to feel that they were taking part in a communal
ritual. One respondent recalled how he joined the Mkhedrioni by literally
being swept up in a crowd on its way to picket and protest the “occupation”
at a Soviet military base.51 Two other Tbilisi youth reported joining after
attending public demonstrations outside the Ministry of the Interior, protesting
the “humiliations and abuses” that the Abkhaz and South Ossetians were
inflicting on “defenseless Georgian minorities,” and demanding action.52
In this anarchic millieu, democratic reforms accelerated in the late Soviet
period, leading to modern Georgia’s first authentically multiparty elections
on October 28, 1990. Zviad Gamsakhurdia’s “Round Table – Free Georgia”

a one-way ticket to Atlanta, Georgia and enrolled him in dentistry school. See Derluguian
(2005), 32 for a related set of insights from the North Caucasus.
49 See Derluguian (2005), 202. Svante Cornell chronicles the historical reasons that the Abkhaz
feared cultural assimilation in Cornell (2001), 155–170, with particular attention to this
period.
50 Census data show that the Russian-speaking percentage of the population declined from
6.3 percent in 1989 to 1.5 percent in 2002.
51 Interview conducted November 21, 2006 in Mingrelia.
52 Interviews conducted October 24, 2006 and November 16, 2006 in Tbilisi. The second
respondent, Data, enlisted in the National Guard, rather than the Mkhedrioni, and left to
South Ossetia to fight for his brethren in December 1990.
Kto Kogo? 67

coalition won convincingly over the Georgian Communists (64 percent to

30 percent). In November 1990 Gamsakhurdia was elected overwhelmingly
as Chairman of the Supreme Council of the Soviet Republic of Georgia.
Gamsakhurdia quickly cut funding to the security and intelligence services
(believing them to be organs of Moscow), discouraged Georgians from taking
part in conscription drives for the Red Army, and explicitly endorsed the
creation of a new locally recruited paramilitary National Guard. He came to
realize that the power of the Mkhedrioni to control the streets was a serious
threat to his legitimacy, and made an attempt to put them down with the help
of the new National Guard, which had been organized by an old friend, Tengiz
Kitovani. (Kitovani had served time in the Soviet prison system, and is discussed
at length in the text that follows.) Gamsakhurdia was assisted with security
detachments from Moscow, which included forty tanks. Ioseliani was taken
into custody and held in a KGB prison until December 1990, but many of
the guards were sympathetic to his cause and his messages filtered out to his
commanders.
Locally recruited militias rushed to fill the void in public order. Elites in the
capital city had no ability to control the violent forces that had been unleashed.
David Darchiashvili recalls that in these early stages the Mkhedrioni and other
groups were quite savvy about the political climate and presented themselves as
a patriotic, order-providing alternative to the corrupt Soviet police forces: “The
theft and murder all came later, of course, but in the early days the Mkhedrioni
checkpoints wouldn’t harass you for your money … they made people feel
safer.”53 It became fashionable for emerging political parties to support parallel
youth movements, which often had paramilitary characteristics. During this
period, rural respondents described migrating to Tbilisi en masse, preparing
for a showdown with the Russian-backed police if necessary. When I asked a
militia member named Vasiil what he thought would happen in the capital, he
answered “I didn’t know … no one told us. We were just told that it was a good
time to be in the capital. That the police and the Russians were scared … and
that it was our time.”54
In May 1991, after the Georgian parliament had declared its independence
from the Soviet Union, Gamsakhurdia was elected President with more than
85 percent of the vote. His popularity began to fade almost immediately.
Dictatorial, willful, paranoid, and self-aggrandizing aspects of his personality
surfaced. Gamsakhurdia’s erratic decision making included a refusal to sign any
defense or economic agreements with the USSR, the haphazard restriction of
free speech by closing of opposition newspapers (lamely citing paper short-
ages), imprisoning underworld figures and parliamentary representatives by
seemingly random logic, stochastically reshuffling his cabinet, and repeatedly

53 Interview conducted November 19, 2006 in Tbilisi.

54 Interview conducted November 20, 2006 in Tbilisi.
68 Jesse Driscoll

declaring martial law to cover his political tracks. The original members of his
“Round Table” coalition began to abandon him. He turned to the streets.
By this point, the fragmentation of Georgia’s security infrastructure had
produced more than seven fully separate militias.55 To some degree, all were
involved in extortion, smuggling, and racketeering. Some of them were led by
individuals with no police background. Tengiz Kitovani, a criminal associate of
Gamsakhurdia, had used the anarchic collapse of the state to reinvent himself as
a soldier of fortune. He achieved fame agitating for Georgian rights in the ethnic
minority enclave of South Ossetia. He transformed his group into a mercenary
armed wing of the Union of Georgian Traditionalists, then shifted his support
to Gamsakhurdia and his “Round Table – Free Georgia” coalition. As he rose
in prominence and multiplied his political connections, Kitovani emerged as an
immensely powerful patron to middle-tier police officers who were looking to
distance themselves from the shame of state affiliation in the pre-1989 period.
One former combatant put his conversion in religious terms, “I had been a
part of an evil machine, and I was not proud … accepting Kitovani’s friendship
was like being forgiven for my sins.”56 Gia Karkarashvili, a relatively obscure
captain in the Soviet army, acquired instant local fame when he and a number of
other disgruntled members of his unit deserted and founded the “White Eagle”
militia group. This group affiliated itself at various points with the Mkhedrioni,
Kitovani’s National Guard, and would eventually form the nucleus of Georgia’s
state army under Shevardnadze.57
During the 1991 August coup in Moscow, Gamsakhurdia used the crisis
as a cover to dismiss Tengiz Kitovani and subordinate the National Guard
under the auspices of the Ministry of the Interior. Gamsakhurdia also sacked
the popular Prime Minister, Tengiz Segua, and radically centralized power by
personally assuming direct control of the KGB, the Ministry of the Interior, and
the Ministry of Justice.58 Kitovani refused to obey the order and simply left
Tbilisi, moving to a nearby armed encampment with the bulk of his troops.59

55 Interview conducted December 6, 2006 in Tbilisi.

56 Interview conducted February 7, 2006 in Tbilisi.
57 Karkarashvili inspired tremendous loyalty on the part of his men, and when he was dismissed
in the late 1990s many members of the unit defected to launch irregular raids across the
unrecognized border with Abkhazia. Interviews conducted October 29, 2006 in Kvimo Khartli,
and November 16, 2006 in Tbilisi.
58 O’Ballance (1997), 104. Amidst the euphoria of epochal social change and the first authentic
election in living memory, Gamsakhurdia’s speeches included a number of unfortunately
phrased comments, in Georgian, directed at non-Georgian minorities. Gamsakhurdia later
denied having actually mouthed slogans such as “Georgia for Georgians.” Parsing speeches
for double-entrendres and hidden meanings to determine if he was, in fact, speaking in code
to fringe nationalist elements of his own coalition, requires a linguistic fluency that I do not
possess. I will simply add that neither did most citizens of listening nervously from Abkhazia,
Kvemo Kartli, or Samtse-Javakheti. Nor did most citizens of Moscow, where most politically
relevant television and radio broadcasts still originated.
59 Wheatley (2005), 55.
Kto Kogo? 69

These recruits formed the core of the militia army that would invade Abkhazia
one year later.
In the interim, Kitovani deployed a few trusted subordinates with orders
to visit Abkhazia “with no guns – just clipboards and notebooks” to map the
locations of waterfront properties and farms with profitable citrus orchards.60
By this point commanders were fairly brazen economic brokers, not even
bothering to hide their ties to organized criminal groups with transnational ties
across the former Soviet Union (the vori v zkone in particular). A fair number
of respondents voluntarily admitted to becoming involved with the National
Guard or the Mkhedrioni in this period because they hoped that it would serve
as a conduit into a glamorous life of international crime. A representative quote:
“There are not so many ways out of my village, if you are not in the right family.
For me, I saw a chance to … get rich, like Pacino in Scarface.”61
A massive hostile demonstration on September 2, 1991 ended with the police
firing into the crowd to disperse it, a chilling parallel to the April 9 martyrdom
of Georgian activists. As a direct result of these events, the regime lost the
support of both the elites and the masses. Tbilisi descended into months of
chaos and anarchy, with open gang warfare between different militias with
vague and rotating political attachments. In rural areas, many non-Georgian
ethnic enclaves had organized militias of their own, or sought protection from
beached Russian military units. These actions were seized on by Georgian
nationalists as proof of disloyalty.
Gamsakhurdia will primarily be remembered as an erratic politician who
needlessly antagonized Georgia’s superpower neighbor and escalated tensions
with ethnic minorities. As a partial corrective, it is important to recall that
as the creeping escalation of violence between fringe Georgian nationalists and
Ossetians threatened to draw Russia into Georgian affairs, Gamsakhurdia kept
dialogues open with the nationalists in Abkhazia and succeeded in striking a
power-sharing agreement to assure Sukhumi its sovereignty. These hard-won
agreements were made irrelevant when a coup removed him from office.62 After
weeks of hiding in a Soviet nuclear fallout shelter, Gamsakhurdia fled the capital
city on January 6, 1992, leaving Tbilisi in the hands of Dzhaba Ioseliani, Tengiz
Kitovani, and Tengiz Segura, who reinstated himself as prime minister.

60 Interview conducted February 7, 2006 in Tbilisi.

61 Interview conducted October 19, 2006 in Chinvali. Viewed through this (admittedly cynical)
lens, Gamsakhurdia’s unforgivable sin may have been shutting down economic relations with
Russia without providing an alternative outlet for either the entitled entrepreneurial class or
for the rural proto-criminal underclass who saw a better labor market in Moscow and St.
Petersburg than Tbilisi. I am grateful to Timothy Blauvelt for this speculation.
62 See Jones (1996) and Zuercher (2005). Gamsakhurdia’s final miscalculation was his attempt
to centralize control over the militias that kept him in power. This backfired. Obviously the
coup-plotters and their intellectual supporters had a more charitable interpretation of their
actions. For an unusually thoughtful reflection on this period, see Areshidze (2007), 29–32.
70 Jesse Driscoll

In the immediate wake of the paramilitary coup, the new republic was
bankrupt, with no credit to purchase food or fuel. The Abkhaz, South
Ossetians, Adjars, and Armenians of Akhalkalaki were de facto independent.
Gamsakhurdia – who for all his faults was the country’s elected leader – had
successfully rallied an insurgent army of “Zviadists” and crossed the border
into Chechnya to launch an insurgency. Tengiz Segura, now prime minister,
went hat in hand to Moscow, then to the Commonwealth of Independent
State (CIS) summit in Minsk, and then to the United States and other Western
capitals, but was time and again rebuffed in his pleas for aid. The coup leaders
had no idea how to resurrect the carcass of the state, no hope of international
recognition, and no money to pay police salaries or retain military soldiers.
Georgia was a failed state, with no obvious path to recovery.

tajikistan: 1990–1992
Tajikistan’s state failure produced, with the possible exception of Chechnya,
the most brutal of the post-Soviet wars, with per capita violence on a scale
unmatched anywhere in the region. For a country containing fewer than
2 percent of the citizens of the Soviet Union, it produced 37 percent of the
casualties of the post-Socialist wars.63 At least a sixth of the population was
displaced by the fighting, including the permanent exodus of virtually every
member of the Russian and East German professional class. The World Bank
and International Monetary Fund (IMF) estimate that today the per capita gross
domestic product (GDP) in Tajikistan is approximately one-tenth what it was
before the slide into violence.64

63 The 37 percent figure was calculated using the conservative 41,300 figure cited in the Lacina
(2006) replication data for all the wars that broke out on the territory of the former Soviet
Union between 1991 and 1993, including in the denominator the estimations of deaths in
wars in Chechnya, South Ossetia, Abkhazia, and Ngorno-Karabakh. Because so many of the
civilian deaths were due to disease and exposure in the frozen mountains of Afghanistan, we
will probably never know the true number of deaths in the Tajik civil war. The 50,000 number
is regularly cited; estimates by specialists range from fewer than 10,000 (cited by Christoph
Zuercher in a 2011 Stockholm International Peace Research Institute (SIPRI) workshop) to
estimates as high as 100,000 (cited by Sergei Gretsky, who can credibly claim to have lived
through the Tajik civil war as a close observer). Thanks to the United Nations and the
International Committee for the Red Cross/Crescent (ICRC), there is much more reliable
data on population displacement. In the opening phases of the war approximately 300,000
ethnic Russians and Russian speakers from other republics – the core of the Soviet-educated
professional class in Tajikistan – emigrated, never to return. More than half a million additional
Tajiks and Uzbeks fled their homes (600,000 is the estimate that the United Nations relief
missions used in their internal correspondences in 1995). This does not include the tens of
thousands of “predatory migrants,” who seized homes in Dushanbe abandoned by fleeing
Russian speakers, or rural kolkhoz property abandoned by the losers of the civil war.
64 Nakaya (2009), 1–2.
Kto Kogo? 71

The social mobilization that led to the Tajik Civil War took place in the
poorest, most rural, and most mountainous corner of the Soviet Union.65
Unlike Georgia, Tajikistan did not achieve independence after years of social
mobilization for independence. It was “pushed out” of the USSR by Yeltsin’s
government after the failed coup and the secession of Belarus and Ukraine.66
In the years preceding the transition to independence, long latent tensions over
scarce resources had begun to emerge. Rural parts of this cotton-monocropping
republic were functionally self-governing. The collective farm (kolkhoz) was
the relevant unit of social organization for many citizens. Formal authority
structures wobbled atop local informal networks, and rural Party nomen-
klatura were difficult to distinguish from the shadow economy elite. A
lucky individual who emerged with the influence to serve as a Party inter-
mediary could profit by navigating the space between official Communist
Party production quotas and local productive capacity. But wealth-generation
opportunities were rare, and largely concentrated in labor-intensive agriculture
(cotton), light industry (military-industrial production), and mining (gems,
gold, and aluminum). Not incidentally, it was also on the military front-line
of a decade-long war, sharing a border with Afghanistan. Tajikistan was the
only Persian-speaking SSR in the Union.
One of the peripheral Tajik adaptations to Soviet totalitarianism was
the tendency to avoid committing very much to writing. Relations between
neighboring communities in rural parts of the country were governed by oral
agreements and informal understandings when possible. Legislative acts and
formal appointments to the hierarchy – the visible state, assigning jobs by
quota to minority groups, so relevant to politics in Georgia – had little to do
with the structures that shaped relevant life choices for most Tajiks.67 Instead,

65 Gretsky (1995), 1. The Panj River runs through the rugged and treacherous Pamiri mountain
range, forming the political border with Afghanistan. In my interviews with combatants, it
was clear that many drew inspiration from their grandfathers who fought in the Basmachi
revolts of the 1920s and 1930s, in the same terrain with the same hit-and-run guerrilla tactics.
I have heard it said that if one places a map over the territories that the Red Army never quite
subdued when they declared victory in the late 1930s, it bears an eerie resemblance to the parts
of Tajikistan that were the last hold-outs on the 1997 Peace Process.
66 Derluguian (2005), 168. Olivier Roy provides a nuanced alternative reading of the historical
data, suggesting that the Central Asian republics, like the republics in the Caucasus, were
bound to shed ties with “a Soviet model that had become exclusively Slavic.” See Roy (2000),
128. Either way, when the Soviet Union broke apart, Tajikistan was dependent on Moscow for
46 percent of its 1991 budget. This figure was actually quite representative of other Central
Asian states, as well – Uzbekistan, with all its urban wealth and relative industrialization,
received 43 percent of its 1991 budget from Moscow. It is with these facts in mind that Georgi
Derluguian writes the following: “To the credit of the Far Eastern Economic Review, when
the Western media were still celebrating the springtime of nations, this journal in 1992 ran the
cover story under blunt title ‘Dumping Central Asia.”’ 357 FN 4.
67 See Rashidov (1993), 296.
72 Jesse Driscoll

kolkhoz heads and state farm (sovkhoz) directors – particularly the “cotton
barons” of Khatlon, Kulob, and Leninobod – were the dominant social actors.68
Most of the literature identifies the roots of the Tajik civil war in a
half-century of divide-and-rule policies that split elites among various regions
and “clans” during the Soviet period.69 Since Stalin’s death, Moscow and
Tashkent had consistently allowed a few families, based primarily in the
region of Khojand, to police entry into the state apparatus.70 These elites
zealously guarded the rents that came from their privileged positions. On the
eve of independence, elites from different geographic regions were divided
by deep personal mistrust. Regionalism (mahalgaro’y) had evolved over
generations. Gretsky (1995) describes the resulting social stratification and
inequality:

[W]hen the Khujandis ascended to top party and government positions in Tajikistan in
the 1940s they endorsed localism as the corner stone of their policy, and kept regional
rivalries boiling, while reserving for themselves the role of arbiter. Under the Khujandis,
localism assumed such proportions that it began to somewhat resemble the Indian caste
pyramid with its division of labor. In Tajikistan, the popular wisdom put it in the
following way: ‘Leninobod governs, Gharm does business, Kulob guards, Pamir dances,
Qurghonteppa ploughs.’71

Political and economic impoverishment went hand-in-hand. Glasnost and

perestroika brought forth latent competitive pressures from elites in Tajikistan’s

68 Rumer (1989). The notion of “family” in Tajikistan is the subject of a dense anthropological
literature that I have not mastered. I inadvertently stumbled into a language problem through
my effort to ask the same questions of Tajik and Georgian militia members. Of the eighty-nine
former combatants who fought in the Tajik civil war, more than half of them claimed to
have at least one family member serving in the same militia group. This was the result of
a translation problem on my questionnaire because “family” in Tajikistan includes a broad
network of extended patronymic affiliations, based on blood ties and strategic marriages. The
“avlod” (family) is hierarchically organized, with elders (who can be male or female) at the top
and junior members obligated to them. Each avlod is identified by its location of origin, and
there are also social rankings of these avolds by prominence, creating what Matveeva (2009)
has described as a “caste system” (7). Unlike the Georgian clan networks, these ties did not
represent viable pathways for social mobility or economic opportunity outside of the republic,
except for a few family networks that stretched across the border to the ancient Tajik cities of
Bukhara and Samarkand (now part of the territory of Uzbekistan). For a general discussion of
Soviet failure to penetrate the clan structures in Tajikistan, see Rakowska-Harmstone (1970).
69 For outstanding overviews of the use and misuse of “clan politics” as an explanation for the
war in Tajikistan, see Roy (2000), 14, 92–100, Rubin (1998), 147–152, and Akiner (2001),
26–27, 65.
70 Kathleen Collins (2006) argues that Tajikistan was unique among the Central Asian republics
in that it was the only republic where interclan “pacts” were never forged in the 1980s. In
my reading of her argument, she gives causal weight to Moscow’s desire to keep a state
free of any groups that might have ties to revolutionaries in Iran or Afghanistan, which
was reverse-engineered by the Khojandi elite in the Uzbek-dominated northern region of
Leninobod. 102–117 and 130–140.
71 Gretsky (1995).
Kto Kogo? 73

impoverished areas around Gharm, who believed they had been forced into
second-class positions outside of communist networks. Not surprisingly, these
rural citizens tended to have more pious religious worldviews than their
urban counterparts.72 Though it is oversimplistic to think of Tajikistan’s
civil war as a clash between different regionally-based family networks,
birthplace provides a good rule-of-thumb for determining which faction a
particular person supported in the war that emerged after independence.
Conservative “neo-communist” forces recruited from regional groups that had
benefitted disproportionately in the late Soviet period – Khojandis, Kulobis,
Hissoris, and Uzbek minority groups. Their opponents recruited heavily from
rural Islamic activist networks that had deep support structures in Gharm
and the rural lowlands, and ethnic Pamiris, who had been overrepresented
in the police and Soviet security structures since the start of the Afghan
war.73
A number of populist Islamic parties mobilized in 1990 and 1991, led by
eloquent spokespersons among the urban intelligentsia. Both factions claimed –
in Tajik and Russian – to represent moderation and democracy. They stood
in opposition to the inherited Communist party apparatus and recruited
primarily from populations traditionally excluded from power. As Akiner
(2001) foreshadows, “[Islamic Renaissance Party] representatives insisted that
they were ‘against forcing people to accept our path,’ but even at this stage their
bodyguards were toting Kalashnikovs.” 74 In February 1990, when Tajikistan
was still a part of the USSR, hundreds of citizens poured onto the streets of
the capital to challenge the police in response to a rumor that scarce housing
in downtown Dushanbe would be set aside for resettled Armenians. Later
accounts suggest that the violence, which produced as many as 25 deaths
and 800 injuries, was organized by criminals – an omen of what was to
come.75
After First Party Secretary and President Kakhar Makhkamov supported the
Moscow putschists in August 1991, he was forced to resign in the face of mass

72 The best account of opposition ideology is probably Dudoignon (1998). A good companion
piece is Whitlock (2005), who describes the allure of “underground Islam” in late Soviet
Tajikistan.
73 Though Pamiri cadres were purged from the upper echelons of power in the Tajik state in 1937,
the unusually strong cultural value placed on education led Pamiris to be overrepresented in
professional classes that were dominated by Russians and East Germans: doctors, academics,
and civil servants. As the Soviet war in Afghanistan escalated through the 1980s, Soviet security
services heavily recruited from this population, which was often linguistically proficient in
Persian and Dari, yet less prone to being cultural sympathetic with the Sunni Tajiks in
Afghanistan. As a result, Pamiris were relatively overrepresented in the state security forces
at the time of independence.
74 Akiner (2001), 34. See Tadjbakhsh (1993) for a first-person account of the party organization
during this period.
75 See Auten (1996) and Nourzhanov (2005), 115.
74 Jesse Driscoll

protests.76 Tajikistan gained independence on September 9, 1991. Rakhmon

Nabiev, a former party secretary representing the traditional ruling families
of the northern region of Khojand, assumed the republic’s presidency. He
dissolved the Communist Party and attempted to legitimize his ascendency in
an October presidential election, where he ran against Davlat Khudonazarov,
a Pamiri intellectual who was the former head of the Soviet Union of
Cinematographers. The election was remarkably open, fair, and rigorously
contested by regional standards. Nabiev defeated his opponent with the support
of a convincing majority, with 85 percent of Tajiks casting their votes.77 The
ascendence of Nabiev was taken as a signal of political continuity.78
Opposition forces from regions traditionally locked out of power, particu-
larly Gharm and the eastern Gorno-Badakhshon Autonomous Oblast (GBAO),
feared that their window was closing. Some members of the opposition tried to
declare the election fraudulent. At some point there was a decision to test the
rhetorical promise that the new state had made regarding freedom of assembly
by using street power to get their way.79 In December 1991, the Speaker of
the Supreme Soviet, Safarali Kenjayev, declared martial law and a state of
emergency. The head of the Ministry of the Interior, Mamadaez, Naujavanov, a
Pamiri from GBAO, refused to implement his orders. In retaliation, Naujavanov
was relieved of his position. At the time of independence, the Pamiris of
GBAO were overrepresented in the Ministry of the Interior.80 Naujavanov’s
dismissal stoked fears among Pamiris that they were being purged from
the state apparatus. To many ethnic Tajiks, his dismissal was surely war-
ranted, and a long overdue correction for Russian affirmative action favoring
Pamiris.

76 Even after Makhkamov stepped down from office, the Ministry of the Interior refused orders
to impose a state of emergency, and allowed more than 10,000 demonstrators to continue
paralyzing the capital city of Dushanbe for weeks.
77 According to Grotz (2001), 466. Nourzhanov and Bleur (2013) reports that 58 percent
casting votes for Rahmon Nabiev, 34 percent for Davlat Khudonazarov, 5 percent for
Saifiddin Turayev, and marginal support for three other candidates (290). Muriel Atkin
(2002) corroborates the 58 percent figure for Nabiev but reports that Khudonazarov achieved
only 30 percent of the vote (100). Gleason (1997) claims 58 percent and 29 percent. Grotz
(2001) reports 56.9 percent and 30.1 percent (466). Khudonazarov was endorsed by the
Democratic Party of Tajikistan, the Islamic Renaissance Party of Tajikistan, La’li Badakhshon,
and Rastokhez. Nabiev was supported by the Communist Party of Tajikistan. The details of
when these various parties were founded, registered, and suspended are well documented in
Babak, Vaisman and Wasserman (2004), 269–321.
78 Collins (2006), 102–117 and 130–140.
79 For a more sympathetic view of the opposition demands their political platform, see
Akbarzadeh (1996), 1108–1112.
80 There were also many Pamiris who had received special military training as a result of the Soviet
military engagement in Afghanistan. Ever fearful that militant ideas would spread north like
a contagion, transmitted perhaps by the charismatic personality of Shah Massoud, Moscow
wanted to make sure that there was a non-Uzbek, non-Tajik force of military specialists in
Tajikistan’s security structures. Roy (2000), 139–140.
Kto Kogo? 75

As the snows melted the following spring, demonstrators from traditionally

disenfranchised regions of the country gathered in Dushanbe and staged a
mass protest in Shahidon (“Martyr’s”) Square, calling for Nabiev’s resignation.
Fearing that the government would give in to the demands of these agitators, a
counter-demonstration formed in nearby Ozodi (“Liberty”) Square, featuring
mobs of Nabiev supporters – men from Kulob, Khojand, Tursunzade, and
the Dushanbe sprawl.81 The President was unable to deploy the security
ministries to disperse these demonstrations because the security ministries were
themselves gridlocked between supporters of the opposition and supporters of
the new regime.
In the midst of this gridlock, Nabiev – who had already announced his
intention to demobilize and disband the Tajik army because he could not
pay them – ordered that approximately 2,000 machine guns be distributed
to conservative “activists” in Ozodi Square. This was one of the final tipping
points to the fragmentation and disintegration of the post-Soviet security
infrastructure.82 The government also released thousands of violent criminals
from prison to swell the ranks of its supporters in Ozodi. Kenjayev resigned.
On May 5, the Kulobi and Khojandi mobs left Dushanbe in protest. Rather
than disband, they organized in their regional strongholds for insurgency,
proclaiming their intent to recapture the state by armed force. With the
withdrawal of the Leninobodis and their Kulobi allies, the new government was
dominated by its loudest and most radical voices. This included a large number
of men who were identified – fairly or unfairly – as Islamic fundamentalists,
flush with confidence from their successful street protests. Armed militias
proliferated in the capital city claiming opaque ties to political parties with
radical platforms. Though Nabiev technically remained the President, in
practice executive power passed to a small cadre of Pamiri and Gharmi
intellectuals headed by Akbarshah Iskanderov, the Deputy President and Prime
Minister. Prominent Kulobi and Khojandi elites refused to recognize the
legitimacy of the new elite. War followed.
There was substantial unrest and fear in rural areas, as different rural farms
organized their young men for war. With no reliable news service, it was
difficult for anyone in a rural area to know what was actually happening outside

81 The question of why agricultural Kulobis saw themselves as natural allies of the old guard
Khojandis of Leninobod is addressed explicitly by Lena Jonson and Akiner (1998). In the
1970s and 1980s, Kulob extorted the regime with the threat of subtly escalating violence and
forcing the attention of Moscow. Members of the “shadow elite” simply started killing police
officers, knowing that if the Khojandis were to call Moscow or Tashkent for help, it might
invite cadre reshuffling on the grounds that they could not manage their own republic. As a
result, the informal arrangements brought the Kulobis into the Khojandi power structure as
“junior partners” – essentially criminal enforcers for wealthy families that owned factories
in Leninobod. For a description of how Khojandi cadres played divide-and-rule games to
monopolize power for generations, there is no better introduction than Collins (2006).
82 See Akiner (2001), 37.
76 Jesse Driscoll

of a very small locality. By the beginning of summer 1992, it became clear to

elites in the capital that the governing bargains that had upheld order in the
Tajik state were disintegrating. But news traveled unevenly through kolkhoz
networks to outlying rural areas. Village defense militias began to form across
the countryside unevenly. From the perspective of most citizens – including
most militia recruits – there was no way to determine what information was
reliable. It was clear that longstanding political bargains were unraveling
against a backdrop of extreme uncertainty and fear.
By midsummer 1992 the war was fully underway in rural areas. The
basic strategy of both sides was essentially identical – to violently liquidate
the rural population bases from which their opponents recruited.83 Local
strongmen cannibalized local military garrisons to create voluntary village
defense units. These militias quickly became the primary actors in the civil war.
In the words of one European observer, “Neither side distinguished itself by
humanitarian conduct in the war; ultimately, the side that won committed more
atrocities.”84 Gharmi militias attempted to blockade the region of Kulob, which
was identified as being the main military stronghold of pro-Nabiev groups.
The blockade failed. Kulobi militias emerged as the dominant fighting force
in this period, and displayed a willingness to employ unspeakably gruesome
anti-civilian tactics.
A recurring theme that emerged through extended interviews was an
awareness that the real threat was that the wrong group would end up
in control of the capital city. Even in isolated rural areas, militia recruits
remembered registering awareness that there were profoundly threatening
changes afoot in Dushanbe that had the potential to affect everyone’s lives in
the long term, regardless of how the war was resolved. Many of my respondents
had never traveled more than a few dozen kilometers from their place of birth,
let alone to the capital city, before the war. When they described the recruitment
dynamics, it was not surprising that they drew on proximate frames of reference
to describe their mobilization. Their decisions were governed by local insecurity
and local power dynamics constituting what Olivier Roy describes as the “war
of the kolkhoz.”85 But local politics in Tajikistan were always in the shadow of

83 The worst atrocities against civilians took place the densely populated rural south where clan
groups were interspersed due to Soviet resettlement policies in the 1940s and 1950s. Use of
population displacement as a strategy to “lock in” a particular demographic group’s power
before a peace settlement is a relatively common and well-documented phenomenon. See, for
example, Posen (1993) and Naimark (2002). For an analysis of war crimes in the Tajik civil
war, see Rubin (1993), Bornet (1998), Brown (1998), Gorvin (1998), and Whitlock (2005).
84 Denber, Rubin, and Laber (1993), xviii.
85 When Olivier Roy describes the Tajik civil war as a “War of the Kolkhoz” he is referring to the
fact that these extremely localized solidarity networks determined social mobilization patterns,
and pitted different village farming communities against one another as they organized militias
and frantically sought allies. In rural areas where state penetration was weak, local feuds
had long been diffused through cross-cutting horizontal networks of elders and influentials
(gaps and gashtaks). Now these same social networks were used to mobilize young men
Kto Kogo? 77

Dushanbe, where the real decisions were made, decisions that determined who
would get rich and whose children would stay poor. Abdulrahim, an ethnic
Uzbek, recalls a speech by “Baba Sangak” Safarov in November 1992, as he
rallied troops to march on the capital:

“Baba Sangak [Safarov] called us together … he knew that most of us didn’t want
to leave our families … we were brave fighters, but we just wanted to defend our
homes … but we all decided to go. He said ‘They won’t come here to block the roads
next time, you know – they’ll just get the Russians to do it next time, and we won’t be
able to beat them. If we don’t take this chance, Kulob is finished, and everything we
just fought for is for nothing. Your daughters will be forced to wear the veil and be
thrown from school. Your sons will work in the cotton fields and die poor. And your
grandchildren – if they are lucky – will have Pamiri or Russian last names.’ That is why
we went. To protect what is ours.”86

These sorts of arguments became commonplace among the literate class. The
weeks of public demonstrations in Dushanbe between April and May of 1992
provided a perfect incubator for this sort of radical propaganda to fester
on both sides. When the crowds were dismissed, they carried these honed
polemics back to their communities. The sense that “everything was at stake”
became a common theme in important public spaces, such as mosques, gap
meetings, kolkhoz association roundtables, and chaihonnas (tea houses), and
in private spaces, like around dinner tables, in the hours spent with close
friends and co-workers. If a mullah or local police chief could quietly convince
the matriarch or patriarch of an influential avlod that it was necessary and
appropriate to contribute young men to a larger cause, most of the work
was already done.87 Strong hierarchical family ties, reinforced by gendered
obligations to “protect the family,” probably removed much of the agency
from the militia recruit. One of the striking differences between the responses
of former Tajik militia members from their Georgian counterparts was that
the Tajiks tended to convey a general discomfort with the idea that “joining”
a militia was actually choice at all.88 The instinctive tendency to deny one’s

for war. The central question – and one that varied within districts and even within rayons
(precincts) – was whether these informal networks had sufficient social standing to persuade
local law enforcement to defect from their loyalty to the regime and lend their arms to the
militias. The general pattern seems to have been that a charismatic field commander, who
often emerged from within local security services or the criminal underworld, recruited a core
of enthusiastic supporters, and then was legitimized by a local authority figures such as the
civil administrators, Islamic mullahs, or kolkhoz heads.
86 Interview conducted July 27, 2005 in Kalininobod.
87 Nourzhanov (2005) describes these linkage mechanisms in villages highlands of Kofarnihon,
Jirghatal, and Komsomolobod, where “revolts” by “opposition militias” were actually just
local police and military garrisons switching sides. Markowitz (2011, 2013) interprets
cross-regional data in a way that is consistent with this account.
88 It was regularly described as a forced choice, or a situation where “everyone had to join.”
Further questioning led to admissions that not everyone joined – that, in fact, many of their
78 Jesse Driscoll

own agency and put the recruitment decision outside of one’s own hands was
prevalent throughout Tajikistan.89
The war activated long-buried community resentments, both within and
between kolkhoz structures. One common cleavage in the cotton-producing
lowlands was between traditional family networks (that could trace their
lineage back many generations) and the “Gharmis” (highlanders that were
forcibly resettled in the 1940s, 1950s, and 1960s as part of a Soviet population
management effort). These groups had long seen themselves as competitors for
scarce land and water resources. This trend was particularly acute for recruits
who hailed from the rural lowlands of Khatlon, where combatants described
intense pressure from family members and community elders to enlist in a
self-defense militia and protect their community. “Every [family] had to send a
son to go with Baba Sangak [Safarov],” said Oqil, a Kulobi soldier whose family
lived in Qabadian about an hour outside of Kurgon-Tubbe. “They [“Gharmis”]
had blocked the roads, and we were under siege. They even sabotaged the
[irrigation canals], so by the late summer we would have no water. We had
to break the blockade.”90 Ghulomjon, a “Gharmi” from the nearby village of
Shahri Tuz, presented the mirror-image of the same fears: “Our Rais [district
head] came to our home one night and asked for us to donate our rugs and
gold, to go to Afghanistan to buy guns. He said that he had heard that the [old
government] had opened the prisons and given the criminals guns, and that the
local police garrison was disbanding. … I anyway knew their commander could
not be trusted since he was from Kurgon-Tubbe, and would eventually go to
be with his family.”91 As Table 3.1 shows, the vast majority of respondents
emphasized these sorts of self-defense and family-defense narratives, which
often segued into emotional accounts of violent tactics inflicted on family
members.
In the highlands of Gharm and Gorno-Badakhshon, militia mobilization was
also stoked by local resentments and cleavages against “outsiders,” who, in
this case, were representatives of urban social networks that had been installed
atop the local avlod by political decisions made in Leninobod or Dushanbe and
backed by Moscow. While walking past the remnants of a now-abandoned
kolkhoz meeting house, burned during the war, a former combatant from
a neighboring town volunteered the observation: “It isn’t like it was their
farm … the Russians gave them that farm, for doing someone some favor fifty

friends had opted not to join – but it was a patterned contrast with Georgian recruitment
narratives.
89 See Whitlock (2005), 165–168, for additional narratives in this vein.
90 Interview conducted February 25, 2007 in Dushanbe. There is clearly coercion at work, though
less explicit than kidnapping or forced conscription. Respected family elders would be “asked”
to “volunteer” one son to join a self-defense militia, with the implicit risk that if this patriotic
service were not fulfilled, the family would be branded as disloyal and lose the protection of
its local warlord.
91 Interview conducted February 13, 2007 in Dushanbe.
Kto Kogo? 79

years ago no one can remember.”92 A striking point of agreement among militia
members is that their group – whichever group it was – had been unfairly
subjected to property expropriation by “the Russians” at some point in the past.
It was common to hear that “the best land” and “the best jobs” that had been
taken from them and gifted to someone else. The same gap and avlod structures
that were so effective at spreading information and diffusing local tensions also
served as powerful receptacles for memories of collective grievance.
Moreover, people intuitively understood that political independence from
Russia would provide the opportunity to renegotiate local power and property
rights structures. Given the role of cotton in Tajikistan’s economic structure, it
was not a wise idea to watch this process unfold from the sidelines.93 A former
commander made this point forcefully, when asked about whether he ever used
religious appeals to recruit people:
“Islam? Sure, we’re all Muslims – but that wasn’t why people followed me … They
followed me because I told the truth about the future. Everyone was saying it was a
new day, that the future would be bright. But I knew the truth. The future was already
spoiled by the cotton. Do you know what cotton does to your hands? Your face? Over
a lifetime? No one wants that for their children, their nieces. … All the while the radio,
the politicians – there is an election in the capital, all Soviet lies – they are talking about
peace and freedom, like we were idiots, like we could not see with our own eyes. The
truth was clear: Some people were going to end up picking cotton, and some other
people would lend them money. Call it ‘the will of Allah.’ Once you say it like this, it
is not difficult to recruit friends. Then all that is left to purchase weapons … and with
the [Afghan] border so close and military garrisons everywhere, it was not difficult to
do what I did … once you talked about cotton.”94

At first, the Russian government, acting through the 201st Motorized

Division, tried to assist in the creation of a government of national unity.
Though Russian troops remained inside the garrison, it was well understood
that Russia’s role as kingmaker would be necessary to legitimize any coalition
that formed. But recognition was not forthcoming. Perhaps without realizing
how it would be interpreted by analysts in Langley and Moscow, Tajikistan’s
state television network scaled back Russian-language programming and began
broadcasting Persian-language programming from the Islamic Republic of
Iran.95 These naive geopolitical missteps, along with general fears that the
government could not be a reliable partner, led Russian elites to reject calls
to aid the interim government in putting down the rebellions.

92 Interview conducted July 30, 2007 in Kalininobod.

93 Cotton mono-cropping dominates the Tajik economy to this day. The most lucrative wealth
creation opportunities come from brokering deals with rural kolkhoz in the unconvertible local
currency, then reselling the final products on the world market and banking the profits in an
offshore foreign account. For a detailed discussion of this system, the general contours of which
were well understood in the late Soviet period, see Van Atta (2008).
94 Interview conducted May 21, 2007 in Bishkek.
95 Roy (2000), 140.
80 Jesse Driscoll

As the Tajik state disintegrated, the scramble for power was abetted by the
intervention of regional governments and nonstate actors. In August 1992,
General Abdul Rashid Dostum, a prominent Afghan warlord and an ethnic
Uzbek, sent men and material to aid Sangak Safarov and the Popular Front.
Ahmed Shah Massoud, a rival Afghan warlord and ethnic Tajik, sent roughly
equal numbers of men to support the embattled regime.96 The government in
Uzbekistan was explicitly willing to choose sides and back the well-established
Khojandi families, hoping that Tajikistan would emerge as a docile client-state
to shield the region from the threat that the chaos of Afghanistan would
spread northward.97 The new Uzbek regime trained, equipped, and provided air
support to the Popular Front throughout the spring and summer, and allegedly
even sent Ministry of the Interior personnel to assault Dushanbe directly in
December 1992.98
Russia’s official policy was far more circumspect. Russian armed forces
stayed neutral, backing both sides indirectly by guarding critical infrastructure
in Dushanbe (e.g., bread factories) to ensure that the violence did not spread
north to the factories in the Khojandi rust belt.99 As it became clear that
Tajikistan was at serious risk for becoming a theater in the Afghan civil war,
the Kremlin finally made the decision to throw its weight behind the Popular
Front.100 In September, Moscow formally declined Iskanerov’s plea to send
troops or weapons to pacify Kurgon-Tubbe.101 Dushanbe would fall to the
militias of the Popular Front just weeks later. Opposition forces would regroup
in Afghanistan and launch an insurgency based in the country’s mountainous
eastern regions. This warlord counter-coalition, politically reborn as the
“United Tajik Opposition,” kept the insurgency smoldering for approximately
five years.
After seizing the capital, the Popular Front soldiers were encouraged to
partake in “punitive reprisals” against supporters of the previous regime –
particularly the Pamiri minority group. It is estimated that at least 4,000
families were driven from their homes during the five-week period between
January and March of 1993.102 In January and February of 1993 – the first few

96 Nourzhanov (2005).
97 Olcott (1997), 123–138.
98 Denber, Rubin, and Laber (1993), xix.
99 Gretsky (1995, 1997).
100 Tajikistan is regularly described as a “center seeking” civil war and Georgia a secessionist
conflict. The narrative in this chapter muddles this clean distinction. The growing influence
of “center-seeking” Georgian paramilitaries in the politics of the capital city induced regional
minority groups to bid for secession. This pattern was replicated in Tajikistan after the Popular
Front seized the capital. The difference is that the bids for “shadow state” independence in
Khojand, Kurgon-Tubbe, and Gorno-Badakhshon failed for lack of a foreign patron, so these
secessionist dynamics tended to go unrecorded.
101 Neumann and Solodovnik (1996).
102 There are high estimates of tens of thousands of Gharmis and Pamiris being driven from
Dushanbe and the surrounding urban and suburban slums during this period.
Kto Kogo? 81

months after Rakhmonov was installed – collective farms in the south that had
not supported the Popular Front were targeted for elimination. Violence was
ghastly. Systematic population displacement tactics included the widespread
use of public torture, mass shootings, and gang rape. Approximately 90,000
civilians are estimated to have been forced to flee south, fording the icy Amu
Darya River to seek refuge in northern Afghanistan. Property and land were
transferred to the family members of Popular Front commanders.103 Many of
the men who orchestrated and participated in this campaign of mass killing
were immediately granted roles in state security services, locking in the new
property rights regime.

the post-soviet frontier: russia after empire

Fighting in both Tajikistan and Georgia was sporadic and disorganized, with
criminal hooliganism and violence against civilians far more common than
pitched battles.104 The impromptu armies that fought in these wars often used
stolen buses and tractor-trailers as troop transports. Military tactics and basic
skills – sighting, shooting, and weapon maintenance – were learned by trial
and error by many participants. Given the haphazard organization of these
impromptu armies and the low quality of the recruits, it is not surprising
that Russian military units intervened to settle almost every single decisive
military battle fought on the former territory of the USSR.105 It was only with
military assistance from Russia that South Ossetian and Adjaran militias held
back Georgian paramilitaries. Tacit aid from nearby Russian bases eventually
became explicit when Abkhaz secessionists were aided by Russian airpower
and paramilitary fighters from the North Caucasus in their rout of Georgian
army and paramilitary units in September 1993.106 Artillery from the Russian
Black Sea Fleet played a crucial role in the defeat of the Zviadist insurgency,
deterring armed resistance in Mingrelia in November 1993.
Having gone to such trouble to define terms formally in Chapter 2, it is
a noticeable step backwards to make sweeping claims about what “Russia
wanted” or how “Russia acted.” Observers continue to disagree on the

103 Olivier Roy (2000) and Charles Fairbanks (2002) have described the neo-feudal arrangements
that emerged in rural Tajikistan as the traditional kolkhoz and sovkhoz structures were doled
out to the extended families of victorious warlords. Gorvin (1998) and Whitlock (2005) hint
that some of the broad patterns of property rights distribution that were conferred violently
in the war persist to this day.
104 The style of fighting in Georgia’s wars are well described in Zuercher (2005), especially
115–152, and Popkov (1998). Credible descriptions of the violence in Tajikistan are rarer,
but accounts in Rubin (1993), Nassim Jawad and Tadjbakhsh (1995), Leeuw (1999), ICG
(2004), and Torjensen (2005) ring true.
105 See de Waal (2003), 200–216 on Russian military influence in Ngorno-Karabakh, which was
fought with grisly attrition tactics by Armenian and Azeri military formations. Images were
broadcast for consumption to Russian-speaking ex-republics.
106 Derluguian (2005), 236–239.
82 Jesse Driscoll

question of how much this Russian military intervention was the result of
deliberate policy in Moscow and how much was the result of improvisational
maneuvering on the ground. Though it is difficult to weave a narrative of
this period without reverting to realist shorthand, it should be emphasized
that Russia was managing its own kind of revolutionary state failure at the
time.107 O’Prey (1996) describes Russian military policy toward civil wars
along its borders as “schizophrenic” (415) during this period, suggesting that
confused policy – even within a single country – probably reflects a breakdown
of command and control.
Was Russia guiding the coalition formation process from behind the scenes?
I shared this conspiratorial hypothesis in early interviews in Georgia and was
generally rebuffed with tongue clicks and disappointed head shakes. Gia, a
former member of the National Guard, said of the Russian role in Georgia’s
internal wars:

“You need to know what color the pieces are before you can play chess, you know?
After Gamsakhurdia, we didn’t even have a board! They knew that they couldn’t just
occupy our country. It’s not like Chechnya – we are our own country … so they knew
they needed Georgian friends they could trust, who would not be crazy like Zviad
[Gamsakhurdia], and that had the real support of the people. … but what was the
support of the people without an army?”108

Russia’s 201st Motorized Rifle Division remained in Tajikistan throughout the

civil war and served as a kind of weapons depot, allowing Tajik conscripts to
desert and fight for their home militia, so long as they left the heavy artillery,
helicopters, and combat airplanes behind. Most of my Tajik respondendents
reportedly believe that Russian strategic planners were tacit partners in war
crimes, allowing mass killings to assess the relative strength of the various
warlord factions. A few went further, suggesting that Russian snipers and
poisoners had acted directly to trim the opposition of its more radical voices –
presumably deterring others from putting voice to dangerous ideas. These
conspirational theories are admittedly difficult to falsify. Shirin Akiner’s blunt
appraisal that “had the Russians wished, they could have decisively altered
the course of the war at any time” (46) was confirmed in author interviews
with current and former government military personnel. A former lieutenant
field commander articulated the general consensus on why they did not: “They

107 There were three different kinds of violent processes playing out simultaneously in Russia.
There was revolutionary change – largely peaceful, but punctuated by a military coup – at the
top of the power hierarchy in Moscow (see Roeder [1993], Remnick [1994], 431–490; Solnick
[1998], and Kotkin [2008]). There was an unreported spike in the Moscow and St. Petersburg
murder rates, as criminal organizations competed to fill the void left by the retreating Soviet
state; (see Handelman (1995), Volkov (1999, 2002), Varese (2001), and Reno (2002)).
And there was violence in the North Caucasus, as Chechnya attempted to secede from
Russia.
108 Interview conducted October 19, 2006 in Chinvali.
Kto Kogo? 83

wanted us to kill each other, so that we would all see which of us was the
strongest ally. Then they step in and smile and talk about peace.”109
With many carrots and many sticks, it is easy to blame every contour of
the outcomes on the Russians. But which units, exactly, were in a position to
stop the Kulobis from using scorched-earth tactics to displace populations in
the Tajik lowlands? Who could have arrested the bloodletting in Tbilisi in the
wake of the Gamsakhurdia coup? Responsible treatment of counterfactuals
must acknowledge the possibility that military victory by alternative coalitions
in Tajikistan would have led to mass killing in the lowlands of Khatlon
with the proper nouns reversed. Perhaps Gharmis would have been the ones
revisiting old humiliations on Kulobi avlod networks. It is easy to imagine
scenarios in which ethnic Georgians in Sukhumi were the ones seizing the
best real-estate and – with assistance from Kitovani and armed patrons in the
capital – pushing former Abkhaz neighbors north across the frozen mountains
into Russia. Russian military professionals could not have been blind to these
scenarios. Casual Slavic racism toward peripheral non-Russian peoples is
well-understood; the ethnic status hierarchies that defined the Soviet Union
were – are – as real as any other sociological construct. Russians, in my
experience, still do not afford the victims or perpetrators of violence in their
peripheral wars equal intellectual, cultural, moral, or political status. Tofik,
who had served in the 201st Motorized Division in Dushanbe, analyzed the
Russian dilemma:

“The thing that you have to understand is that the [Slavic] commanders … they didn’t
want to stay here. Their families were scared. The Germans and the Russians … everyone
who could leave was leaving! They wanted only to leave the house in order, to stop the
Afghan war from coming north. Things would never have gotten so bad if Nabiev had
been able to keep a strong hand, like Karimov [in Uzbekistan] … the Leninobodis, the
Kulobis – they were so clever! They knew that the Russians, the Americans, the Uzbeks –
no one wanted this country run by clerics like in Iran … [B]ut why would we [the army]
take sides when things are so uncertain? Our commanders had orders to stay out – they
said, ‘if you want to go to your families and take your guns, you go – but the machines
stay, and you can’t come back. That would put us all at risk’ … After the Popular Front
had already won, then they knew which side they had to be on.”110

Russia – the former imperial power and future third-party intervener – would
be forced to broker with whatever local coalition emerged from the violent
crucible. Consider the blunt characterization of a Russian soldier, Sergei,
speaking on Russia’s role in settling the Tajik civil war:

“After Afghanistan nobody in Moscow had time for another complicated war. [The
Tajiks] are so friendly and passive, so I think most people were surprised at how quickly
things fell apart here. And once everyone had guns, how could you tell who was who?

109 Interview conducted August 9, 2007 in Dushanbe.

110 Interview conducted February 23, 2007 in Tbilisi.
84 Jesse Driscoll

We got a briefing from a KGB officer in Bishkek before we came down here – he told
us that we shouldn’t even bother trying to figure this place, with its inbred family
politics … It was really a no-win for Russia – if we left, things would have gotten very
bad, and we would have just had to come back … The Tajik government – if you know
them, you know they are peasants and thieves. But what could [Russian Foreign Minister
Yevgeny] Primakov do? He said ‘Let them have their broken little state, with their cotton
and their stupid hats.’ After all, war makes its winners … We care only for the border.”111

Sergei’s quote captures the basic dilemma of Russian military intervention

amidst state failure in Central Asia and the South Caucasus. Even when
a strong state has clear interests at stake in the outcome of a conflict, no
difficulty projecting power, and tremendous access to local intelligence, the
unpredictability associated with state failure makes it very difficult to track
or shape outcomes. Unlike in Chechnya, where the Russians were willing to
fight dirty to keep the population in the Federation, there were no serious
plans to occupy and repress the Georgian or Tajik populations directly. Russia’s
dilemma in both theaters was that although their military units could decisively
defeat local armed actors whenever they amassed troops to fight, and thereby
secure territory for shadow-state secessionists, it was difficult to sustain the
claim that anything achieved would be of lasting strategic value from a Russian
point of view. What was needed was for a coalition of Tajik and Georgian
violence specialists to emerge capable of managing their own pockets of lower
class radicals. The strategic dynamics that emerge once this insight is embraced
are modeled in the previous chapter. The details of what this process actually
looked like as it unfolded are outlined in the next chapter.

111 Interview conducted April 16, 2007 in Kalikhum.

4

Warlord Coalitions and Militia Politics

For many, the answer to the puzzle of short wars along the post-Soviet
periphery is overdetermined: Russia intervened, only Russia intervened, and
the totalitarian legacy of scientific socialism bequeathed institutions hard-wired
for centralization. Large, slow-moving structural variables – inculcated beliefs
and geopolitical realities – made for quick and stable war outcomes. But
this chapter provides evidence that the great powers did not send money or
guarantee regime support until after strong clients emerged, through local
agreements among armed groups. In the language of the model: Russia was
essential to establishing the stability threshold s; outsiders dangled v∗ , and
locals adjusted. Aid produced new rents, which incentivized warlords to cartel
the production of street violence, establish local order in the capital, secure
international borders, and keep violence out of sight. Understandings between
militia captains had to emerge on the ground before foreigners could help to
shore up the fledgling regime with aid and assistance.
This chapter begins by describing the process by which heads of state
were selected and installed. The presidents who were selected to head the
warlord coalitions in Georgia and Tajikistan were nationalist technocrats with
established reputations for honesty and fair-dealing. The people doing the
selection were warlords – and in a few cases war criminals with notorious
reputations. The figurehead presidents then managed shifting, overlapping
coalitions of these criminal interests.
The remainder of the chapter explains how warlords came to understand
which of them were “winners” and which were “losers” in the consolidation
process. Ceding power to a civilian figurehead carried risks for warlords.
Almost everyone had a criminal background and knew that his political
immunity could be revoked if the political winds changed. In the interim
period after installation but before the “coup-proofing” process was complete,
political uncertainty translated into street violence. The number of armed actors

85
86 Jesse Driscoll

expanded during this period. To the extent that there was any disarmament, it
was by the losers of the consolidation process who were abandoned by their
soldiers once it was clear that their patrons would never be in a position to
give them jobs. A model that treats all warlords as essentially interchangeable
in this process – subordinating the causal weight of social networks, warlord
ideology and charisma, and clandestine foreign interventions – is sure to be a
messy fit to either the Tajik or Georgian reality. Whether it is a worse fit than
rival two-player “incumbent vs. insurgent” models is for the reader to judge.

equilibrium selection: puppet presidents

In this book, order emerges when some warlords recognize that there are rents
to be gained (v∗ ) from creating a territorial cartel of violence. The central
observable implication of the theory is that the postwar government ought to
be composed of a large number of warlords and a figurehead president. This
president is forced to do the bidding of the warlords who install him because
these warlords continue to control the guns, and the president understands
that if he does not, he will be replaced by some other public intellectual or
representative of the nomenklatura. How does this account square with the
historical record?
Consider Table 4.1, which displays the coalitions that emerged out of
anarchy in Georgia and Tajikistan. In this table a “warlord” is a socially
recognized actor who commanded at least thirty armed men at a time.1 In
the inaugural cabinets there was cosmetic attention to creating a sense of
political continuity for civilians, with familiar Soviet ministry titles doled
out and familiar faces appointed to positions of power. But many of these
appointed figures were notoriously characters from the criminal underworld.
In Tajikistan, Sangak Safarov, the head of the army, had served twenty-three
years in jail for murder. Deputy Prime Minister Rustam Mirzoev had been
previously convicted of gang rape – not once, but twice. Yakub Salimov, the
face of the Dushanbe mafia, was selected to be the Minister of the Interior. Rauf
Saliev, a well-known Dushanbe racketeer and established drug kingpin, was
selected to head the new Secret Police. Even the spokesperson for the political
opposition – Prime Minister Abdumalik Abdullojonov – had a reputation for
being friendly with Khojandi organized crime networks. This was a coalition
of conservatives, assembled with an eye to foreign audiences in the emergent
Commonwealth Independent State (CIS) security framework. They hoped to
signal the emergence of a stable polity that could reassume its place in the
old Soviet ethnic hierarchy, subordinated to the suzerainty of Uzbekistan and
remaining under the watchful eye of the KGB.

1 The “criminal” moniker is applied conservatively only to individuals who are regularly
referenced as criminals in the secondary literature.
Warlord Coalitions and Militia Politics 87

table 4.1. Who Was in Charge after the Post-Soviet Wars?

Name Warlord? Criminal? Position in Government

The Presidium State
Council of Georgia
Tengiz Kitovani Yes Yes Head of National Guard
Dzhaba Ioseliani Yes Yes Head of the Mkhedrioni
Eduard Shevardnadze No No “Head of State”
Tengiz Sigua No ? Prime Minister
First Cabinet of
Emomalii Rakhmonov
Emomalii Rakhmonov No No Head of State/President
Abdumalik Abdullojonov No ? Prime Minister
Rustam Mirzoev No Yes Deputy Prime Minister
Yakub Salimov Yes Yes Minister of the Interior
Ghaffor Mirzoyeva Yes Yes Deputy Minister of Interior
Sangak Safarova Yes Yes Head of Armed Forces
Alexander Shishlyannikov Nob No Minister of Defense
Saidamir Zukhurov Noc No Chairman of National Security
Council (NSC)
Rauf Saliev Yes Yes Head of Traffic Police (GAI)
Shurob Kasimov Yes Yes Head of Special Forces
Mahmadnazar Salikhov No ? General Prosecutor
? = Reasonable people disagree.
a
Unofficial cabinet members (attended meetings in “advisory” capacity).
b
Ethnic Russian, Armed Service Representative of Uzbekistan.
c
Representative of KGB.

The same basic story plays out in Georgia, but without the Russian influence
and with power much more centralized in a celebrity head of state, Soviet
Foreign Minister Eduard Shevardnadze. In Georgia, Dzhaba Ioseliani, the
head of the Mkhedrioni, was responsible for hatching the plan to invite
Shevardnadze back to his homeland to serve on their military council as the
head of “his” newly independent state.2 The conversation between Kitovani
and Ioseliani, which debated the wisdom of bringing Shevardnadze back, has
spawned many urban legends, but it is clear that the main actors involved in
the conversation were anxious about the possibility of being marginalized in a
future coalition. Recalling the tactics Shevardnadze used to ascend to the post of
the First Secretary of Georgia’s Communist Party, Kitovani predicted that they

2 Wheatley (2005), 69–70. Jonathan Wheatley had the opportunity to personally interview
Ioseliani. Ioseliani claims that he was the one who personally telephoned Shevardnadze, and
that Kitovani wanted to bring Shevardnadze back as foreign minister only (98, FN7).
88 Jesse Driscoll

would be “dogs on a leash or jailed” within five years.3 Ioseliani carried the
day, arguing that Shevardnadze had spent a lifetime cultivating a reputation
for honesty, chaste loyalty to the ideals of the Communist Party, and taking
care of his friends. Despite their initial reluctance, both Tengiz Segura (another
architect of the military coup against Gamsakhurdia, but with no particularly
strong social base) and Kitovani were swayed by the logic that they were more
likely to stay afloat financially with Shevardnadze at the helm than with some
unknown alternative. There was agreement, however, that Shevardnadze was
a potential danger: Ioseliani recalled an understanding between himself and
Kitovani that the old man’s “hands must be held.”4
When Shevardnadze returned to Georgia in the spring of 1993, these
men were immediately promoted within the power structure and merged
their militias into the state armed forces and security services. At this time
Zviad Gamsakhurdia was technically still the president of Georgia – it was
not until the constitutional referendum of 1995 that Shevardnadze could
claim that title for his own. He was initially just one member of an ad
hoc military council, a consensus body composed of Ioseliani, Kitovani, and
Gamsakhurdia’s former Prime Minister Tengiz Sigua.5 Shevardnadze made
symbolic gifts to the warlords, distributing uniforms and titles. Kitovani
became the Minister of Defense. Ioseliani held many formal titles including
Head of the Emergency Reaction Corps (an autonomous subdivision of the
armed forces). One of Ioseliani’s hand-picked lieutenants, Temur Khachishvili,
became the Minister of the Interior. Kitovani was permitted to parade
with sophisticated military hardware procured from high-placed Russian
military connections, such as a computerized T-72 tanks. Ioseliani sat beside
Shevardnadze in the State Council. Shevardnadze also expressed his friendship
with Ioseliani and Kitovani in regular public statements, suggesting that he was
personally insulted when people referred to those men as criminals. One of
the first things Shevardnadze did was politically rehabilitate the coup-plotters
in the most public manner possible, stating in an interview with the Moscow
News:

Discussing the criminal records of certain people who are my partners now is offensive
to me. One should not be reminded of sins committed in youth. On the contrary, I admire
the people who had enough strength, will power, and courage to overcome all and make
a new start in life. I categorically disagree with those who keep reminding them of their
past. Now they are great statesmen. … Before returning to Georgia I resolved to forget
old grudges and abstain from witch hunts.6

3 This particular quote is almost surely apocryphal, but I reproduce the quotation marks because
it has the ring of truth. Interview conducted December 14, 2006 in Tbilisi.
4 Wheatley (2005), 70.
5 Wheatley (2005), 69.
6 Quoted in Goldenberg (1994), 93.
Warlord Coalitions and Militia Politics 89

The Presidium State Council was run as a consensus body: Every member
had a veto over every decision. Every member also had proposal power. This
was an unwieldy way to get much of anything done in terms of domestic
policy. Not surprisingly, it did not last long. What was most clear from this
arrangement was the experiment being embarked: Shevardnadze was allowed
to take responsibility for forging an autonomous foreign policy for Georgia,
delegating to a small number of warlords the task of keeping the capital city
from rioting. This provided a window of time to see what Shevardnadze could
do. Five months later, on a hot day in August, Kitovani’s militias – now acting
in their capacity as the Georgian national army – would invade Abkhazia under
the pretense of taking the war to Zviad Gamsakhurdia and his “Zviadists.”
In Tajikistan, there was no analogous celebrity figurehead to serve as a
focal point for coordination. Safarali Kenjayev, a Hissori who was the former
Speaker of the Supreme Soviet, was the founder of The Popular Front of
Tajikistan (PFT), a loose coalition of paramilitary groups united by a desire
to defend the interests of the groups that had gained the most during the late
Soviet period. His faction of the Popular Front was actually the first to capture
Dushanbe in October 1992. For a window of about eighteen hours, his men
occupied the relevant buildings in the capital city. But other warlords did not
coordinate on his ascendency. His soldiers were driven from the capital when
it became clear that neither the Russian garrison in Dushanbe nor PFT troops
from the south were rallying under his flag.
In mid-November 1992, after three different national rulers had been forced
to flee the capital city, the 16th Session of the Supreme Soviet convened in
the unusual location of Arbob Kholkhoz, a collective farm on the outskirts of
Leninobod. This city was the political stronghold of the traditional Khojandi
families who had encouraged the formation of the Popular Front. PFT field
commanders were visibly in attendance. “Baba Sangak” Safarov himself
guaranteed security for the event, and was present at every major meeting.
Many of the Gharmi and Pamiri representatives refused to make the journey
owing to security concerns and were thus not present for voting.7 It was decided
at this meeting that Nabiev would resign as president and that the office of
the presidency would be temporarily abolished. Parliamentarians in attendance
voted Emomalii Rakhmonov, a completely unknown figure, into the dual offices
of Head of State and Head of Government.
There was no doubt in the minds of any of the assembled representatives that
Rakhmonov was Safarov’s candidate. They were both Kulobis, both from the
same region of Kulob, Dangara, and even from the same subregion of Dangara.
In fact, Rakhmonov had been elevated to the post of Sovkhoz Chairman of
the Kulob Soviet only a few weeks before this meeting, after the previous
Sovkhoz head had been murdered by Safarov on the 28th of October.8 A

7 Whitlock (2005), 177 and Nourzhanov (2005), 118.

8 See Nourzhanov (2005), 117.
90 Jesse Driscoll

government representative recalls a conversation with Rakhmonov the night

before he was installed as Head of State in the following way: “He made it clear
that the country needed peace, and that peace would require difficult choices.
But those choices were not going to come any more at the expense of Kulob.
He wanted to build a strong state … based on the values he had learned in
Kulob, and to give charity to the people who had helped him climb.”9 In the
prophetic words of Barnett Rubin, the installation of Rakhmonov represented
the first signs of “a shift from ‘those who held the factories and party personnel
committees’ to ‘those who held the guns.”’10 Approximately one month
later, on December 10, demoralized Pamiri militias finally capitulated and
Popular Front forces entered Dushanbe. The capital would not change hands
again.

no disarmament and no external “security guarantees”

The model in Chapter 2 assumes that after installing a president, warlords will
maintain effective control of men and weapons. Power is still about controlling
violence, and militias provide leverage to extort the civilian president for
ministry jobs. Warlords are permitted to reinvent themselves as reformed
political figures, but they keep soldiers on call, ready to mobilize at a moment’s
notice.
This account closely matches the empirical reality of these post-Soviet
settlements. Voluntary warlord disarmament was not an option. In both
the hinterlands and secessionist “shadow states,” combatants dug in. In the
capitals, as civilians watched police forces become saturated with militia
members, there were perverse incentives for young men to try to reinvent
themselves as warlords and shoot their way into the ruling coalition. Friction
between the police and militias was often emblematic of friction between
individuals on different sides of the semipermeable state membrane. Weapons
proliferated. Many joined the state security services expecting that they would
get access to better guns and newer equipment, making it easier to return to the
mountains.11 Abdullo, a Tajik field commander, reflecting on why his United
Tajik Opposition (UTO) unit allowed itself to be integrated earlier than those
of some of his rivals, recalls that when he had his men take a vote, one of
the arguments that carried the day was “wanting to try the new American
AK-47s.”12 It was widely understood that field commanders expected to keep
some sort of face-saving fallback position if promises were broken down the
road. The pervasive culture of street violence and the demonstrated inability
of the regime to keep its promises meant that retreating to the mountains

9 Interview conducted March 13, 2007 in Dushanbe.

10 Rubin (1998), 129.
11 In model parameters: lowering the cost of fighting c and/or raising the reservation value r.
12 Interview conducted August 11, 2007 in Dushanbe.
Warlord Coalitions and Militia Politics 91

15

Militias in Dushanbe
Militias in Tbilisi

10

Rakhmonov Shevardnadze

0
1, 1
p 991

1 91
1, 1
2
p 992

1 92
1, 2
3
p 993

1 93
3
1, 4
p 994

1 94
4
1, 5
p 995

1 95
5
6
p 996

1, 96
1, 6
97
n 99

ar 99

n 99

ar 99

n 99

ar 99

n 99

ar 99

n 99

ar 99

n 99

ar 199
ec 9

ec 19

ec 19

ec 9

ec 9

ec 9

19
Ju , 1
Se 1

D ,1

M ,1

Ju 1
Se , 1

M ,1

Ju 1
Se , 1

M ,1

Ju , 1
Se 1

D ,1

M ,1

Ju , 1
Se 1

D ,1

M ,1

Ju , 1
Se , 1

D ,1
1,

1,
1

1
1

1
ar
M

M
D

figure 4.1. Nonstate Militias in Dushanbe and Tbilisi, 1991–1996.

was never far from anyone’s mind in this period. As Koba, one of my key
Georgian informants put it, “the Gamsakhurdia coup sent a clear message: If
you can shell the statehouse … you probably won’t be the one starving in the
mountains.”13
Rather than disarmament, what can be easily observed in Figure 4.1 is
a proliferation of active militias in the immediate aftermath of Rakhmonov
and Shevardnadze’s ascensions to power. This was a period of unprecedented
militia expansion and fragmentation.14 The locus of bargaining had shifted to
the capital city, and the rush for spoils created perverse incentives for militia
expansion.

13 Interview conducted December 3, 2006 in Tbilisi. This is not the same “Koba” referenced
earlier in the text.
14 The data on the number of militias in the capital over time were coded using newspapers, public
records, and trusted third-party reports from the period. Interviews with former combatants
were used to clarify the identification of marginal cases where the autonomy of a particular
field commander was contested. For a militia to qualify as an independent observation, it had
to have at least thirty members, a name, a socially recognized leader who spoke on behalf
of a group, and a presence inside the capital city. I constructed a timeline at three-month
intervals for both countries; they are superimposed on the same graph for comparison. Coding
decisions related to the month-by-month timing of fractionalization of the “umbrella” groups
and subordination within ministries were somewhat arbitrary; the broad temporal patterns
were uncontroversial when I displayed them to key Georgian and Tajik respondents. The
specific names of factional commanders are omitted for reasons discussed in Appendix C.
92 Jesse Driscoll

State weakness is often presented as a permissive condition for the prolifer-

ation of violence entrepreneurs – and in one sense it was in these cases. After
all, there was no one to make arrests. But note that, in this account, it was
actually the expectations of a strengthening state that led to the expansion
and fragmentation of militias. The promise of future bilateral aid from foreign
donors increased the value of the prize being fought over. Violence between
factions was replaced by violence that was essentially within the winning
faction. Sustained and intense intraethnic violence followed.
Militias did eventually disappear, not because they were disarmed, but
because their memberships disintegrated or switched commanders as warlords
were forced out of the violence game. This happened rapidly. Though none
surrendered their weapons, many groups “dissolved” – whatever social capital
that sustained them broke apart, and recruits went their separate ways. Viewed
through this lens, the violence during this period is best understood as a product
of a high-stakes tournament between paramilitary militia groups, competing to
be part of the winning coalition selected by the installed president. Aided by
young men recruited with promises of plunder and patronage, some militia
captains became valuable assets for the regime, while others were driven from
the capital city at gunpoint.
It is difficult to sustain the case that any of this was the result of plans
hatched in Moscow. As a Tbilisi diplomat put it: “Independence for these
republics meant that the angry people with the long last names just weren’t
Moscow’s problem anymore.”15 Eduard Shevardnadze was an agent of the dis-
credited Communist Party and the even-more-discredited Gorbachev regime.16
Rakhmonov was a complete unknown – a political wildcard compared to
the fully vetted prime minister, Abdumalik Abdullojonov. This is not to say
that Russian preferences for foreign aid, military assistance, and diplomatic
recognition were absent from the minds of Georgian and Tajik warlords. It is
just that local actors were more than capable of anticipating Russian demands
and providing foreigners with much of what they expected to see before anyone
in Moscow thought to publicly ask for it. A common strategy employed by both
Rakhmonov and Shevardnadze was the practical move of installing a Russian
general as the Minister of Defense (or an ethnically Russian Uzbek general in the
Tajik case), which pre-positioned them to leak information back to Moscow.
I interpret this move as evidence that the new governments had a potential
agency problem with Moscow that they were anxious to solve.

15 Interview conducted April 28, 2006 in Tbilisi.

16 Eduard Shevardnadze tried to go “over the heads” of local militia actors and secessionist
warlords alike and deal directly with players in Moscow to reincorporate Abkhazia and South
Ossetia. He was rebuffed, partially for refusing to join the CIS and partially because of personal
disputes with elites in Moscow who wanted to see him fail. Later, forces in Moscow would be
implicated in multiple assassination attempts against him.
Warlord Coalitions and Militia Politics 93

Chapter 2 provides an account of “full incorporation” after insurgents’

reservation values drop and they scramble to cut the best deal that they can.
The basic logic of the mechanism suggested by the model – declining reservation
values – can occur with or without third-party security guarantees to a postwar
regime. This part of the account also fits the post-Soviet empirical record quite
closely. Russian elites did not even bother articulating an interest in serving
as honest brokers, and were overall neither capable of, nor interested in,
punishing regimes in newly sovereign states for reneging on local promises
to insurgent warlords. Moscow’s consistent rhetorical promise, to the extent
that it was coherent, was to aid the Tajik government as it pressed hold-outs
and to defend secessionist minority enclaves against the threat of the Georgian
paramilitaries.
In Tajikistan, reasonable people have confused the actions of prominent
Russian field commanders or members of the Soviet 201st Motorized Rifle
Division, who would give or sell weapons to both sides, with official Russian
policy. In my view, Russian policy was calibrated toward cauterizing violence
and stopping the chaos from spreading beyond Tajikistan’s borders. Russia’s
nightmare was that Afghan warlords and American-funded Mujahideen would
ally with the UTO and infect the entire region with radical ideology.17 Russia
was also interested in a great many other things. Russia needed to manage
the relationship with the newly sovereign state of Uzbekistan, which wanted
to continue supporting traditional Khojandi client networks with strong
Uzbek ties. Russian diplomats prioritized establishing legal precedents for
CIS intervention into the near abroad. As will be discissed in detail later,
diplomatic efforts by Moscow to facilitate peace talks (often jointly with
Iran and the United Nations) occurred alongside direct military support for
counterinsurgency. Keeping the Tajik state intact as a buffer zone, and hoping
that radical Islamists would self-select into that mountain theater to be killed,
passed as strategy in the early 1990s. The Yeltsin government otherwise had its
hands full trying to establish its own domestic authority at home.
What possible “security guarantees” could the Russians have credibly
offered to Tajik civil war participants under these conditions? There was
no indication that anyone in Moscow thought it was even worth learning
the names of the main actors in the Tajik civil war until sometime midway
through the 1997 Peace Process. As a former high-ranking UN member and
observer of the inter-Tajik peace talks stated, “The main misunderstanding of

17 Akiner (2001) reproduces rumors of more than thirty military training camps, mostly on the
territory of Afghanistan, run by Mujahideen and Arab instructors (FN 16, p. 43). Ahmed
Rashid (2002) suggests that by the end of the Tajik civil war, the UTO began to act like a
real army, mostly because of the training and support from the Uzbek warlord Abdul Rashid
Dostum, who was operating in northern Afghanistan. Most former Russian military and civil
war participants interviewed tended to be skeptical of Rashid’s position, emphasizing that
different field commanders remained essentially self-financing and autonomous by colluding
with state militias to transport narcotics.
94 Jesse Driscoll

the United Tajik Opposition was that they were ‘united.’ The second major
misunderstanding, at least by the end, was that their ’opposition’ was political
in any way. The war gave these men cover, but their motives never matched well
with their rhetoric. … These were gangsters, pure and simple.”18 There was little
political coherence binding together the opposition militias of the UTO. Many
warlords used noms de guerre or sent cousins or nephews to negotiate in their
stead. From the perspective of civilians living in contested regions, this was a
time of unchecked banditry, looting, and terror.
The situation in the South Caucasus, where the conflict broke down over
territorial lines, was different. Russian policy in Georgia raised the reservation
value for Abkhaz and South Ossetian warlords so high that no offer from
Shevardnadze could make them better off than they already were behind the
shield of Russian guns and UN flags. This did had the effect of providing
security guarantees to certain non-Georgian ethnic insurgents who had seized
administrative structures in Abkhazia, South Ossetia, and Adjara.19 But the
direct result of this support was the ethnic cleansing of Georgians by Abkhaz
and South Ossetian militias. The next two decades of cold tension make
it clear that Russian policy did not provide an efficacious framework for
general disarmament or conflict deescalation. Just as Russian backing for
ethnic minority militias had served as a recruiting tool for the Mkhedrioni
and National Guard units (who could then claim to be acting to protect, or
avenge, their brethren), ongoing Russian support for the territories reinforced
the necessity of a strong Georgian army to stand up to Russia, undo military
humiliation, and recapture lost territory.

urban warfare: the time of troubles

By January 1, 1993, war reporters would describe both the Tajik and Georgian
capitals as controlled by a militarily dominant coalition of paramilitaries.
The Popular Front had driven their opponents across the Tajik border into
Afghanistan or into the impassable Pamiri mountains of Badakhshon to freeze
to death. The forces of the Mkhedrioni and the National Guard – which would
soon prove incapable of imposing their will on the Russian-backed secessionist
enclaves of Abkhazia, Adjara, or South Ossetia – were in control of Tbilisi.
While war had raged in rural areas the state apparatus had continued to
provide many public goods for most civilians in the capital. Buses continued
to run. Schools continued to operate. Citizens kept their heads down and
went about their routines. Bread factories kept pushing out non (bread) in
Tajikistan, and the metro subway system kept hauling young men and women

18 Interview conducted December 4, 2005.

19 Darchiashvili (1997a), 2-7. It practically goes without saying that certain Georgian nationalists
tend to see Russian fingerprints everywhere. This approach whitewashes the nihilistic
anarchism that took root among the Georgian political elite, the ethnically exclusivist radical
populism that seized the streets, or the criminal behaviors of the new military caste.
Warlord Coalitions and Militia Politics 95

to dates across Tbilisi. Most of the militia members believed that they were
owed something by the civilians back in the capital who were the beneficiaries
of their protection and sacrifice. A key respondent voiced a common complaint
heard by many veterans of Abkhazia:

“Back in Tbilisi life went on pretty much as before. There were parties, wine, weddings,
graduations … People bought cars and planned for their future. But we’d been fighting
for their behalf! For their kids! I watched friends get shot! For them! And then it was
over and we came home they were like ‘Oh, that war was such a tragedy.’ They didn’t
help, and didn’t know a thing about it! So yeah, we were mad.”20

Militia members confronted an unpaid Soviet-era police structure that was cut
off from political and material patronage in Moscow. Urban citizens recall two
distinct waves of urban violence after Dushanbe and Tbilisi fell into the hands
of ruling paramilitary factions. The first was a chaotic reign of terror when
militia members were encouraged by commanders to claim their share of the
spoils of victory in the form of looting.21 Bahron, who was a pastoralist in
Kulob and never had visited Dushanbe before the war, remembers a sense of
betrayal and resentment that he felt toward the civilians in the capital:

“We came into this city and realized that they had everything, these Pamiris and
Leninobodis. We finally saw with our own eyes that [their commander] Hussein had
been telling the truth – the Russians had been sending money to Tajikistan for years,
but these men in the capital had stolen it! All their homes had glass windows and gas.
We knew … that we had been fools to spend so many winters in the cold.”22

This relatively short period of unchecked looting, lasting only weeks, was
replaced by a very long, very tense stand-off between the armed groups. The
second wave of urban violence came only many months after the carnival of
violence had subsided after the end of major combat operations. With both
capitals awash with cheap weaponry and angry young men, militia captains
became power brokers, managing the resentments and expectations of their
soldiers. Insecurity gripped the urban centers of Dushanbe and Tbilisi for
months. As warlords began to merge their forces with the security services
of the captured regime apparatus, appropriating for themselves positions in
the Ministry of Defense, Interior, State Security, and other power ministries, it
became impossible for civilians to tell police from criminals. Residents of Tbilisi
and Dushanbe recall this period as “The Time of Troubles.”
And then the violence stopped. Lifetime residents of these capital cities agree
substantially on the month – sometimes down to the week or the day – that
order was restored in their neighborhood. Violence broke out in these states
slowly, as a result of accumulating processes of escalation through the late

20 Interview conducted December 4, 2006 in Tbilisi.

21 This resonates with more general accounts of “carnival” and “revenge” following military
victory by criminalized armies in Yugoslavia and elsewhere. See Mueller (2000), 54–56.
22 Interview conducted August 14, 2007 in Dushanbe.
96 Jesse Driscoll

Soviet period. Peace, by contrast, broke out virtually overnight.23 Figure 4.1
suggests that by the end of 1995, most militias had retreated or dissolved.
Dushanbe and Tbilisi were patrolled by uniformed police officers, some of
whom did not feel the need to carry weapons. To the delight of the international
aid community and foreign diplomatic corps, prominent militia commanders
were arrested. Anarchy on the streets subsided.
Driving militias from the capital city was considered a tremendous victory
for state-building. But where did these police forces come from? What processes
were sustaining urban violence during the “Time of Troubles,” and why did
those processes come to an end? And if it was possible for the forces of order
to triumph so quickly and completely, why did it take so long for the “tip”
toward order and security to occur?
The “Time of Troubles” occurred in the relatively brief period after
Dushanbe and Tbilisi had changed hands, but before the emergence of a
secure regime, when militia members had the opportunity to convert their
short-term social capital into long-term life opportunities. It was a window
of time characterized by deep uncertainty about the final shape of the postwar
coalition. It was also a period when the foot soldiers felt the window closing on
their ability to make heroic demonstrations of their worth to their commanders.
Competition became violent, as the wartime coalition splintered apart and
different militia groups cannibalized each other’s memberships. Political actors
supported large militias as insurance against being trimmed from the spoils
of victory. As described in the previous chapter, many militia members were
lured into service by non-collateralized promises of future employment by
their patron. Once external events clarified membership in the ruling coalition,
large militias became obsolete. The promises of many would-be patrons were
suddenly worthless. It took only days for militia recruits to switch commanders
or quit the streets.
Scholars’ failure to account for the sudden disappearance of urban militias
reflects a more general failure to properly account for the motivations of
combatants in the post-Soviet wars.24 That the bargains between warlords

23 The autumn of 1994 for Tajikistan and the summer of 1995 for Georgia are typically cited. In
Dushanbe, stability and order did not emerge seamlessly; there was territory just outside the
city limits where government forces could not safely traverse until the late 1990s. But there
was a second-tier drop-off in urban violence in late 1994 when the departure of the militias
removed the feeling of an urban war zone, making the city center safe for foreign embassies,
intergovernmental organizations, and NGOs to open their doors.
24 There are two main reasons that the spectacle of urban violence has received relatively short
shrift. The main reason is that the patterns of urban violence cut cleanly across the “master
cleavage” of the civil war: Georgians were murdering other Georgians, and Tajiks from the
same region were killing each other in the streets of the capital. The urban violence simply did
not fit the story that most people were telling. As such this period of intrafactional violence
tends to be treated as an afterthought in otherwise excellent case studies. It is typically described
as a period of general state failure, criminalized “bandit patriotism,” or (most commonly of
all) “anarchy.” Even the best descriptive accounts of the period tend to describe it in vague
Warlord Coalitions and Militia Politics 97

and their subordinates could disappear overnight undermines the idea that
the sustained presence of militias can be explained by theories that rely on
slow-moving structural factors such as state weakness, easily accessible cultural
repertoires of violence, deep primordial solidarities, or psychopathy.25 Up
until the moment that they departed the streets of the capital, the departing
militia members explicitly considered themselves part of the state, and believed
themselves to be well positioned to contest their share of the rents of
statehood. Rather than imagining these groups through the various analogies of
“ethnic armies,” “mafia businessmen,” or “criminal thugs” during this period
these groups can be productively thought of as participants in a high-stakes
tournament. Warlords struggled with the problem of recruiting and retaining
volunteers in an environment where money had lost value. Militia members
were trying to help their paramilitary captain secure a good position in the
consolidating regime among a complex and uncertain environment.
How were the stakes of this violent tournament understood? Having lived
through the rampant corruption of the late Soviet period, the emergent
violence entrepreneurs understood that the best opportunities for a better
life required affiliation with the government. Interviews with militia members
from rural areas revealed shared understandings with their urban counterparts
on this point: life was better inside the state. Foot soldiers watched as
warlords and their lieutenants secured virtual empires inside the new ministries,
inheriting centralized proto-industrial economies, with numerous bottlenecks
in the provision of public goods, each an opportunity for rent-seeking. The
ability to issue passports or transit permits, or operate buses or trucking
businesses without fear of being stopped at checkpoints, blurred the line
between smuggling and trade. Consumer goods of all kinds traveled by road to
the post-Soviet capitals. Liquor, gasoline, drugs, and cigarettes were the most
profitable trade goods, but food was imported as well. Entrepreneurs with

terms (e.g., “militias ruled the streets”), then provide a description of a changed situation
with no reference to what mechanisms produced or sustained the change (e.g., “order was
restored”). For Georgia, see Aves (1996) 5, 54–55; Darchiashvili (1997b), 3; O’Ballance (1997)
112, 133, 152–160; Leeuw (1999) 182–183; Demetriou (2002), 26; Devdariani (2005), 167;
and Zuercher (2005) 137, 148; There is virtually no scholarship on the urban aspect of the
war in Tajikistan. The narrative of state failure and anarchy emerged as a constant theme in
conversations with current and former European embassy employees and local academics – see
Akiner (2001), 37–44; Whitlock (2005); and Lezhnev (2006), 51–72. A secondary reason is
that most of the men who were the targets and perpetrators of violent tactics were disposable
and anonymous. The bilingual elites of the Soviet academy were (understandably) unwilling
descend into the streets and survey the opinions of the angry, violent, unpredictable young men
from rural areas who had taken over the streets.
25 For explanations that rely on state weakness, see Gambetta (2002), Volkov (2002),
and Lezhnev (2006). For explanations that rely on cultural repertoires of violence, see
Mardin (1978) and Shabad and Liera (1994). For explanations that rely on psychological
predispositions, see Fanon (1968), Stinchcombe (1968) and especially Mueller (2000) and
Petersen (2002).
98 Jesse Driscoll

money, property, or connections could buy police divisions wholesale with

a few well-placed appointments. Private armies could mediate contract and
wage labor disputes. Criminal kingpins and warlords were the pivot players,
as they managed the armed forces that could bring the whole system down. As
the security services formed and re-formed, commanders used their armies like
lobbyists to jockey for power, influence, and privileged jobs.
The fighters, who had often made great personal sacrifices in an attempt
to transform their society, wanted these jobs. The wars terminated in a way
that settled none of the underlying grievances that motivated young men to
fight. If they could not achieve their deepest desires, they could at least be
realistic about compensation for their sacrifices and labor. Not all jobs were
equal. The most coveted positions were in subsections of the Ministry of the
Interior and the Border Guards, as it was well understood that these jobs
could be immediately converted into tax farming and smuggling rents. Because
the politics of privatization were being worked out at the same time as the
civil wars were being settled, being in the police was a path to nearly instant
racketeering wealth. The liquidity of property rights made mutually beneficial
transactions between criminals and residual nomenklatura transparent. In both
countries, the traffic police was a regulated system of roadblocks, controlled
stops, and bribes. The worst jobs tended to be in the army, as it often left soldiers
stationed outside the capital or tasked with conducting grinding, thankless
counterinsurgency operations in the mountains. Particularly in Tajikistan, the
army came to be seen as a dumping ground for the least politically connected
militias. But at least it was a job.
The evidence that follows suggests that there was a powerfully simple logic
to the violence in this period. Militia members were recruited into militias by
either the lure of short-term benefits (e.g., excitement, racketeering rents) or the
hope of long-term jobs in the consolidated regime. These benefits were weighed
against the probability of injury or death that came with entering the violence
game under the leadership of a particular militia captain. Promises of jobs were
always conditional on the militia captain being incorporated into the regime.
Because it was impossible to predict what the final membership in the ruling
coalition would be, it was very difficult for warlords to make promises of spoils
credible to their memberships.
One of the main strategic behaviors that militia recruits engaged in when
faced with this uncertainty, was switching militias. As Table 4.2 makes
clear, side-switching during and after the civil war was relatively common,
particularly after major combat operations were concluded. It was common
for small groups of men recruited in the same neighborhood or village to
“move together,” as this would give them protection and bargaining power
within the new militia structure after switching commanders.26 Side-switching

26 As Table 3.1 from Chapter 3 reports, identical percentages of Tajiks and Georgians (about
80 percent in both subsamples) expressed that one should never really trust anyone “not in
Warlord Coalitions and Militia Politics 99

table 4.2. Evidence of a Street Violence Market: Militia Members Switched

Patrons

Variables All (%) Georgian (%) Tajik (%)

Switched Commanders at Least Once during
Civil War 35 45 25
Switched Commanders at Least Once after
Civil War 31 47 17
Quit His Militia When He Felt His Life Was
at Stake 38 52 24
Observations 173 84 89

was more prevalent in Georgia than in Tajikistan. In Georgia, these close trust
networks could be reinserted as a full-module into a different commander’s
network (allowing “horizontal” movement between militias). In Tajikistan
the avlod obligation networks that connected recruits and commanders
(a “vertical” relationship, sustained by the shared shame of disappointing an
invisible audience of second cousins and distant matrons) yielded systematically
different behaviors: Tajiks in my sample were unlikely to desert from a militia
even when their lives were at stake, and members tended to remain in the service
of the commander who recruited them. The obvious impetus for side-switching
was when a commander was killed. In Georgia, by contrast, almost half of the
respondents in my sample switched commanders at least once during the war,
and nearly four-fifths switched at least once.27
As discussed in the previous chapter, all of this manifested as an emerging
futures market in political favoritism. Violence during this period was an
indirect byproduct of high-stakes bargaining between paramilitaries competing
over the spoils of victory. Some militia captains – with the assistance of young
men recruited with promises of plunder and patronage – became valuable assets
for the regime and earned shares of black market goods and services. Others
were driven from the capital city at gunpoint. Understanding how militias were
recruited can provide insight into changes in the meaning of the violence as
the war progressed. The price of social order, at least in the minds of many
soldiers, was making sure that the distribution of wealth and influence squared
with leaders’ demonstrated ability to organize violence. If it could not, there
were strong incentives for newcomers to try and break into the system, hoping
to be bought out.

your family,” which speaks to similar cultural scripts on this matter across both of the study
populations.
27 Tajik respondents were also more than twice as likely to report that they trusted their
immediate commander. My data make it impossible to discern whether this was a cause or
effect of the pervasive Georgian militia-switching.
100 Jesse Driscoll

a political economy of militias: expansion and

contraction in four phases
In the post-Soviet wars, the decision to join a militia amounted to risking
one’s life in exchange for a short-term mix of security and loot, and the
long-term promise of better life opportunities if a recruit aligned himself with
a worthy patron. The participants in this violent scramble understood that
some warlords would be in the coalition assembled by the president and others
would be marginalized. The men who were the most likely to join these militias
tended to be either serially unemployed or temporary true believers. There was
an overabundance of young men willing to forego safer life opportunities for
the excitement of holding a gun. Separating the wheat from the chaff required
a sort of tournament among militia captains: a sorting process to determine
which individuals had the most talent for convincing others to fight on their
behalf.
One of the central characteristics of the post-Soviet wars is that they were
fought with voluntary labor in states with few meaningful institutions. There
was no trusted national currency, no banking system, and no courts to enforce
contracts. The nature of the spoils being fought over required political capital
– but the political center had already been shown to be vulnerable to rapid
disintegration. The leaders (“warlords”) and followers (“recruits”) had come
of age in a social system where kin networks were necessary for everyday
life. Friends and allies drew on dense networks of social capital, and joined
militias on the intuition that getting a foothold inside the new state bureaucracy
represented their best chance to increase their life opportunities. In one sense,
there were a few analytically distinguishable kinds of uncertainty: uncertainty
by a recruit over the character of the warlord (whether a patronage relationship
was incentive compatible), uncertainty by a recruit over which warlords
would emerge inside the “shadow state” at the end of the coalition-formation
process (whether the warlord would ever actually be in a position to dispense
patronage), and uncertainty by the patron warlord about the quality of a
recruit.
The brutal tournament cut through these layers of complexity to address the
central question: When the chips were down, which warlords could organize
violence most efficiently? This question could be answered only in real time,
and with deeds, not words. The brutal urban warfare that emerged during
the “Time of Troubles” resolved the three kinds of uncertainty in the emerging
market for militia labor. It reduced uncertainty for militia members about which
warlords were able to recruit large groups of capable fighters, driving charlatans
and incompetents away. It reduced uncertainty for militia captains about the
loyalty, fighting capability, and overall resolve of their men. And, in the end, it
reduced everyone’s uncertainty about which warlords were likely to end up in
the winning coalition and be able to deliver on their promises of patronage. The
Warlord Coalitions and Militia Politics 101

gang war in the streets accomplished what the war itself had not – it screened
the unfit members from the mob, let everyone gauge their relative strength
accurately, and finally allowed them to select a minimum winning coalition.
Most civilians’ primary memory of this period is being caught between
different spheres of authority. The practical effect was often competitive
racketeering: double and triple taxation for those unlucky stragglers who could
not, or did not, hide their wealth. Consider this story from a Dushanbe-based
musician:

“I had a friend who loved rock music. After independence … he could make good
business on the street just selling copies of his Led Zeppelin tapes. The problem was that
guys who worked for the government … they never paid. They’d come and pick over his
merchandise, take what they wanted. Then the tax police would come and do the same
thing. Then guys from the army. Then some guys came who claimed they were with the
mayor? Then the same local police came by again – guys from [his neighborhood]. He
tried to complain. They said there was nothing they could do – that everyone had to
chip in and be patriotic. So after a while, he just stopped. He figured somebody would
steal from him no matter what, but he didn’t count on lots of different people stealing.28

With no functioning state authority capable of centralizing control of the

militias and paying wages, there was not an obvious solution to the problem of
public order. Disputes over who exactly was permitted to profit from informal
tax structures led to infighting between warlords. Directors of profitable
national industries began to be killed with regularity. Informal taxation schemes
were improvised to divide up territory and spoils between warlords. There was
no recognized authority to split tax rents. As the various warlords ascended to
ministry positions, new lieutenants had to step up and prove themselves.
Various militia captains – the de facto providers of urban order – were
suddenly thrust into the midst of everyday governance dilemmas that could
not be delegated or deferred. How would food be distributed to families who
could not afford to pay? Where would the influx of refugees from rural areas
be housed? What sorts of punishments should be meted out against people
accused of crimes? Who gets to carry a gun? What should be done with aspiring
journalists who are bearing witness and naming names?29

28 Interview conducted February 21, 2007 in Dushanbe.

29 It was especially unsafe time to be an aspiring journalist or photographer. Armed actors,
especially in Tajikistan, had an interest in keeping certain details from being recorded. A partial
list of murdered Tajik journalists from this period, assembled in Panfilov (2003): Murodulla
Sheralizoda (May 5, 1992), Olim Zarobek (May 6, 1992), Shirindzhon Amridzhon (May
26, 1992), Emilia Podobed (June 1992), Turadzhoni Kobil (October 1, 1992), Olimdzhon
Iorasen (October 1992), Tabakkal Faizullo (October 1992), Ahror Sharif (December 9, 1992),
Saidmurod Iorien (January 1993), Abdulhakim Shukurzoda (April 21, 1993), Pirimkul Sattori
(May 28, 1993), Hushvaht Haidarsho (May 19, 1994), Hamidzhon Hakimov (November 18,
1994), Muhiddin Olimpur (December 12, 1995), Zuhuruddin Suiiari (April 26, 1996), Viktor
Nikulin (March 28, 1996).
102 Jesse Driscoll

The solutions to many of these problems required cooperative arrangements

between different neighborhoods. This meant either cooperation between
militia captains or expansion of one militia at the expense of another. So long
as the residual fears about the potential for a reignited wider conflict justified
their existence, there was substantial social support for popular captains to
keep their armies intact and throw their weight around the capital. But some
inevitably began to gaze nervously up the hierarchy, trying to determine who
had actually secured a “roof” and who had not.30 Competitive pressures led
militia captains to expand their recruit pool and to target each other.
There were two different sorts of newcomers to the violence game: new
foot soldiers (essentially “new cheap labor,” seeking to prove themselves as
muscle) and new organizers (essentially “new violent social capital,” seeking
to establish themselves as political entities and potential patrons). Because
all of these groups lacked formal uniforms and many commanders relied on
pseudonyms and noms de guerre, it was relatively easy to put on a black
leather jacket and take to the streets. Some of these men came from various
corners of the capital, and some came from rural areas or other towns.
Once a few promising up-and-comers had been incorporated into the police
force, there was no obvious way to limit entry into the violence market. The
general tactic was to promise permanent jobs (with pensions) in exchange for
service. This futures market in political favoritism depended on a recruit’s
belief that his patron would eventually be in a position to make good on this
promise.
It is important to note that there are two separate audiences for the theater
of militia arms-racing. First, the different militia commanders were competing
with each other for limited space at the political trough. Political parties
were brokering with armed groups and choosing which militia factions could
make good on offers of security. Weak commanders were potential targets for
dismemberment by established captains or by new organizers trying to break
into the violence game by making a name for themselves. Second, commanders
had a similar incentive to signal to potential militia recruits that they were
capable patrons. Recruitment was based on promises – “cheap talk” – mediated
by potential members’ expectations that the warlord would be capable of
paying them down the road. This was a probabilistic assessment. Was a
captain’s star rising or falling? A militia captain who could not convince recruits
to join up or recruit allies would be identified as weak and vulnerable. The sense

30 The language of the “roof” is inherited from Soviet times. As Fairbanks (1996) notes, strong
informal rules from Soviet times prohibited the betrayal of patrons. Advantage was gained by
loyalty to the party apparatus and deference to a small number of “administrative gatekeepers”
(369–372). The well-studied tendency for “cadres” to move vertically and horizontally through
the Soviet party structure is a phenomenon that has analogies in most armies, political parties,
and bureaucracies. This is perfectly compatible with prominent models of clientelism or
patronage politics developed in other parts of the globe: ideally, someone two or more tiers
higher in a hierarchy will ensure that a lower-ranked individual’s interests are represented.
Warlord Coalitions and Militia Politics 103

that “winners win,” and that success was the best predictor of future success,
drove many militia captains in Tajikistan and Georgia to escalate neighborhood
violence. Captains took serious risks to provoke responses from other militia
captains so that they could establish a reputation for courage and staying power.
It was better to be an agent of a strong, politically ascendant warlord than a
weak one.
The processes unfolded in five analytically distinguishable phases.

Phase 1: Warre
The breakdown of social order that accompanied the first months of inde-
pendence provided an opportunity for unemployed youth to shoot guns,
escape the boredom of their daily lives, experience the thrill of taking part
in demonstrations, and commit petty criminal acts. As noted earlier, all of
the armed groups that fought in the post-Soviet wars developed criminal
characteristics and recruited from the urban and rural underclass. Yet to treat
these groups as nothing more than roving bands of criminal alcoholic youths
intentionally misunderstands the nature of the social bonds that kept these
groups in the field.31 A Tajik field commander, explaining why he recruited
primarily from his karate dojo, stated: “I needed men who were serious. I
knew things would get bad, and I wanted men who I trusted to watch my
back.”32 A former Mkhedrioni member echoed this – “Of course there were
kids and drunks around. But that’s not what [the bosses] needed. They needed
reliable people. We were building an army, not a gang. We were reclaiming
the nation from the Russians.”33 Interview respondents in both countries often
freely admitted that other men in their unit were alcoholics, drug addicts, and
troublemakers who liked violence for its own sake – but they were also usually
quick to note that those men tended to be bad soldiers who could not see the
big picture or keep their peers’ respect.34 No one wants to risk his life for a
drug addict or sadist.

31 My attempt to describe these groups is found in Chapter 3. For an eloquent presentation of a

simpler alternative perspective, see Mueller (2000).
32 Interview conducted July 26, 2007 in Dushanbe. Later in the interview, he added: “Not
everyone likes to hit.”
33 Interview with Nicholas Rurua, currently serving in the Georgian parliament. Interview
conducted December 03, 2006 in Tbilisi.
34 This is broadly consistent with a number of sociological works on the organization and
composition of gangs. Peripheral members are often relatively violent and difficult to control,
but the members of this disorganized milieu tend to be trying to impress the core of competent
and disciplined organizers who manage economic and political relationships. See Jankowski
(1992) and Venkatesh (2007, 2008). Almost none of my respondents identified themselves as
free-riders, of course. It is possible that this reflects a systematic bias in my interview pool, but
I believe it is more likely that interview respondents presented idealized versions of their own
motivations and actions and attributed less favorable characteristics to others.
104 Jesse Driscoll

The “master cleavage” of the war dominated targeting decisions and pogrom
behaviors in this period. In Dushanbe, Gharmi, Karategini, and Pamiri
neighborhoods were essentially cleansed through targeted campaigns of murder
and rape. The property deeds of the victims’ homes – sold at a tremendous
loss – were transferred to real estate speculators with political connections
to the Popular Front or the new regime.35 In Georgia, though there was a
similar deluge of anarchy in which ethnic minorities faced harassment and
humiliation on the streets, neighborhoods were not purged in the same way.
Families could largely stay indoors and wait out the chaos. In response to this
wave of urban violence, many citizens closed their businesses, stayed indoors,
or fled the capital. Many people kept their distance from the militias in the
1991–1993 period. The collapse of public order and economic life meant that
the spoils of victory were uncertain, and joining one of these groups required
participating in a culture of violence that was repulsive to civilized people.

Phase 2: Opportunistic Joiners in the Aftermath

Recruiting dynamics changed once the capital city was under the control
of a strong militia coalition. Potential recruits realized that there was a
real possibility that victorious militia commanders would actually be able
to make good on their promises. But the final ruling coalition was still in
flux. Everyone knew that political fortunes were uncertain. On one hand,
it was very difficult for anyone to determine which commanders were lying
(to themselves, as much as to potential recruits) about their future political
fortunes, or to determine which commanders were simple opportunists. On the
other hand, capturing the capital city or acquiring a position of status was no
longer hypothetical – political power was suddenly fluid and tangible. Ministry
portfolios were actually being doled out, often to well-known criminals, as
political acknowledgment of certain individuals’ armed strength. The stakes of
victory in the consolidation game were clarified as foreign aid began to trickle
into state coffers.
Without any functioning state bodies to arrest them, and without any real
risk of dying in war to deter them, many young men were attracted to the
seductive glamour of life with a nationalist militia.36 A tertiary milieu of young

35 The character of the ruling coalition that supported Rakhmonov was never clearer than in the
treatment of “enemy” ethnic groups after the state was seized. Militia members in Tajikistan
were encouraged to “drive [the Pamiris] back to their mountains” by their captains, leading to
a highly organized campaign of civilian displacement and property expropriation by the new
Kulobi rulers. Interview conducted August 4, 2007 in Kulob. This had the effect of raising the
political stakes of politics significantly in Tajikistan, making control of the state apparatus an
all-or-nothing game. It was a tactic that was quite useful in uniting the various factions of the
Popular Front for a time.
36 Interviews with participants and “opportunistic joiners” revealed three broad sets of expla-
nations for missing the war. First, many were simply deterred by the threat of dying in the
Warlord Coalitions and Militia Politics 105

men – the sort who already knew that they were not likely to grow up to be
middle-class professionals – were always available, anxious to ascend through
violent rituals to manhood. In the chaotic environment sketched previously, it
was relatively easy for a newcomer to “pass” as a member of almost any group,
if a few others would vouch for him.37
At the same time that the supply of youth on the streets was expanding,
field commanders, militia captains, paramilitary lieutenants, and other wartime
coalition members who controlled the capital were increasing their demand for
new recruits. With the spoils of victory being divided in real time, everyone –
militia captains, paramilitary lieutenants, and foot soldier recruits alike – came
to understand that the process of moving from a “winning wartime coalition”
to a “minimum winning coalition” would be competitive. There would be
relative winners and relative losers. Some of the warlords would be pushed
out of the ruling coalition and others would reappropriate their hard-earned
wealth. The short civil wars meant that the militia coalitions had seized the
capitals without developing any institutions to redistribute spoils equitably.
Though the concern of being written out of the coalition was more acute in
Tajikistan than in Georgia, militia members in both countries who had fought
since the beginning recall this period of uncertainty largely the same way: They
went from not being sure if their group could win the war to not being sure if
they could keep what they had won. In a few short weeks, the inherent tensions
relating to the divisibility of state spoils started to splinter the unity of the
umbrella militia coalitions. Identifying oneself as “Mkhedrioni” or “Popular
Front” was increasingly meaningless and redundant – what mattered was
one’s particular factional commander, and groups began to identify themselves
to each other by referencing their patronage relationship. Fears of being
manipulated and discarded were rampant.
Urban militias were, at this point, a crude but well-understood kind of
political insurance. Their primary purpose was to remind other social actors
that there would be serious consequences if the warlord, or the group he
claimed to speak for, was cut out of the distribution of pork. But the outlines
of a stable core coalition were beginning to take shape, and individual militia
members could take concrete actions – such as recruiting friends or switching
commanders – to improve their lot.

fighting. Second, respondents who had families had to see to their safety during the war, which
often meant fleeing with their children or parents to rural safe zones or to live with urban
relatives. A third explanation was simply that their offers of service were rejected by whoever
was recruiting in particular area, usually because they had clan or ethnic criteria for group
membership. A Georgian respondent reported with great seriousness that he was “too fat” to
join the Mkhedrioni in his neighborhood in 1992, though this did not stop him from eventually
signing up a year later after the war in Abkhazia was settled. Interview conducted November
11, 2006 in Kutaisi.
37 Certain tattoos helped distinguish long-time members of the vori v zkone (“Thieves in Law”)
in Georgia; there was not an analogous signaling mechanism in the Tajik prison system.
106 Jesse Driscoll

Phase 3: Attrition and Selection in the Time of Troubles

Urban violence escalated. In Tajikistan, once it became obvious that armed
groups from the region of Kulob had begun to consolidate their position
inside ministries at the direct expense of other allies in the Popular Front, a
steady stream of Hissori, Khojandi, and Lakai foot soldiers began to drift into
Dushanbe. As a former combatant from the region of Khojand recalled this
period, “We knew we needed more men, so that Kulov [the captain] could
speak with a louder voice. …[S]o we brought friends.”38 Georgians could also
draw on clan and family networks that stretched to rural areas. Oleg’s anecdote
is representative: “My brother told me that he was sure to get a job in the
Special Reaction Emergency Unit if his group was strong enough to get noticed
by Dzhaba [Ioseliani, the head of the Mkhedrioni]. … He said that it would
be good for the whole family, so I sold one [of the family horses], took my
hunting rifle, and went to get a car in Rust’avi [to come to Tbilisi].” Oleg was a
self-described family man who had fled with his family, staying neutral during
the worst phases of the civil war violence, but his family pressured him to help
his brother get a good job during the scramble for post-Soviet spoils.39 “The
state” may have been just a handful of half-empty buildings in the capital city,
but in the words of a former Popular Front member in Tajikistan, “Everyone
got it … whoever controlled the police … controlled the bazaars and the streets.
And whoever controlled the streets could whisper in the ear of the big man
[Rakhmonov], who would stand like a statue.”40
As it became more and more obvious that the symbols of the old state,
symbols of the Georgian church, and political slogans were providing cover
for open banditry, militia commanders’ moral authority declined. This opened
the door for a third group of militia recruits: local self-defense militias that
formed as a response to the disorder and chaos that the “first wave” of
militias unleashed. These men, many of whom were career police officers,
self-consciously described themselves as cut from a different cloth than those
in the militias that emerged from anarchy. Rather than being motivated by
abstract fears (e.g., Russian influence), they often described their fears as
local and tractable. They often reported disgust at the criminal elements
within the mass movements. These urban formations described themselves
as protectors of innocents from arbitrary violence, providers of local public
goods (most notably security), and generally fair providers of order – similar
to how the Mkhedrioni presented themselves in the pre-independence period.
Their challenge to the Mkhedrioni and National Guard formations was met
with predictable violence. Various “neighborhood defense committees” formed
out of youth gangs emerging from the slums of Tbilisi and the outskirts of
Dushanbe, and the members began to operate in unpaid parallel positions to the

38 Interview conducted May 14, 2007 in Bishkek.

39 Interview conducted November 2, 2006 in Tbilisi.
40 Interview conducted July 17, 2007 in Dushanbe.
Warlord Coalitions and Militia Politics 107

lowest levels of the power ministries in both capitals. The basic arrangements
described were strikingly similar: a leader was given permission to collect taxes
on behalf of a particular militia commander in exchange for local autonomy
and free access to the electricity and water grid, with the vague promise that
their positions would become salaried at some point in the future.
It did not take long for newcomers to realize they might be able to shoot
their way to more power. A common strategy was building up a competent
group of violence specialists, taking territory, and eventually merging forces
with a captain who would share his veteran nationalist credentials. Levan, a
neighborhood organizer who was later incorporated with his men into the local
police, understood that holding firm would eventually allow him to be “bought
out” and described the bargaining matter-of-factly:

“The Mkhedrioni were serious at first … but then some of the best of them died in the
war, I think. Anyway, after a while, their men asked for too much … there were beatings
and then shooting at night and then a rape, and it got to be too much to bear. So we
organized ourselves … and eventually their bosses came to ask me if I wanted to work in
the Ministry [of the Interior]. I said no … [but then] they offered again, and I accepted.”41

Militia members who were active in Dushanbe and Tbilisi tend to recount the
interfactional violence of the 1993–1994 period through one of two self-serving
narratives. The first story was the hardened, virtuous “true believer” militia
members fighting against upstarts, criminals, and opportunistic newcomers. A
member of the Georgian National Guard put it like this:

“Kids from the city … they watched movies, they thought they wanted to be war heroes
like Rambo. But they had missed the war. They knew nothing. We needed another war
to scare out the kids [who] never went to war. So we needed to bring the war here.”42

The alternative version was that it was the “real” soldiers who were the
goons and cowards, less equipped for this new period of violence than new
recruits coming up from the streets, who had ties to the solidarity communities
emerging in various urban slums and squatter camps. To hear them tell it, these
newcomers were just as ruthless and organized in their application of violence
as the first wave of fighters but possessed better discipline, better intelligence
thanks to ties to the community, and perhaps simply a greater willingness
to stand and fight to hold their turf. What both groups agree on is that this
period was far more dangerous than participating in the rural phase of the civil
war. As one veteran-turned-nightclub-bouncer put it: “There were frontlines in

41 Interview conducted November 17, 2007 in Tbilisi. Readers might note the casual reference to
the Ministry of the Interior as “The Ministry” – clearly its own center of institutional gravity.
42 Interview conducted November 3, 2006 in Tbilisi. Abdullo, a Dushanbe police captain, stated:
“No one would back down without a fight. So, we fought.” Interview conducted February 2,
2007, in Dushanbe.
108 Jesse Driscoll

Abkhazia, and if you weren’t brave you could hang back. But here [in Tbilisi],
there were no frontlines. Everyone is just a drive away.”43
The first few months of violence took their toll on many of the prominent
warlords and militia armies. Militias organized around charismatic authority
were vulnerable to decapitation. In Georgia, Deputy Minister of Internal Affairs
Gregori Gulua and Deputy Minister of Defense Nika Kekelidze were brutally
assassinated within weeks of each other in the spring of 1994. Assassination of
prominent militia leaders induced rapid and unpredictable shifts in the political
fortunes of subordinates. In March of 1993, two of the most prominent
Tajik warlords – Sangak Safarov and Faizali Saidov – were both killed in a
shootout.44 Until that time, Rakhmonov had been seen as the hand-picked
representative of Safarov, the most prominent commander of the Popular
Front.45 Suddenly the seat behind the throne was vacant.
This kind of violence had two effects. The first effect was attrition because it
raised the costs of organizing based on raw opportunism. Men who joined
these groups with a “live only for today” attitude were poorly suited for
this new crucible. Even more important than weeding out the unfit soldiers
was weeding out unfit warlords. If a leader could not sustain optimism
about his ability to make it into the state, or maintain a reputation for
generously distributing spoils to his men, his men were likely to switch to a
different commander (in Georgia), replace him with a better commander (in
Tajikistan), or simply quit. A secondary effect was the creation of common
knowledge in the violence marketplace regarding the relative strength of
different warlords. Both Tajikistan and Georgia are honor-based societies.
Blood-feuds are enforced. Public actions that could risk spirals of violence were
costly signals of willingness to endure extended retaliatory violence. The strong

43 Interview conducted October 28, 2006 in Tbilisi.

44 Experts still disagree on the precise details of Safarov’s death. Olivier Roy (2000) and a few
others believe that the argument between Safarov and Saidov escalated over accusations that
Saidov was “too soft” on returning Gharmi refugees coming down from the mountains of
Afghanistan (49). Safarov, Roy claims, feared that with the Gharmis driven from the region,
the ethnic balance would shift in favor of the Uzbeks in areas around Kurgan-Tupe; Saidov’s
mother was Lakai Uzbek, and he rejected any right of return for the Gharmis. Gretsky (1995)
plausibly claims that this was a dispute about control of the new national army. Nourzhanov
(2005) articulates the consensus belief on the deaths of Safarov and Saidov: “the whole accident
was planned in Dushanbe … Kulyobi commanders were liquidated by the very same people
whom they had put in power” (118).
45 President Rakhmonov deftly handled the political situation in the wake of this unexpected
event to demonstrate that he enjoyed the support of a plurality of remaining Kulobi militia
commanders, heading off the first serious intra-Kulobi power struggle. As the various militia
captains chose sides, or hedged their bets by pushing for increased regional autonomy outside
of the capital city (the strategy of Ibodullo Boimatov and Mahmud Khudoiberdiev, two
extremely influential non-Kulobi warlords), the Tajik army “became nothing more than an
arms depot for the new political party.” Interview with UN representative, conducted August
16, 2005 in Dushanbe.
Warlord Coalitions and Militia Politics 109

were sorted from the imitators, the confident from the brash. After a year of
this, there were no imitators.
The most common flashpoints for symbolic violence were the checkpoints
and roadblocks that various militia members established inside the urban areas.
Checkpoints at major street intersections made it possible to control which
vehicles moved in and out of particularly wealthy neighborhoods, black- and
gray-market bazaars, or areas with large amounts of foot traffic. These areas
provided a key source of income for self-financing urban militias and were
signs of prestige and strength for the captains who controlled them. A murder
taking place in a protected bazaar would drive off merchants, lowering the
tax base for the warlord offering protection. A shooting or public beating
at a roadblock would test the mettle of the soldiers who remained. Different
paramilitary factions would intentionally march their troops through parts of
town controlled by other militias, openly risking provocation and escalation. In
Georgia, one Mkhedrioni described his clash with another sub-group of “new”
Mkhedrioni:

“They came at us at night, when we were walking home. … They surprised us, and
[aimed – gesture] guns on us. Then they took turns beating us with boards. … One of
them broke my finger kicking me while I tried to [protect – gesture] my head. When they
were done, one of them said that they didn’t want to have to come back, but that it was
payback, and that if we came back next time they would use the guns. … [W]e wanted
to take the fight to them, but Davit [our boss] calmed us down. They both went and
talked to Dzhaba, and he sorted it out. They didn’t come to our street anymore, and we
never saw each other again.”46

Militia membership based on vague associations had gone from being a

relatively fun and even lucrative lifestyle choice to being an extremely
dangerous vocation. As noted previously, the point of these acts of violence
was not just the control of the checkpoint or the taxes from the bazaar – the
point was to see which groups were actually capable of convincing their men
to stick around and endure punishing losses and which groups were trying to
fake their way through the consolidation process. A Kulobi foot soldier said of
this period, “We stayed in the streets because we needed to show Rustam [our

46 Interview conducted November 14, 2006 in Tbilisi. This account highlights the important
role of informal institutions in conflict resolution – in particular the charismatic authority
of Dzhaba, the head of the Mkhedrioni. The form that the resolution took (division of
territory) also reflects a political bargain between commanders to collude to share the rents
(in the literal sense of apartment rents) in different parts of the capital city. It is also clear
from this anecdote that Dzhaba had no ability to control the day-to-day operations of the
lower-ranks of “his” hierarchy. Wheatley (2005) observed that, whatever its origin myth, by
1994: “[S]maller criminal gangs (often referred to as ‘Mkhedrioni’ but quite clearly beyond
the control of Jaba Ioseliani and other leaders of the ‘official’ Mkhedrioni) dominated at local
level, typically offering protection to local communities against other marauding gangs. Very
often the leaders of these gangs would assume nominal state positions, such as that of mayor
or district administrator (gamgebeli).” 80.
110 Jesse Driscoll

boss] that we were capable of staying in the streets. We were told ‘no one has
use for cowards.”’47
Clashes gradually became more and more dangerous. Competition for urban
territory was often settled with guns or pipes, and bodies were often left
on display to show the cost of being affiliated with a weak faction. Armed
shootouts with high-powered weapons – and, in more than one case, even
the shelling of urban neighborhoods with mortars – were the end result of
these escalating displays of paramilitary strength. Sasha, a member of a Hissori
faction of the Popular Front, described the summer of 1994 as the most
dangerous period of the war:
“It was the worst time … it was so hot during the days, and there was no gas, no
water. And every week, we would stand at our post in the heat. And in the night,
there was always gunfire … One night our wall [checkpoint] was attacked by men with
grenades and explosives, who sprayed bullets all down the street to drive us away –
but we took shelter and returned fire into the darkness … Those men that ran away
weren’t welcomed back … It was very dangerous, and no one was making money on the
job … But we couldn’t leave … not without just all going home [leaving Dushanbe and
going to Hissor]. There were three other factions [of the Ministry of the Interior] that
were waiting for an opportunity to take our taxes, to show that [our commander] was
weak, to take his job and send us all into the army. We did not fight a war to be unpaid
infantry for Kulobi generals.”48

Or consider this quote from Georgia:

“Everyone knew which was our restaurant; it was Levan’s space – Mkhedrioni
space … we even had a picture of Dzhaba on the wall … we would all meet up there
after work [at the Ministry] to relax. But everything changed when Levan got pulled up
to work [in the Ministry of Justice] with [Tedo] Ninidze … Then he was big time … there
was one night … it was February 1995 … we realized that the room was thinning
out … people were all going back to the toilets but not returning. Men in ski masks
had come in through the back, and were were waiting for us there, one at a time. They
put our faces in the toilets and … knocked out my teeth on the porcelain. … They said
that we should tell Levan what happened. … when we did, and he told us that we should
think about joining the Army, that we could get hot meals and better work there … and
that we wouldn’t be able to stay at the Ministry.49

Some recruits, expecting good jobs in the state bureaucracy, could convince
themselves to endure these kinds of risks.50 Dense social networks and ties
47 Interview conducted August 15, 2007 in Dushanbe.
48 Interview conducted February 7, 2007 in Dushanbe. Note the explicit ranking of jobs in terms
of perceived enrichment opportunities: The Ministry of the Interior, where one could expect
relative autonomy and many opportunities to interface with merchants and extort civilians, is
the prize. Valued much lower is a uniformed job in the military hierarchy.
49 Interview conducted October 25, 2006 in Tbilisi.
50 Generalizations in this section are drawn from a statistical analysis of only the sub-sample
of respondents who reported being present in the capital city for the “Time of Troubles,”
excluding fighters who fought only in the rural parts of either conflict.
Warlord Coalitions and Militia Politics 111

table 4.3. Which Militia Members Quit When Their Lives Were at Stake?

(1) (2) (3) (4)

Variables Full Favored Only Georgian Only Tajik
Georgian 0.853# 0.957∗
(0.695) (0.495)
Joined after Fighting 2.311∗∗∗ 2.390∗∗∗ 2.237∗∗∗ 2.777∗∗
(0.600) (0.502) (0.639) (0.931)
∗ ∗ ∗
Joined to Protect Family −1.465 −1.470 −1.703 −0.608
(0.727) (0.716) (0.832) (1.710)
Operated Close to Home 1.465∗ 1.558∗∗ 1.321# 1.996∗
(0.609) (0.599) (0.765) (1.021)
Switched Units during War 0.141
(0.516)
Switched Units after War 0.273
(0.657)
Trusted Commander −0.707# −0.658# −0.522 −0.710
(0.496) (0.474) (0.603) (0.851)
Stayed for a Better Job 0.043
(0.611)
Stayed for Fun −0.837# −0.863# −0.397 −1.949#
(0.545) (0.531) (0.624) (1.243)
Wanted Influence in Capital −0.568
(0.661)
Constant 0.169 −0.254 0.751 −1.139
(1.111) (0.848) (0.875) (1.884)

Observations 115 120 69 51

Pseudo-r2 0.281 0.286 0.229 0.351
Logit regression results with Quit When Things Got Violent as dependent variable.
Standard errors in parentheses. *** p < 0.001; ** p < 0.01; * p < 0.05; # p < 0.25.

of reciprocal obligation convinced many militia recruits to endure with their

commanders through this violent tournament. But which of the two self-serving
narratives sketched at the beginning of this section was more accurate?
What kinds of militia members had the comparative advantage during the
consolidation phase?
Table 4.3 is an attempt to address these questions systematically. In my
sample 43 percent of respondents – half of the Georgians and about a third
of the Tajiks – reported that they retired from militia activity because they
112 Jesse Driscoll

feared for their lives. Because this variable is dichotomous, a logit estimator
can be employed to see what individual characteristics made a militia member
relatively likely to quit in the face of violence compared to his opponents who
stayed in the fight. The independent variables included in the regression are also
coded as binary variables for ease of interpretation. Models 1 and Model 2
present the cumulative and favored models. In my sample, “opportunistic
joiners” – new entrants into the violence game who joined the civil war after
the bulk of the violence was completed – were more likely to quit the streets
when violence intensified. Social ties between commander and recruit predicted
staying power. Men who fought through the civil war had established close ties
to their patrons and had a higher expectation of a good job (conditional on their
militia being a part of the solidified winning coalition, of course). In addition
to the simple “opportunism” dummy, two additional indicators of dense social
ties emerged as statistically significant, though only marginally so in some
specifications. The first was a dummy variable for Georgian paramilitaries –
which, as discussed earlier, were substantially more prone to defection and
side-switching – and the second, a dummy variable for the question of whether a
militia member reported “really trusting” their commander. Plainly, Georgians
in my sample were more likely to quit when faced with violence. Recruits
reporting that they trusted their commanders also reported that they were
willing to take risks for them.
Members of a militia that formed out of self defense – who reported “joining
primarily to defend their family” – were relatively willing to endure violence.51
This variable can be interpreted as sorting “opportunistic” joiners from “late”
joiners. Urban self-defense militias that formed during the postwar chaos
were well-incentivized to stick around. Urban hangers-on who joined after the
fighting, but were not actually acting in defense of the neighborhood, were
likely to quit in the face of determined terrorism. The minority of respondents
in both states who reported carrying out militia operations close to home
were more likely to retire from militia life when faced with serious violence,
which can be explained by a closer examination of these thirty-nine cases. In
twenty-eight of them, the “home region” that the respondent described was
a particular neighborhood of the capital city, meaning this was an additional
proxy variable for (largely late-coming) Tbilisi- and Dushanbe-based gangs. My
interpretation of this trend is that militia recruits whose homes were in urban
areas were more likely to have friends and family that could help get them visas
out of the country or resettle them in a stable job. Many urban militia recruits
had better “exit” options. Rural migrants were, as a rule, more desperate, and,
as a result, more willing to endure violence to avoid returning home.
A disturbing and unexpected pattern appears in the data: militia members
who reported “fun,” “excitement,” or “enjoyment of the experience” as

51 I attribute this trend to a higher level of psychological satisfaction and cognitive consistency
associated with being “on the defensive” in an anarchic social environment.
Warlord Coalitions and Militia Politics 113

a motivation for joining a militia or sustaining militia membership were,

statistically speaking, more likely to sustain ties with a militia when confronted
with violence. Given that it was the opportunistic Georgians (“Mkhedrioni
Youth”) who, for the most part, were reporting having joined for fun, my prior
expectation was that multicollinearity would wash out statistical significance,
or that the coefficient would be signed in the opposite direction. To make sense
of these trends, Models 3 and 4 subdivided the data further into Tajik and
Georgian militia members. A few new patterns emerge. The “opportunist”
and “close to home” variables provide powerful, if indirect, evidence that the
militia members who were most willing to risk their lives were those who
joined early and were contributing their social capital to a warlord’s political
project on behalf of their civilian social network. Those networks were stronger
when they originated in rural, rather than in urban, areas. In Tajikistan,
avlod ties bind tight. Georgian responses to the questions about “protecting
one’s family” and “really trusting the commander” drove the general results in
Models 1 and 2. In Georgia, only recruits who were successful at finding militia
commanders whom they trusted enough to persevere through the “Time of
Troubles” stuck around. There was insufficient variation in the Tajik responses
to these questions to generate meaningful statistical significance in the face of
controls.52 And the puzzle of the “good times” variable is resolved: the effect
was driven by the small number (11) of Tajiks who responded in the affirmative,
rather than the large number of Georgians (33) in the various subsamples.
Returning to the interview transcripts, I realized that this trend was driven
by a mix of hardened veterans of the Soviet campaigns in Afghanistan and a
few very scary characters who reported “enjoying” their time in the militias
because they enjoyed the power that came to them from the sudden reversal of
long-established hierarchies.
Militia members considered their life opportunities inside a politically
aspiring militia in contrast to the relevant counterfactuals. These calculations
took place in a complex social environment where there was never enough
information to confidently assess whether one was placing the right bet.
Men were weighing the costs and benefits of sustained membership, taking
calculated risks, and implementing their best strategies on a day-to-day basis.
This is not to imply that these men engaged in hyperrational behavior on
a day-to-day basis, nor is it to dismiss the emotional and psychological
complexity that surely surrounded every individual decision to enlist in a
militia, participate in violent actions, take beatings, watch friends be shot at
close range, or eventually attempt to return to civilian life. Patterns in the data
simply suggest that they understood themselves to be embedded in a high-stakes

52 An alternative explanation is that the same sorts of psychological and emotional mechanisms
in the laundry list above (e.g., the tyranny of sunk costs, shame, etc.) may have made it difficult
for the respondent to admit that he had never really trusted his commander in the first place –
but only in Tajikistan.
114 Jesse Driscoll

bargaining game and weighed the consequences of their actions. Players at the
bottom of the hierarchy were watching the political consolidation process at the
top with a great deal of attention and focus. Foot soldiers were keenly aware
of the week-to-week and even day-to-day shifts in the fortunes of prominent
militia commanders as they navigated the thicket of nomenklatura politics, and
were constantly reevaluating their own safety and prospects at the bottom of
the power structure based on their expectations of what was happening to
their patron. In general, the relative power of different leaders came to be
measured by their official title in the regime. Moreover, there was a rough
consensus that the best jobs would go to the men who demonstrated an ability
to absorb punishment and stand firm, and if a leader acquired a reputation for
weakness, his men would surely seek a different patron. In dozens of interviews,
respondents said that the violence gave them a chance to see whether their
commanders were “real men.”

Denouement and Deterrence in the “Time of Troubles”

After a few months of urban attrition, militia competition normalized. As the
supply of recruits dried up, some of the youth gangs disbanded completely.
Yet, as large militias continued to fragment, it became clear that there was,
in fact, a limit to the number of potential contenders who could compete in
this tournament. Only those warlords with access to deep reservoirs of social
capital and powerful friends could survive for long. Groups became smaller,
better organized, and their leaders began to view each other differently. This
new period featured smaller numbers of determined and committed men, all
acutely conscious of the fact that they might, at any time, be forced to prove that
they were worthy of inclusion in the ruling coalition. Cosmetic displays of force
became less necessary, but trust between warlords did not emerge automatically.
And no one really disarmed.
In this new environment, fierce, well-armed, and unpredictable militias
on every street corner were gradually replaced by “connected men” who
traded on their reputation for friendship with other respected and prominent
militia heads. Positions inside the government became an acceptable metric
for keeping score of political favors, as tactical bargains calcified into strategic
alliances. Many “deposed” or “excluded” warlords during this period were
still making decent sums of money managing drug transit lines, sitting on
import–export bottlenecks, or managing bazaars – high reservation values r,
in the language of the model from Chapter 2. The loss of an esteemed position
in a ministry, however, indicated a decline in a warlord’s long-term bargaining
power vis-à-vis other warlords, as it indicated that he had lost his protection
from the president and his time in the winning circle would eventually draw
to a close. It was a sign of true strength to openly flaunt the authority of the
president, normally only undertaken by warlords who had a firm territorial
base and foreign guarantees of their security, such as Aslan Abashidze in
Warlord Coalitions and Militia Politics 115

Georgia or Mahmud Khudoiberdiyev in Tajikistan. These early moves toward

a formalization of expectations, transparent assessment of risks, and stable
political relationships between a fixed number of militias began to congeal into
something recognizably state-like. A Mkhedrioni member in Rustavi summed
this transition from violence to connections well: “At first none of us knew who
was who … [but] after a few months, it was really clear who the big men were.
Then it just becomes about who you know, and if you have respect.”53
As it became obvious that the umbrella organizations had outlived their
usefulness as mechanisms of patronage, and affiliation with a large number
of undisciplined street hooligans began to be seen as a liability, militia
formations began to restrict their membership to the core of original friends
and allies. The threat of violence was still ever-present, but as reputation
began to take the place of brute force, it became possible for militia members
to cease carrying weapons, which reduced the frequency and intensity of
armed brush-ups on the streets.54 This made it easier for the captains to
make credible promises to their subordinates. As family networks reasserted
themselves and unaffiliated opportunists fell away, another mechanism raised
the costs of violence: blood feud deterrence. Once all the players knew each
other, individualized attribution for murders was relatively easy. And since,
at this point, no one took seriously arguments about “revolutionary times”
or “anarchy,” the escalation of blood feuds was reported to be a plausible
deterrent for respondents in both countries. Efforts were made to keep civilians
in their neighborhoods where they could be easily controlled and taxed. To this
day, most adult residents of Dushanbe can recall the name of the warlord who
controlled their neighborhood bazaar through this period.
It also became clear during this time that allegiances between militias,
and across different ministries of the new government structures, provided
a better source of funding and resources than the unaffiliated racketeering
projects. Friends in high places could make sure that their subordinates got
access to government supplies of ammunition and intelligence and allowed
them easier coordination with police and army forces to arrest, intimidate, or
disappear opponents. Control of state ministries also lowered the transaction
costs associated with smuggling gasoline, narcotics, tobacco, and weapons.
In fact, practically all of the economic activities that had allowed militias to
be self-financing during the initial period of state failure were fairly natural

53 Interview conducted November 22, 2006 in Rustavi.

54 When asked about this, most Georgian militia members insisted that this was because of
leadership effects at the highest levels. Both Kitovani and Ioseliani were already dreaming of
a future where they would rehabilitate themselves and transmogrify into “legitimate players”
in Moscow’s emergent class of businessmen, playing down their criminal ties. They impressed
the need for restraint and a culture of moral virtue for their men. There is a famous (though
possibly apocryphal) story that Kitovani ordered one of his junior lieutenants to be publicly
tortured and killed for raping (or perhaps simply sexually harassing) a young woman whose
uncle was a Georgian Orthodox priest.
116 Jesse Driscoll

candidates for economies of scale, best achieved through a state monopoly

inside the Ministry of the Interior.55
One of the major prizes during this transition period was downtown
real estate. A number of neighborhoods adjoining the main downtown
thoroughfare (Rudaki in Dushanbe, Rustaveli in Tbilisi) contained apartment
buildings that had been granted by the Soviet state to members of the local
intelligentsia, academics, artists, and local party notables. In Dushanbe, many
of these individuals had used their connections to simply flee the country with
whatever they could carry, leaving their apartments in the hands of friends and
relatives. In Tbilisi, by contrast, deeds, titles, and property rights tended to pass
without much incident to their previous Soviet owners, though keeping one’s
family apartment ultimately required friends in high places. By midway through
1994, a number of innovative militia captains in the Ministry of the Interior
(notably Kakha “Black Panther” Tamunindze, who also had connections in
the Office of the Prosecutor General) began to force families near Rustaveli
and in the Old City to pay protection money, gradually raising informal taxes
until they finally acquired title transfers. The only protection against this sort
of expropriation was for a family member to acquire political protection from
someone in the political hierarchy even higher than the militia captain. Often
this was Shevardnadze himself, who took special care to cultivate favors among
the intellectuals. In Tajikistan, the intimidation of civilians who happened to
be sitting atop valuable real estate was often more crude and cruel.56 In both
capital cities, many of the refurnished houses that are today rented to Western
Embassy personnel, aid workers, and diplomats were acquired at gunpoint
during this period.
Coalitions of strong militia captains colluded to protect common interests.
Together they were capable of acting collectively against opponents or new
entrants into the violence game. In time, trust and reputation became resources
that were just as important as men or weapons, and political relationships
gradually replaced visible demonstrations of group strength. A more orderly
sort of bargaining was possible, with smaller numbers of disciplined individuals
as enforcers. The situation in the capital city moved from something that
could be described as “anarchy” to something that looked far more organized.
Essentially, it was a stand-off between competing armed factions, each with
powerful patrons within the government. This process favored the militias that

55 Akiner (2001) and Zuercher (2005). Shevardnadze brought warlords with interests in the black
markets for oil futures, scrap metal export, drug trafficking, and currency speculation – all
businesses that thrived in periods of stability – into the Ministerstvo Vnutrennikh Del (MVD),
the old Soviet Ministry of the Interior. The highest level of the Ministry of the Interior was
composed of a mix of old friends from Shevardnadze’s time in the ministry and co-opted
Mkhedrioni organizers.
56 In Tajikistan, brazen quasi-legal land grabs by politically protected individuals persist to this
day. The 2006 bulldozing of one of Central Asia’s oldest Jewish synagogues for the construction
of a third presidential palace is an exemplar of this trend.
Warlord Coalitions and Militia Politics 117

could play politics, and it was the crafty leaders, those who were able to foster
close ties with other militias, build tactical and strategic alliances with other
states’ intelligence services, and secure access to financiers and commercial
front companies to launder money and distribute illicit goods, who came to be
recognized as the dominant players. Violent political competition had reached
an equilibrium. Thus, the ruling coalition was born in the streets.

uncertainty and updating

For Georgian or Tajik citizens living through this period of transitory anarchy,
the view from 1994 was bleak. Neither civil war seemed to be on the path
to resolution. Enemies had been driven into the mountains, but with foreign
support they could easily return. As a veteran noted: “At first, we thought that
Gamsakhurdia and his army would return … After the debacle in Sukhumi,
my friends and I returned to the capital city to see what [the Russians] would
do next to undermine us. Would they stir trouble in Adjara? Javakheti?
Akhalkalaki? Or try again in the city [Tbilisi]? No matter what, we had to be
vigilant and ready.”57 In Tajikistan, the UTO had regrouped in the mountains
of Afghanistan and were attempting to reinsert themselves into Tajik politics.
The important questions for the victorious militias revolved around distributive
politics, and how the new state would be organized. Everyone knew that
Rakhmonov and the Kulobi coalition that installed him would claim the state
apparatus for themselves at the direct expense of the Leninobodi families that
had run the state continuously since the 1950s. But the situation was still fluid,
and most predicted that Kulobi rule would be a passing phase, with politically
connected Leninobidis returning to their traditional positions at the upper tier
of the state. If the Kulobis of Leninobod made a serious bid for power with
backing from Uzbekistan and Moscow, the war could easily have restarted.
The drawdown in militia membership was precipitated by a consensus
among militia members on the streets that the window for being “bought
out” by the state had closed. In both states, this new expectation was driven
by a focal event that shored up uncertainty on the same two questions. It
became public knowledge that certain militia captains would be excluded
from the future state coalition, which rendered a large number of promises
from the upper ranks worthless. Many hopeful street-level organizers and
runners were suddenly operating without “a roof”; they no longer had political
protection from other factions, de facto immunity from the fumbling legal
system, or a patron to advance them through the ranks. At the same time,

57 Interview conducted December 3, 2006 in Tbilisi. This particular respondent also captured the
mood at the time, which was that the humiliating defeat in Abkhazia needed to be revenged.
“We needed a new military, a strong military! We needed to build ourselves strong, to return
to Abkhazia! Otherwise, the Russians would just come again and again. [Russians] needed to
be shown.”
118 Jesse Driscoll

it also became public knowledge that “insider” militias were sufficiently well
organized to jail members of fringe militia groups (and if it came to some
last-ditch confrontation were probably capable of annihilating them). In other
words, a minimum winning coalition had emerged, and the oversupply of
militia members burdened this coalition. Given this transparent collapse of
“demand” for militia groups, the pool of ambiguously affiliated gunmen dried
up overnight.
What was necessary for the general decline of militia politics was an exoge-
nous shock – something completely unexpected that would force everyone
to update their beliefs. This event had the effect of clarifying which warlord
commanders were going to be “in” and “out” of the warlord coalition, which
allowed their soldiers to cut their losses and quietly flee the scene. Just as
important, the response of the “inside” militias had to be sufficiently coherent
and convincing to demonstrate that they were actually capable of winning
against any viable coalition of enemies. This sort of certainty would have been
impossible before the “violence market” had opened up and provided different
warlords with the ability to demonstrate that they could recruit and maintain
a militia at a relatively low cost. It is beyond the scope of any general theory to
predict something that locals could not predict themselves.
In Georgia, the event that shored up certainty about coalition membership
was a failed assassination attempt on the president, Eduard Shevardnadze,
on August 29, 1995. Most agree Dzhaba Ioseliani and the Russian-affiliated
Minister of Defense, Igor Giorgadze, orchestrated the attempt. Shevardnadze
survived the car bomb by pure luck. When the president made it a public
fight between himself and the coup-plotters – standing minutes after the car
bomb before a hastily assembled parliament session – everyone wanted to be
on his side. His survival changed everyone’s calculations about the future.
It became clear that his regime would persevere in the near term and the
public outcry associated with the assassination attempt provided a buoyant
month of popular legitimacy. Ioseliani was finished. Mkhedrioni members who
had integrated themselves into the Ministry of the Interior turned on Dzhaba
immediately, practically falling over one another to arrest him. Street-level
Mkhedrioni enforcers and Giorgadze’s loyalists in various subsections of the
Defense Ministry immediately left Tbilisi. Giorgi, a former member of the
National Guard who now works as an embassy driver, spoke for many: “We
were running the streets … we were kings of this city. And then, overnight, we
all just ended up in jail.”58
The assassination attempt also put to rest the idea that Shevardnadze’s
survival was dependent on networks in Moscow. At least some Russian actors –
likely well-placed political players given Giorgadze’s conspicuous decision to
flee to Moscow in a Russian military aircraft – had the means and wherewithal
to attempt to install a different puppet president. There was a strong sense

58 Interview conducted October 24, 2006 in Rustavi.

Warlord Coalitions and Militia Politics 119

that Georgia was under siege, but also a sense that, if Georgians rallied behind
Shevardnadze, he might be able to purge the state of the Russian vipers.59 The
contingency of everything described in the next chapter must be reemphasized,
however. If Shevardnadze had died, Ioseliani might still be running large parts
of Tbilisi. In the words of Alexander Rondeli: “If those had been East Germans
who designed the bomb and timer instead of Georgians, we’d be just like
Armenia now.”60
In Tajikistan, Rakhmonov organized the 1994 election to shore up his claim
to power. A defining feature of Tajikistan’s politics in the mid-1990s was the
uncertainty over the degree of foreign support that the government would
receive from Russia and Uzbekistan. To return briefly to the language of the
model, one might say that players were uncertain of the stability threshold s. No
one knew whether Rakhmonov and his Kulobi backers would really be allowed
to rule without sharing meaningful power with Uzbekistan’s traditional clients,
or if some sort of tacit Russian (or Uzbek) “security guarantee” would kick
in. The election broke down along predictable regional lines.61 Everyone
anticipated that Rakhmonov would have an advantage in the capital city and
its environs, as well as his traditional power base in the south. Seventy-nine
percent of the population of Khojand (31 percent of the population) and 96
percent of voting Badakhshonis (3 percent of the population) chose Abdumalik
Abdullojonov, the Prime Minister from a well-known Khojandi family. But
these opposition votes were swallowed under the demographic weight of
the newly created superdistrict of “Khatlon,” which was created by merging
three oblasts in the southern region – Hissor (5 percent of the population),
Kulob (12 percent of the population), and Kurgon-Tubbe (21 percent of
the population). The entirety of this newly created district of Khatlon was
delivered to Rakhmonov, with 99.5 percent of the votes counted for the sitting
president.62

59 Recall from the model: A very high stability threshold s increases the probability of any single
warlord being pivotal, making presidential promises to pay off warlords more credible than
they would be otherwise.
60 Interview conducted at the Georgian Foundation for Strategic and International Studies
(GFSIS) in Tbilisi, November 2, 2006.
61 Whatever one thinks of the validity of these numbers, the ability to conduct an election under
conditions of abject state failure is an impressive display of the administrative capacity that the
Soviet legacy bequeathed. Voter turnout was reported at 88.33 percent; the percentage of the
total voting age population who participated was 77.28 percent. Though Gorno-Badakhshon
was functionally independent at the time, it is important for the dynamics that followed to
to note that they did vote in this election. The election was also a referendum on the new
constitution, and 90 percent of voters chose to adopt it. See Grotz (2001), 20, Atkin (2002),
104, and IDEA (2011).
62 See the Current Digest of the Post-Soviet Press: 1994–1995, vol. 46, no. 45, pp. 20. Population
numbers are from the 1989 census, reported in Nourzhanov and Bleur (2013), 285. It should
be emphasized that the Khatlon region of Tajikistan was the site of horrible anti-civilian
tactics aimed at terrorizing civilian populations out of their homes, and many of the people
120 Jesse Driscoll

After the stunning results, Rakhmonov announced he would be reshuffling

his cabinet, elevating his core Kulobi supporters at the direct expense of the
Hissoris, Khojandis, and other traditional Uzbek clients within the Popular
Font. From the perspective of domestic power consolidation, it was a risky
and brilliantly brazen move, presenting the Russian government with a clear
take-it-or-leave-it offer. They could either explicitly endorse the presidency of
Rakhkmonov or declare the elections invalid and return the country to war.
Moscow opted for stability on Rakhmonov’s terms. Fifteen billion rubles were
sent to the Dushanbe regime, which paid government salaries for the first
time since independence. Tajikistan’s presidential election was hailed as proof
of “progress towards democracy.”63 All of this outraged the government of
Uzbekistan, and it was certainly not the end of challenges to Rakhmonov’s
rule, as we shall see in Chapter 5. But, in general, militia captains who wished
to remain in the capital city were forced to acknowledge that Rakhmonov was
there to stay, not a temporary placeholder before the Khojandis reclaimed the
presidency. Factional fighting shifted substantially to clashes outside of the city
limits of Dushanbe from this point forward.
Both regimes also laid the groundwork for the peaceful withdrawal of militia
forces by lowering the stakes of politics for losing militia captains and even
providing different safety nets for the paramilitary foot soldiers who were
shut out of the consolidation process. Both states passed blanket amnesty laws
for acts of wartime violence, absolving militia members of responsibility for
crimes during the transition period. Ongoing insurgencies in rural areas of both
states meant that militia groups could withdraw from the capital and carry
the fight to “the enemy” – integrating with the state army or border guards –
without anyone losing too much face. Ongoing border friction in Abkhazia
and the insurgency in the highlands of Tajikistan provided convenient places to
send unruly militias for both post-Soviet presidents throughout the 1990s. The
opportunities for war profiteering and narcotics smuggling made these sorts of
internal exile options relatively profitable for the militia captains. Emigration
from the state was another common safety valve, especially in Tajikistan, for
violent paramilitary fighters who wanted to start new lives. Relatively stable
employment opportunities emerged for these men as bodyguards or private
security for the new urban businessmen; emerging transnational smuggling
organizations that were operating with tacit regime approval provided another
exit strategy. Both regimes were careful to cultivate a reputation for allowing
former militia commanders to retreat from the political arena with dignity and
keep their wealth without the risk of future retaliation, so long as they kept out
of politics. It took time and demonstrations for warlords to believe that they
could simply walk away from the violence game, and sometimes these promises

conducting the election were the beneficiaries of these land-grabs. The election effectively ended
all conversation about postwar land redistribution in Tajikistan.
63 Atkin (1997b), 303.
Warlord Coalitions and Militia Politics 121

were revoked capriciously. But many doors were left open for unlucky warlords
to exit.
Merging militia forces into the rump bureaucracy was a violent, uncertain, and
competitive process. Militia commanders convinced their memberships – all of
whom wanted to be compensated for their sacrifices – that they had a shared
interest in eliminating imitators and claiming the lion’s share of state spoils for
themselves. New contenders could not be deterred, however, as there was an
overabundance of foot soldiers who saw militia membership as their best way
out of the subproletarian underclass. “War of attrition” dynamics helped to
create common knowledge about the balance of power between warlords. The
upsurge in violence was the result of calculated, violent competition. Warlords
were trying to gauge each other’s actual strength before agreeing on how to
divide state spoils, and militia recruits were trying to gauge warlords’ access to
state spoils before committing to militia membership. None of this had anything
to do with “anarchy” or security dilemmas, except as a permissive cause. It was
the fact that all actors could see the end of transitional chaos on the horizon
that made urban violence between militias necessary.
These militas became increasingly indistinguishable from patronage-based
political parties or lobbying groups. Wage competition between various militia
groups gradually sorted the violence market. The memberships of these militia
groups expanded, and then contracted, based on a few parameters: the overall
urban risk for potential new militia recruits, the attractiveness of “exit options”
from militia life, and expectations about whether a captain would stay a part of
the ruling coalition. Militia members watched anxiously as power consolidated,
knowing that they were bit players in the drama of consolidation politics. Focal
events made it clear that in the halls above the streets, a stable coalition had
emerged. But nothing in this account implies that “the state” suddenly became
strong enough to take on militias and restore order. Rather, in Tajikistan and
Georgia “the state” was itself formed from a subset of militias and warlords
who had their armed checkpoints, bazaars, shadow-economy enclaves, and
local tax collection legalized by the decrees of a government they installed.
Many of the ministers, deputies, and uniformed members of the Georgian and
Tajik security services were affiliated with militias at the time when the streets
went quiet.
Thus far, the narrative has treated civilian elites – including the president –
as interchangeable and disposable. Multiple presidents had already been
removed in coups. The initial bargain that brought warlords into the state was
predicated on the promise that the president cede warlords positions in the
government. Though formal institutions were vital to serve as a go-between
for aid, legitimacy, and policy concessions, violence entrepreneurs could always
threaten to smash these institutions if they were not given their fair due. As they
began their tenures as heads of state, neither Rakhmonov nor Shevardnadze
had any real control over the patchwork of paramilitaries that had penetrated
the state apparatus. At some points, both rulers were literally prisoners in
122 Jesse Driscoll

the presidential palace. Regional experts bluntly characterized civil–military

relations as “feudal.”64 The same bonds of trust that were used to survive the
war and the “Time of Troubles” could be used to organize a coup against the
president, and this fact was well understood by all parties behind the throne.
Even in the late 1990s the presidents were sometimes described by militia
members as “marionettes.”65 How these “puppet presidents” cut their strings
is the puzzle that motivates the next chapter.

64 Jones (1996), 46, Fairbanks (2002), 136–150.

65 I heard this term used to describe Shevardnadze in my very first life history interview with
a former Mkhedrioni member, conducted February 27, 2006 in Tbilisi. In one of my last
interviews in Tajikistan (conducted August 16, 2007 in Dushanbe), a Kulobi member of the
Popular Front dismissed Rakhmonov in the early 1990s as a “kukla” (puppet).
5

Coup-Proofing

Georgia and Tajikistan were essentially ruled by criminalized military juntas

in the early 1990s. Presidents were installed to act as figureheads. The same
insivible bonds of trust that were used to recruit soldiers and coordinate
military actions during the war could be – and had been – used to organize
coups in both states. Because this fact was well understood by the president and
the warlords behind the throne, the president implemented policies to satisfy
this critical constituency. Ministry appointments were handed out to cronies,
closed-bid privatization schemes were implemented to benefit mafia captains,
and arrest warrants targeted businessmen who could not acquire sufficient
political protection. At first, the presidents were puppets, with powerful
armed groups and underworld financiers holding the strings. A long-time
associate of Eduard Shevardnadze, who served in the government during this
period, recalled the daily ritual of humiliation and extortion in the following
terms:

“Dzhaba [Ioseliani] or one of his lieutenants walked in every day with a bunch of papers
for him to sign – usually documents of ownership or deeds and titles to houses and
apartment flats. … There were arrest warrants [for political enemies] as well. They didn’t
walk in with guns, but … there were guns in the building. He had to sign them. He had
to sign them all.”1

Yet by the late-1990s, both presidents had successfully placed themselves

securely in the center of the web of relationships holding their respective states
together, and it was difficult for anyone to imagine either state functioning

1 Interview conducted in Tbilisi November 29, 2006. For excellent descriptive accounts of this
period, see Gretsky (1995); Aves (1996); Brown (1998); O’Ballance (1997); Demetriou (2002).

123
124 Jesse Driscoll

without these men at the helm.2 The focus of this chapter is on how Eduard
Shevardnadze and Emomalii Rakhmanov managed to ascend to the top of their
respective political hierarchies.
The analytic narrative presented in this chapter describes how the two
presidents, in their weakly institutionalized postwar settings, contended with
multiple potential challengers simultaneously. If the logic of the n-player model
is correct, there should be evidence of the warlord coalition changing over time
to accommodate shifting domestic and international pressures. The figurehead
does not change: In postwar settings, individual leaders become focal points,
bottlenecks in the “two-level games” that are so critical to international
diplomacy.3 In the context of the politics described in the previous chapter,
this meant managing one public persona for the benefit of foreign powers,
important transnational donors, and trilingual elites to assure v∗ , and a separate
persona for the shadow-elite who can make good on private, overlapping, and
sometimes contradictory promises while the pie is divided.
The domestic institutions that might in times of peace be used as commit-
ment devices to hold coalition memberships steady were captured by militias
and divided as conflict spoils. There was no one to punish presidents who
reneged on promises. Proposal power in the coalition formation process created
an oligopoly centered on the president, who controlled the timing of cabinet
reshuffling and could thus choose a moment when a warlord was politically
alienated or ascendant to declare it was time to reshuffle a ministry. This
chapter emphasizes that what “divide-and-rule” actually looked like is long
periods of presidential inaction: waiting for warlords to alienate or embarrass
themselves, then selectively paring them from the coalition. The equilibria that
emerged in Georgia and Tajikistan were quite different, and most of the chapter
is devoted to describing them. Two parameters – the reservation value (r)
available to nonconsolidated warlords and total available amount of wealth
that the president has to distribute (v∗ ) – explain the shift from a situation
where the president was the puppet of warlord interests to one where he held
the strings.
Though most of the chapter focuses on the leadership styles and strategies of
the two presidents, it concludes by returning to the warlord perspective: Were
warlords actually interchangeable as coalition members or armed agents of the
state? Qualitative data from Georgia and an analysis of warlord biographical
data from Tajikistan suggest that, broadly speaking, they were. This chapter
also presents the summary results of systematic statistical analysis to reveal
which observable characteristics of Tajik warlords – home region, which side
they fought on, how many soldiers were under their command – predicted

2 See O’Ballance (1997) and King (2001b). Similarly, by 1999 Emomalii Rakhmonov started to
relentlessly centralize as much power as possible, finally turning on his own inner circle starting
in about 2002. See ICG (2004).
3 Putnam (1988).
Coup-Proofing 125

political fortune in the consolidation phase. In my interpretation, these data

reinforce the appropriateness of the lottery metaphor that anchors Chapter 2.
At the time of decision, warlords could not predict who would be politically
rehabilitated as a nationalist war hero, who would fade quietly, and who would
be violently liquidated. Evidence of warlord miscalculation – warlords joining
the president’s coalition and later regretting their choice – is consistent with this
book’s theory. Who would win and who would lose as consolidation unfolded
was simply not predictable.

the model redux: how did “puppet presidents”

cut their strings?
If there were difference in coup-proofing strategies between the two states,
the difference cannot be the result of inherited Soviet institutions. Both
Georgia and Tajikistan states shared a Soviet institutional legacy. Kotkin (2002)
cynically observes that it is “clear from today’s Russian-speaking populations
in most of the newly independent states, as well as the self-declared additional
Trashcanistans within them: the various post-Soviet nations emerged deeply
Soviet.” Twenty years after independence, institutions and behavioral patterns
in post-Soviet states share important similarities. One similarity is very strong
executives. As summarized in the first chapter of this book, post-Soviet leaders
inherited populations with restricted choice sets and all the national symbols of
state authority, including a time-tested party system to manage the aspirations
of talented elites, uniforms and statue-lined streets for parades, museums and
archives, a school system for the education of children, a national television
network, and much more.4 Across most of post-Soviet Eurasia, self-reinforcing

4 A centralized radio and television network, broadcasting a single narrative from the titular
capitals, is often underappreciated in the laundry list of compliance-generating technologies.
Anderson (1991) argues that technologies that disseminate information at low cost – such
as radio, newspapers, and television – allow modern states to manipulate the social sense
of time. The mechanism is sociocognitive. A mass reception of the same narratives in the
same language at the same time – watching the same television serials as they air, hearing the
president speaking on the radio, or downloading the same “Internet memes” – is a modern
national ritual: An “extraordinary mass ceremony” in which “each communicant is well aware
that the ceremony he performs is being replicated simultaneously by thousands (or millions)
of others” (35). In social science jargon, one could say that public and synchronized mass
communication technologies lower the cost for political leaders to create “common knowledge”
(images and narratives that every citizen knows that every other citizen knows, as summarized
in Chwe [1998]). In previous epochs, this task had been the sole domain of religious authority.
Warren (2015) and Herman and Chomsky (1987) provide theory and empirics on how these
technologies produce docility among state populations. I recall being in the rural village of
Kalikhum when the lights were turned on for the first time in months, in concordance with
the springtime Navruz holiday. The national television channel played songs of peace, and
Rakhmonov’s face was plastered on the screen. I knew that if it were not for that man, the
lights would still be off, and I knew that my host family understood this in the same way.
126 Jesse Driscoll

processes of single-party hegemonic stability were unleashed once the chaos

of the transition subsided and the electricity was turned back on. Rural votes
could be traded for access to scarce public goods. Once foreign relations
were secured and neighbors stopped backing coup-plotters, all that was left
was to neutralize the risk of insurgency from the internal hinterlands. This
task was accomplished with ritual elections, occasional violence, and constant
surveillance. Potemkin institution-building in the capitals was paired with a
return to Soviet vote-farming in rural areas. Strong presidents, in this account,
emerged via a path-dependent logic to take the place of Party Secretaries.
Emphasizing the constraints of inherited culture and path dependence comes
at the expense of human agency. The model that anchors this book, by contrast,
has no functioning institutions at all. Institutions are assumed to be captured
by violence specialists. Mishandled distributional politics can escalate to the
violent overthrow of the regime in power, as Georgian and Tajik history
show. Inherited institutions did not prevent the removal of Gamsakhurdia or
Iskanderov. The “coup-proofing” strategies of Shevardnadze and Rakhmonov
were necessary to avoid a similar fate.
As a starting point for comparing these strategies, the previous chapter
revealed an important similarity: both men paid off warlords in order to stay
in power. To capture the vulnerability of presidents in this period, violence
entrepreneurs are given the last move of the game in the model presented in
Chapter 2. In essence, warlords are given a collective veto over distributional
politics. If the president fails to democratize, provide public goods, punish
corruption, or allow for a flourishing civil society, the cost to him is losing
potential support from civil society or donors.5 If, on the other hand, the
president fails to pay off enough warlords, there will be a violent coup and
he will be replaced. The militia-permeated institutions described in Chapter 4
are evidence of how these payoffs manifested.
Part of the explanation for variation in coup-proofing strategies is the
different geopolitical situations that the two states faced. I argue that
international system-level constraints set the structural parameters of the game,
and in particular the stability threshold s, which changed the way the game
was played in Georgia and Tajikistan. Russian policy toward the newly
independent states was haphazard and sometimes improvised, to be sure. But,
in broad brushstrokes, Russia pursued different strategies in response to state
disintegration in the South Caucasus and Central Asia. In Georgia, Russian
foreign policy had dual justifications: keeping leverage over the new state
by threatening to support local ethnocracies and protecting innocent civilians
from ethnic cleansing.6 Thus, indigenous Georgian state-building efforts were

5 This could be parameterized as a lower v∗ .

6 The well-known secessionist regions of Abkhazia, South Ossetia, and Adjara fit into neat
spaces on preexisting Soviet maps, and the lesser known Armenian and Azeri enclaves of
Akhalkalaki, Dmanisi, Ninochminda, Marneuli, and Bolnisi all managed to trade votes for
Coup-Proofing 127

undermined. In the language of the model: the stability threshold was higher
in Georgia, where a regime was consolidating in explicit opposition to, and in
defiance of, Russian interests.
However, in Tajikistan, where a strong bulwark was needed against the
radicalism in Afghanistan, state-building efforts were simply not allowed to
fail. The potential for irregular infantry units and dangerous ideas to filter
north across the Panj River to infect the other states of Central Asia was
terrifying. Russia’s foreign policy was chaotic and disorganized in the early
1990s, but Boris Yeltsin’s 1993 statement that the Tajik–Afghan border is
“in effect, Russia’s” left no doubt that Russian leaders identified a vital
interest in containing the consequences of Tajikistan’s state failure. If Iranian
or Afghan-backed guerillas had been able to use the impenetrable Pamiri
Mountains as a base to spread war into Kyrgyzstan and Uzbekistan, the
Uzbek government would likely have redrawn the map of Central Asia by
incorporating the Tajik region of Khojand – or perhaps all of the Ferghana
Valley – as a buffer state. The expansion of Uzbekistan would have had
unpredictable effects on Turkmenistan and Kazakhstan, with their abundant
oil and gas reserves. These were all terrible outcomes that would have come
at the expense of the regional security arrangements Russia was frantically
sculpting.7 To return to the language of the model: the stability threshold was
lower in Tajikistan, where a regime was consolidating in explicit support of
Russian interests.
Differences in the consolidation projects of Shevardnadze and Rakhmonov
are consistent with what the model would predict if one country had a
high stability threshold and the other had a low threshold (see Figure 5.1).
Revolutionary nationalism in post-independence Georgia had a distinctly
militant and anti-Russian flavor. Shevardnadze’s foot dragging on joining
the Commonwealth of Independent States (CIS) and quick expansion of ties
to Western European states was seen as provocative in Moscow; Georgian
paramilitary attacks on ethnic minorities could have led to destabilizing refugee
crises spilling across borders. Many Russians made no secret of their desire to
see Shevardnadze removed. As a result, Shevardnadze kept his friends close and
his enemies closer. He incorporated many Georgian warlords quickly, such that
his Ministry of the Interior became a virtual state-within-a-state.
In Tajikistan, by contrast, Russia lent its armed forces and intelligence
to support the Rakhmonov regime against domestic challengers. Alexander
Shishlyannikov, a representative of the armed services of Uzbekistan, and

de facto independence throughout the 1990s. The Mingrelians in western Georgia were
also slow to forgive the elites in Tbilisi for the overthrow of Gamsakhurdia and the brutal
counterinsurgency campaign against the Zviadists.
7 This is my own distillation of events, but for associated insights see Gretsky (1995),
Nassim Jawad and Tadjbakhsh (1995), Atkin (1997a), Centlivres-Demont (1998), Dudoignon
(1998), Zviagelskaya (1998), and Akbarzadeh (2001).
128 Jesse Driscoll
Typical Warlord Reservation Value (r )

High Low

v* increases via aid and investment; r “Potemkin Democracy”: Nearly

declines with broad consensus around All Militias Are Incorporated.
Shevardnadze as inevitable leader and a Persistent Secessionism in
good patron to supporters in the capital. Abkhazia, Adjara, South Ossetia.

Georgia (late -- Chapter 5)

High State Failure/Puppet President:
Violence in Capital over Coalition
Membership, Anarchic Periphery
Stability Threshold (s )

(“Time of Troubles”)

Georgia and Tajikistan (early -- “Neo-Feudalism”: Rule by a Very

Chapters 3 and 4) Narrow Coalition, Supported by
Insurgency until 1997. v* increases via Russian Aid. Slow Incorporation
Low bilateral strategic aid. r drops when UTO of Rural Holdouts. 9/11 brings
lose safe havens across Afghan border, U.S. military power into region.
forcing warlords to settle and join the
state even without security guarantees. Tajikistan (late -- Chapter 5)

Basic logic of parameter shift, as v *

increases and r decreases

figure 5.1. Divergent Consolidation Pathways: How do the parameters shift?

Saidamir Zukhurov, a representative of the KGB, were both members of

Rakhmonov’s first cabinet. There was a shared consensus among CIS states
to keep the Afghan civil war in Afghanistan – and to keep American-funded
Islamic fundamentalists or Iranian clerics from coming to power. As a result,
I model the stability threshold as comparatively low, which is only to say that
Russian support gave the president of this fragile new state more latitude to pick
and choose among various local warlord allies than his counterpart in Georgia.
Relatively low lottery odds also translated into high-stakes in Tajikistan, which
manifested as fierce friction between various members of the insider coalition.
The president was nearly removed from power by warlords who could see that
as the peace process unfolded and additional warlords were incorporated, their
bargaining position would be eroded.
But what did it mean for their position to be eroded? What were the warlords
really after? Did they simply wish to skim cash off foreign charity money
coming in (v∗ )? Perhaps, but they also sought foreign recognition of their
coalition to keep a stream of rents coming from both foreign aid (v∗ ), on top of
all of the rents that came from future domestic sources (v).8 In the medium, and

8 This issue is somewhat confused in many political science treatments, with much piled into the
term “spoils.” See, for instance, Collier and Hoeffler (2004), particularly their interpretation of
the “primary commodity exports” variable. Markowitz (2013), in an argument that resonates
with the one in this book, shows that the ability of weak states in Central Asia to survive depends
on local elites’ ability to divide up “immobile capital” among themselves.
Coup-Proofing 129

certainly in the long run, many commanders came to realize that domestic rents
(v) would greatly outweigh any short-term incoming foreign cash (v∗ ). There
are bribes, and then there are longer-term ways to avoid taxes and control
regulation, permits, and other forms of rents.9 The distributional politics that
allow for v and v∗ to be distributed are messy and unpredictable, which is why,
to return briefly to the jargon of game theory, any equilibrium that “sticks”
needs to be subgame perfect.
The president is a focal symbol of order and stability in a state recovering
from failure. A confident and secure president sends a signal to investors and
foreign donors that a country is open for business, and v∗ follows. Warlords
who are embedded in the state infrastructure – siphoning off wealth to pay their
private armies, which they have not disarmed – have an interest in participating
in the theater of civilian governance. The logic of the model is that gangsters
realize that they can do better laundering money from naïve foreign donors
and investors through a corrupt state apparatus than they can fighting among
themselves to rule the state. Recall also that the partial incorporation and full
incorporation equilibria both describe a peace that is self-enforcing without the
need for an external guarantor. Order may be financed by outsiders, but it is
monitored and enforced by locals.
But how does political power become centralized in the president’s hands in
this noninstitutionalized bargaining situation? The model in Chapter 2 answers
this question. Warlords face a collective action problem once the president is
installed. If all the security ministries act together, they can remove a president
from power without bloodshed, but a warlord acting alone against a president
is likely to fail. In the model, a warlord usually cannot carry out a coup alone –
defection of multiple warlords from the president’s coalition is almost always
necessary to bring down the state. Tactical alliances between warlord factions
do not easily congeal into a stable opposition coalition. In the absence of
an institutionalized bargaining environment, and with no rules to structure
coalition-building, no authority to enforce contracts between warlords, and
the constant possibility that intrigue or accidental gunfire could eliminate a key
coalition member, blocking coalitions can dissolve overnight.10 The president
is well positioned to take advantage of splits in warlord coalitions as they

9 Some of the criminals who gained power probably did not actually think very hard about the
long run; some of these actors were actually driven by short-term payouts. It is likely the case
that least two of the Tajik warlords analyzed in this dataset “exited the data” because they
became addicted to heroin, stopped showing up for their jobs, and simply disappeared into
the gray maze of street stories.
10 In the absence of formal institutions capable of structuring relationships between armed actors,
warlords needed to use informal institutions with high transaction costs. For an excellent
overview of the various sorts of efficiency losses associated with the use of informal institutions,
see Dixit (2004), especially chapters 3 and 4. A partial exception to the “no counter-coalitions”
rule was the United Tajik Opposition, but as the case study that follows demonstrates, it was,
in the end, highly vulnerable to divide-and-rule strategies from the center.
130 Jesse Driscoll

emerge, using divide-and-rule strategies to become an indispensable member

of the winning coalition. And though it is extra-model, the president’s role
as the primary spokesperson for the country’s foreign policy means that he
has a central role in negotiations with international companies, banks, and
donor organizations. The fact that his signature is required on privatization
documents, and that he is granted the symbolic power to appoint ministers,
deputy ministers, and fill other state positions at a timing of his choosing, puts
him in a position to distribute favor selectively and shape the overall bargaining
space.
The model shows that, once in power, the president should always be
able to buy off a coup. Warlords should understand that once a president is
installed, they will lose leverage over him, even if they keep their guns. This
is a central problem of civil war settlement in this framework. All players
realize that different order-providing coalitions are sustainable equilibria.
Donor-funded institution-building projects are important sources of liquidity,
but v∗ is divisible and fungible. The pie can be cut to accommodate myriad
possible distributions.11 The decision to install a president in the first place is a
risky gamble.
Warlords weigh their odds of success in this unpredictable coalition
formation lottery against their reservation value r for staying outside the state.
As warlords’ reservation values drop, presidents in consolidating states will be
able to offer deals to incorporate warlords selectively into the state apparatus.
The warlords who install a president at the beginning may be trusted allies, but
the model assumes that they are functionally interchangeable with late-joiners.
Though the strategy profile in Appendix B emphasizes that the president should
pay exactly s warlords exactly what they would get in a coup, in the real world a
very good strategy for staying in power is for the president to assemble multiple
warlord coalitions, with overlapping memberships, that could each pass the
stability threshold, and to distribute the spoils of statehood widely. As outside
options fall, he can assemble a cheaper coalition and gradually keep more of
the distribution for himself.12
As the warlords slowly capitulated to presidential authority, the president
began to create a coalition of armed men with a shared interest in keeping
him in power. Most of these men were still not loyal to the president in
any meaningful way, and many of them would probably have liked to see
him replaced. But there was sufficient uncertainty about who exactly would
replace the president, and what the warlords’ position would be in the new
government, that most of these men were willing to opt for the relative
safety of the current regime. Everyone understood that, unless it succeeded
dramatically, any challenge to the president’s coalition meant becoming a target
for dismemberment by other militia leaders whose loyalty had been proven. Bit

11 See Proposition 4 and the associated proof in Appendix B.

12 Ferejohn (1986) presents a lucid and elegant exposition of the rationale for this assertion.
Coup-Proofing 131

by bit, the old militia structures were replaced by sprawling and bloated power
ministries, staffed with a mix of untested newcomers and co-opted militia
members, opaque patronage networks, overlapping spheres of authority, and
an earned reputation for corruption and unaccountability.
With wartime militia networks diluted in the new ministry structures,
friendships and obligation chains that led upwards to the president were critical
for long-term survival. Rather than a coup threat, new police forces became a
constituency for the status quo. And with foreign military support, economic
aid, and diplomatic recognition, this new warlord-backed regime became
impossible to dislodge. Some warlords outside the coalition initially found they
could thrive by operating independently at the periphery of state control, but
eventually rural holdouts were squeezed into submission. Presidents sometimes
used these newly incorporated warlords against the old, taking advantage of
splits in the warlord coalition. Ultimately, both men insulated themselves from
warlord challenges and emerged dominant.

summary of strategies
Eduard Shevardnadze ascended to power in Georgia at the behest of a
paramilitary junta, but quickly realized that gaining leverage over the leaders at
the top of the power pyramids would be insufficient. There were actually dozens
of self-financing militia factions within the Mekhedrioni and the National
Guards, each backing a different leader (or, as the side-switching analysis
in the previous chapter discussed, dividing their loyalty among a number of
potential leaders). Shevardnadze was also keenly aware that as he deployed
his unique asset – his celebrity status in Western capitals, with the promise
of delivering huge bilateral and multilateral aid flows – whatever successes
he achieved would provoke Russian hard-liners in Moscow. Georgian militias
invited Shevardnadze to lead because of his unique potential to enrich the state
(high v∗ ). Owing to external pressures, he needed the support of virtually all n
of the militias to rule (implying a high stability threshold s).
Shevardnadze’s strategy was to incorporate as many warlords as quickly as
possible. He used the hierarchical structure of the ministries to break down
“umbrella militia groups,” which claimed huge memberships, into smaller
constituent parts by promoting sub-lieutenants and lieutenants independently
of each other. They were lured into these positions by the possibility of amassing
personal wealth. Most were given their own semi-independent fiefdoms, housed
within the labyrinthine Ministry of the Interior. By explicitly committing to
policies such as rapid privatization, a restricted state role in economic policy,
an uncompromising position toward the secessionist territories of Abkhazia
and South Ossetia, and, most of all, non-interference into the affairs of his
own Ministry of Interior, he lowered the political stakes between potential
coup-plotters. The situation was made even more attractive once it became
clear that, so long as he was at the head of the state, stable order and “Potemkin
132 Jesse Driscoll

democracy” would allow his state to extract huge rents as a regional pivot
player between Russia and the West.13 The equilibrium that emerged gave the
president the upper hand in bargaining with individual warlords, but little room
to maneuver or steer domestic policy. This situation endured until the Rose
Revolution.
A different kind of strategy, pursued effectively and ruthlessly by Emomalii
Rakhmonov, was to raise the stakes of politics, heaping favors upon a
pivotal warlord constituency to the exclusion of others. This strategy emerged
naturally in Tajikistan once it became obvious that Russia would support
any coalition capable of guaranteeing the security of the southern border
with Afghanistan, and able to check the hegemonic aspirations of Uzbekistan
and Iran. In the language of the model, Russia’s support allowed the game
to be played with an artificially low stability threshold s. Other parameters
differed from the Georgian situation as well: the relatively low gains to stability
(v∗ > v, but only just) and the high reservation wage r for noncooperative
warlords (stemming from the Afghan drug trade, the mountains, and the
support from across the Uzbek and Afghan borders) slowed the speed of state
consolidation.
Full incorporation emerged because, by 1997 (and even more so by 2003),
two global parameters had shifted in ways that eventually changed warlords’
strategic calculations. The first was a gradual rapprochement between Russia
and Uzbekistan over the composition of the Tajik government, which lowered
the reservation value r for warlords previously able to hide across international
borders, or hold out hope of foreign support to keep the fight alive. As r
dropped, and more warlords opted into the state, the inclusion of new warlords
into the ruling coalition gave the president the ability to choose his allies, setting
off violent opposition by marginal members of Rakhmonov’s original coalition.
A condominium between Uzbeks and Russians emerged only after the warlords
who constituted the Tajik government demonstrated that they could fulfill
certain basic functions, especially guaranteeing traffic through the overland
routes necessary for the Afghan heroin trade and marginalizing, or silencing,
radical Islamic voices. But it was the U.S. invasion of Afghanistan in 2001 that
really cemented Rakhmonov’s hold on power. Security assistance increased v∗
available to the Tajik government. The reservation value r for most warlords
was high when they were operating in the impenetrable Pamiri mountain range
against ground forces, but modern air power changed the calculations. A shy
man, whose only political experience before ascending to the position of head
of state was working as the head of a collective farm in the agricultural region
of Kulob, “Rakhmon” proved himself an adept political survivor. Today he
heads of the only relevant political dynasty in Tajikistan.

13 I borrow this clever phrasing from King (2001b).

Coup-Proofing 133

eduard shevardnadze: the indispensable georgian

It is no secret why Eduard Shevardnadze was selected as the figurehead
to pull the Georgian state back from the abyss: in the West, the recently
retired Soviet Foreign Minister was probably the second most popular Soviet
politician (after Mikhail Gorbachev) and the second most famous Georgian in
history (after Josef Stalin). Shevardnadze was simply far ahead of the typical
post-independence learning curve when it came to matters of international
diplomacy, geopolitics, and the practical necessities of using international
norms and rules to a new state’s advantage. His personal friendships with
foreign leaders and dignitaries – especially Bill Clinton and James Baker –
were critical for getting international recognition of Georgia with territorial
boundaries defined to include the territories of Abkhazia and South Ossetia.
He was the first president from the Caucasus or Central Asia invited to visit
the White House, and on his visit, Clinton promised $70 million in bilateral
aid and signed policy directives for the CIA to train and equip a special
unit of bodyguards to protect the new Georgian president.14 More broadly,
Shevardnadze’s unique set of diplomatic contacts gave him the ability to seek
multilateral and bilateral aid from both Russia and the West simultaneously.
He proved successful in this task beyond anyone’s expectations. On a per
capita basis, Georgia received more democracy promotion assistance from the
United States than the rest of the former Soviet Union combined. From Western
European states, Georgia received approximately $172 million per year during
the Shevardnadze era.15
Shevardnadze understood that the West wanted a geopolitical foothold in
the Caucasus and that he was considered a “known quantity” in a confusing
and chaotic region. He successfully branded Georgia as a potential energy
corridor, securing contracts with multinational oil companies to build pipelines
connecting Caspian oil to Western markets by way of a route that bypasses both
Russia and Iran. In the imaginations of certain Western defense intellectuals
this transformed Georgia from a forgettable ethnic oddity on the far side of the
Black Sea to the geopolitical pivot of the Caucasus. Shevardnadze’s status as a
celebrity diplomat, and his personal involvement in all of these negotiations,
made him seem larger than life. If he was not literally irreplaceable, he
certainly was a unique national asset for a state confronted with a bewildering

14 See O’Ballance (1997), 138–139 and Areshidze (2007), 36.

15 Between 1995 and 2000, more than $700 million of bilateral aid arrived from the United
States. In 2000 USAID’s budget amounted to $200 per Georgian citizen, as compared to
merely $1.25 per Russian. In addition to being the fourth-largest per capita recipient of
USAID the 2002–2003, Georgia also received some 400 million Euro in the decade before
the Rose Revolution from the coffers of the European Union, with additional contributions
from many individual member states. Tudoroiu (2007), 323. See also Stephen Jones (2006),
41–42.
134 Jesse Driscoll

set of challenges.16 In the words of Georgian political scientist Ghia Nodia,

Shevardnadze’s unique skills and autonomy in matters of foreign policy were
the starting point for his domestic power grab: “It was on the image of this
indispensability that he started building his internal legitimacy – and was later
able to translate this into real power.”17
At the same time he was cultivating relationships with the United States and
Europe, Shevardnadze had a great deal of experience and many relationships in
Moscow. He made the most of his ties to the Russian foreign policy apparatus
to extract critical concessions and aid from Russia. This meant devoting a great
deal of personal energy and political capital to trying to convince Russians that
a Georgian government funded by the West would not be a threat to Russia’s
traditional sphere of influence. He was substantially successful in this task –
certainly more successful than any credible alternative would have been. He
used his contacts in the Russian military to make sure that the terms of the
Tashkent Agreement (which divided the spoils of the Soviet military machine)
were actually implemented, thus transferring tanks and heavy weapons to
the Georgian state despite its “less than legitimate credentials.”18 Later he
apparently convinced Yeltsin to authorize the use of the Black Sea Fleet to help
rout the Zviadist insurgency.19 At Moscow’s insistence, he also appointed a
Russian protege (Vardiko Nadibaidze) as the Minister of Defense and a Russian
general (Igor Giorgadze) as the Head of the Internal Security Services. Placing
Russian agents at the head of the army and internal police meant that Moscow
could closely monitor Shevardnadze’s actions, exert an independent role in
military and foreign policy, and have an obvious route to a military coup if
it so desired.
Shevardnadze gambled that if he allowed a system to evolve with multiple
spheres of influence, he would be the most capable of navigating the resulting
maze of coalition politics. He became famous in Georgia in the early 1990s
for developing overlapping spheres of authority, a tactic that undermined

16 One of the often overlooked aspects of the Shevardnadze legacy was his quiet promotion of
the idea among influential Georgians that this kind of aid and investment – the sort that does
not come with Russian strings attached – should always be preferred to Russian bilateral aid
or investment from Russian banks. Interviews with Vladimer Papava, Conducted in Tbilisi,
February 7, 2006 and November 4, 2006. See also Papava and Chocheli (2003), generally.
17 See Nodia (2002), 428.
18 See Trenin (1995), 137. On the same page the author editorializes further: “[T]he Russians
gave the Georgians more than enough weapons in 1992 to impose a military solution in all
internal disputes. It is certainly not the fault of the Russian army that Tbilisi made such poor
use of them.”
19 It remains unclear whether these sailors were acting under orders from Moscow. A credible
alternative tale of how the Black Sea Fleet was drawn into the conflict was that Shevardnadze
took advantage of the fact that Moscow and Kiev both claimed the fleet as their own. As the
tale is often told, Shevardnadze traveled in person to bribe the Ukrainian Admiral Baltin, the
naval commander at the time, to move his artillery into position, reportedly with a briefcase
filled with cash and two cases of Armenian cognac.
Coup-Proofing 135

his credentials as a reformer, but also made his friendship and cooperation
completely indispensable for anyone who wanted to actually get anything done.
He was a relentless, tireless worker, and most of his work was done in meetings
where he would sit down to build relationships, face to face, with individuals
from every possible walk of society. He was notorious for postponing difficult
decisions and shrewdly “maneuvering between different political forces … to
broaden his power base and to neutralize major warlords.”20
Although he eventually centralized political authority in a hegemonic party
structure, the Citizens’ Union of Georgia (CUG), for his first two years in
office he exercised power in a personalist and deinstitutionalized fashion,
preferring to “balance between opposing political factions.”21 This allowed
the militia rank and file to convince themselves that they could accept favors
from Shevardnadze without betraying their militia brethren or blood oaths. So
long as there was a common external enemy in the form of Russia, it provided
an impetus for warlords to work together. From the perspective of individual
psychology, there was nothing particularly difficult about being anti-Russian in
a political environment that was so virulently nationalist and anti-communist.22
In mid-September 1993, at the height of the Abkhaz crisis but before the final
loss of Sukhumi, Shevardnadze appointed himself the Minister of the Interior,
and asked the Georgian State Council to dissolve itself and approve a state
of emergency. He was essentially proposing that the coup-plotters abrogate
their roles, granting him what amounted to sweeping dictatorial powers. When
this proposal was met with resistance from Ioseliani and other Mkhedrioni
loyalists, Shevardnadze – in what was apparently a brilliantly improvised piece
of political theater – publicly resigned and stormed out of the chamber. He
resumed his position only after massive street demonstrations and a revolt of
the professional state bureaucracy made it clear that no one wanted to see
him go. He could, in fact, bring the whole state down with him if opponents
rejected his ultimatums. But as the model suggests, for an equilibrium to be
possible, most of the wealth must ultimately fall into the pockets of the warlords
who control the raw tools of violence and those most symbolic institutions of
statehood, the security ministries. Shevardnadze was the leader of a state with
no army, other than the National Guard and the Mkhedrioni. The ongoing

20 See Jones (1997), 527 and Nodia (2002), 419.

21 Wheatley (2005), 85.
22 A few interview respondents went further, stating that once Shevardnadze came onto the
scene, they realized that they had the possibility of provoking a serious war with Russia then
waiting for the United States and Europe to come to their aid as a state-building strategy. It
was a familiar kind of argument: state-building requires reducing a lot of the complex issues
– including the transition from communism to capitalism, emotional decolonization for a
multiethnic and multisecretarian society, and more – down to “all of us vs. them.” The problem
(though I did not interrupt the interviews to point this out) was that the Russian language was –
and continues to be – the way that many non-Russian non-Georgians communicate with each
other. For the 30 percent of Georgia’s citizens who did not speak Georgian at home, persistent
war between Georgians and Russians leaves them trapped between worlds.
136 Jesse Driscoll

battles with Zviadist insurgents and Russian-backed secessionists demonstrated

that the Georgian state required both. Having secured foreign help to keep
the state afloat, yoking the military strength of these private armies without
bowing to the demands of the warlords was the core governance problem for
Shevardnadze.
One of the first effects of Shevardnadze’s celebrity status and foreign
policy prowess was that events in the South Caucasus played out in a way
that was a bit more “public” than they would have been otherwise, in
terms of international news reporting. When Shevardnadze declined Russia’s
offer to have Georgia join the CIS – claiming to be executing the will
of the consensus governing body – the challenge was understood violently.
Clandestine arrangements introduced paramilitary armies from the North
Caucasus to fight an undeclared war against the Georgian state, infiltrating
through the secessionist regions.23 Russian troops enforced a cease-fire in South
Ossetia, provided support for the Abkhaz against the Georgian militias, and
threatened to do the same in Adjara and the Armenian enclaves. Russia’s
threat to keep the small mountainous country fragmented indefinitely was very
credible.
As diplomatic efforts to mend fences with Moscow occurred against the
backdrop of Russian military and diplomatic support to Abkhaz and South
Ossetian secessionists, this two-level gamesmanship required all the skill of a
master politician. Shevardnadze needed to convince the Russians that he was
extracting costly concessions from Georgian nationalists, while simultaneously
convincing those same skeptical Georgians that he had not “gone native”
during his years inside the Kremlin. Against the hopes of many in the United
States and the demands of Georgian nationalist warlords, Shevardnadze finally
opted to join the CIS. Though none of these positions were ratified by
Parliament, Shevardnadze acceded to the presence of Russian peacekeepers
along the unrecognized border with Abkhazia, the deployment of Russian
border guards along the Turkish border (encircling Georgia), and negotiating –
but, in the end, not signing – the Treaty on Russian Military Bases on the
Territory of the Republic of Georgia, which would have granted Russia’s
military basing rights in Georgia for 25 years.24

23 See Derluguian (2005), 260–273.

24 At the end of many days of negotiating the minutiae of the treaty session, Shevardnadze
apparently agreed, shook everyone’s hand, and left the room without actually affixing his
signature to the treaty. The Russian delegation left with a verbal agreement and publicized
the treaty as a foreign policy success. The matter was thought settled, and the news cycles
marched on. But it turns out that with no signature the treaty was technically nonbinding
under international law. Apparently no one from Moscow’s delegation felt they had the gravitas
to “make him” sign. This technicality allowed Shevardnadze to renegotiate the content of the
treaty at an OSCE meeting in 1999 – now with no war in the background and support from the
West. He negotiated the withdrawal of troops by July 2001 rather than 2020. If he had signed,
Coup-Proofing 137

In the mid-1990s, there was a general sense that the “frozen conflicts,”
which had left unrecognized states in Abkhazia, South Ossetia, and Adjara,
would be reincorporated into the Georgian territory after patient diplomatic
negotiations.25 This would not be the case. The Russian leadership was not shy
about using facts that emerged haphazardly in these conflicts to press foreign
policy interests. As it became progressively clearer that Shevardnadze was not a
disposable figure in the Georgian political scene, and that he had aspirations –
indeed tangible plans – to move Georgia out of Russia’s sphere, the Kremlin’s
position hardened. Military and political assistance to ethnic minority regions
was provided. In the language of the model, this lowered the cost of war
and raised the reservation value r for Abkhaz, South Ossetian, and Adjaran
warlords. This might have begun as a strategy for punishing Shevardnadze
for not joining the CIS and weakening his coalition formation project, but,
over time, Georgian coalition politics evolved in the face of these new
constraints.
The citizens of Abkhazia, South Ossetia, and Adjara became bargaining
chips in a geopolitical game played between Russia and Georgia, which was
trying mightily to reframe the issue as a conversation between Russia and
the West. Protected from Georgian interference by the presence of a Russian
military garrison, Aslan Abashidze and his family accrued fantastic personal
wealth as the lords of Adjara during the 1990s (and, in a lesser known part
of the transaction, would regularly lend his party’s parliamentary votes to
Shevardnadze’s party in exchange for non-interference).26 Abkhaz and South
Ossetian “shadow states” persist to this day, now with established customs
houses, foreign ministries, flags, national histories, and many other trappings
of statehood.27 These frozen conflicts provided Russia with a substantial source
of leverage over the successively more Western-leaning Georgian governments.
Though these regional hold-outs had no ability (or ambition) to seize the capital
of Tbilisi, they remained territorial outposts for Russians. More importantly,
they were lightning rods for Georgian nationalists, who fed off the anger of the
displaced refugees who had been ethnically cleansed from the conflict zones.
In May 1998, a six-day war broke out in the Gali district, the buffer region
that straddles the unrecognized Georgia–Abkhazia border. The frozen conflict
went hot in hours, yielding around 250 casualties and temporarily displacing
between 30,000 and 40,000 Georgians from their homes. Shevardnadze was
blamed by one faction for having coddled the guerrillas and blamed by the other
faction for not doing enough to protect ethnic Georgians in Gali, who had been
driven from their homes twice in five years. These events led almost immediately

the South Caucasus would be different today. Interview with Archil Gegeshidze, October 17,
2006. See also Larsson (2004), 406–409.
25 For two examples of prominent area experts who committed to this position in print, see Suny
(1995) and Laitin (2001).
26 See Derluguian (2005), 230–233.
27 See King (2000).
138 Jesse Driscoll

to a military uprising in Senaki (Mingrelia), reportedly led by former Zviadists

– now incorporated into the Georgian army and Interior Ministry. The mutiny
was put down the same day, but the entire set of events served as a reminder of
the fragility of the arrangements that kept order in Georgia.
Meanwhile, back in the capital city of Tbilisi, processes of gradual alignment
with the West were yielding huge social dislocations. The high-stakes politics
of privatization that accompanied the end of socialism were usually decided
behind closed doors. I asked a key informant within the Georgian Ministry of
the Interior (“Levan”) how former militia members got their share of foreign
aid, which seemed (to me) to be accruing in the bank accounts of Europeans
and the new class of nongovernmental organization employees (“NGO-nicks”).
His response is worth reproducing here at length:

“So, you have the big men at the top in the Ministries. They are the bosses, and they
make sure that things get divided fairly, so that nobody causes trouble. If there’s trouble,
there’s sometimes a scandal, and then someone from Parliament or from the Chancellery
[Executive Branch] will have to get involved, and then maybe everyone loses their jobs.
Those of us at the bottom pay up, and if we have problems [with another Ministry
employee] we take the problem to our bosses and let them sort it out … But most of the
time, everyone knows who is connected, and they pay up to make sure that they keep
a roof, you see? We are like tax collectors. Say you’re a rich man, you speak English,
you run an NGO – you still have to rent your building, yes? You have someone to drive
your car? You still eat at restaurants? You have cousins and brothers and family, and you
want them to all get jobs and stay out of trouble? Then you need to know people, to do
favors. … Everyone needs soap and clothes and [music] tapes, and even the pensioners
have kids sending money home from abroad, so the police who get [to work] the bazaars
are always fine … but everyone does better when the Tiblesei [Georgian elite that have
lived in the capital for more than 3 generations] are getting rich. Then, it’s just a matter
of dividing up the plate.”28

As a former Minister of Public Order and Minister of the Interior for Georgia,
Shevardnadze immediately understood what most Western journalists covering
the region did not: although the large militia factions (the Mkhedrioni, the
National Guard, the White Eagles, etc.) pretended to be organized based on
strong ideology and dense social capital, their unity and hierarchy were based
more on rhetoric than reality.29 As described at length in the previous chapter,
these umbrella organizations were composed of much smaller networks,
recruited using family and criminal ties. The Mkhedrioni in particular claimed

28 Interview conducted in Tbilisi, November 1, 2006.

29 Shevardnadze was tremendously sophisticated as a student of contemporary street politics and
local hierarchies of deference – he had a reputation for going out of his way to pay respect to
the pecking order that militias established, address individuals by their self-appointed rank,
and ask permission from superiors before speaking with subordinates. This was primarily a
way of signaling to the militia members that he took their honorifics very seriously – and it set
him apart from most of his peers, who could not convincingly hide their class biases.
Coup-Proofing 139

an enormous membership but, by this point, most of the members were tied
to a particular protection relationship with a captain, a local political boss,
or a politically connected “businessman” organizing imports and exports.
Shevardnadze understood that if he were to pull at loose threads and quietly
deal with militia captains individually, he could centralize authority without
rocking the boat or directly confronting the celebrity leaders of the umbrella
factions, Tengiz Kitovani and Dzhaba Ioseliani. He slowly tightened this web
and let warlords turn on each other, while simultaneously assuring “insider”
warlords that he would look the other way while they looted the state. His eyes
could not be everywhere, but a lot of different kinds of people were beginning
to think of Shevardnadze as their “roof.”
One of the central problems that Shevardnadze faced – a problem that
motivated the model in Chapter 2 – is that militia captains who were already
controlling lucrative bottlenecks in the informal sector were skeptical of the
idea that they could do better by giving their loyalty to Shevardnadze. Recalling
a conversation with Georgi Karakashvili (the commander of the White Falcons,
a group that was a relatively early convert to Shevardnadze’s consolidation
project) about promotion into the Border Guards, a former Mkhedrioni
lieutenant told me:

“[The offer] was a better and safer life, yes, with a salary and good opportunity to get
noticed … but it would take me away from my home and friends. And once I was all
alone, how could I be sure that they would not later say ‘him – he’s a criminal, and
I do not need to keep my word to him.’ Who would blame them? … But I took the
promotion, in the end. It was a risk, but what was my choice?”30

This captures two facets of the general dilemma warlords faced: a leader cannot
credibly pre-commit to a particular distribution of wealth, so warlords have
no real reason to take him at his word. But if the warlords foresee a time in
the future when their reservation values will slip toward zero, they might be
convinced to throw in their lot with the president anyway, being forced to take
a risk just to keep what they have. This process of gradually lowering profitable
options for warlords outside of the state was the core of the state-building and
incorporation project in Georgia. The offers came in the form of promises of
autonomy within the ministries, which was valuable, even if it was understood
that the autonomy could, in theory, be revoked at any time.
All the while, Shevardnadze and his economic team used the transition to
a market economy to make the office of the president indispensable to the
functioning of the domestic economy. New aid projects and foreign investment
meant new opportunities for entrepreneurship, especially for Georgians who
could speak English. The influx of wealth also brought new opportunities
for graft, corruption and personal enrichment. The creeping formalization of

30 Interview conducted in Kvimo-Kartli on November 13, 2006.

140 Jesse Driscoll

the economy brought a proliferation of permits and regulations, as well as

the inevitable back room payments and bribes that could circumvent the new
regulatory regime. As the organization on the front lines of enforcement, the
Ministry of the Interior was well positioned to tax the informal bottlenecks in
this new era of entrepreneurialism. But it was the office of the presidency that
was the true beneficiary of this new culture of corruption, as the president was
a focal point for the emerging class of criminally connected businessmen who
made huge profits from skewed privatization processes. In a transitional society,
where the national currency was worthless and property rights could not
really be enforced outside of personalist connections, the president’s continuing
friendship and patronage was invaluable. Once all the major players came
to realize they had a shared stake in their friendship with the same man –
and that no one else could offer the same degree of wealth creation that
Shevardnadze could – contemplating a coup became much more difficult.
Former Minister for Economy Vladamir Papava described the situation well:
“While the laws were sometimes ambiguous, Shevardnadze was quite careful to
not renege on promises regarding a person’s [privatized] property. Investors and
wealthy Georgians needed to trust the rules, trust the system. And, of course,
because sometimes his promises overlapped, he was the system.”31 (L’état,
c’est moi.)
Shevardnadze’s ability to promote militia members within the shells of the
formal security institutions inherited from Soviet times was also a powerful
commitment mechanism. Portfolios with independent hiring and firing author-
ity were particularly coveted, as they allowed warlords to formalize their patron
relationships with their subordinate clients and screen recruits for personal
loyalty. When loyalty to a gang leader was the paramount consideration for
promotion and advancement, it led many career bureaucrats and competent,
honest armed agents to retire. This led to a strange situation where superficially
similar security institutions – with the same buildings, names, and even
uniforms as in Soviet times – emerged with completely different personnel and
a different sort of corporate culture. The institutionalization of unaccountable
pockets of authority at the top levels of the bureaucracy had a corrosive
effect on overall morale, with terrifying long-term consequences for Georgian
civilians.32
While the Constitution of 1995 gave Shevardnadze power to selectively
dismiss individuals anywhere in any ministry, this sweeping power was less
threatening to warlords than it might have appeared for a few reasons. First,

31 Emphasis in original interview February 6, 2006.

32 Charles King has observed that in some respect this culture was not new: underpaid traffic
police using their positions of authority to extract bribes from drivers’ security was a fixture
of late Soviet life in the Caucasus, and those technologies of extortion were replicated in the
new security ministries (2006). See Zverev (1996), Nodia (1998a,b), and Areshidze (2007) for
descriptive accounts of this phenomenon.
Coup-Proofing 141

it was assumed that if a particular individual was dismissed after a few years,
it would only be after the ousted person had used his time in the ministry to
install loyal subordinates who could continue to make payments to him or his
family. Second, it was fairly well understood that micro-managing the complex
personal relationships that greased bureaucratic wheels inside the Ministry
of the Interior would have been an impossible task for anyone, especially a
president whose attention was mostly focused on foreign relations. This gave
lieutenants a great deal of practical autonomy. Third, constant reshuffling of
portfolios would have created perceptions of instability and uncertainty, which
would be as bad for Shervardnadze’s rule as it was for the warlords themselves.
This third point is discussed in more depth in the text that follows.
Over time, the ability to set up miniature fiefdoms within the Ministry
of the Interior freed members from the “umbrella” militia hierarchies that
had emerged in the 1988–1991 period. Shevardnadze understood, perhaps
better than anyone in Georgia, how to pit the career incentives of different
short-sighted men against each other. Alex Rondeli recalled: “He was a
man who saw the whole board … always calculating, always thinking about
promoting with one hand and installing your worst political enemy one step
below you in the hierarchy at the same time. He could manage relationships
with the best of them. He was the finest product of the [Soviet] system.”33
He met with a great number of police captains, militia lieutenants, and other
subordinates in the Mkhedrioni and National Guard hierarchies, always for-
mally, being careful to respectfully ask the permission of the State Council first.
But very slowly, using his powers of appointment over the security ministries,
Shevardnadze began to promote lieutenants and sub-lieutenants, re-shuffling
security personnel at almost monthly intervals during the 1993–1995 period.
He created entirely new security sub-bureaucracies within the Ministry of the
Interior with overlapping mandates, including the Border Guards, the Special
Emergency Response Corps, Tbilisi Rescue Corps, Government Guard, Internal
Troops of the Ministry of Internal Affairs, the American-trained Special Unit
Alpha and a CIA-Trained Presidential Guard, and several even smaller forces.
A zone of autonomy within the security services, even if its mandate overlapped
with someone else’s authority, could be transformed into lucrative business
opportunities. Most leapt at the opportunity to guarantee themselves a piece
of the pie.
What this meant was that real power increasingly lay outside the official
institutional framework. Stephen Jones described politics during this period as
“feudalized,” in the sense that it was “fracured politically and geographically
into unofficial power centers [which were] dominated by holders of important
political office but who run their spheres of influence on the basis of informal
networks, mutual favors and obligations.”34 Formal institutions were evolving

33 Interview in Tbilisi on November 19, 2006.

34 Jones (1997), 525.
142 Jesse Driscoll

during this period in a way that made Georgia appear like it was following
an Eastern European script of legal and economic reform, parliamentary
accountability, and engagement by civil society groups. However, as in
Soviet times, these formal structures worked only intermittently without the
aid of personal networks. Because these personal networks and informal
institutions are difficult to monitor or map, particularly for outsiders, it is
hard to make inferences about anyone’s actual influence during this period by
following official posts, or any metric other than their network distance from
Shevardnadze. Dzhaba Ioseliani, for instance, spent virtually all of 1993 as a
mere Parliamentary Deputy. In the aftermath of the 1992 elections, Kitovani
was forced out of his position as Minister of Defense, but, in practice, retained
influence by naming his own replacement and by that point was managing a
controlling stake in Georgia’s partially privatized energy sector.
What of the non-Georgian ethnic minorities? On the whole, their warlords
fared well under this system. Georgia is often presented by political scientists
as a laboratory to observe how ethnic boundaries can be constructed as static
and militarized.35 Ethnic secessionist regions – Abkhazia, South Ossetia, and
Adjara – all fit neatly into spaces on old Soviet maps and can provide extensive
historiographic evidence of their cultural distinctiveness from Georgians.
Recall, however, that a large number of lesser known Armenian and Azeri
enclaves (Akhalkalaki, Dmanisi, Ninochminda, Kvemo-Kartli, Marneuli, and
Bolnisi) and the Mingrelians in Western Georgia, were all accommodated by
the Georgian state. All were potential sources of leverage for Russia, but, in the
end, most of these enclaves organized arrangements functionally similar to that
of Adjara’s “warlord” Aslan Abashidze, brokering their votes in parliament
to Shevardnadze’s party in exchange for food, fuel, and non-interference.36
Shevardnadze’s guaranteed voting constituencies were ethnic minority areas,
which would deliver their votes for the CUG en masse. In the early stages
of power consolidation, he appointed Otar Pacacia and Avatandil Margiani,
symbolic representatives of Mingrelian and Svan ethnic lobbies, respectively,
to the State Council and then to high positions in his government, assuring
them that the informal governance structures those communities were using
to distribute favors would be protected in the new state. By the 1999 par-
liamentary elections, even Aslan Abashidze’s “Union for Democratic Revival”
was boxed into defining itself as a national party and part of the opposition
coalition.37
What evolved over time was an “off-the-books” economy where between
60 percent and 70 percent of economic activity in the country – certainly

35 See for instance, Toft (2005), 1–33, 87–126.

36 See, Marten (2012), 68–86.
37 Jonathan Wheatley suggests that the police “cleared away the local warlords” (88) in some
districts, but in my reading of the period it was far more common for these men to remain in
place (with their authority legitimized) and then find themselves enmeshed in the president’s
favor economy.
Coup-Proofing 143

much of the trade in manufactured goods, alcohol, tobacco, and food – was
unreported and untaxed.38 Former Prime Minster Zurab Zhvania estimated
that smuggling had cost the state 60 million lari ($29 million) from uncollected
tobacco revenue alone in 2003 and almost $200 million from nondeclared oil
products.39 Thomas de Waal of International War and Peace Reporting quotes a
corruption specialist who put the issue in even starker terms, saying that “The
[Georgian] government could eliminate its fiscal deficit if it would just fully
collect the taxes on two products, imported gasoline and cigarettes.”40 Despite
the fact that the government retained the bloated state apparatus of Soviet
times and continued to promise a huge range of social services (health care,
pensions, etc.), tax collection was about 9 percent of reported gross domestic
product (GDP). It was a system where no state employees were paid enough
to live on, and so they had to marketize their place in the bureaucracy. It was
structural corruption on a societal scale, which made it impossible to invest
money without high-level personal connections. Given that the Ministry of the
Interior was incorporating criminal militias, it is not surprising that this sort
of corruption was particularly pervasive in the security services. Svante Cornell
claimed that the Georgian police forces in the late Shevardnadze era were the
most corrupt in the region, thoroughly penetrated by criminal interests.41
An overlooked part of this strategy was turning the lower tiers of the militias
against their factional commanders. Once the warlords were substantially
co-opted, it was possible for Shevardnadze to order the arrest of prominent
warlords (such as Kitovani in September 1994 or Ioseliani in 1995). The
deeper proof of Shevardnadze’s political prowess was not when Ioseliani’s
Mkhedrioni forces were deployed to dismantle Kitovani’s National Guard, but
when co-opted Mkhedrioni members later turned on Dzhaba himself. Jailing
the spiritual head of the Mkhedrioni would have been extremely threatening to
the remaining militias, had the president not already demonstrated – repeatedly
– that certain new rules of the game were fixed. The regime had demonstrated
that, even given the opportunity, it would not liquidate the assets of new
businessmen, who would be permitted to keep the wealth and property that
they acquired during the scramble for Soviet spoils. The commitment by
the president to keep the state extremely weak in certain areas, especially
public administration and taxation, was made credible by handing entire
bureaucracies over to incorporated militia members as explicit side payments
for their loyalty. In this regard, the incremental moves to “crack down” on

38 King (2004b), 16 and Slade (2007).

39 Ratiani (2004).
40 Quoted in de Waal (2005), citing MacPhee (2004).
41 Cornell (2003), 33. Though drugs are neither produced nor consumed in Georgia in
commercially interesting quantities, according to Shelley (2004), Interpol estimates that 10
percent of Afghan heroin moving to Europe moves through the South Caucasus. Georgia was
hyperbolically identified as “the most” corrupt Soviet state by Ekedahl and Goodman (1997),
280.
144 Jesse Driscoll

prominent Mkhedrioni leaders who had kept aloof from Shevardnadze in early
1995 were more important than local scholars have tended to recognize, as
most of the police lieutenants who enacted the arrest warrants had themselves
sworn blood oaths to the Mekhedrioni a few short years before. It was a
public warning by Shevardnadze that Dzhaba Ioseliani was no longer above
politics. Shevardnadze could order one group of Ioseliani’s men to attack the
other group, and they would listen. Whatever arrangements were enforced,
they would be enforced primarily through the Ministry of the Interior.
The failed assassination attempt against Eduard Shevardnadze was an
important turning point for Georgia. According to rumor this assassination
attempt was a direct response to Shevardnadze’s affront on Ioseliani’s leader-
ship. Shevardnadze used the event to dismiss the Russian head of his military,
Igor Giorgadze (who had fled the country in a Russian aircraft), order the arrest
of Ioseliani and his lieutenants, and pass a new constitution – finally formalizing
his presidency and established the current legal framework for the Georgian
state. The legitimacy conferred through his mass-based political party was used
to channel latent anti-Russian resentments into an energetic, Western-oriented
foreign policy on the European model. But what Shevardnadze did not do
proved just as important. He did not use the assassination attempt to purge
the lower ranks of the security agencies, despite the fact that he would have
had public support in doing so. Again, my interpretation of this inaction is
that the non-prosecution of former regime opponents who had joined the
ranks of the state sent a clear signal that certain rules of the game would
be sacrosanct. As a high-placed former cabinet official wryly commented,
“He was magnanimous in his victory after the car bomb. That was why he
survived [instead of Gamsakhurdia]. He understood that there was plenty to
share.”42
As the model suggests, the promise of increased wealth opportunities for
warlords who colluded with the state rather than challenging it was what
brought unruly militias to heel. Importantly, Shevardnadze was removed from
the day-to-day functioning of the shadow economy that managed to enrich the
Georgian militia captains. This is consistent with the model, given the extremely
high stability threshold (s). Shevardnadze gave his armed agents a relatively free
hand domestically, so long as they did not take actions (generating violence or
scandal) that would disrupt the flow of aid and threaten the entire system. Only
when an individual stepped out of line would a punishment materialize, and
even then, only with the firm backing of various other power centers in society
poised to take advantage of the deviant’s fall. The mechanism that was used
to punish failure was kompramat (compromising information), a strategy from
the Soviet era of maintaining a secret file of illegal dealings for the purposes of
blackmail. For all of the disadvantages of the rampant corruption in the favor
economy described previously, it had the virtue of creating a paper trail. Almost

42 Interview conducted in Tbilisi, November 29, 2006.

Coup-Proofing 145

anyone who had grown rich and powerful in Georgia during the 1990s had
done so by running roughshod over some obscure legal barrier. With everyone
implicated in the same ponzi scheme, there was a constant threat that the
government would expose an individual criminal’s misdeeds and expropriate
all of his family’s wealth. Zurab Zhvania, then chairman of the Parliament and
manager of Shevardnadze’s 1999 electoral campaign, reported to Keith Darden
that all of the so-called “anticorruption” campaigns carried out in Georgia were
actually thinly veiled purges.43 To understand how this worked in practice,
consider one militia member’s account of Shevardnadze’s “magic desk”:
“American democracy is old … you do not know court politics. Here [in Georgia] we
have to make our democracy work in the shadow of an older history, in the shadow
of the Tsars. … Ioseliani was at first like a court magician – like Rasputin! – using his
eyes and his voice to make Soviet bureaucrats lose their balls … but I think he [Ioseliani]
didn’t know what he was up against … Shevardnadze had those gifts too, you see: He
had a magic desk, with a permanent memory. If you worked for him, he would have a
file on you. All your history, every mistake. And when it came time for you to go, he
would call you in, open the drawer, show you that he had you. But if they quit, and
went quietly, his way, it would stay safe and forgotten in the desk. … The magic trick
was that you believed that he would actually forget and forgive, and leave the file in the
desk, if you did what he said. He had that power. The power priests have. The power
to bless and forgive.”44

This equilibrium was propped up by huge sums of Western development aid.

But life grew more difficult for most Georgians. In spite of the proliferation
of NGOs and the rhetorical promises that Western European–style welfare
socialism was around the corner, it was evident that the individuals most
capable of paying bribes were also the individuals with the connections to
place money in offshore accounts or purchase property. By the end of 2003,
in spite of huge infusions of Western aid, internal debt in unpaid salaries and
monthly pensions for state employees (which had by this point dropped to less
than $7) had reached $120 million. Half of the population was living below
the international poverty line.45 Although there was great care taken with the
facade of safety and normalcy in downtown Tbilisi, economic transactions
took place against a social backdrop where violence was just below the
surface, precisely as one would expect, given the fact that many of the police
officers were rehabilitated militia members.46 The relationship between citizens
and the police during this period harkened back to the late Soviet period:

43 See Darden (2008), 52.

44 Interview conducted in Tbilisi, May 8, 2007.
45 To give a sense of the size of this untaxed and illegible “shadow economy”: the reported
per capita GDP of the state divided by the best population estimates gave a per capita GDP
figure roughly on par with that of Swaziland. See King (2004b) and Papava (2006).
46 Wheatley (2005) reports that “on at least six occasions between 1997 and 1999, suspects are
said to have ‘committed suicide’ by jumping out of the top windows of MIA detention centers
during police interrogation.”
146 Jesse Driscoll

representatives of the Ministry of the Interior were dangerous, and you did not
want them to think you owed them a favor. According to a survey conducted
in the waning months of the Shevardnadze administration, only 32 percent
of Tbilisi residents said they would turn to the police after being victims of a
violent crime.47

emomalii rakhmonov: the slow talker

When the militias seized the capital city of Dushanbe in December of 1992 in
the name of the Popular Front of Tajikistan (PFT), Emomalii Rakhmonov had
been the Head of State for just a few weeks. Rakhmonov was named head of
state in a special session of the Supreme Soviet convened in territory held by
the Popular Front. This former kolkhoz collective farm head from Kulob was
a compromise candidate, in that both the armed groups from his home region
of Kulob and Khojandi financiers with political ties to Tashkent and Moscow
believed that they could control him.48 For years afterwards, Rakhmonov was
commonly referred to as Rais Rudaki Prospect (“President of Rudaki Avenue”)
because his administration was unable to enforce its edicts even a mile from the
presidential palace.49 Rakhmonov faced a situation even more harrowing than
that of Shevardnadze in Georgia. He was literally a hostage to Kulobi war
criminals in his own inner circle, while the Khojandi elite schemed to replace
him and recapture high state offices. A rebel army launched guerrilla raids from
the mountains in the country’s east, where the elites were toying with secession.
The economy was nonexistent.
Shevardnadze’s tactics of exploiting factional and interpersonal divisions
within paramilitary umbrella groups to seize control “from below” were
foreclosed to Rakhmonov because of dense avlod ties within militias. In a con-
versation with a high-ranking member of United Nations Mission of Observers
in Tajikistan (UNMOT) with firsthand knowledge of the consolidation process,
I described Shevardnadze’s strategy of dismembering large umbrella factions by

47 Wheatley (2005), 132–133 reports that this survey also reported that only 20 percent of citizens
(across the country) would involve the state in a land or property dispute.
48 Various Lakai Uzbek and Hissori militias – both minorities – were wary of returning a Khojandi
to the regime’s top office, and rationalized the rise of a Kulobi. Most hoped Kulobi rule would
be temporary, and some – Kenjayev in particular – even held out hopes that they would
eventually be able to emerge as pivotal in the new power structure, if the country aligned
closer to Uzbekistan.
49 Even after the 1997 Peace Process, many warlords continued to control territory in which
neither the government nor UN observers could enter without the permission and escort of
these warlords. The most prominent of these locations were Darband (controlled by Mullo
Abdullo), Tavildara (home of the Minister of Emergency Situations Mirzo Zioyev), and even
the Leninsky District, twenty kilometers from downtown Dushanbe (controlled by Rakhmon
“Hitler” Sanginov).
Coup-Proofing 147

exploiting frictions between sub-commanders for comparison. His response to

this is worth quoting here at length:

“He [Rakhmonov] did the same thing here – you peel away the label of the Popular
Front and you find Kulobis, Hissoris, Lakai, Leninobodi financiers, Uzbeks – a real
mixed bag. And he exploited those divisions, certainly … but I can’t think of a time
when he successfully turned a lieutenant against a commander if they were from the
same region … that just couldn’t happen, you see. I know that it sounds strange, but
these guys are all related to each other. … That’s part of why the competition got so
fierce and bloody, and why it was so difficult to reach compromises at first: Everyone
knew that everyone else just wanted to hand things over to their brothers-in-law. … Even
today, everything worth owning is owned by someone who has sealed the deal not just
with a joint business venture, but by arranging a marriage between someone in their
family to someone with ties back to Dangara [the President’s home region]. That’s how
it is here.”50

Despite these obstacles, Emomalii Rakhmonov developed a strategy for

political survival. He realigned domestic institutions and economic policy to
make it difficult for elites from different regions to cooperate. Rakhmonov’s
coalition came to power as a client of Uzbek and Russian interests: essentially
to hold the line against the civil war in Afghanistan from moving north. When
frictions between Uzbekistan and Russia began to emerge, Rakhmonov allied
his small nation strongly with Russia and proceeded to purge Uzbek-backed
factions from the Communist old guard. The eventual settlement of the
Tajik civil war came at the prerogative of Moscow and at the expense of
Tashkent. Being a peacemaker strengthened Rakhmonov’s legitimacy with the
population and introduced new players into the governance shell-game. A
combination of symbolic autonomy and large shares of the Afghan narcotics
trade smoothed this transition. Eventually, he freed himself of dependence on
the Kulobi warlords who put him into power. Cumulatively, the first decade of
Rakhmonov’s rule saw the fragmentation of the Popular Front coalition and
the political disenfranchisement of every region other than the president’s own.
Unlike Shevardnadze, there is little temptation to call him a genius. But if power
politics in for Tajikistan were a scored game, Rakhmonov – now “Rakhmon” –
won it.
Emomalii Rakhmonov lacked Eduard Shevardnadze’s gravitas, celebrity,
external confidence, and foreign policy experience. He came to power
supported by politically untested rural criminal-clan networks, with no friends
in Moscow or Tashkent, let alone Europe or the United States. But he proved
a remarkably quick student of geopolitics. Tajikistan was unique among
the Central Asian republics in that it was spared the cadre purges of early
1980s because the Soviet Union was fighting its war in Afghanistan.51 What

50 Interview conducted in Dushanbe August 15, 2005.

51 Going after cotton corruption was a luxury that Moscow could afford in republics like
Uzbekistan, but in Tajikistan, Moscow was more or less willing to leave Khojandi elites in
148 Jesse Driscoll

Rakhmonov took from this historical period is that Moscow could be counted
on to come to the aid of his government so long as Islamic radicalism was kept
on the other side of the border. The civil war demonstrated that Rakhmonov
and the Kulobi mafia networks that backed him were capable of controlling the
demographically dense lowlands of their country, making these Kulob-based
militias indispensable coalition partners for the time being. The key to the
centralization of power in the hands of a shrinking clique was international
public relations with Moscow – “He was a fast learner and a great salesman.”52
In broad brushstrokes, the story of Tajik politics in the 1990s is the
story of Emomalii Rakhmonov slowly learning the political skills that Eduard
Shevardnadze had spent a lifetime perfecting. He became adept at mapping
the frictions between different coalitions, exploiting divisions to split his
enemies, and acquiring a sense of timing and political theater. Russian military
power gradually eroded the UTO’s will to fight. Rakhmonov’s foreign policy
gradualism facilitated a slow rapprochement between Russia and Uzbekistan
over the composition of the Tajik government, which finally strangled the
ambitions of certain Uzbek-backed militia captains who dreamed they could
capture the state apparatus with Tashkent’s blessing. The Tajik government’s
control of the major north–south roads gave the Ministry of the Interior a
virtual monopoly on the rents from the Afghan heroin trade. The eventual
arrival of American strategic bilateral aid after the invasion of Afghanistan
gave the regime new resources to buy off warlords. As the logic of the model
anticipates, while warlords were losing outside options this pulse of external
funding led to a broader ruling coalition – not a democratic outcome, but a
stable outcome that expanded the president’s opportunities for divide-and-rule.
For the first few years in office, Rakhmonov acquired a reputation as a
patient, methodical, nonthreatening force in Tajik politics – not at all ruthless,
not at all relentless, very interested in listening, and not quick to give voice
to his opinions. Like Shevardnadze, he found that delaying difficult tradeoffs
and acquiring a reputation for indecision was a decent strategy for building
trust between warlords. Also like Shevardnadze, he developed a remarkably
consistent foreign policy orientation that allowed him to make the most of a

power so long as they served as a loyal “buffer state” against the chaos in their southern
neighbor. Kathleen Collins argues that the social solidarity against these purges was what
created interclan elite pacts in other regional states. She hints that in 1991, these pacts allowed
Uzbekistan and Kyrgyzstan to avoid Tajikistan’s violent fate with clandestine clan pacts. See
Collins (2006), Chapters 4 and 5.
52 Quote from former Ministry of Foreign Affairs official. A member of the United Nations
Tajikistan Observation Mission (UNTOP) observed the following: “Rakhmonov always gave
a warning before he went after you. He’d warn once, twice … he’d wait for months or years,
giving people a chance to remove themselves from politics once they were becoming too
threatening. He always gave everyone opportunities to get away. … He was not a genius,
no … more of a ‘slow talker,’ if you know what I mean.” Interview conducted in Tajikistan,
July 22, 2007.
Coup-Proofing 149

bad geopolitical hand. He succeeded in convincing Russia, Uzbekistan, and

the United States that it was better to prop up his corrupt and fragile state
than to risk state collapse that might infect the region. Rakhmonov needed no
urging to commit his country to the CIS Collective Security Treaty. The 1993
Russian–Tajik Agreement on Friendship, Cooperation and Mutual Assistance
made Tajikistan a Russian military outpost. Rakhmonov also proved to be
extremely skilled at controlling the flow of information to the Russian security
services, gradually convincing Moscow that power sharing with Hissori, Lakai,
or Khojandi groups would only produce an Uzbek puppet state, which would
be adverse to Russian geopolitical interests.
Russian support – manifesting in the language of the model as a low
stability threshold s – facilitated a hard line against challengers to Rakhmonov’s
regime. Despite a number of high-profile meetings and international mediation
efforts, the internationally recognized government in Dushanbe refused to
make concessions to the United Tajik Opposition (UTO). Some elites in
Gorno-Badakhshon (GBAO) declared independence in 1993, but failed to
achieve any hint of international recognition.53 As they grew more and more
desperate, the Islamist rhetoric of the UTO increased in tone and frequency in
an attempt to attract financing from Iran, the Gulf States, and Afghanistan.
This backfired, making the worst-case threat of the Kulobi hard liners more
credible to distant elites in Moscow. Any doubts that the elites in Dushanbe
were the last line of defense against militant mujahideen, potentially poised to
export revolution to the soft Muslim underbelly of the former Soviet Union,
were belied by intelligence reports that included quotes from certain radical
clerics affiliated with the UTO.54
Similar tactics were used to trim the membership of the Popular Front
coalition of anyone whose influence network stretched to Uzbekistan. Promi-
nent Khojandi, Hissori, and Lakai warlords would cry foul to distant cousins
and one-time patrons in Uzbekistan, hoping that Uzbekistan’s president Islam
Karimov would be able to intercede on their behalf to Moscow. But appeals
through Tashkent backfired in the same way: they were reframed as evidence
that Uzbekistan was trying to direct Tajikistan’s political development and limit
Russia’s role.
Because the president inherited an impoverished state apparatus and faced
militias recruited with dense social capital, it was not easy to use patronage

53 Recall that GBAO has cast votes in the 1994 election with 96 percent for Abdullajanov and
only 4 percent for Rakhmonov.
54 I honestly have no idea how one would, with the full benefit of hindsight, assess the voracity
of these sort of claims. I suppose it is possible that there were pockets of angry Salafi jihadist
radicals lurking in the mountains, poised to seize fragile states. What can be stated with
certainty is that many journalists in the immediate aftermath of the September 11th attacks
found it convenient to take this propaganda at face value. See, for example, Rashid (2002) –
and take special note of the cover art.
150 Jesse Driscoll

opportunities to subdivide the militia captains.55 The first priority was the
restructuring of domestic institutions to formalize the bargaining advantage
that the Kulobis had earned with their military victory. It took no time at
all for the Kulobi ruling class to grasp that independence meant sovereign
debt. Functionally, the war’s winners were in a position to borrow money
using the bodies of the war’s losers – laborers residing in the rural, but
economically productive, cotton-producing regions of the south – as collateral.
The internal boundaries of the state were redrawn to give Kulobis semiperma-
nent dominance over the cotton-producing agricultural southwest, essentially
reappropriating the institutional advantage that the Khojandis had used to rule
the country unchecked for decades.56 Khojandis were passive participants in
the disenfranchisement of the other non-Kulobi members of the Popular Front,
apparently willing to gamble that the disenfranchisement of other groups with
ties to Uzbekistan would give them an effective monopoly on ties to Tashkent.57
As Rakhmonov and his network of Kulobi apparatchiks began to monop-
olize power within the state apparatus, this centralization came first at the
expense of groups with ties to Uzbekistan. Time and again, Uzbekistan would
have the opportunity to play the part of the spoiler. The political maneuvering
that followed took on many aspects of a proxy war between Uzbekistan and
Russia, for reasons outlined at the beginning of this chapter. Briefly, over

55 The fact that the warlords’ families were clustered by geographic region and connected by
political marriages meant that wealth transfers had some positive spillover effects. As such,
warlords from Kulob could convince themselves that their fortunes were lifting together, at
least for a time, even as individual Kulobi warlords were purged from the ruling coalition.
Over time, however, this intra-Kulobi optimism proved misplaced. At the time of this writing,
it has been nearly two decades since the official end of the civil war and more than two decades
since the Kulobis unambiguously seized control of the state apparatus, yet is not clear that the
so-called “Kulobization” of postwar power structures led to much public investment in the
region of Kulob. In winter of 2007, the city center of the regional capital of Kulob still did not
have a constant power supply.
56 Successive waves of population displacement had transformed the demographics around
Kurgan-Tubbe in a way that favored Uzbek-affiliated ethnic groups. As it was articulated
to me, the fear was that Kurgan-Tubbe could ally with the Khojandis in the north, giving
Uzbekistan potential influence over two of Tajikistan’s five electoral districts. A military coup
or popular revolution at the center by Uzbek-backed fifth columnists, it was feared, could be
3/5. The solution was simple: redraw regional boundaries to create only four regions. This
crude gerrymandering was achieved by revising the constitutional framework to combine the
administrative districts of Kurgan-Tubbe and Kulob into a single district of “Khatlon,” an
ancient name of Kulob. This effectively disenfranchised Uzbeks and Hissori populations in the
south under the demographic weight of Kulob, reified in the 1994 election when 99.5 percent
of Khatlon was reported as having voted for Rakhmonov (see Chapter 4). The capital and
surrounding regions remained a district, but one under control of Kulobi militia formations.
Khojandis could then be played against the Pamiris. The government cemented this advantage
further by resettling many Kulobis in western Khatlon, installing them at the heads of collective
farms after the murder or displacement of previous district heads. See Roy (2000), especially
97–98, and ICG (2005).
57 Interview conducted in Dushanbe January 28, 2006.
Coup-Proofing 151

generations at the top of the Tajik status hierarchy, many Khojandis had
come to see themselves as closer, culturally and economically, to Tashkent
than to Dushanbe. In recognition Khojand’s special status in Moscow, the
Russian foreign policy apparatus was convinced to keep a consulate open in
Leninobod until 1996. The mass expulsions of the Gharmis during the civil
war had left about half of the population of Kurgan-Tubbe as ethnic Uzbeks,
who could have emerged a natural power base for Uzbek-backed factions.
Kurgon-Tubbe, whose commander Mahmud Khudoiberdiyev maintained close
contact with Tashkent, simply did not pay taxes to Dushanbe for years. The
Uzbek government made no secret of the fact that they did not trust the Kulobis’
“soft” approach to incorporating Islamists, and resented Russian efforts to
cultivate a client in the region that was not linked through Tashkent.58 An
embassy representative described the problem in the following way:

“So, today, we know how the story ends, and we look back and think ‘you’d have to
have been crazy to think Rakhmon was an Islamic fundamentalist, or sympathetic with
Iran, or anything like that! He’s old-school! He’s as conservative as they come! He’s a
Party man!’ But we were getting fed those lines all the time [from the Uzbeks]. The truth
is that, midway through the 1990s, nobody had any idea who was really running this
country behind the scenes. Everything got mixed up in the war: Tajik nationalists who
were talking about redrawing the map to get the old cities [Bukhara and Samarkand]
back … Iranian money … nutty ideas floating in from Pakistan by way of Afghanistan
along with the drugs. And all we knew, really, was that these rural guys from Kulob
had the whole golosovaniya [voting] machine locked down. But what did they believe,
really? It was hard to know.”59

Military reforms to institutionalize Kulobi military control were another way

to delay confronting the Khojandis. Control of the military was primarily a
fight between Uzbeks, Hissoris, and Lakai, which was not directly threatening
to the Khojandi industrialists. A former PFT fighter opined: “The Khojandis
thought it would be like power sharing, where they ran the party and the
Kulobis ran the army.”60 In early 1993, Rakhmonov announced plans to form
armed forces exclusively out of Kulobi units under the command of “Baba”
Sangak Safarov, the most influential of the Kulobi warlords. This was a terrible
blunder. It alienated non-Kulobi members of the Popular Front and led many to
publicly consider abandoning the regime, which would have effectively opened
a third front in the ongoing civil war. But soon after, on March 30, Sangak

58 For a brief overview on how Uzbek domestic policy considerations shaped their approach
to the Tajik civil war, see Horsman (1999). Uzbekistan’s President Karimov encouraged
uncompromising, brutal punishment toward any faction linked to transnational Islamic
networks. In part this was surely because many of these same leaders had, in the heady days of
Tajik nationalism, voiced irredentist aspirations toward the Tajik cultural capitals of Bukhara
and Samarkand.
59 Interview conducted in Dushanbe, February 8, 2006.
60 Interview conducted in Dushanbe, March 3, 2007.
152 Jesse Driscoll

Safarov was killed in a shootout with Fayzali Saidov. It was extrordinarily

convenient for Rakhmonov to be able to plausibly claim that the whole
ham-fisted centralization of the state had been Safarov’s idea in the first place.
Starting in 1993, representatives of Russian intelligence “essentially set up
shop” in Dushanbe, sharing maps and military intelligence to aid Rakhmonov’s
counterinsurgency against the United Tajik Opposition. These security ties
provided a critical conduit for the Kulobis to feed information to Moscow’s
military and political elites on Uzbek-backed coup plots, completely bypassing
the Ministry of Foreign Affairs and other state organs where the Khojandis
kept power. “Rakhmonov convinced the Russian military intelligence that they
could work with his people … they let the military make the case to Yeltsin.
The Leninobodis [Khojandis] in the Foreign Ministry thought that their friends
in Moscow’s Academy of Sciences would save them … [T]hey got completely
outmaneuvered.”61
Between 1993 and 1995, the Kulobis set about the more ambitious task of
purging Khojandi influence from the state. The reversal of status hierarchies
was sudden and total. The second round of Deputy Ministry appointments
pushed out virtually every Khojandi with ties to the old guard from the
security forces, tax administration, and the presidential inner circle. Faced
with the prospect of being completely excluded from the seat of power in
the republic, Khojandi militias and political parties took a number of steps
to disrupt power consolidation. There were assassination attempts, terrorist
bombings, and urban riots. In August 1993, Rakhmonov deployed several
Kulobi “police units” to the northern regional capital of Leninobod, which
led the Khojandi regional governor, Abdujalil Khamidov, to blow up the
only bridge on the only road connecting Dushanbe and the northern regional
capital.62 In response, in the space of a single week, Rakhmonov fired
Abdumalik Abdullojonov (the Khojandi Prime Minister since 1992) along with
his brother Abdugani (the mayor of Khojand) and Governor Khamidov (also a
Khojandi).
Russian military intelligence moved quickly to quash secessionist aspira-
tions, urging the Khojandis to challenge Rakhmonov within the framework
of the Soviet-inherited state borders. They acceded. Abdumalik Abdullojonov,
who had been in the government alongside Rakhmonov since the beginning, ran
against Rakhmonov for the office of president in 1994. The older generation of
Khojandi nomenklatura gambled that with the regime in Dushanbe controlling
only some 35 percent of the land mass of the country, former patrons in
Moscow and Tashkent would engineer the reunification of the state under their

61 Both of the preceding quotes from the same interview conducted in Dushanbe, February 24,
2007.
62 Although reasonable people disagree, many believe that this was an effort to “test the waters”
at the opportunity of secession in the Abkhaz or South Ossetian model with the military
backing of Uzbekistan. See Neumann and Solodovnik (1996), 97 and Nourzhanov (2005).
Coup-Proofing 153

leadership. Their confidence in foreign patrons proved misplaced. The day after
a dubious election in which 99.5 percent of the inhabitants of Khatlon were
reported to have voted for Rakhmonov, the Yeltsin government legitimized
results with a statement declaring elections “free and fair.”63
It did not take long for Abdullojonov to be charged with embezzling state
assets and being barred from holding public office or standing in future
elections. Just after the elections, Rakhmonov dismissed the heads of the
KGB, Ministry of the Interior, and Prosecutor-General in Leninobod, as well
as thirteen out of sixteen district chiefs who were seen as being sympathetic
to Abdullojonov. The influx of Kulobis to fill these posts provoked large
demonstrations in Khojand, and the ensuing wave of arrests and prosecutions
finally crushed Abdullojonov’s network. General Mamajanov, Khojand’s
provincial military commander, led an abortive armed uprising that was put
down in January 1996, with aid from Russian secret police forces.64 Residual
Khojandi elites were forced to content themselves with symbolic and cosmetic
posts, mostly in the Ministry of Foreign Affairs. Kathleen Collins goes so far as
to state that they have been so cut off from political and economic assets that
they are today “the poorest faction in the state.”65
An obvious disadvantage to this transparently maximalist bargaining strat-
egy was that it made it difficult for Rakhmonov to credibly offer anything to the
UTO. Memories of brutality, atrocity, and broken promises were still fresh in
the minds of many opposition elites. In 1995, opposition field commanders saw
little reason to make concessions. But as we shall see, over the next two years,
it became obvious that enough warlords had quietly committed themselves in
the same kind of order-providing bargain. Previously uncommitted warlords
began to recondition their strategies, aligning with Rakhmonov’s regime in a
cascade of collusion and collaboration.
In the freezing Tajik winter of 1996, a coordinated mutiny from within
the PFT coalition nearly collapsed the state. The forces of two prominent
field commanders surrounded Dushanbe. They seized the airport, cutting off
supplies of electricity and food. The revolt was led by Ibodullo Boimatov,
who had at one point controlled the lucrative aluminum factory at Talvidara,
and Mahmud Khudoiberdiyev, who ruled the southern agricultural hub of
Kurgan-Tubbe as an independent city-state. Both warlords claimed to speak
for powerful constituencies. They had helped to install Rakhmonov, but feared
that as Rakhmonov’s “peace process” unfolded, they would lose influence
in the ruling coalition. Acknowledging that their fears had merit, President
Rakhmonov offered generous compensation to the blackmailers. Boimatov was
appointed to the position of special trade envoy for cotton and aluminum, the

63 For a description of the circumstances surrounding this election, see Chapter 4.

64 Akbarzadeh (2001).
65 Collins (2006), 280.
154 Jesse Driscoll

country’s two largest export items. Khudoiberdiyev became the First Deputy of
the Presidential Guard.
Starting in the summer of 1996, there were special shuttle diplomacy meet-
ings between representatives of the Rakhmonov government and individual
UTO warlords, trying to convince them to switch sides and incorporate their
soldiers into the Tajik army. These individual negotiation sessions were replaced
with large working groups in the immediate run-up to the comprehensive peace
process, with meetings conducted in Khojand and Kulob.66
Just over one year after his first attempt to collapse the state, Khudoiberdiyev
raised the stakes with another uprising. Bolstered by foreign support from
Uzbekistan and promises of support from shadowy financiers, he declared the
establishment of the “Autonomous Defense Council of Central and Southern
Tajikistan in Kurgan-Tubbe and Hissor.” Few allies rallied to his cause this time.
Boimatov, in particular, let Khudoiberdiyev flail in the wind. An improvised
counter-coalition of rehabilitated rebel field commanders, many of whom had
been branded as “criminal insurgents” and “Islamic terrorists” only weeks
earlier, drove Khudoiberdiyev from the country in humiliation. Reports of
what happened next are contradictory, but a common story is that he fled to
Afghanistan and took refuge with the Uzbek warlord Abdul Rashid Dostum.
When Dostum was defeated in 1998, Khudoiberdiyev tried another uprising
in Tajikistan, this time in Khojand. It failed. He was finally forced to flee in
disgrace.67
In 1996, Khudoiberdiyev was pivotal and could extort the president. By
1998, this was no longer the case. His final abortive coup was put down quickly
and decisively by recently rehabilitated opposition warlords. This is wholly
consistent with the interchangeability assumption in Chapter 2. That warlords
in the initial ruling coalition feared that the integration of new armed groups
into the state would come at their expense also fits the logic of the model.68
But the main lesson that most observers drew from these abortive coups was
that Rakhmonov was a permanent fixture. Even at the height of the 1996
uprising, it is telling that neither Khudoiberdiyev nor Boimatov, nor any official

66 Interview conducted in Dushanbe, June 18, 2007.

67 It should also be noted that the 1996 uprising was not first time Khudoiberdiyev and his
army switched sides in the war. By my rough count, it was the third of six switchings. His
uneven political fortunes provide a lens through which to view the kaleidoscopic complexity of
Tajikistan’s warlord politics in the 1990s. Khudoiberdiyev served in the Soviet military until its
collapse, contracted his services to the elected postwar government (1991), joined the Popular
Front and played a key role in installing Rakhmonov as president (1992), withdrew from
Dushanbe politics and installed himself as a feudal lord in Kurgan-Tubbe (1993–1994), led an
Uzbek-backed revolt against the state (1995–1996), became a first deputy in the presidential
guard (1996), led a doomed revolt in 1997, reinvented himself as an Afghan “frontliner” in
1997–1998, and finally led his doomed insurgency in 1998. As of this writing, reasonable
people disagree about whether he is alive and living in Uzbekistan or dead.
68 Recall from the model: Because s is static, warlord payoffs are decreasing with k, meaning that
as the coalition size increases, every warlord within it does strictly worse.
Coup-Proofing 155

spokesperson for the regime in Tashkent, called for Rakhmonov’s resignation.

Instead, they demanded that certain hard line Kulobi warlords be dismissed and
that the district of Khatlon be divided into its two original halves, returning to a
framework for regional economic redistribution that was not weighted toward
Kulob. Even with no meaningful formal institutions and distributional politics
being decided at gunpoint – literally – by positioning himself as an intermediary
between Mosow and the Tajik state organs, Emomalii Rakhmonov had become
a permanent fixture in Tajik politics.
This power would only grow as Rakhmonov widened his governing
coalition during the peace process. Promises to redraw district boundaries were
forgotten. On June 27, 1997, Rakhmonov and the head of the UTO, Said
Abdullo Nuri, officially concluded a “General Agreement on the Establishment
of Peace and National Accord.”69 The executive branch of the government,
which included all state ministries, regional and local government departments,
and law enforcement bodies, pledged to incorporate UTO representatives on
the basis of a 30 percent quota.
Parallel to the high-level diplomatic negotiations was a process of warlord
incorporation negotiated between various Tajik warlords between 1995 and
1997. Opposition field commanders struck separate deals with government
negotiators or Russian military forces, allowing them to become state repre-
sentatives of their region, delivering votes and keeping order in exchange for a
share of the emerging state spoils. These agreements were reached piecemeal at
first, but once the state had reached a number of side-agreements and the UTO
warlords began to feel that the writing was on the wall, they began to come
forward to negotiate collectively in two separate, yet informal, roundtables
(the “Gharm protocol,” which recognized the central role of decentralized field
commanders in the forthcoming Peace Process) in the summer of 1996. The
rebels chose to agree to a settlement because the warlords who were fighting
on their behalf had abandoned them, having struck side deals to keep their drug
routes, rural fiefdoms, and positions in the lucrative state ministries.
In virtually all cases, these arrangements explicitly allowed warlords to keep
their armies intact. Commanders handed over almost none of their weapons,
or merely transferred them (with documentation) to “the government,” where
they were assigned to the weapons locker of the particular unit where they

69 See, for instance, Brenninkmeijer (1998) for a representative review of the International
treatment of the conflict up until the time of the 1997 Peace Process or Akiner (2001) for
what is certainly the most-cited general review of the politics surrounding the Peace Process.
For a magisterial and critical insider’s view of the implementation of the Tajik Peace Process,
see Nakaya (2009) and Heathershaw (2009). The centerpiece of the agreement was the creation
of a twenty-six-member Council on National Reconciliation, composed of thirteen opposition
and thirteen government representatives. This body was tasked with the practical details of
monitoring and upholding the ceasefire, dealing with refugees, amnesty, and key aspects of
local government and representation. See “Key Points in the 1997 General Agreement,” in
Abduallev and Barnes (2001).
156 Jesse Driscoll

were assigned. Most military commanders, regardless of standing, were offered

positions in the Ministry of the Interior MVD (Ministerstvo Vnuternnikh Del),
the Army, the Ministry of Defense MOD (Ministerstvo Oborony), the Ministry
of Emergency Situations MChS (Chrezvychainykh Situatsii), the Ministry of
State Security MB (Ministerstvo Bezopasnosti) or the State Committee on the
Protection of Borders KOGG (Komitet Oborony Gosudarstvennoi Granitsy). In
the aftermath of integration, the same wartime hierarchies persisted, with the
army in particular “operating much like a collection of militia groupings rather
than as a vertically organized militia structure.”70 The Peace Process made
certain field commanders’ de facto control of rural populations de jure, coupled
with informal understandings that opposition commanders would be allowed
to rule the mountainous highlands (specifically the Rasht and Tavildara Valleys,
and Gorno-Badakhshon), with no real interference from Dushanbe. Most of
the security guarantees of the peace accord were simply never implemented, or
implemented and then flagrantly reneged upon.71
The Peace Accords succeeded in their primary mission, which was to provide
a face-saving framework by which opposition militia commanders could be
incorporated into the state, and for warlords to legalize the holdings they had
acquired over the course of the civil war. Many field commanders transitioned
to the class of “businessmen,” having reappropriated Soviet state property
(apartments, land, factories, bazaars, cotton-fields and processing facilities,
and even shopping centers) for themselves and their families. Sometimes
these opportunities emerged out of ministry appointments, and sometimes
they were simply legalizing what had been seized during the war. Torjensen
and MacFarlane (2007) document a number of examples of these sorts of
arrangements. Hoji Akbar Turajonzoda, a religious figure who moved between
the military and political wings of the UTO, ascended to the post of Vice
Prime Minister. A few years later, it was reported that he acquired “a
cotton-processing plant in Vakhdat (Khatlon), the main shopping center in
Vakhdat (the Univermag), and two flats in downtown Dushanbe,” as well as
years of rents from monopoly contracts importing wheat from Kazakhstan.
Shurob Kasimov, a Popular Front Field Commander, owns a string of businesses
and properties, including the rights to many of the properties in the Varzob
district where embassy and NGO workers picnic in the spring. After Ghaffor
Mirzoyev fell from Rakhmonov’s favor in 2005, his assets were seized. It was
revealed that he owned of over 30 apartments in Dushanbe, the bank Olimp, a

70 Torjensen and MacFarlane (2007), 319.

71 It could be argued that the UN provided political cover for domestic purges of incorporated
militia members who officially joined the state in the framework of comprehensive peace.
When the UN Observer Mission criticized corrupt practices and intransigence on the part of
incorporated UTO militia members in 2001, Rakhmonov seized the initiative and fired anyone
suspected of disloyalty from the ranks of the Ministry of Defense and Ministry of the Interior.
Coup-Proofing 157

few factories, a major Dushanbe casino, and the Sultani Kabir car bazaar (the
largest in the country at the time).72
And what of the Islamic fundamentalist ideology that was so threatening to
conservatives in Tashkent and Moscow? For an overall summary response to
this important question, it is hard to improve on Matveeva (2009):

On a provincial level the Islamic orientation of seemingly irreconcilable field comman-

ders subsided substantially, even in Karategin Valley. Following the peace agreement the
former mujaheddin became directors of state farms and heads of enterprises, or obtained
appointments in the local authority. These new roles deprived them of the advantages
of being in opposition, made them share the burden of everyday management with
the secular authorities and forced them to act within the secular law. In this context
their ideological positions and behavior changed radically. It appears that the Islamic
movement was unable to sustain a mass following with the civil war was over. For
example, Mirzo Zioyev, a UTO commander (his family originated from Talvidara, but
Zioyev himself came from a resettlers’ stock in Vakhsh) transformed himself into an
Islamist warrior and cultivated a strict Wahhabi rule in the Tavildara district, enforcing
Islamic order by force of arms if necessary. The terrified population obeyed. Alcohol,
cigarettes, music, civic marriages and secular dress was prohibited. As soon as Zioyev
was gone and became a minister in the new government, the local people returned to
their normal practices. Across the former opposition areas, as the power of Islamist
commanders withered away after the civil war, the population returned to the ‘degree’
of Islamism it had practiced before; i.e., socially conservative, but not extreme.73

The 1998 parliamentary election might have provided the opportunity for
various factions who remained outside of Rakhmonov’s direct control to form
a coherent counterweight to Rakhmonov’s party. These parliamentary elections
were delayed, then cancelled, by presidential fiat. The 1999 presidential election
proceeded as scheduled, reifying Rakhmonov’s authority in a crude spectacle of
consent. By this point, virtually all armed groups in the country were implicated
in Rakhmonov’s favor economy in one way or another; none of them had any
interest in giving “the people” the opportunity to strangle their golden goose.
Pro-regime terrorism cast a long shadow over the election – a barely contested
exercise that included death threats, total media control by the president’s party,
and over a dozen publicized political assassinations of prominent candidates,
journalists, and financiers. The only contender, Davlat Usmon of the Islamic
Renaissance Party, was listed on the ballot only days before the vote. Usmon’s
son was kidnapped the day before ballots were cast. One of the most notable
and symbolic deaths in the midst of this election was that of the original founder
of the Popular Front, Safarali Kenjaev. Kenjaev was widely believed to have
been planning to run for the November presidential race. He was shot and
killed by unidentified gunmen on his front porch.

72 The Financial Times, August, 2005.

73 Matveeva (2009) 40.
158 Jesse Driscoll

By early in 2000, it was possible to observe that “a new coalition of warlords

from Kulob, Gharm, and the Pamirs has presided over a massive redistribution
of wealth … [and] secured dominant positions in rent-seeking activities such
as distributing foreign aid and drug trafficking.”74 As one would expect in
a relationship fundamentally based on armed extortion, dynamics between
Rakhmonov and the new warlord coalition were complex and contentious
throughout this period. Just like Shevardnadze in Georgia, the president was
constantly rotating the portfolios of important security ministers so that none
of them could grow too powerful. As the peace process unfolded and more
warlords joined the coalition, Rakhmonov gained flexibility to purge, promote,
and reshuffle to his liking. Consider the career of Yaqubjon Salimov – the
head of the Dushanbe mafia turned Minister of the Interior (1992–1994).
Salimov was dismissed in 1995 and replaced by Saidamir Zukhrov (himself a
former head of the Ministry of Security in Kulob) because Rakhmonov became
concerned with the extent of Salimov’s influence within the Dushanbe police
force and the amount of wealth he had acquired. It was amicable: Salimov
was appointed ambassador to Turkey, then reinstated in 1996 as Chair of
the Customs Committee. Just over a year later, Salimov saved Rakhmonov’s
life in Khojand, pushing the president out of the way when a grenade was
thrown at him. But Salimov’s decision to throw his support behind Mahmud
Khudoiberdiyev’s August 1997 Rebellion was a terrible miscalculation. After
the coup was put down by Kulobi warlords and incorporated UTO units,
Salimov was forced to flee Tajikistan – first to Saudi Arabia, then to Moscow.
He was arrested in February 2004, extradited to Tajikistan, and sentenced to
15 years in jail.
The fates of the three most prominent Kulobi warlords that spearheaded
the government assault on Khudoiberdiyev in 1997 are also illustrative of how
fragile coalition politics were during this period. Ghaffor Sedoi Mirzoyev, who
guaranteed security at the largest bazaars in Dushanbe, had been the head of the
Presidential Guard since 1995. Sukhrob Kasimov, a commander of the Special
Forces of Dushanbe Police in 1992–1994, was promoted to the head of the
Ministry of the Interior Rapid Deployment Brigade by Rakhmonov to “keep
Mirzoyev under control.”75 Sulton Kuvvatov, the former head of the Ministry
of the Interior’s Unit of Combatting Economic Crimes, had been head of the
Tax Committee since 1995. All were “Popular Front Men” who managed to
rise from relative obscurity to prominence during the consolidation phase. And
all three have since been purged. Kuvvatov was dismissed in 1998, stripped of
his Parliament seat in 2002 and placed under arrest for the vague charges of
“insulting Rakhmonov” and “inciting ethnic hatred” in 2005. Mirzoyev was
arrested and sentenced to life imprisonment in 2004. Kasimov was dismissed
in 2005 and quickly disappeared.

74 See Nourzhanov (2005), 124.

75 Nourzhanov (2005), 120.
Coup-Proofing 159

Pamiri warlords, in general, were capable of making good on their bargains

with the state where their Kulobi and Hissori counterparts were not. The
impassable mountains make self-governing arrangements a practical necessity,
which reduces a warlord’s fear of the regime reneging down the road. On the
regime side, it is unlikely that an uprising by a coalition of police chiefs in
distant mountain villages poses the same existential threat as a coup by military
generals in the capital city. Pamiris are also deeply resented by many ethnic
Tajiks in Dushanbe, who are quick to fall back on old stereotypes of the insular
religious group as wealthy elitists. The events of the early 1990s suggest that
it would be very difficult for this group to seize national-level political power.
The overland narcotics trade route from Afghanistan provides strong incentives
for cooperation and opportunities for personal enrichment. The risk that the
profitable drug transit corridor would be disrupted by violence outweighs any
possible gain to Rakhmonov that could come from targeting a particular field
commander’s career. There is little in Badakhshon worth taxing or harvesting
other than this drug trade or hydropower rents. In both cases, security is
guaranteed by local agents with arrangements brokered in Dushanbe.
How did an oligopoly of violence emerge in Tajikistan? Warlords willing to
pledge personal loyalty to Rakhmonov and commit their troops to supporting
the Tajik government were integrated into the state. There could be no long
term political guarantees, as the warlords themselves were the enforcing
agents for contracts. Field commanders observed Rakhmonov as he carefully
and methodically applied divide and rule logic to this coalition, occasionally
inducing warlords to turn on each other in exchange for a larger share of state
spoils. This risk was weighed against alternatives, which changed substantially
over the fifteen year consolidation period. In general, warlords were offered the
same kind of deal regardless of their characteristics: they were given amnesty,
permitted to make large sums of money for a few years, and were then squeezed
out of the governing coalition. The relative winners in the shell game were the
Pamiris of Gorno-Badakhshon and actors with allies in the KGB. The relative
losers were the Hissoris, who could not shed their association with Uzbekistan.

the lottery: evidence of warlord interchangeability

The differences that emerge between the consolidation projects of Shevard-
nadze and Rakhmonov are consistent with what the model would predict if
one country had a high stability threshold and the other had a much lower
threshold (recall Figure 5.1). In Georgia, where Russia was interested in
undermining the government and backing a coup, Shevardnadze needed to
keep his friends close and his enemies closer (maneuvering while constrained
by an extremely high s). Shevardnadze had enormous success at incorporating
Georgian warlords quickly, but was then unable to steer the state apparatus
once they were all inside his Ministry of the Interior. In Tajikistan, by contrast,
Russia lent its armed forces and intelligence to support the Rakhmonov regime
160 Jesse Driscoll

table 5.1. Former Paramilitaries in the Georgian Ministry of the

Interior

Faction Ministry Affiliation Ministry Affiliation

1992–2003 (%) 2004–2007 (%)
Mkhedrioni (n = 55) 89 51
National Guard (n = 18) 67 44
Forest Brotherhood (n = 6) 67 50
White Eagle (n = 8) 88 75
Zviadists (n = 13) 77 69
Adjaran Police Force (n = 5) 0 100

against domestic challengers in a straightforward effort to keep Islamists or

clients of Uzbekistan from coming to power. This allowed Rakhmonov to pick
and choose among various local warlord allies (maneuvering with a much
lower s). The high stakes in Tajikistan also led to fierce friction between the
members of the insider coalition and the other warlords, as well as between
these members and the president himself, over the peace process and efforts to
incorporate additional warlords.
In Georgia, one can observe the rapid incorporation of most of the warlords
into the state apparatus. Once Eduard Shevardnadze was the face of the
franchise, near-universal buy in to the state-building project was the trend.
Much of Shevardnadze’s institution building was aimed at reassuring as many
people as possible that he was a servant of the state and the nation, subordinate
to the interests of the people. The Abkhaz, Adjar, and South Ossetian warlords
were the exceptions, not the rule. Though the junta’s State Council endured
for months, in practice it did not last long as a consensus body. The statute
that created the State Council allowed new members to be admitted by
a two–thirds vote, and by May 1992 it had evolved into something like
a “pseudo-parliament” of warlords, fragmented political party elites, and
representatives of ethnic enclaves.76 Shevardnadze assumed the pivotal role of
coalition formation, acting both from behind the scenes and from the bully
pulpit, taking responsibility for forging consensus and voting super-majorities
from different interest groups, civil society figureheads, and warlords. He did
all of this without a hint of subterfuge.
As a result, official membership in the Ministry of the Interior – headed
symbolically by Shevardnadze himself for a time – swelled to nearly 30,000
by 1995, with at least as many unofficial affiliates. Much of the personnel for
this new Ministry of the Interior – so often cited as the key to Shevardnadze’s
centralization of power – came from the ranks of various nationalist militias.

76 Wheatley (2005), 70.

Coup-Proofing 161

Table 5.1 reports the percentages of interviewed militia members who reported
Ministry of Interior affiliation during this time.77 The trend toward joining
the Ministry of the Interior held for urban and rural members, for members
whose time in the militia preceded the Gamsakhurdia coup (e.g., in the
National Guard), for groups that fought against the state (e.g., the Zviadists),
and for opportunistic joiners (e.g., in self-identified “Mkhedrioni” units in
Tbilisi during the Time of Troubles). The exception is the Adjarans, who had
seceded in the 1990s with the help of Russian aid and were part of their own
parallel police force under the direct command of Aslan Abashidze. Table
5.1 also reveals that despite the Rose Revolution and recent anti-corruption
campaigns in the Georgian state, the institutional footprint of the Shevardnadze
compromises with nationalist militias persists in the composition of the modern
police force.78
The civilianization of militia structures slowly eroded the power base
of those militia commanders who could not learn new tricks. Sociopathic
behaviors, radical political slogans, and anarchism were shed by many
members, given the chance at a secure job and a pension in a patronage
hierarchy. A labyrinth of corruption emerged in Shevardnadze’s Ministry of
the Interior. In the words of Mark Mullen, the former Georgian director of the
National Democratic Institute and Transparency International:

“A state within a state is what it was … They collect[ed] their own taxes, elect[ed] and
promot[ed] their own officials through a process that nobody saw, and [weren’t]
accountable to anyone for anything … it was the same Mkhedrioni, the same
guys … everybody knew who they were, what they were capable of. That was the
point. That was their power.”79

Co-opting challengers by bringing them into the Ministry of the Interior

ultimately led to the creation of a bloated, factionalized, and inept police force
– but one whose members saw the president at the head of a hierarchy. The
empirical record suggests that, by 2003, many in the security services believed
they would keep their jobs even if the president was forced to step down from
power in the face of street protests. The Georgian polity had a centralized police
apparatus a decade after Sheverdnadze ascended to power, with substantial
authority vested in the head of state. This is where our narrative will pick up
at the end of this chapter.

77 Responses add up to more than eight-four interviews – an artifact of side-switching between

groups.
78 Fairbanks (2012) forcefully articulated this argument just days before the election that removed
Saakashvili from power.
79 Interview conducted in Tbilisi on March 10, 2006. This runs in tandem with the descriptions
of Jonathan Wheatley (2005, 78–83), Ghia Nodia, and many others who misleadingly describe
Ministry of Interior “crackdowns” against criminals and isolated Mkhedrioni leaders as if the
two groups were distinct.
162 Jesse Driscoll

In Tajikistan, consolidation was slow and grinding. An original dataset

of warlord biographies allows a statistical assessment of how much time
passed before an incorporated warlord was purged from the ruling coalition.
I use statistical regression to compare warlords’ political fortunes in the
consolidating Tajik state. Using Russian- and Tajik-language newspaper
archives and embassy reports, a team of Tajik researchers employed by the
Small Arms Survey compiled a list of the most prominent “field commanders.”80
With the aid of two research teams based in Bishkek and Dushanbe, I revisited
the Small Arms Survey’s secondary source materials and identified an additional
twenty-seven field commanders, for a total of ninety-seven observations.81
Between 1992 and 1997, fifty-seven of these ninety-seven Tajik warlords
accepted positions in the new bureaucracy, subordinating their armies to
Emomalii Rakhmonov’s political hierarchy.82 About half of these warlords
represented the core of the Popular Front coalition that seized the capital city
in December 1992. Another major wave of warlord integration came with
the 1997 Peace Accords, which guaranteed 30 percent of the positions in the
“Power Ministries” (State Security, Defense, and Interior) to members of the
UTO. Other field commanders integrated into the state apparatus piecemeal.
The column on the far right of Table 5.2 displays the number of years,
rounded up, that a warlord served alongside his militia, after he had been
incorporated into the state in an official capacity but allowed to retain his
men. A few of the warlords departed voluntarily and retired as militia captains,
removing themselves from politics and withdrawing from public life. Others
were transferred to a non-security ministry or fired outright, which meant
they effectively lost their private army. These purged warlords were often
killed. When they were not, they were usually imprisoned, exiled, or simply
disappeared once it was clear that they had lost their political protection.83
The final column in Table 5.2 reveals that warlords from some regions fared
systematically better than others. A warlord’s home region was a good predictor
of which side they fought on in the civil war. Of the fifty-seven warlords
who joined the state through the 1992–1997 period, by December 2006 only

80 Torjensen (2005).
81 For a discussion of coding rules and criteria, see Appendix C.
82 When a warlord did not join the state it usually meant that he was killed in the course of
civil war violence, though a small number of non-joiners fled to Afghanistan, disappeared
into the criminal underworld, or simply refused to disarm or compromise. Even after 1997,
many warlords continued to control territory in which neither the government nor United
Nations observers could enter without the permission of these warlords and armed escort
of their subordinates. The most prominent of these locations were Darband (controlled by
Mullo Abdullo), Tavildara (home of the Minister of Emergency Situations Mirzo Zioyev), and
even the Leninsky District, about twenty kilometers from downtown Dushanbe (controlled by
Rakhmon “Hitler” Sanginov).
83 I estimate that in roughly a quarter of the cases, the warlord’s exit from state security services
involved his death, but there are so many cases of “disappearances” that this generalization is
mostly impressionistic.
Coup-Proofing 163

table 5.2. Tajik Warlord Summary Statistics: Average Time Until Purge Varies by
Region

Region Warlords Warlords Warlords Joined Joined Average Years

of Birth (Total) (UTO) (PFT) 1992–1993 1996–1997 Until Purge
Kulob 24 5 18 13 3 6.7
Gharm 18 13 5 3 8 6.2
Hissor/Uzbek 10 1 9 8 0 2.5
Badakhshon 25 21 0 0 14 7.6
Other 20 3 8 7 1 7.3

sixteen remained. In the cabinet reshuffle of December 2006, the final remaining
high-level UTO warlord in the dataset, the Minister of Emergency Situations
Mirzo Ziyoev, was discharged. Anecdotally, it is well established that every
prominent warlord in the Popular Front who installed Rakhmonov has either
retired from politics, fled the country, landed in jail, or died under mysterious
circumstances.
Appendix C provides an expanded statistical analysis of these data, but the
main results can be easily summarized. Warlords from the president’s home
region of Kulob were purged at a rate that is statistically indistinguishable
from the average. Subsections of the winning and losing sides of the civil
war tended to be more likely to survive, but not in a manner that was
predictable in advance. Indeed, whether one believes that the civil war was
defined by competing ideologies (“Islam vs. Rump Communists”) or regional
politics (“Kulob, Hissor, and Khojand vs. Gharm and GBAO”), in the
aftermath of the war, the coalition formation process yielded strange and
unpredictable bedfellows, political alliances that transcended easily-codeable
divisions. Strong warlords, defined as those who controlled large armies, were
ultimately no more likely to emerge as members of the ruling coalition than
weak warlords who controlled small armies. The inference that I draw from
these patterns is that intracoalitional commitment problems were at least as
serious as the “master” commitment problem between the winners and losers
in the civil war. In retrospect, it is strange to remember that the Hissori/Uzbeks
were the ones who looked like they had “won” the war back in 1992. As
these statistical patterns show, they proved the most vulnerable warlords in the
coalition formation process. By contrast, the most insulated warlords – Pamiris
from GBAO – were clearly on the wrong side of the war in the early 1990s. I
see this as evidence that the warlords could not predict themselves who would
be a winner or a loser over the long run, once the president was installed.
What happened? The short answer is that all of these people were
interchangeably awful, and recognized as such by civil society. Despite Russian
ethnocentric stereotypes, Dushanbe and Tbilisi are cities that are full of
164 Jesse Driscoll

thinking, educated, civilized people. Most of them understood that it was

morally unnatural for rapists and murderers to hold positions with social
power, respect, legal immunity, and institutionalized protection. Most people
did not want criminal warlords managing property disputes. They understood
exactly how outrageous a hand fortune had dealt them when they saw civil war
commanders rising to positions in the state and handing out patronage. Tempo-
rary exceptions to this basic moral rule, which were justified by the temporary
anarchy of the early 1990s, did not actually void the rule in the minds of most
Georgians and Tajiks. So once the worst periods of violence had passed and
normalcy returned, first the perceived Uzbek clients were pushed out, then the
obvious sociopaths were trimmed, then the patrons who hired the sociopaths,
and finally the leftovers. Rakhmonov simply rode successive waves of popular
opinion and doled out what these men already “had coming” to them.
The long answer is that over the course of the fifteen-year observation period,
membership of the ruling coalition changed substantially. In the early 1990s,
friction between the new Tajik and Uzbek governments manifested as friction
between Kulobis and Hissoris in coalition politics. The Hissoris were purged.
In the late 1990s and early 2000s, the Pamiri Opposition commanders, who
had reinvented themselves as police officers in rural Badakhshon, were kept
in the state as subsequent rounds of purges removed former UTO combatants
from Gharm and elsewhere.
The story of the “de-Uzbekification” of the Tajik ruling structure has already
been previewed in this book. The district of Hissor, to the west of Dushanbe,
produced a number of prominent warlords in the Tajik Civil War – most
famously Safarali Kenjayev, the original founder of the Popular Front, and
Ibod Boimatov, who led a 1996 mutiny against Rakhmonov’s peace process and
brought the state to a standstill.84 The Soviets also relocated several semino-
madic Uzbek tribes (Lakai, Kungrat, and Durman) to the agricultural lowlands
of southern Tajikistan near Kurgon-Tubbe, creating local frictions over access
to scarce land and water. The most prominent warlord to emerge from this
region was Mahmud Khudoiberdiyev, the commander of the First Brigade at
Kurgon-Tubbe. During the heady days of Tajik nationalism in the late Soviet
period – when it became clear that Russian-enforced cosmopolitanism was
breaking down and no one could predict what would replace it – these Uzbek
groups were natural recruits for counterrevolutionary militias. Conservative
elites in Tashkent, who feared that a new regime in Tajikistan might invite the
conflict in northward or revisit territorial claims to Samarkahd and Bukhara,
were interested in helping local co-ethnics gain influence and establish a
protectorate. Yet once the Popular Front established itself in the capital city, the

84 The district of Hissor is identified by Olivier Roy as a textbook example of the irrationality of
the Soviet Border system – it is a district with approximately 60 percent Uzbek speakers facing
directly across the border with the region of Sukhan Darya, a majority Tajik-speaking enclave
in Uzbekistan (68–69).
Coup-Proofing 165

dominant Kulobi faction saw these field commanders as a potential fifth column
for Uzbek influence in the consolidating state, or as likely allies for a power
grab by the northern Khojandi faction. Hissori warlords found themselves early
targets for purges and persecution. To revisit the metaphor from Chapter 2, they
had played the state-formation lottery and lost.
Ten warlords in my data are identified as being born in one of the
Uzbek-dominated southern regions of the country, and nine of them were
affiliated with – indeed were founding members of – the Popular Front. These
Uzbek warlords formed nearly a quarter of the warlords in the initial ruling
coalition. The Uzbek government made no secret of the fact that they saw
their weak neighbor Tajikistan as a natural protectorate, and these warlords
thought that their affiliation with a powerful foreign patron would act as a
security guarantee. The strategy backfired. After 1995, these warlords were
singled out for elimination.85 The longest surviving warlord from this group,
Safarali Kenjayev, was murdered outside his home (along with his bodyguard
and driver) on March 30, 1999. With a seven-year tenure in the state, he was
still purged faster than the average warlord in these data, despite the fact that
he literally founded the political movement that captured the state apparatus.
Kenjayev miscalculated, and Kenjayev was liquidated.
Contrast this narrative with the fates of the other significant ethnic minority
group, the Pamiris. The GBAO makes up 45 percent of the land area of the
Tajik state, but only contains 3 to 4 percent of the population. This inhospitable
Mountain range is the home of the Pamiris, a distinct ethnic group distinguished
by language (Shugni), religious sect (Ismaili), and physical appearance. After
the Popular Front seized the capital city in December 1992, the militias that
had supported the previous government fell back to the impenetrable Pamiri
mountains, while in the capital hundreds of Pamiris were targeted for ethnic
cleansing, with their homes and property expropriated by the victorious Kulobi
militias. Shortly thereafter, the autonomy of Gorno-Badakhshon was rescinded
in Dushanbe. Representatives in the regional capital of Khorog responded by
declaring independence. In the spring of 1993, a compromise was reached.
GBAO agreed to recognize the suzerianty of Dushanbe in exchange for a return
to autonomous status, a promise to not disrupt the activities of the Aga Kahn
(who was delivering aid to the starving populations), and a set of soft promises
that the region would be allowed to manage its own security. This bargain
generally held through the civil war, and its broad contours persist as of this
writing – though it is not advised to pay too much attention to the particulars of
the security arrangements. Formal autonomous status is one of the few formal
protections of the civil war settlement that can said to have any teeth. Of the
twenty-five Badakhshoni-born warlords identified in the data, none of them
served in the Popular Front. Yet returning to Table 5.2, these warlords were

85 The estimated median and mean survival times for Uzbek warlords are 2.4 and 2.5 years,
contrasted with 8.3 and 8.4 for other warlords.
166 Jesse Driscoll

clearly capable political survivors.86 In the consolidation lottery, they were

relative winners.
But all of this is an ex post rationalization of trends that could not have been
forecast by actors at the time they made their decision to back Rakhmonov or
keep fighting. Many warlords could not predict coalition winners and losers.
They were playing the odds in a high-stakes consolidation game.

a coda for v∗ : aid and politics after coup-proofing

The visualizations in Figure 5.2 demonstrate the degree to which the economies
and state budgets of Georgia and Tajikistan have been “floated” by foreign
aid (v∗ ). Although there were important differences between Shevardnadze’s
and Rakhmonov’s governance styles, the countries have also been socialized by
different geopolitical audiences. The content of targeted aid programs varied
by country.87 Tbilisi continues to pivot between Moscow and Washington, and
Western aid to Georgia aimed at “strengthening civil society” and “democracy
promotion” has geopolitical overtones. Dushanbe continues to be judged by
both great powers on the same criteria: its ability to “hold the line” against
radical Islam. Both states have grown into these roles.
Georgia is probably the best Eurasian example of a country where democ-
racy promotion activities by the United States and Western Europe can
be said to have “worked.” With 55 percent of discretionary spending in
the 2003 state budget filled by Western donors and a constant stream of
rhetorical commitments to democratization, Shevardnadze could not easily
interfere with the actions of foreign democracy-promotion NGOs in his own

86 The estimated median and mean survival times for former field commanders from
Gorno-Badakhshon are 8.8 and 10.1 years, contrasted with 5.1 and 6.4 for other warlords.
Half (8/16) of the warlords who were still members of the state security apparatus in December
2006 were Pamiris – most of whom had, one point or another, taken up arms against
Rakhmonov’s regime.
87 Whether and how foreign aid works to transform the institutions of target societies is the
subject of a vast literature. For a useful review of this literature’s findings, see Wright and
Winters (2010). For a good review of how aid applies to fragile states, see Girod and Tobin
(2015) and Girod (2012). For more on how political debates over the effectiveness of aid have
coevolved with best practices as understood by the discipline of economics, see Easterly (2001)
and Helpman (2004). The central difficulty for analysts is that institutions in societies adapt
dynamically to expectations of foreign assistance: governments show donor states what the
donor states expect to see in order to receive more aid. As Moscow’s central planners came
to appreciate: local actors become quickly aware of the incentives created, and in a general
equilibrium framework they can act strategically. At some risk of cultural stereotyping, one
of the legacies of the Soviet experience in Central Asia and the Caucasus is a mentality of
“learned helplessness” from a century of colonial condescension. The myth of the Potemkin
village captures an important cultural response to totalitarian rule: peripheral republics became
extremely sophisticated about figuring out what metrics the social scientists in Moscow were
using to judge “success” and producing those metrics. They are practiced at telling program
evaluators what they need to hear and generating data to confirm theoretical priors.
Coup-Proofing 167

(a)
Georgia
0.18
Tajikistan
Non-study average,
0.16 no Russia

0.14

0.12
Ratio of ODA to GDP

.01

0.08

0.06

0.04

0.02

0
1991 1996 2001 2006
(b)
2
Ratio of ODA to Government Expenditures (in current USD)

Georgia
1.8 Tajikistan
Non-study average,
1.6 no Russia

1.4

1.2

0.8

0.6

0.4

0.2

0
1991 1996 2001 2006

figure 5.2. Georgia and Tajikistan: wards of international charity.

168 Jesse Driscoll

country. Georgian citizens themselves, taking their president at his word

when he invoked the promises of democracy and legitimacy, were ultimately
unwilling to tolerate the regime’s perceived manipulation of elections in 2003.
Civil society groups found themselves aided substantially by foreign-funded
NGOs, including active participation by embassy employees and American
and European expatriates. The gridlocked police apparatus did not respond to
orders to dismiss crowds of demonstrators.88 Peaceful protests swept Eduard
Shevardnadze and his regime from power in November 2003 – a set of events
that have come to be known as the Rose Revolution.89
In the years since the Rose Revolution observers have witnessed a reen-
ergized presidency governing over what even cynics agree is a substantially
more transparent version of the same superpresidential state structure. The new
president, Mikheil Saakashvili, rode the nationalist euphoria to make a few
institutional changes that earned him credentials as a reformer – notably firing
the entire 60,000-person “traffic police,” which made it possible for citizens to
drive around the streets of Tbilisi without constantly stopping to pay bribes.
He dramatically simplified the tax code. Saakashvili dominated the political
landscape as completely as his predecessor through a similar mix of informal
networks and formal institutions, particularly the judiciary and Parliament.90
Shortly after the change in government, the breakaway republic of Adjara was

88 The historical record on this point remains muddled, but the bulk of the evidence supports
Fairbanks (2004) account, that Shevardnadze gave orders to disperse the demonstraters which
were not carried out (117). I strongly suspect there was an uncomfortable psychological
resonance with the 1989 attack on Tbilisi citizens which initiated the cycle of violence in
late-Soviet Georgia.
89 By far the most comprehensive and responsible review of the structural and political conditions
that created the Rose Revolution, both in terms of strategy and tactics of all the major actors,
is found in the work of Jonathan Wheatley (2005), chapters 6 and 7. For useful extension and
descriptions of the event itself, see Fairbanks (2004), King (2004b), Miller (2004), Stephen
Jones (2006), and Papava (2006). Mitchell (2004) reports that Shevardnadze appealed to Aslan
Abashidze to send a goon-squad from Adjara to organize a counterdemonstration in front of
the Parliament – to no effect (344). Protesters eventually stormed the (ungarded) doors of
parliament. As he fled, Shevardnadze apparently attempted to declare martial law but he was
informally abandoned by his security forces on November 22 and then formally repudiated
by them on November 23. He was then forced to resign. One interpretation of these events,
offered by Lincoln Mitchell, is that Shevardnadze “became aware that he no longer controlled
the military and security forces. Bloodshed was avoided largely because the president was too
politically weak to command it.” 348. The facts are also consistent with an alternative theory
of which I am fond: That after the death of his wife and a decade of thankless labor, the old
man was tired. The determined opposition in the Rose Revolution forced his hand, leaving
him the relatively easy choice to leave office as a temporarily disgraced hero for historians to
rehabilitate, or as tinpot despot who fired into crowds of his own citizens.
90 After the 2008 Parliamentary Election was resolved, Saakashvili’s party, the United National
Movement, controlled 119 seats in the 150-member Parliament, providing a comfortable
cushion of safety above the 100 votes that would be necessary to change the constitution.
The second largest faction – the United Opposition, consisting of eight small parties and a few
individual politicians – has only seventeen seats.
Coup-Proofing 169

reincorporated into Georgia proper. The Rose Revolution had the effect of
rehabilitating the regime in the imaginations of many Western defense analysts,
who rediscovered their optimism for the regime and doubled-down on their
aid packages. Georgia’s defense budget saw an unbelievable 50-fold increase
between 2002 and 2007. To demonstrate support for Georgia’s government
in the opening phases of its disastrous war with Russia in 2008 over disputed
secessionist territories, the United States pledged $1 billion in reconstruction
and development aid.
Under Saakashvili, NATO membership became a political totem for Geor-
gian citizens. While continuing to mouth the same promises of peaceful
compromise and negotiation in English, in Georgian the tone of discourse
changed. Talk of “internationalizing the conflict” – that is, introducing U.S.
and European mediators into the conversation, backed by a muscular military
presence – was thought to be the best route to a permanent settlement on
Georgia’s terms. Georgian troops were quick to volunteer for the coalition of
the willing in Iraq, and the Georgian Train and Equip Program (GTEP) was
extended to create pockets of professional leadership within their Ministry of
Defense. But the events of 2008 brought to the fore just how easy it would be
for an accidental war between NATO and Russia to break out over a forgettable
flashpoint in the South Caucasus.
As we approach the two-decades mark from the end of the “Time of
Troubles,” the fundamental commitment problem that fueled the conflict in
the first place is still present in Georgian society. In Georgia today, the events of
the early 1990s tend to be recounted with a selective memory that borders on
collective amnesia. In a nationally representative survey designed by the author
and implemented through the Caucasus Research Resource Centers in Georgia
in fall 2008, just after the war, an overwhelming majority of Georgian citizens
prioritized reclaiming control of the “lost” territory of Abkhazia and South
Ossetia over all other political issues, including good relations with the West
(prioritized by only 7.1 percent of respondents) or relations with Russia. More
than three-quarters of the respondents would reject any peace settlement that
did not include a “right of return” for displaced Georgians.91 Two generations

91 It is the official position of the Saakashvili government that reintegrating Abkhazia would
come only with provisions of autonomy. Yet, given the broad, deep trends in Georgian public
opinion on minority issues, the Abkhaz in particular can be perhaps forgiven for being skeptical
of claims that they would be treated magnanimously by Georgians if the frozen conflict were
resolved in Georgia’s favor. When Georgian citizens were asked a variety of questions related
to the practical meaning of “autonomy” for the Abkhaz, 71 percent of respondents believed
that “Georgians should be able to purchase whatever property they wanted regardless of the
wishes of the provisional Abkhaz government,” 61 percent of respondents believed “Abkhaz
and South Ossetians should be denied the right to serve in the Georgian military, police, or
border guards,” and 78 percent of respondents believed “schools and universities in Abkhazia
should not be allowed to teach classes in only Russian or Abkhaz.” These data are evidence of
what most Georgia watchers have identified through anecdotes for years: a deeply immature
170 Jesse Driscoll

of Georgian politicians have promised that ties with the West would help the
Georgians reclaim territory, with the unspoken implication that this territory
was stolen by traitors at the behest of Russia.
What Russia’s policies have incentivized is an equilibrium in which national-
ist Georgian parties – which no longer have paramilitary wings, but one senses
that they could be recruited at any time – hold pivotal coalition roles, and
the West provides liquidity (v∗ ) in exchange for basing rights, a non-Russian
oil pipeline transit route to get Caspian oil to Europe, and the satisfaction
of helping to create a pro-Western political environment in a traditionally
Russian-speaking region. From the perspective of many in Moscow, the
failure to reach a lasting diplomatic settlement to these longstanding territorial
disputes in Georgia has had the unanticipated – but tangible – benefit of halting
NATO expansion to the South Caucasus. Every time the frozen conflicts thaw,
European and North American observers are reminded that it is probably not
in the national interest of any NATO member state to be obligated by treaty
to consider defending distant Georgia against Russian “invaders.” As the 2008
war between Georgia and Russia clarified, Georgia has no hope of reclaiming
these statelets by force. Furthermore, its desire to join NATO, participate as
a full member in European development and security dialogues, and present
itself as a fully functioning member of the community of “normal” states will
continue to be held hostage.
The presidency of Saakashvili was criticized as failing to live up to the
promises of the Rose Revolution. The incumbent president and his party
enjoyed a number of structural benefits, including favorable media coverage
and free time on state television networks. Critics charged that Saakashvili’s
party, the United National Movement, used dirty tricks to monopolize political
power. The incumbent party benefitted from an inherited Soviet culture of
bureaucratic compliance, and a broader cultural expectation that private
behaviors can and will be monitored and reported to authorities.92 Local
bureaucrats – particularly in poor and rural areas – know that they are expected
to deliver their district to the ruling party in order to prove their loyalty and
keep their jobs. In a pre-election survey conducted in the lead-up to the 2008
Parliamentary Election, 32 percent of voters reported that people had to “vote
a certain way to keep their jobs.” An additional 34 percent of voters admitted
that this happens “sometimes.”93

and self-referential rhetoric has been allowed to fester in the Georgian political discourse. The
survey was nationally representative, stratified by region, with N = 2,700 respondents. For
additional information and open-use documentation contact the Caucasus Research Resource
Centers.
92 See Harris (2004) and Paxson (2005).
93 About a quarter (23 percent) of survey respondents in the average study precinct reported that
these beliefs were widespread, with a standard deviation of 22 (e.g., in some precincts no one
reported this fear and in others it was reported by a plurality).
Coup-Proofing 171

For all the disappointments of the post–Rose Revolution era, however,

Saakashvili’s decision to concede electoral defeat to Bidzina Ivanishvili’s
Georgian Dream party in 2012 bucks a powerful regional trend. His is one
of the only examples of a post-Soviet leader in Central Asia or the Caucasus
losing an election and voluntarily stepping down from power. The experiences
of Georgia in the last decade have me wondering whether the Soviet experience
may have created favorable conditions for the transition to an “open access”
political order.94 Time will tell. For all the conspiracy theories of rampant
corruption and offshore accounts that motivated angry protesters during the
Rose Revolution, Eduard Shevardnadze passed away in July of 2014 in a small
house on the outskirts of Tbilisi. As historians debate his legacy, it should be
remembered that many contingent outcomes – a peaceful regime transition,
progress toward democratization, and a Western-oriented foreign policy – are
all, at least in some small part, a result of his decisions.
A peaceful regime transition is much more remote in Tajikistan. The totality
of power consolidation by Rakhmonov became transparent in 2003 when
the government proposed fifty-six amendments to the Constitution. Hidden
among the list was a clause to extend the current presidential term to the
maximum of two seven-year periods, making it possible for Rakhmonov to
stay in power until 2020.95 In 2006, the final high-profile UTO representative,
Mirzo Ziyoev, was discharged. In 2007 the president dropped his surname’s
Russified suffix and demanded to be named in all official documents by
his purer Persian name “Rakhmon.” Virtually every prominent warlord
in the Popular Front who installed him has either retired from politics,
fled the country, landed in jail, or died under mysterious circumstances.
The warlords who survive in the Tajik state are mostly co-opted Pamiris,
managing neighborhood politics in GBAO. Russia underwrites the security
of the whole arrangement while keeping plausible deniability over any of the
particulars.
Tajikistan’s geographical position and high mountain ranges made it a
natural corridor for the Afghan narcotics trade. The rents of the drug trade
came from guaranteeing low-risk transit from its southern border to the
northern border on the way to the lucrative Western European market. This
meant control of key roads and border checkpoints, and the formal institutions
of the state – the military, police, and border guards – had every incentive to
make sure that particular trucks were not harassed at road blocks or stopped
for not having the right papers. Though in principle this suggests myriad

94 North, Wallis, and Weingast (2009), 148–154. The authors identify three “doorstep
conditions” for a transition to an “open access” society, while being very clear that they do not
have a predictive model explaining when exactly a country will transition from a closed to an
open access society: “DC#1: Rule of law for elites. DC#2: Perpetually lived organizations in
the public and private spheres. DC#3: Consolidated control of the military.” 151.
95 The government claimed 96 percent voter turnout and 93 percent approval of the amendment.
172 Jesse Driscoll

opportunities for entrepreneurial youth to try to smuggle heroin independently,

in practice any sophisticated international syndicate capable of buying heroin
by the truckload probably wants to deal with a local partner in the Ministry of
the Interior.96
Then, practically overnight, NATO military forces were in Afghanistan.
For a time, a great deal more Western attention was devoted to the region.
Was this increase in security assistance responsible for keeping Afghanistan’s
contagion of improvised explosive devices (IEDs) from spreading north to infect
Tajikistan? Or was Tajikistan inoculated against this sort of violent extremism
even before the influx of United States Agency of International Development
(USAID) and European security aid? One cannot prove a negative, and many
different theories can be laid atop an observational dataset of zeroes. What
can be said, with confidence, is that the new security assistance strengthened
Rakhmon’s hand.
Tajikistan occupies the same place in the imagination of most American
and Russian diplomats: a hardship post at the edge of the map. If not for the
small border with China on the far side of the Pamiri mountains, it would
be landlocked by other landlocked countries. Development aid continues to
be frustrated by a “culture of corruption.”97 What is important to the great
powers, in this context, is that a Salafi Jihadist subculture is not permitted to
thrive. Certain Tajiks in the security infrastructure are digging in for a long
war.98 Most Tajiks have difficult but predictable lives. Pockets of extreme
poverty and structural violence are in plain sight for anyone who cares to
observe. But there is no war today, and probably there will be no war tomorrow.
For that, most Tajiks are grateful. These observations are unlikely to become
obsolete any time soon. Change comes very slowly in Tajikistan.

96 At the time of this writing, the website of the U.S. embassy proudly declares that it helped
provide the Tajik government with a $28 million state-of-the-art border checkpoint facility for
the Tajik Customs Service.
97 Mirzoev (2002) reproduces survey data revealing nearly universal acknowledgment by Tajik
citizens that the state is highly corrupt (358, 360), and indeed that basic life is impossible for a
household – regardless of income – without paying bribes for virtually everything (365, 375).
98 The strategies and technologies that will be used to cauterize future threats are well understood.
The cultural response for many powerless people, in my limited experience as an ethnographer,
is to retreat from reality into imagination and conspiracy. Conspiracy theories allow people
to find meaning, even in silences. Some of my Tajik informants never talked on the phone.
Nothing I could say would convince these men – who suspected my motives, and who had
been raised on a vision of the CIA gleaned from some mix of Alias, The Dark Knight, and
their own experiences living in an Orwellian police state – that there were not “ghosts in the
phones.” But this paranoia does not capture the daily troubles of most Tajik citizens, or find
resonance with their lived social reality.
6

Implications

The processes described in this book are hard to square with the picture of peace
building that is typically put forward by optimistic liberal interventionists.
The “root causes” of the wars were in no way resolved. Conflict identities
between factions were not transformed in any fundamental way. The winners
in the consolidation process did not demobilize so much as they became the
state. Soldiers watched the daily drama of the consolidation process anxiously,
hoping that they were backing the right horse. Many switched factions based on
perceptions of whose star was on the ascent. Aid and investment increased the
value of the “prize” being fought over in the streets, creating perverse incentives
for militia expansion. A ruling coalition formed out of the violent sorting
through processes of attrition on the streets. Losers in this process demobilized
when they could no longer improve their life chances through violence, but
not because there was any sort of security guarantee making them feel safe.
Presidents gained power not by legitimizing or strengthening the state, but by
subtly influencing the distributional politics associated with its cannibalization.
All of this went on under the noses of foreign observers and Russian military
peacemakers.
The post-socialist presidents of Georgia and Tajikistan inherited situations
of intense institutional weakness. Paramilitary warlords and private armies
emerged as power brokers in the chaotic aftermath of state collapse, and
the new presidents were forced to incorporate armed factions and known
criminals into ruling coalitions. These warlords and their men proceeded
to pillage state institutions from the inside, rendering the state increasingly
weak and useless. This entailed a great deal of violence and side-switching as
coalitions formed and re-formed according to kaleidoscopically complex local
logics. Yet over time both Shevardnadze and Rakhmonov were able to use a
fairly weak set of institutionalized powers – their ability to bestow selective
favors, create redundant ministries with overlapping mandates, bypass the

173
174 Jesse Driscoll

written rules governing transactions between state bodies, and play a focal
role in international affairs – to insulate themselves from the threat of a coup.
Substance followed form in the post-socialist states. The symbolic functions of
the head of state transmogrified into real power.
The theoretical contribution of this book is a model of state-building as
a problem of coalition formation by amoral and interchangeable violence
entrepreneurs. The process of dividing the rents of sovereignty has been
modeled formally. Thinking of civil war settlement through the lens of coalition
formation, rather than two-player problems of bargaining or commitment,
emphasizes the ability of some warlords to extort rents from inside the coalition
by threatening to remove the president and return the country to a state of
war. Foreign patrons and multinational corporations desire order, not war,
but this requires that they recognize someone to take responsibility for what
occurs inside the territory of an internationally recognized state. To shape
local incentives, they dangle aid. Warlords swap their fatigues for business
suits to gain access to these income streams. They do not disarm. Wealth is
distributed in the form of ministry positions, tacit nonenforcement of the tax
code, closed-bid contracts, and rigged privatization schemes. When the guns go
silent, order is self-enforcing without third-party policy enforcement. Peace is
the result of jointly determined warlord strategies.
The empirical contribution of this book is a description of a process that
unfolded in Georgia and Tajikistan, straightforward in its logic and banal
in the brutality of its implementation. A small number of militia captains
realized that they could do better by working together to install a president and
reinventing their careers as security officers and politicians in the new regime
ministries. They estimated that, once installed, they would be able to continue
self-financing through racketeering, smuggling, and contract enforcement,
while also extorting the president through the constant threat of removing
him from power in a coup. The struggle over ruling coalition membership
was violent, shifting from rural to urban areas once the “conventional” phase
of the war was resolved. Collusion and coalition formation among various
militia captains, based on the incentive to eliminate imitators and purge
opportunists, was critical, as it revealed the bargaining power of the warring
factions. Destructive urban warfare eventually ended, and a semblance of order
reemerged in the capital cities. Shevardnadze used his celebrity to petition the
West for multilateral aid, but doing so provoked Russia and required that he
buy off most of his opponents simultaneously. In Tajikistan, by contrast, the
president’s strategy for political survival rested on playing both sides of the
Western-Russian divide, narrowing the social bases of state power and raising
the stakes of politics.
But if all of that is correct, then certain realists will feel entitled to ask difficult
questions: Why persist with the idealist fictions of the liberal state-building
canon? Why not be uncompromisingly honest about the processes that are
likely to unfold? A cynical answer is that democratic legislatures in wealthy
Implications 175

states could not justify voting for humanitarian assistance if they believed that
they were financing the kinds of militia politics described in this book.1 Even
if aid rents are just a form of divisible postwar spoils (v∗ ) that the warlords
divide up, more is better than less if consolidation is the desired outcome.
An ideological script for how civil war settlement ought to unfold – with an
optimistic role for loans, grants, and training – may be a political necessity to
securing humanitarian aid from donor states.
A better, more optimistic answer is that the university-educated class of
multilingual elites who manage the modern humanitarian aid complex see
themselves in solidarity with innocent people whose rights are trampled by
very bad post-war states. These privileged, well-meaning bearers of v∗ have
social protections that are often not afforded to the citizens of the countries
in which they temporarily reside, and they believe they can write a script
to a better future. And perhaps they can. Because this book focuses on
the warlords, their armies, and the gritty and corrupt distributional politics
between them, the tone of the narrative implies a tradeoff between stability
in the short term and strong and accountable institutions in the long run. But
perhaps there is no such tradeoff. Perhaps we should acknowledge that our
social scientific models of nonviolent social transformation have low predictive
power. As recent Georgian political history has demonstrated, large numbers
of NGO-funded idealists can, occasionally, completely overhaul the constraints
of a warlord-saturated political system and induce lasting social change. It
is sufficient to acknowledge that the liberal state-building canon, with its
normative emphasis on democratic institutions in the Weberian mode, is at
odds with processes described in this study. But that is not a sufficient warrant
for torching the canon.2
Some liberals, anxious for an endorsement of their missionary spirit and
disappointed with this stubbornly amoral and materialist account, are sure to
ask a different set of equally-difficult follow-up questions. Is there truly no
relevant role for democratic accountability in peace-building? No role at all
for public goods? For holding government agents accountable to some kind
of social contract? What are aid professionals supposed to imagine they are
engaged in, if not trying to move societies toward a Weberian state?3 And for

1 A humanitarian functionalist might be able to look the other way, justifying large aid flows
by arguing that short wars save lives, even if the money ends up lining the pockets of unsavory
characters. But it is unlikely these arguments could survive sustained debates in a democratically
accountable legislature. Olivier Roy, the OSCE head of mission to Tajikistan, who arrived on
the scene fully fluent in Persian, left a digital treasure trove of English- and French-language
emails for some future historian of the Tajik civil war. The fact that these emails are archived,
not part of the official OSCE reports, is evidence that he believed plain speech could undermine
the process that the OSCE mission was involved in: helping to oversee a transition to peace.
2 Even Halford Mackinder (1942), a hardened geopolitical realist, recognized a role for idealism
and values. 6.
3 Weber (1953), following Trotsky, defined the state as “a human community that (successfully)
claims the monopoly of the legitimate use of physical force within a given territory.” 78. At
176 Jesse Driscoll

that matter, what of civil society? What of the content of all the different aid
programs, troop trainings, education exchanges? Is v∗ from the U.S. actually
interchangeable with v∗ from the Russia – or, for that matter, from Iran?
Of course not. The model in this book is merely meant to provide an account
of how foreign involvement in the closing phases of a violent civil war might
facilitate settlement even if there are no security guarantees and no value added
by the foreign aid programming specifics. The model does not differentiate
between different sources of v∗ in the interests of simplicity, readability, and
parsimony. Chapter 3 makes clear that the fear of what social institutions would
look like in state where the v∗ came from Iran motivated many Tajik militants
to take up weapons. The fear of what social institutions would look like if
they were forcibly integrated into a state where the v∗ comes from the West
motivates many Abkhaz and South Ossetians to remain outside of Georgia’s
state-building project to this day. In both states expectations of future foreign
aid unleashed tremendous violent social energy precisely because the motives
of foreign audiences were so easily anticipated.

implications for policy

At the broadest level, the prognosis implied by this analysis is optimistic. If the
leader of a postwar state can quickly tap into international aid (in the form of
loans, charity, and security assistance) and then credibly commit to distributing
it to warlords (who are prepared to carry out a coup against him if he does
not make good on promises), a president may be able to quickly “buy” peace.
Though this account is primarily meant to apply to small and centralized states,
the model in this book provides an integrated lens through which to interpret
the decline of civil conflict since the end of the Cold War. It is more difficult
for warlords to survive outside of a state without superpower funding (e.g.,
the end of the Cold War lowered reservation values r).4 There is also a greater

least since William Reno (1998), it has been difficult to square the Weberian benchmark of
“monopolizing violence within recognized borders” with what everyone understands about
political institutions in weak states. What we really mean when we talk about strengthening the
state, in this account, is often associated with that elusive variable of “development.” To quote
Robert Bates (2001): “Development involves the formation of capital and the organization of
economic activity. Politically, it involves the taming of violence and the delegation of authority
to those who will use power productively.” 25. But a “monopoly of violence” sets the bar too
high. Many states have ungovernable rural or sub-proletarian pockets. Many leaders remain
in power by exploiting their pivotal roles as the mediator between the international system
and key providers of violence. In the tail of the distribution, some state leaders clearly thrive
while ignoring – or even terrorizing – the majority of their citizens, and selling partial (or
probabalistic) protection from the state’s own agents. The empirical work of Karen Barkey
(1994) and Ferguson (2005), and the theoretical work of Jackson (1990) and Wagner (2007)
develop these intuitions further.
4 See Kalyvas and Balcells (2011). Face and voice recognition software, the proliferation of
camera phones, mobile money – all of these make it more difficult for war criminals simply to
Implications 177

availability of credit and multilateral aid (e.g., higher levels of v∗ ) for states that
make cosmetic efforts to democratize.
The threat of mass-casualty terrorism has raised the stakes of getting
good intelligence. This in turn requires assistance from local partners and
cooperation between the permanent veto players in the United Nations Security
Council. It has also raised the stakes of inaction. The idea that pockets of toxic
beliefs are going to simply be permitted to metastacize strikes many military
professionals, and this author, as quaint. I cannot imagine a constellation
of political circumstances that would allow the great powers to outgrow the
temptation to tinker in the top-left “Postmodern Imperialism” corner of Figure
1.1, in the first chapter of this volume. Technology is rapidly expanding the
scope of governments’ ability to monitor individuals. Neither social scientists
nor intelligence analysts in the employ of the great powers can credibly commit
to ignoring what goes on in distant war zones. What the great powers can
credibly commit to ignoring, as an empirical matter, is the indefinite persistence
of very bad governance practices, so long as those governance practices serve
the function of keeping threats contained. But common diagnosis of a shared
problem for the great powers is in some sense good news for interstate
cooperation at a global level.
Regionally, the picture is murkier. With China and Russia as permanent
veto-players in the United Nations Security Council, UN Peacekeeping Oper-
ations (UNPKOs) in Eurasia are relatively rare.5 As a new generation of
post-Soviet elites come of age, it remains an open question whether these states
will remain economically and culturally dependent on Russia. Human capital
could easily begin to flow East or South. Geopolitically speaking, Eurasia is
impossible to ignore: the rise of China and India, the persistent risk of Pakistani
state failure, and the long-term viability of the nonproliferation regime are
all challenges that will pull the attention of Western analysts. Over the next
century, as Russian gradually contracts into a regional power centered on
the ancient Christian city-states of the Rus, it will become more tempting
for regional and great powers to assertively press their advantage in Russia’s
backyard.6 I have argued that part of the key to quick and decisive civil war
settlement in the post-Soviet wars was an absence of proxy-war dynamics in
the early 1990s, out of deference to creaky politics in Moscow. In the future,
proxy wars – of the sort that may be emergent in Ukraine – are more likely. As
such, the prognosis for quick settlements to future Eurasian civil wars is not
clear. If there were another episode of state failure or insurgency in Tajikistan,
or a new conflict in Uzbekistan, Kyrgyzstan, or the South Caucasus, it is not

fade away into state security structures via the processes described in this book. I am not certain
whether this ought to be modeled as increasing or decreasing the reservation wage of staying
outside of the state. I suspect that all of these technologies will vastly increase the capacity of
relatively weak states to manage unruly populations.
5 See Stedman (2003).
6 Cooley (2012), especially 16–29 and 162–178.
178 Jesse Driscoll

obvious that Russia would be capable of dictating the terms of the settlement in
quite the same way. Bullying and bribing the warlords in Afghanistan and the
government in Uzbekistan to stop supporting would-be clients from across their
borders worked reasonably well as a strategy for Russia to end the Tajik civil
war on its own terms in the mid-1990s. Many doubt whether these strategies
could, or would, be replicated today. It is even more unlikely that the United
States could play a unilateral stabilizing role from across an ocean. Russia’s
gradual disengagement from the region may have discouraging implications
for human security in Central Asia.
In Tajikistan, a partial incorporation equilibrium gave way to a full
incorporation equilibrium with time. I have argued in this book that this
is largely due to the complimentarity of great power interests, with Russia
acting as the lead state, to shield Central Asia from the chaos in Afghanistan
or ideological infiltration from Iran. Reservation values for unincorporated
warlords dropped gradually through the 1990s, and precipitously when NATO
forces entered Afghanistan. Interest convergence between the great powers
saved many lives, and kept the borders intact.
But the Georgian case is also worth considering, given that the weak states
of Central Asia are boxed in from every direction by nuclear states that have
demonstrated a willingness to support proxy wars in their neighbors. In the
case of Georgia, reasonable people disagree about whether the contemporary
situation represents a partial incorporation equilibrium (in which certain
warlords in Abkhazia and South Ossetia remain stubbornly outside state
control) or a full incorporation equilibrium (with revised boundaries that the
Georgian government refuses to recognize).7 But what almost no one disagrees
upon is that in Georgia today Russian and American strategic assistance
represent substitutes, not compliments.
This ambiguity in how one ought to even describe the Georgian case suggests
that a “partial incorporation” can endure as an equilibrium indefinitely, or
at least for a very long time. Nonconsolidated states – not only Georgia,
but also contemporary Somalia and Afghanistan – may represent stable,
self-enforcing equilibria. What we observe is persistent low-intensity conflict
between one coalition of criminal gangsters in the capital city and various
potential counter-coalitions of rural actors that have built their own shadow
state institutions with assistance from foreign actors (e.g., Russia, Ethiopia,
Pakistan, respectively). Insurgent warlords may desire political reconcilia-
tion, but calculate that there is more to be gained at the fringes of state
control.
I am often asked for my opinion on whether and how soon Georgia should
be admitted to the NATO alliance. Alexander Rondeli, the gregarious head of

7 For a useful description of the overlapping zones of political control that define Georgia’s
internal hinterlands, see Marten (2012), 86–101.
Implications 179

the Georgian Foundation for Strategic and International Studies, once put the
matter to me in the following way:

“The problem is easy to diagnose. The problem is that there are certain people in charge
in Moscow who act like they don’t know where Russia’s borders are. … America says
it wants Georgia to develop independent of Russia. We agree! We don’t want to be
like Armenia, like Uzbekistan, like your Tajikistan … we want to be free, to be like the
Baltic states in Europe. And you [Americans] say you want the same thing for us. We
have democracy. We train with your army. We sent troops to Iraq. What is left for us to
do?”8

I will violate a Georgian taboo by answering Rondeli’s well-asked question

in plain language: If Georgian leaders wanted to get serious about getting
NATO membership, a good first step would be to abandon territorial claims
to the break-away territories. If Saakashvili had been the visionary he imagines
himself to be, he would have used the afterglow of the Rose Revolution to
do this in 2004. He instead took a number of provocative steps that led to
war. Now the territories are almost certainly gone for good anyway, though
it will fall on some future statesman to pay the domestic costs of admitting
it. The blunt truth is that many in the Georgian political class continue to
plan for a war with Russia. Their enthusiasm for this outcome is frankly
disturbing. Most Americans do not want to fight a war with Russia at all,
and certainly not over who owns what hotel in Sukhumi. As I am not myself
a Georgian, it is not appropriate for me to pass judgment on Georgians’
collective unwillingness to consider giving up their historical claims to the land
of Abkhazia and South Ossetia. But I hope that my Georgian friends will
extend me the reciprocal courtesy of acknowledging that, from the point of
view of American national interests, the events of the August War of 2008
are exemplary of why Georgia is a potential security liability.9 Russia has
demonstrated a willingness to keep reservation values for Abkhaz and South
Ossetian warlords higher than any offer from the Georgian center – even when
foreign aid was flowing into Tbilisi at an unsustainable rate. Georgia has been a
stalwart ally to America, especially since 2003, but if Georgians want to be part
of a Western security community and the NATO defensive alliance, it a problem
to be floating in quantum uncertainty between a full incorporation equilibrium
and a partial incorporation equilibrium, depending on whose version of the
map prevails. Georgia does not need Russian help or permission to withdraw
from disputed territory. It could unilaterally transform itself into a better ally, if
only Georgians wanted to. But I fear that a plurality of Georgians do not want
to. Many believe they can wait Russia out. Many believe that they can spend
another few decades continuing to enjoy massive flows of foreign aid, slowly
shame the West into providing security guarantees, and eventually reclaim their

8 Interview with Alexander Rondeli Febuary 27, 2006 in Tbilisi.

9 See King (2008a, 2008b).
180 Jesse Driscoll

lost territory by some combination of military coercion and Western mediation.

In my view, this kind of thinking is the heart of the problem – the worst kind
of violent utopianism, celebrating escalation scenarios that culminate in war
between Tbilisi and Moscow. And the problem is increasingly understood in
exactly the same way in Brussels and in Washington.
One of the main policy-relevant lessons that should emerge from this book
is that locals have strong incentives to misrepresent regional history. Given this,
it important for policy audiences to be skeptical when either regional experts or
armchair grand strategists claim to know what armed groups in a distant part
of the globe are “really fighting for.” Interview data do not cut this knot if the
soldiers themselves either do not know why they are fighting or are strategically
misrepresenting their own interests for the camera. And answers to the question
can be tailored to the desired policy prescription in very predictable ways. If
armed groups reflect deeply held political beliefs or social divisions, perhaps
no international gendarmerie will be able to resolve long-term grievances or
even stop the killing.10 But if, on the other hand, armed groups rest on thin
kinds of social capital that are just cover for criminal opportunism, perhaps
it is morally incumbent on military professionals from the developed world to
either buy them off with aid – as the model in Chapter 2 suggests is possible –
or destroy these groups and spare civilians the horror of war.11 Between these
two extremes lie a universe of gradations.12 But a healthy dose of humility is
justified when it comes to predicting outcomes internal to civil wars, or the
high-stakes coalition politics that accompany the settlement.
When it comes to predicting outcomes, Russia’s failure to shape Georgian
and Tajik domestic politics in the wake of their “order first” policies in the
early 1990s can also serve as a cautionary tale. Russia and CIS forces had
access to regional intelligence networks – acquired over decades of patient
penetration and cultivation – that the United States can never really hope
to acquire in Eurasia. Yet in both cases, Russia’s foreign policy goals were
thoroughly subverted by locals. Recall that Moscow’s overarching goal was
to build a strong buffer state and stable power-sharing in Tajikistan and to
undermine Georgian efforts to orient their foreign policy toward the West
and NATO. Russians superficially succeeded on both accounts in the short

10 See Laitin (1999b) and Luttwak (1999).

11 John Mueller has diagnosed the central problem of the twenty-first century military interven-
tions as one of “policing the remnants of war.” He posits that many of the armed groups
responsible for civilian deaths in the developing world are disorganized criminal affairs run by
gangs with very fragile ties to the society for which they claim to speak. Given the huge moral
costs associated with standing on the sidelines while local criminals terrorize their populations,
and given that any minimally competent military force should be able to rout and dismiss
disorganized bands of “thugs,” he argues that a liberal attitude toward intervention could save
countless lives. See Mueller (2003) and Mueller (2004), 8–23 and 117–160.
12 See Kalyvas and Kocher (2007) and Fearon (2007) for an example from recent history of
how subtly different interpretations of these questions translate into radically divergent policy
recommendations, holding geography and time constant.
Implications 181

term, but in a longer view, their policies have backfired in certain respects.
Efforts to weaken Georgia led Shevardnadze to respond by building a broad
coalition of social actors oriented toward the West, leading to an increasingly
confident and independent foreign policy. In its efforts to make Tajikistan an
outpost of Russian interests in the near abroad, Russian foreign policy has been
complicit in the creation of an extremely brittle state with a narrow governing
coalition. This is not to say that Russian policies were doomed to fail, or that the
United States ought to give up on local partnerships and good intelligence. It is
simply to note that in the scramble to establish friendly regimes in Georgia and
Tajikistan, the Russian government was incapable of monitoring distributive
politics or having influence over who got what. Georgian and Tajik agents got
their way more often than they did not.
In Georgia and Tajikistan, warlords were not just taking advantage of state
weakness to advance criminal agendas – they were trying to become the state.
Any model that does not account for the possibilities of tactical cooperation
and long-term collusion between the state and militia actors is incomplete.
Although most authors writing on this period imply that warlords were
disarmed by “the state” once it became strong enough to confront the militias,
this simplification actually obscures the defining political problem of the time:
that violent militias had moved from the streets to become explicitly part of
the state, well positioned to contest their share of the rents of statehood. This
is a disturbing insight for policymakers and scholars. When distant observers
see brutal killing of civilians during the closing phases of civil wars, there is a
tendency to want to send aid to shore up state weakness. The argument in this
book should be interpreted as a caution against the impulse to simply increase
bilateral aid to provide more resources to the consolidating regime. Models that
focus on state capacity – with rebel militias locked in zero-sum competition
with state institutions to provide public goods – are certainly relevant to
counterinsurgency and asymmetric warfare, but they can be misleading when
the state itself is up for grabs. Though often the political instinct is to “prop
up” our ally (the recognized government) with an increased flow of bilateral
and multilateral aid (to signal our commitment and to strengthen the police and
military), if violence is occurring because various armed actors within the state
are attempting to secure a larger share of state rents, targeted aid transfers may
just increase the value of the prize being contested. Chapter 4 demonstrated
that in the closing phases of both the Georgian and Tajik civil wars, militia
members were engaged in months of grueling and destructive attrition warfare
in the hopes of being admitted to the state apparatus with a sufficiently
large side payment. Expectations of aid increased the stakes in the war of
attrition.
To summarize: The extortion dynamic described in this volume are necessary
considerations for elites hoping to “buy peace.” This has relevance for
contemporary policy toward Ukraine, where a war by warlord proxies is
underway.
182 Jesse Driscoll

implications for ukraine

At the time of this writing there is an ongoing civil war in Ukraine.13 Various
militias and state forces are clashing. The Russian government’s position is
that the proximate cause of this violence is Ukrainian state weakness, and the
permissive cause of the violence is Western meddling in Russia’s traditional
sphere of influence. The Ukrainian government’s position is that this war
amounts to an undeclared invasion of Ukraine by Russia. Prominent Western
opinion leaders have invoked comparisons to the Anschluss.14 Although the
root causes of the war are disputed, the costs are not. Thousands have died,
many thousands have been wounded, tens of thousands have been displaced,
and tens of billions of dollars of lost productivity and damage to critical
infrastructure are estimated. Formal ceasefires have come and gone. What
happened? And does the theory in this book shed light on what might be
expected in the future?
Though Russian- and English-language narratives diverge quickly, the basic
sequence of events leading to the current conflict is not disputed. In the fall of
2013, President Viktor Yanukovych of Ukraine declined to sign an association
agreement with the European Union in favor of exploring membership in
Russia’s Eurasian Economic Union (EEU). Many Ukrainians believed the
window was closing on the possibility of labor mobility to, and economic
and cultural integration with, the West. Broad-based opposition emerged,
galvanized in opposition to the prospect of a geopolitical pivot to the East.
A wide coalition of civil society groups organized protests. Weeks of escalation
finally culminated in prolonged and violent clashes with state security forces
in Maidan Square at the heart of Kyiv – a bloody, high-stakes, ideologically
charged reenactment of the 2004 Orange Revolution.15 On February 22, 2014,
Yanukovych shocked domestic and foreign audiences by fleeing the country.
He left behind a paralyzed, headless government. Russia’s state media outlets
decried the situation as a Western-backed coup d’etat against the legitimately
elected government in Ukraine. Ukrainian state security services stood down.
Ad hoc voices from civil society promised new elections.
Political power passed to the parliament, which quickly passed a resolution
that would have made Ukrainian the sole state language at all levels. If
enacted, parents of Russian speakers would have had to enroll their children
in Ukrainian-language schools for them to have any chance at any job as
a government clerk. The bill was vetoed a little more than a week later

13 I employ the analytic category “civil war” not to imply that the causes of the war are indigenous
to Ukrainian politics, or that third-party engagement is irrelevant to the conflict – simply to
note that according to the coding rules employed elsewhere in this manuscript the events of
2014 qualify as a civil war. The death toll is disputed, but at least 1,000 people have died, and
more that 100 of the deaths are on the government side.
14 Rucker (2014).
15 For an analytic postmortem on the Orange Revolution see Beissinger (2013). His analysis
emphasizes the nonideological nature of the 2004 mobilization. See especially 16–17.
Implications 183

by the interim president. But the threat to force linguistic assimilation, and
strip life opportunities from Russian-speaking citizens who refused to give up
their native tongue, increased the stakes of the political conflict.16 Self-defense
militias began to form.
For narrative coherence, at this point it is necessary to posit an “internal
explanation” for the actions of Vladimir Putin – to speculate about one
particular human being’s emotions, psychology, and state of mind.17 My best
guess into Putin’s psychology is that his actions with respect to Ukraine were
driven primarily by domestic considerations – namely a desire to maintain his
position as the most popular person in the Russian-speaking world, and to
thereby write himself into Russian history – and secondarily by a desire for
symbolic confrontation with the West. I suspect he understands himself to be
engaging in a battle of wills against hypocritical geostrategic competitors, and
is practically daring them into a proxy war.
Though Putin’s reasoning will remain the subject of speculation, what is
known is that he seized on the anarchic situation in Ukraine as a proximate
justification for intervention: more or less to remind everyone that he could, he
broke one of the most basic rules of international sovereignty. On February 27,

16 The acting president of Ukraine, Oleksandr Turchynov, vetoed the bill on March 1, effectively
stopping its enactment. In my opinion, the decision to remove Russian as an official language
can best be understood as an emotional decision, driven by spite and aimed at punishing the
eastern half of the country for electing Poroshenko in the first place. The best that can be
said of this rash move by parliament is that the vote was quickly overridden – but it validated
the worst-casing of Russia’s propagandists. For a useful discussion of spite, see Petersen (2011)
49–50. Petersen (2002) and Petersen (2011) describe various mechanisms by which anticipated
reversal of status hierarchies produce violent actions by individuals.
17 Recall from Chapter 1: Internal explanations, according to Ferejohn (2004), come from within
an individual, such as psychological biases, emotional responses, and culturally contingent
belief structures. For expert testimony on Putin’s psychology, see McFaul (2014), especially
8:00–12:10 and 20:20–21:15. A different kind of expert testimony, drawing on different
sources and articulated with different motives, can be found in Remnick (2014). Maybe
Putin feared that Ukraine’s normal political institutions – regular elections that facilitated
slow rotations of power between Western- and Eastern-oriented governments – were in the
process of being bypassed by street politics, to the disadvantage of Russia and the advantage
of opportunists in the West. While a vast strata of the Ukrainian population was interested in
orienting their futures toward the West, many other Ukrainians, especially in the East, felt their
votes had been invalidated and the promises of democracy betrayed by the nonconstitutional
transfer of power. Many feared being reduced to second-class citizens in an emergent status
hierarchy that rewarded English-language skills, and worst-case scenarios were validated by
the rash actions of Parliament, maybe Putin felt a genuine obligation to come to the defense of
“his people.” Maybe Putin was angry that clandestine Western agents – whom he imagined to
have been orchestrators of Maidan – were getting away with regime change in his backyard.
Maybe he feared that the United States was planning to use the same kinds of tactics in Moscow
on him. Maybe he was interested in probing the limits of favorable media coverage, testing if
he could “wag the dog” to push his own popularity so high that conversations about election
falsification would be absurd. It is likely a combination of these, but no one can produce
evidence to validate or falsify alternative hypotheses.
184 Jesse Driscoll

Russian forces seized the Crimean Peninsula. Many Crimean citizens greeted
them as liberators. Others, particularly those of Ukrainian or Tatar ethnicity,
were far more apprehensive. A referendum to join Russia was organized quickly
and, on March 16, a substantial majority of Crimea’s citizens voted in favor of
unification – although interpreting these results was complicated not only by
Russia’s military occupation, but also by the fact that the ballot did not include
an option to remain within Ukraine.18 These actions were declared illegal by
Western governments, and, as such, the map remains contested. In a number of
public speeches, Putin made it clear that he considered these moves a correction
of historical mistakes made by Soviet map makers under Khrushchev, and
did not consider the internationally recognized borders of Ukraine to be the
business of anyone except Russians and Ukrainians. He repeatedly reaffirmed
his duty to come to the aid of Russian speakers, and painted Crimea’s citizens
as helpless hostages trapped in the anarchic and proto-fascist Ukrainian state.
Shortly after, he began to invoke “Novorossia,” which were – are? – imperial
Russian territorial claims on southeastern Ukraine, which may extend to the
entire north Black Sea coast.19
In response to Russia’s military adventurism, and recognizing the impo-
tence of the Ukrainian state to do anything at all about it, many more
pro-Russian militias self-organized.20 Pro-Ukraine self-defense militias orga-
nized in response.21 Charismatic warlords used the same kinds of persuasion
sketched in Chapter 3 of this book: the certainty of masculine glory and
the chance to play soldier in the short term, vague promises of a good job
in the security sector in the medium term, and romantic hopes of social
recognition and a respectful place in an emergent nationalist narrative in the
long term. The backbone of the labor for many pro-Russian militias seems to
have been Afghan civil war veterans, though there was apparently no shortage
of disaffected youth ready to join.22 As weeks turned into months, dozens
of militias declared autonomous “People’s Republics” around their city or
oblast in the hopes of being recognized and annexed by Moscow. Though
diplomatic recognition did not materialize immediately, what did materialize
was barely covert Russian assistance aplenty.23 Over the next few months,
pro-Russian separatists inflicted humiliating losses on the Ukrainian military,
sometimes in full sight of Organization for Security and Cooperation in Europe

18 The official vote tally was 96 percent in favor with 86 percent reported turnout.
19 New York Times: August 29, 2014.
20 New York Times: April 30, 2014, New York Times:June 4, 2014.
21 New York Times: May 23, 2014.
22 “One rebel group, Oplot, comes from the Ukrainian city of Kharkiv. Another, the Russian
Orthodox Army, is composed of Russians and Ukrainians. A third, named for a river, Kalmius,
is made up mainly of coal miners. This motley mix forms just part of the fighting force of
Ukraine’s eastern uprising. It is more patchwork than united front: Some groups get along
with others. Some do not. And their leaders seem to change with the weather. ‘I can’t keep
them straight anymore,’ said a fighter …” New York Times: July 09 2014.
23 New York Times: May 27, 2014.
Implications 185

(OSCE) observers.24 Clumsy attempts at counterinsurgency by the Ukrainian

military destroyed entire cities. A quarter of the country’s industrial and mining
output was disrupted. The Ukrainian currency devalued 60 percent.25 Tens of
thousands of Ukrainian civilians on both sides have been brutalized by war and
will carry bitter memories for the rest of their lives.
As of the fall of 2014, Ukraine was clearly at war. An intergenerational
struggle for national reunification, and a deep sense of bitterness at being
abandoned by the West, was the product being peddled to the next generation
of Ukrainian light infantry recruits. In a September trip to Kyiv, I observed
dozens of young, athletic, masked men recruiting for the “Azov” militia in front
the ancient St. Sophia Cathedral. I observed a similar initiation ritual just hours
later, as stoic young men prayed over machine guns near the Monument to the
Martyrs of the Artificial Famine of the 1930s. My life history interviews provide
a basis for informed speculation as to the motivations of these young idealists.
They were angry. They were anxious to prove their manhood – to themselves,
to each other, to their families, and to their girlfriends. They wanted to claim
their place in a nationalist narrative, fighting the war of independence that was
denied to their fathers and uncles when the Soviet Union broke apart peacefully.
They wanted to grease the skids toward a respectable career as soldiers or police
officers, and doubted that volunteering would hurt their chances.
The theoretical model from Chapter 2 can be laid atop the Ukrainian
empirical record to generate a short sketch of an analytic narrative in the service
of providing policy recommendations. State failure in Ukraine was imminent in
early January 2014. Russia could have responded by sending military assistance
to prop up Yanukovych’s embattled regime, but it did not. As a Ukrainian
academic who was part of the Maidan Coordinating Council put it,

“If he [Yanukovych] has gotten on the phone to Putin and said ‘I need help, I need
tanks, I need police battalions, maybe military divisions, I am the elected head of
state … come arrest these criminal protesters,’ there is absolutely nothing anyone could
have done. … but he [Yanukovych] fled, so we’ll never know. … I think he was worried
about being strung up, like Ceaus, escu in Romania.”26

Returning to the language of the model: Russia responded to the collapse

of the government by pro-Western demonstrators by raising the stability
threshold (s). In practice, this meant demonstrating its economic and energy
leverage, flaunting its superiority in military hardware, and engaging in a
disciplined propaganda war. Russian media portrayed the new regime as
dominated by fascists. The voices of Pravii Sektor brownshirts were broadcast
as if they were representative of the modal member of the Maidan protests.
Consonant agreement and numbing repetition linked “NATO” with “Nazis”

24 New York Times: September 24, 2014.

25 Al Jazeera: September 28, 2014.
26 Interview conducted with Oleksiy Haran in Kyiv, June 23, 2014.
186 Jesse Driscoll

as a rote cognitive script in the minds of Russian-language media consumers.

As for establishing facts on the ground, Russia demonstrated the efficacy of
a variety of very low-cost military technologies – logistical support to locally
recruited militias and bike gangs, unregulated border crossings for irregular
infantry units from the North Caucasus, and plausibly deniable support from
the Russian regular military – to impose costs on the interim government
in Kyiv.
Consistent with the logic of the model, the warlords with the highest
reservation values (r), the Russian-speaking warlords in the East, were the
first to torch the Ukrainian social contract. Subsequent Russian support
raised the reservation value for pro-Russian militias.27 To keep order, the
Ukrainian government would have had to commit to higher transfers of v∗ than
before, striking new deals with newly mobilized various warlord constituencies.
The problem was that there existed no functioning government capable of
striking politically difficult bargains. A new president, Petro Poroshenko,
was elected in late May 2014, so finally there was someone to negotiate
with Putin and appeal to Western states for assistance. Ukrainian army
forces were hastily deployed, the word “invasion” was constantly repeated,
mandatory conscription into the army was discussed, and the fledgling coalition
did what it could to demonstrate to warlords in the East that a war of
independence would entail horrible violence (high costs of war c). Early results
were not promising. And at the time of this writing, Ukraine appears to be
sliding toward either a partial incorporation equilibrium, with functionally
independent statelets in the East analogous to Abkhazia or South Ossetia. A
full-fledged state-failure equilibrium is not out of the question. In anticipation
of the latter, capital flight set in and the economy contracted substantially
(v < v∗ ).
What would be necessary to peacefully resolve this conflict, restore order,
and reestablish domestic sovereignty in Ukraine? In the language of the model:
What would it take to get back to a full incorporation equilibrium?
Georgian history suggests that pragmatism does not emerge in the absence
of leadership and sustained political will. Applying this lesson to Ukraine, my
opinion is that Ukrainian domestic actors will have to work together to “get
over Crimea” as quickly as possible. It is imperative that Ukrainian historical
claims to the Crimean peninsula not be allowed to become a litmus test for
participation in Ukrainian politics. There is a pragmatic, nationalist rationale
for a gag rule on the Crimea issue: The question of “whose history is more
right” has no objective answer. The 2014 Parliamentary elections provided
some cause for hope in this regard, and I hope Ukrainian politicians will be able

27 Many armed spokesmen in Donetsk and Lukhansk gave statements making clear that they
saw in Russia’s actions a once-in-a-lifetime opportunity to shape an independent future for
themselves. They believed that even if they failed, conditional on taking up arms and organizing
militias, they would gain access to a future where they could credibly demand far more
functional autonomy from Kyiv than would be possible otherwise.
Implications 187

to run for office successfully without having to go on the record establishing

their nationalist credentials on Donbas and Crimea. If positions harden, and
hostile position-taking on Crimea becomes a prerequisite for electoral viability
in Ukrainian politicians, Georgia’s past will likely be Ukraine’s future.
Although the West has every reason to encourage the emergence of a
pragmatic conservative coalition willing to compromise and strike politically
difficult bargains, there are two huge problems that do not, at the time of this
writing, have obvious solutions. The first is that many Ukrainian voters feel,
with reason, that if they acquiesce on Crimea, then they will practically be
inviting Russians into Odessa. The second is that the main tools that the West
has at its disposal to shape domestic politics are financial incentives (v∗ ).
The general hope in Kyiv, after the Maidan events and the annexation of
Crimea by Russia, was that Western states would come to the rescue with
low-interest loans, generous International Monetary Fund (IMF) packages, as
well as manipulation of the global price of oil and sanctions on Putin’s inner
circle in Russia. On the diplomatic front, negotiators in the winter of 2015 were
hard at work trying to hammer out arrangements that would make credible
federal guarantees acceptable to both domestic constituencies in Ukraine and
Russian-speaking audiences.
Optimists hoped that Russia would stop waging its undeclared proxy war
via warlords in breakaway provinces or contribute to the rebuilding effort by
lending the assistance of its security forces to quickly settle the matter. If Russia
were to close its border and begin forcibly disarming warlords, the kinds of
collateral civilian damage on display in the summer of 2014 would not need to
be repeated. But as a Ukrainian military veteran observed:
“So long as the conversation is about NATO and the EU … I doubt the West is really
going to be able to build things here in Ukraine’s faster than Russia can tear them
down … it would be nice if we could get money like Georgia for a few years, but
Russia … they are so close, and we are so close [taps his heart] to them, here, and you
[Americans] are so far away. … Even Germany, France – they have no interest here.
Only sentiment.”28

For its part, the current Russian government has demonstrated to future
generations of Ukrainians the kinds of costs they could expect if they continued
to drift toward the cultural and economic embrace of the West: Russia can
simply flood the region with irregular soldiers and biker gangs, with occasional
clandestine support from special forces, to fight an undeclared war. And
at any time, Putin could decide to march his forces further, link up with
Transdeniestria, and claim all of the north shore of the Black Sea for Russia.
It is next to impossible for Russia to credibly commit to not do such a thing,
given Putin’s recent actions. This possibility is widely recognized, and is itself
a strong argument for withholding security guarantees to Ukraine. As in the
South Caucasus, the benefit that Russia accrues from keeping its neighbor

28 Interview conducted in Kyiv September 14, 2014.

188 Jesse Driscoll

destabilized – keeping the NATO alliance out of its immediate neighborhood –

may outweigh the costs of absorbing refugees and enduring scolding from
international diplomats and op-ed writers. Ukraine could face many years in
an ambiguous partial incorporation equilibrium.
But what if the West did take a principled stand against Russia, pouring
postwar reconstruction aid (v∗ ) into Ukraine in the kind of localized Marshall
Plan that some have advocated? It is not impossible to imagine a consensus
emerging to stand up to Putin’s aggression. The thought leaders would
presumably be residents of the Baltic states, supported by offshore Anglophone
idealists. The costs of persistent conflict would be borne by German, French,
Dutch, Turkish, Spanish, Polish, Italian, and many East–Central European
constituencies (measured in gas prices at the pump) and by many Ukrainian
families (measured in acute human suffering). It is even possible to imagine
scenarios in which Ukraine is invited into NATO without first settling the
Crimea border issue. This would set the stage for a new Cold War–style
standoff, with Crimea playing the role of East Germany. The argument in this
book should remind policymakers that if the West decides to approach the
problem in this way, two things should be expected.
First, once the aid arrives, the model emphasizes that Western donors will
immediately lose control of how it will be distributed. The aid will become a
source of rents, distributed in the service of keeping the government in power.
Whatever our intentions, much of the v∗ will almost certainly end up lining
the pockets of the most loathsome kinds of intermediaries: not only warlords,
but also cynical vote brokers, war profiteers with multiple passports, and real
estate speculators. Georgian and Tajik history suggests that trying to purchase
peace with foreign aid is an ugly business.29
Second, even if it were possible to target the aid toward reconstruction
in the shattered cities, some of the main beneficiaries will be local warlords
in the East – individuals who can, by force of arms, veto the particulars
of distributional politics. Increased transparency and contemporary forensic
accounting techniques may make it possible for tomorrow’s social scientists
to trace the side payments as they accrue to local mayors, regional governors,
and police chiefs. These are the social actors best positioned to drag their feet,
demonstrate their power to be disruptive, get their bribes from Kyiv, and then

29 In the case of lethal aid, the risks are substantially higher. In my estimation there is a substantial
risk of that Western military aid to Ukraine, advocated by many hawks in the West, could
backfire. If Putin can credibly claim that the dead Russian soldiers who come back in body
bags are dead because they are being killed by Americans, or by American guns, he may
be able to ride his recent wave of wartime popularity indefinitely. It would actually be an
ex-post validation of the Kremlin’s most virulent propaganda claims. It is frustrating that many
self-styled “realists” either do not understand this, or pretend as if they do not. It is very hard
to falsify the conspiracy theory that the Maidan events were engineered by Western security
agencies. Sending military technology one year after unconstitutional regime change is not
exactly “disconfirming evidence” for Putin’s theory.
Implications 189

decide whether or not to play their part to keep their districts from joining
the Ruble zone.30 And it is already obvious to anyone paying attention that
the security forces of the Ukrainian state plan to fight this war by working
in tandem with patriotic militias representing a broad coalition of anti-Putin
Russian-speakers of every ideological stripe. If this is to be a new Cold War
over the future of Ukraine, it will be a war largely fought to the benefit of
warlord proxies, using violence to build autonomous quasi-states in the zones
of contestation.
It may be valuable, as an analytic exercise, to try to view this crisis from the
East and not the West, and employ the value-neutral metaphor of bargaining, or
a game of nerves. It might be the case that Putin is doing all of this to signal that,
when it comes to the defense of core interests the Russian Federation in 2014,
he is willing and able to fight dirty. Russia may no longer be a superpower, but
even as a regional power it has many potential sources of leverage over Ukraine,
and deep historical interests in the area – interests that precede the existence
of the United Nations Charter by centuries.31 These core interests were never
at stake in Central Asia or in the South Caucasus.32 If this interpretation
of Putin’s psychology is correct, the chorus of scolding voices claiming that
the Crimea precedent will unravel the post–World War II international order
may be somewhat off the mark. By redrawing the map conservatively, and
provoking a containable war in a part of the world that is in Russia’s sphere of
influence, perhaps Putin was communicating where post–Soviet Russia’s revised
security red lines actually are.33
A Levada Center poll in late 2014 estimated Putin’s popularity at 87
percent.34 At that time, Western governments were imposing economic
sanctions as the instrument of punishment – a completely different kind of
Western response than that seen to Russian moves in Georgia or Tajikistan.
These sanctions inflicted substantial costs on the Russian economy, felt across
a wide stratum of Russian society.35 It is too early to tell whether these sanctions
will change Russian policy. I would be surprised if they did. It is easy to imagine
scenarios in which Russians “rally ‘round the flag” in response to the sanctions,

30 See Donetsk People’s Republic Plans to Switch to Ruble (2014).

31 Putin’s tactics also had the effect of drawing attention to asymmetry of interests between the
United States and Russia in this particular part of the globe. Overall, there are simply too many
issues related to global governance in which there is a natural complementarity of interests
between the United States and Russia: maintenance of the nuclear nonproliferation regime,
a grand bargain with Iran, future relations with China, the functioning of the UN Security
Council, and more. Putin factored these geopolitical realities into his bargaining position.
32 Whether or not they exist in the Arctic or Siberia is a matter for another day.
33 I am heartened by Foreign Minister Sergei Lavrov’s calls for a “Reset 2.0.” “Now there’s a need
for what the Americans might call a ‘reset’ … The current US administration is today wrecking
much of the cooperation structures that it created itself along with us. Most likely, something
more will come up – a reset No. 2 or a reset 2.0.” BBC (2014).
34 See “Putin’s Approval Rating Soars to 87%, Poll Says”: August 6, 2014.
35 See Birnbaum (2014).
190 Jesse Driscoll

tightening their belts and blaming the West for their hunger. It is easy to
imagine scenarios in which countersanctions on food and agricultural imports
from the West ultimately deepen Russia’s economic and security relations with
post-Soviet republics.36 It is difficult, by contrast, to imagine scenarios for
either regime change in Russia, or for a policy reversal on Crimea without
regime change. I am afraid that the standoff has no end in sight. Putin has
demonstrated that he can engineer a situation in Eastern Ukraine analogous to
the partial incorporation equilibrium in Georgia: an ambiguous posture that
keeps Ukrainian politics permanently destabilized.
The Georgia analogy is not lost on Ukrainians. I asked a prominent
Ukrainian intellectual how he thought his country would be different in thirty
years. He did not hesitate to respond that he did not think his country would
exist in thirty years. Surprised by his candor, I asked him what he meant. He
explained that in the future “[T]here will be three Ukraines, just like how today
there are three Georgias.” I asked him if he would share his theory of what
the borders of the three Ukraines would be; he gamely sketched the map that
appears as Figure 6.1.37 He went on: “We’ve already lost the East, with the
destruction of Donetsk and Lukhansk … our politicians just don’t know it yet.
But those people will never think of themselves as part of Ukraine again. We’ll
never bring them back, just like Georgians will never get back Sukhumi or
South Ossetia.” He envisioned a rump state, or a collection of ambiguously
federated city-states, emerging east of the Dnieper River to provide Russia with
the strategic depth it demands.38
A final provocative observation: Much in this book contradicts the standard
account of war termination that is often championed by advocates of the
International Court of Justice. It is worth recalling that one of the first things
that Shevardnadze did was publicly pardon and forgive Ioseliani and Kitovani.
Political rehabilitation of war criminals is at the core of the consolidation game
as presented here. An uncomfortable social fact is that the decent people and the
bad apples work in close proximity in postwar state security ministries. In both
the Georgian and the Tajik cases, processes of local justice worked themselves
out, albeit slowly and imperfectly. But criminals were not actually locked in
zero-sum competition with legitimate state institutions so much as the criminals
were shielded from prosecution by these institutions. That was the whole point.
In reference to Ukraine: I would be surprised if either the relevant heads of state
(Vladimir Putin, Petro Poroshenko) or any of the prominent warlords (such as

36 See Schenkann (2014).

37 Interview conducted in Kyiv, September 8, 2014. Author name withheld by request. He was
quick to caveat his prediction: “If Putin dies, then everything in Russia unravels, and it could
easily be very bad … and then there will be more than three Ukraines.”
38 Another memorable excerpt from the interview: “I do not pretend to know what will happen
up here … I hope we [in the West] keep Kyiv … and I think we will. And probably the whole
territory out there, and maybe parts of the suburbs of Kyiv, would be contested for a while –
this is to be expected.”
Implications 191

Sketch Produced: 9/8/2014

A C

E
D B

A = “Polish Ukraine” — Cities North of Romania, up to/including Kyiv (?) D = Moldova

B = Novorossia — Essentially a straight latitude line from Transdniestria E = Transdniestria
C = East Dniepr — Borders ambiguous; likely site of persistent civil conflict F = Romania

figure 6.1. Ukraine 2045: a predictive map.

Alexander Zakharchenko, the premier of the self-proclaimed Donetsk People’s

Republic) are ever indicted under the mechanisms of the Rome Statute. There is
a gap between the universal rhetorical promises of international human rights
law and the cynical reality of selective enforcement that is unlikely to be bridged
in this part of the world any time soon.

implications for theory

Civil wars are often treated by political scientists as contests between an
“incumbent” and an “insurgent” player. This approach can shed light on
attrition, asymmetric bargaining, and commitment dynamics, all of which are
understood with a great deal more clarity today than they were two decades
ago. Two-player models are attractive for their parsimony and tractability.
These models generate policy-relevant insights that can be applied to many
different settings. Because assisting incumbent regimes as they fight grinding
wars against insurgents is a persistent problem, there is a permanent demand
for analysts trained to deploy insights from these models.
192 Jesse Driscoll

For certain policy-relevant debates, there may be no real alternative to

beginning the conversation by positing, by assumption, the existence of A and
B: a legitimate state A, and an illegitimate rebel group B that is attempting
to seize that state. But as scholars, we should notice that beginning the
conversation in this way effectively forecloses the possibility of even asking a
different question that may be even more policy-relevant in the closing phases
of civil war: How is the practical boundary between “state” and “rebel”
constructed by local actors trying to make peace?
My book is a direct attempt to answer this important question. All of the
politics I describe revolve around strategic actors trying to manipulate the
boundary of the coalition that would ultimately constitute “the state.” Rather
than just assuming Georgian and Tajik incumbents into existence, my approach
draws attention to violent and cooperative processes that allowed corporate
state actors to emerge.
The central problem with two-player “incumbent vs. insurgent” approaches
is that they rely on assumptions that are inappropriate to the analysis of politics
in failed states. Treating “insurgent” and “incumbent” players as fixed entities
reifies the master cleavage of the civil war. It is not only postmodernists and
radical constructivists who have noticed that treating ethnic groups as the
unit of analysis – especially when combined with the language of “zero-sum
conflict” – plays primarily to the advantage of military hard-line factions
in the incumbent state.39 As an empirical matter, civil wars since 1945
have rarely featured professional armies with coherent and unified lines of
command-and-control. Analogies to two-player firm or interstate bargaining
problems are therefore strained. This is especially the case in situations of acute
state failure, when the incumbent regime faces off against multiple challengers.
Side-switching in civil wars – especially highly fractious civil wars – is far
more common than local politicians, foreign pundits, or political scientists have
generally acknowledged. One of the main insights that emerges from the model
in Chapter 2 of this book is that even if the figurehead president remains the
same, the constellation of social forces that support a regime at the end of the
civil war may not be the same constellation of social forces that supported the
regime at the beginning of the war.
But once one abandons the analytic simplification of the two-player
approach – the state vs. the insurgent terrorists, the red team vs. the blue
team, our guys vs. their guys – where does one go next? It is one thing for
aspirational policymakers to note in passing that “security problems for the
great powers now emanate from weak states,” but quite another for social

39 Treating the ethnic group as the unit of analysis blurs distinctions between ruthless elites –
who often generate fear and hatred cynically, with the expectation that they will be able to
benefit in the aftermath of violence – and the ethnic masses who pay the costs of fighting.
This critique has been well-treated by Gagnon (1995), Brown (1997), and De Figueiredo and
Weingast (1999).
Implications 193

scientists to grapple with what this pithy observation actually implies for the
future of security studies. This book, with its theoretical and empirical emphasis
on the strategies of individual people (warlords), contributes to a literature that
emphasizes agency over structure in civil war zones. But the research design
endorsed by this book does, I fear, contribute to a dark trend: The size of
the social unit that can be credibly described as a security threat is relentlessly
shrinking, and the state’s focus on these smaller units is relentlessly growing.
Chapter 3 describes how and why certain individuals succeeded in politicizing
grievance by quickly assembling private armies. To anticipate and cauterize this
kind of threat in an era of mass casualty terrorism, some state agents already
take it upon themselves to worry about individual people, map their social
networks and institutional endowments, read their private correspondence,
and ultimately judge whether or not they hold the “wrong beliefs.” It is not
clear where this line of inquiry should end. As a scientist, I am afraid that
there is no end in sight. Sometimes it is possible to infer intent by the content
of speech; sometimes language is intended to be “cheap talk.” It is trivial to
justify defensive war and preventative strikes if one takes certain angry people
at their word. The world is full of angry voices and crazy ideas. Though I am
not optimistic that there is a viable research program in studying grievance
narratives in a comparative fashion, I am positive that the content of speech
by people in the “belief creation” business – elected officials, clerics, teachers,
and activists – is going to be the subject of sustained scrutiny by state-funded,
computer-literate social scientists. I predict that epidemiological techniques will
be brought to bear to track the prevalence and virulence of “dangerous ideas.”
All of this will occur in the name of national security.
These observations and predictions are not meant to be an apologetic for
authoritarian practices. But I do not think liberal idealists or civil libertarians
do themselves any favors by pretending that the problem is going to disappear.
Scholars interested in understanding how to cauterize violence quickly and
efficiently have no real alternative than to study autocratic practices, even
if only in comparative context, even if only as a component of long-term
peace-making processes in the liberal mode. I predict that in the coming
decades, our field will devote considerably more attention to the direct and
indirect technologies autocratic governments use to manage their populations.
In so doing, scholars will surely notice analogous practices in well-governed
“open access” societies.
The problem with making responsible inferences, testing theories rigorously,
and generally advancing a research program on these kinds of questions is that
those sort of data tend to be tremendously sensitive and difficult to gather. Even
open, liberal, and strong societies enforce gag rules on topics that can inflame
popular imaginations. The machinations of the secret police and domestic
intelligence agencies are state secrets everywhere. All states have good reasons
to keep researchers at arms’ length from the inner workings of their police and
military. It is not surprising that competent, strong states are much better at
194 Jesse Driscoll

doing it.40 But if we acknowledge this to be the case, we must also acknowledge
that this creates a real inference problem for the research community. If the kind
of research that generates theory and powerful narratives can be gathered only
in the subset of countries where security services have broken down and allow
researchers in, or where many of the basic tasks of governance is carried out
by international NGOs under the watchful eye of the United Nations, we may
be building theory based on a subset of civil wars that are not representative of
what civil war settlements are likely to look like in the future. It may be telling,
in this regard, that most civil wars occur in Asia, but most of the best social
science on rebel groups and civil war dynamics since the end of the Cold War
has emerged from countries in Africa.
Many Russians remember what it was like to live in a police state
whose ideology was validated by confident social scientists. Today, all ethnic
Russians live in close proximity to angry nihilists from different moral
communities – potentially violent people who nurse real historical grievances.
Their experiences may have something to teach the West about the limits of
power in relation to populations that do not particularly want to be policed by
a foreign-funded gendarmerie (or, one might add, be studied by well-meaning
foreign social scientists). A Russian officer, when I asked what he meant when
he said “all Chechens are crazy,” met my eyes, and with great seriousness traced
his finger in a line across the table:

“We don’t try to understand them. That’s what you do … To try to understand what
is in their heads – that is an arrogant thing, you know. We have our culture. They
have theirs. Maybe they are like us, and they don’t always understand themselves. In
war … we decide where the line is, and then we kill who crosses it. Let them believe
what they want about us. If they don’t cross the line, it’s good for me. Who cares the
‘why.”’41

If I were certain that I agreed with these sentiments, I would never have gone
so far away from my home or stayed gone for so long. But if I were certain that
these sentiments were irrelevant to the Russian cauterization of its peripheral
wars, I would not reproduce them here. At the risk of cultural stereotyping:
Russian foreign policy professionals were more willing than Americans to
assume the worst about Georgian and Tajik social actors. Russian military
professionals took for granted that their presence would be exploited in cynical
and unpredictable ways by criminals and savages. Russia’s ability to credibly
commit to ignoring local nuance may have been an asset at times, if the logic of

40 It may be that there is no practical solution to this problem, at least when dealing with ongoing
civil wars. Collaboration with historians who can manage state security archives is where
I anticipate new frontiers of empirical research. Alternatively, it is not difficult to imagine
parallel sorts of coding and data aggregation projects, using crime statistics and field data on
police and military characteristics across a variety of states facing low-intensity insurgencies.
41 Interview conducted in Chicago, May 18, 2010.
Implications 195

this book is correct.42 What interested Moscow, and what their policies actually
incentivized, was the creation of a minimum threshold of stability and order. A
government receptive to their interests could not be guaranteed. The civil wars
provided an opportunity for Russia to sort the military strength of potential
local partners and crudely leverage emerging coalitions in both states. They did
not really try to do much more. Behaviors were judged, not motivations. This
conservative approach, and a demonstrated willingness to allow a very ugly and
cynical drama to play out under their noses in Georgia and Tajikistan, had the
effect of limiting Russian liability for outcomes on the mountainous periphery.
When things went wrong, the fault did not necessarily lie with Russia. That
was important.
But over the long term, given the high stakes associated with mass-casualty
terrorism, I fear that deterrence and in-group policing will only be able to take
security studies up to the line traced by the Russian officer. Crossing that line
requires very detailed kinds of individualized data. Many eager scholars and
graduate students will surely talk themselves into helping the security forces
of weak states collect and analyze these data. This will happen in the name
of providing global public goods, in the name of combatting terrorists, in the
name of helping the trains run on time. The Russian officer may have been
ahead of the counterinsurgency curve. Somewhere on the far side of his line is
what George Orwell called “thoughtcrime.”
This has implications for how our research community thinks and talks
about the people we study – and particularly how we discuss the phenomenon
of “weak states.” The experiences of fully collapsed states in West Africa and
Yugoslavia in the mid-1990s are not representative of what goes on in most of
the world’s civil war zones. Yet they generated a powerful body of discourse –
appeals to “anarchy” and “state failure” – which elites in the developing world,
and agents of foreign security organs who are necessary intelligence partners,
now cynically reappropriate. They too often use this language to absolve
themselves of responsibility for their own actions. State weakness is real, of
course, but “the absence of the state” can be the result of strategic design. State
weakness can even occasionally be fully theatrical. What is publicly justified
as an absence of capacity is often a lack of political will. Selling guns to local
elites, sowing fear, and then saying “anarchy” to international observers is a
common governance strategy in some parts of the world. It is important for
researchers working in conflict zones to consider a variety of analogies and
rival explanations for violence, rather than taking strategic elites at their word.
Social scientists are beginning to develop understandings of how violence
markets work in weak states. This volume contributes to that literature,

42 Roy (2000) argued persuasively that Russian commitment to neither monitor nor meddle in
post-independence affairs was made credible by generations of institutionalized pro-Slavic
racism – but the multicultural hegemony of the United States does not leave its agents with
that easy option.
196 Jesse Driscoll

emphasizing that relationships between militia captains and recruits involved

a set of calculated risks, renewed as conflict parameters changed. I emphasize
that outsiders’ ability to model or second-guess how these processes would
actually play out was very limited. The narratives that will endure in post war
social memory are probably stories where local heroes had agency in resolving
their own wars. The task will fall to future multilingual historians with a real
stake in cultivating local social memory – whose first languages are Russian,
Georgian, Abkhaz, Azeri, Armenian, Tajik, Shugni, and Uzbek – to decide what
empirical details are worth remembering and passing along to their students.
Meanwhile, if American social scientists cannot credibly commit to ignoring
distant war zones, or to wondering what goes through the heads of very angry
subalterns, we can at least tell other peoples’ stories in a way that emphasizes
their agency.43 I have tried to remain carefully cautious about claiming to know
what other people believed, or to passing judgement on the appropriateness of
those beliefs. This book’s approach is anchored in a “thin rationality” that
does not depend on ideological or religious metaphors. This middle ground is
aesthetically pleasing to me, and facilitates the cross-cultural conversations that
I understand to be the essence of the comparative project in political science.

43 I am increasingly of the opinion that if there are normative reasons to “get the story right,” then
there are also normative reasons to rewrite that story in the universal language of mathematics.
Fearon (1997) argues that a great deal of diplomacy, translation, and communication might
be reduced to an assurance game, where uncertain actors are trying to determine whether they
understand the essence of the strategic situation in the same way. Haas (1992) and Satz and
Ferejohn (1994) suggest that as a scholarly shorthand develops, ideas transverse disciplinary
and linguistic barriers at progressively more rapid speed. Given that rational choice approaches
show no sign of collapsing under the weight of accumulated contradictions on their own
accord, the forced perspective of different roles, games, and constraints has independent value.
Appendix A

Case Selection and External Validity

This short appendix is meant to supplement empirical claims made in the first
and last chapters of this book. It has two purposes. The first is to demonstrate
that the countries in this study are outliers in terms of civil war duration.
Even with numerous statistical controls, the wars resulting from the breakup
of the USSR were unusually short. The second is to demonstrate that the broad

Legend
Border of Former USSR
Tajikistan and Georgia (national claims –
Abkhazia and South Ossetia not shown)
Kilometers
0 125 250 500 750 1,000

Cau
casu
s Mo
Black unta
ins
Sea Caspian
Sea

Pamir Mounta
ins

Pe
rs ian
Gu
lf

figure Appendix A.1. Georgia and Tajikistan: unusually mountainous terrain.

197
198 Case Selection and External Validity

50

War Length (years–right censored) 40

30

20

10
TAJIKISTAN
GEORGIA
0
0 1 2 3 4 5
Log of Mountainous Terrain

figure a.2. Civil wars since 1945: mountainous insurgencies last longer on average.

contours of the settlement of the wars in these two countries can be squared
with patterns observed after other civil wars in other parts of the globe since
the end of the Cold War.
The coding conventions employed in this study follow Fearon and Laitin
(2003), defining a civil war as a conflict taking place within a recognized state
that results in 1,000 or more deaths, with more than 100 occurring on the
government side (thus excluding one-sided massacres). These coding rules may
overaggregate the phenomenon of “civil war,” pooling very distinct processes
together to gain statistical power. In an important article, “Why Do Some
Civil Wars Last So Much Longer Than Others?” James Fearon (2004) notes
that when one inductively examines the cases in standard datasets, one finds
remarkable heterogeneity of the kinds of political violence that are counted
as civil wars. When similar kinds of events are clustered, the “type” of war
emerges as the best predictor of war duration.
As Figure a.1 shows, both Georgia and Tajikistan contain formidable
mountains. To consider the effect of mountains on civil war length, Figure
a.2 displays a simple scatterplot of mountainous terrain plotted against civil
war length, using the Fearon (2004) dataset. Mountains are very good places
for insurgents to hide from the state, so guerrilla wars tend to go on longer
in very mountainous countries.1 Tajikistan and Georgia – the cases that

1 For this reason, mountainous terrain is a critical component of the identification strategy for
the latent “weak state” variable in Fearon and Laitin (2003). The authors demonstrate that
mountainous terrain is a statistically robust predictor of civil war outbreak.
Case Selection and External Validity 199

table a.1. Why Do Some Civil Wars Last So Much Longer Than Others?

Variables (1) (2) (3)

Fearon Control Controls, without
(2004) Variables Post-Soviet Dummy
Coup or Revolution 0.382∗∗∗ . 0.398∗∗∗ 0.402∗∗∗
(0.083) . (0.090) (0.093)
Sons of the Soil 2.135∗∗∗ . 2.060∗∗∗ 1.992∗∗∗
(0.510) . (0.490) (0.485)
∗∗∗
Drugs 1.935∗∗∗ . 1.892 1.948∗∗∗
(0.459) . (0.454) (0.475)
Mountainous Terrain 1.162∗∗ . 1.145∗ 1.130∗
(0.082) . (0.081) (0.081)
Post-Soviet Wars 0.302∗∗∗ . 0.364∗∗∗
(0.120) . (0.153)
Anti-colonial Wars 0.806 0.826
(0.242) (0.253)
Ethnic Fractionalization . 1.255 1.374
(0.427) (0.468)
Population 1 1
(5.20e−10 ) (5.24e−10 )
Cold War (pre-1991) 1.223 1.359∗
(0.243) (0.258)
p 1.073 . 1.088 −0.665
SE(p) 0.071 . 0.074 −0.665
Observations 149 . 149 149
Observations (ended) 135 . 135 135
Weibull regression results with civil war length (measured in years) as the dependent variable.
Coefficients in the table report estimated multiplicative effects of a one-unit change in the
independent variable on the average length of civil war. ***p < 0.01; **p < 0.05; *p < 0.1.

constitute the bulk of this book – do not conform to this trend. Contrary
to the predictions of area specialists at the time – who confidently predicted
long-running, hard-to-end conflicts based on hardened clan and ethnic hatred –
the wars were resolved with unusual speed. When relevant control variables
are taken into account, the puzzle of this speedy resolution emerges even more
clearly. Model 1 of Table a.1 replicates the empirical findings in Fearon (2004)
using an updated dataset of civil wars between 1945 and 2008.2 The statistical

2 This model adds a continuous mountainous terrain variable that emerges as statistically
significant in the Fearon duration replication data, but does not feature in the write-up because
of its low predictive power. In addition to expanding the temporal range of analysis to 2008, this
dataset makes a number of small modifications to the original dataset based on evolving coding
decisions since the original dataset construction. Most of variables presented in this model are
coded according to variable descriptions in Fearon (2004). For Coup or Revolution, a country
200 Case Selection and External Validity

and substantive significance of the coefficients is unambigious: postcommunist

wars lasted about a third as long as civil wars in other parts of the world. The
modal civil war since 1945 has been a rural insurgency in a relatively poor
state, pitting ill-equipped regime militias against rebel armies. Mountainous
terrain and narcotics production are associated with longer civil wars because
they lower the costs of initiating or sustaining rebellions. The so-called “Sons of
the Soil” rebellions – land disputes between an indigenous minority group that
fears cultural extinction and state-supported immigrants hungry for land – are
difficult to end. Coups and urban revolutions, by contrast, tend to be centered
in the capital city and feature “tipping” dynamics. The rebellion either succeeds
in toppling the government or it fails – but whatever happens, it happens
quickly.3
Model 2 in Table a.1 demonstrates that many structural features of the state
have little or no predictive value once the general type of civil war is taken into
account.4 Model 3 in the same table, which does not include a dummy variable

is assigned a “1” if the civil war is initiated with either a split in the military (“coup”) or popular
social movement in the capital in the style of the French or Iranian revolution (“revolution”),
and “0” otherwise. For Sons of the Soil, a country is assigned a “1” according to Fearon coding
rules, which relate to the internal immigration dynamics that define the identities of incumbents
and insurgent groups, and “0” otherwise. Drugs is a binary variable that has been recoded in a
number of cases, coded “1” in cases where a rebel group in the country funds itself substantially
through the use of narcotics or other contraband and “0” otherwise. The Post-Soviet Wars
include only the conflicts that emerged from the disintegration of the Soviet Union proper,
excluding wars in Eastern Europe. In these data, this variable does not include Bosnia and
the wars of Yugoslav secession (including Kosovo), the Romanian revolution, or the second
Chechen war. This leaves the first war in Chechnya, Tajikistan, Georgia’s various wars, the
Karabakh conflict in Azerbaijan, and Moldova. To avoid arbitrary coding decisions, I included
control variables that do not vary over the course of the civil war itself, and did not include
variables that are clearly endogenous to the course of the war itself (and hence the dependent
variable), such as fragmentation of rebel groups, battle deaths, GDP loss due to the civil war,
and the like. Other variables are reappropriated from Fearon and Laitin (2003), Fearon (2004),
or hand-coded. Following convention, I use 1997 as the end-point for the Tajik civil war.
3 Fearon admits that his model does not persuasively account for the post-Soviet wars quick
end: “In the model, increasing one side’s probability of decisive victory shortens expected
war duration. However, the thrust of the analytical results on relative military capabilities
is that matters are complicated, since imbalanced capabilities tend to reduce prospects for
a negotiated settlement while balanced capabilities increase them. … In addition, the model
highlights the problem of untangling relative capabilities from the propensity of different
capabilities to produce decisive victory or stalemate.” Fearon (2004), 298. Political scientists
tend to reason that foreign military aid to rebels against a weak government should prolong
civil wars. Nicholas Sambanis (2002), in a review of the large-n statistical literature on war
settlement, makes this point forcefully: foreign interventions into civil war zones are often
thought to strengthen rebellions that would otherwise “quickly be crushed by the government.”
222. A stable settlement pattern allowing for prolonged “frozen” conflicts is not discussed in
Sambanis and is rarely considered by civil war specialists in political science.
4 Anti-Colonial Wars are wars fought against a European metropole, primarily during the period
of decolonization. Earlier statistical probes suggested that these wars were systematically shorter
than their counterparts. Various working papers discussed in Sambanis (2002) support the
finding that ethnic fractionalization increases civil war length, so an Ethnic Fractionalization
Case Selection and External Validity 201

for the post-Soviet wars, is included to show that although post-1991 civil wars
tend to be resolved faster than wars fought against a backdrop of superpower
competition, this trend is statistically attributable to the rapid resolution of
wars along Russia’s new frontier.5 It would be reassuring to believe that
benign aspects of the post–Cold War world system – such as the replacement
of superpower “proxy wars” with the emergence of regular multilateral
peacekeeping missions under UN auspices – are responsible for this statistical
trend. Though these data are not inconsistent with that interpretation, the
mechanisms suggested by this volume may be an understudied components of
the system-level pattern.
In the formal model that anchors this book, coups in postwar states should
usually not occur. Presidents should be able to anticipate them and buy them
off. In the real world coups do occur. Approximately one in five of the
“Peace Processes” in the Doyle and Sambanis (2006) dataset are interrupted by
coups. That means that 80 percent of postwar governments manage to insulate
themselves from coups, by either good fortune or good strategy. Are the subset
of postwar governments that are capable of coup-proofing their regimes less
likely to return to war?
Yes, they are. Empirical data from the Doyle and Sambanis (2006) dataset,
which uses the “Peace Process” since 1945 as its unit of analysis, is well suited
to address this question.6 The primary explanatory variable of interest in this
analysis is whether the peace process was interrupted due to coalition infighting
and instability, punctuated by a coup. I developed an original coups variable.
Using the Goemans, Gleditsch, and Chiozza (2009) leadership dataset as a
baseline, I hand-coded coups when leaders left power in an irregular manner,
internal power struggle, or an uprising by armed agents of the state.7 The

control was added. The Cold War variable is coded “1” if a conflict was initiated before 1991,
and “0” otherwise. Owing to space constraints, additional model specifications were omitted.
Including regional or decade dummy variables, GDP-per capita measures (for the beginning
and end of the war) and other control variables had no statistically significant impact beyond
cluttering the model.
5 Various mechanisms by which superpower competition in the Cold War affected civil conflict
are reviewed in Kalyvas and Balcells (2011).
6 Their data codes a variety of relevant characteristics, including data on the host state, the war
itself, and the interveners. I reproduce several of their measures as control variables here. The
Politicide variable is a binary variable for whether the civil war included instances of mass
killing directed against an ethnic, religious, or political group. Civil War Duration is measured
in months. Rebel Victory is a binary variable for whether the peace process was initiated at
the end of a decisive military victory by a nonstate actor. GDP is a measure of per capita
GDP from the start of the war, taken from Fearon and Laitin (2003). Ethnic Heterogeniety
is a 0–1 scale for ethnic fractionalization developed by James Fearon (2003). Strong UN
indicates the presence of a transformation United Nations mission – a cumulative category
for Multinational Peacekeeping Operations, Peace Enforcement Operations, Transnational
Administration, etc.
7 The variable correlates with the dependent variable (Peace Process Failure) at 0.18, with a
variety of cases (Argentina, Guinea Bissau, Laos, Pakistan, Paraguay) having coups but no civil
war resumption and many more having civil war resumptions without coups.
202 Case Selection and External Validity

table a.2. Which Civil Wars are Most Likely to Restart?

(1) (2) (3)

Variables Basic “Decisive Victory” Best Fit
Coup (Goemans, & Gleditsch, and 0.989∗∗ 1.249∗∗ 1.065∗
Chiozza, 2009) (0.477) (0.618) (0.626)
Politicide 1.224∗∗ 1.741∗∗∗
(0.505) (0.602)
Civil War Duration −0.006∗∗ −0.007∗∗
(0.003) (0.003)
Rebel Victory −0.954∗ −1.208∗
(0.559) (0.701)
Per Capita GDP −0.276∗∗
(0.107)
Ethnic Heterogeniety 2.679∗∗∗
(0.824)
Strong UN Mission −3.378∗∗∗
(0.876)
Constant −0.073 0.109 −0.665
(0.208) (0.336) (0.517)
Observations 138 133 133
Pseudo r2 0.026 0.111 0.270
Logit regression results with Failed Peace Process as dependent variable Clustered Standard Errors
by Country. ***p < 0.01; **p < 0.05; *p < 0.1.

outcome variable for the models displayed in Table a.2 is a binary indicator
Peace Process Failure, which measures whether the civil war restarted within
five years of the beginning of the peace process. The statistical model used to
analyze these data was a simple logit regression with clustered standard errors
by country.
During the fragile transition after a peace process, a coup within the
incumbent government makes civil wars more likely to restart. Table a.3
displays the substantive impact of varying a few “usual suspects” from the
civil war literature.8 These results validate general intuitions: the small fraction
of civil wars lucky enough to receive strong international peace enforcement

8 Because the substantive interpretation of logit coefficients in Table a.2 is not straightforward,
I use CLARIFY to run Monte Carlo simulations of the model to produce the parameter
values, using Model 3 in Table a.2. The change in outcome values for binary variables is
displayed for switching the independent variable to 1; for continuous variables the change
comes from switching the independent variable from the 25th percentile to the 75th percentile.
For simplicity of presentation, these results are interpreted in terms of their individual impact
on the probability that a country will “return to war,” while holding all variables at either their
mean or median value.
Case Selection and External Validity 203

table a.3. Model 3: Internal Politics and Civil War Resumption (Simulations)

Proposed All at All at

Variable Mean Value Variation Mean (%) Median (%)
Return to War (DV) 0.57 0–1 49 44
(Baseline) (Baseline)
Coup (G&G) 0.20 0–1 +19 +22
Politicide 0.33 0–1 +24 +31
Duration (months) 0.73 14–111 −19 −17
Rebel Victory 0.22 0–1 −21 −27
Per Capita GDP 1.79 0.59–2.34 −12 −13
Ethnic Heterogeneity 0.56 0.33–0.77 +29 +26
Strong UN Mission 0.09 0–1 −43 −49

missions – generally cases from the 1990s – are good candidates for peace
to endure. International interventions matter, third-party peace enforcement
works, civil wars settled by rebel victory at the end of a long civil war are likely
to stay settled, mass killing during the civil war raises the probability of the
war restarting, and countries that are poor or ethnically heterogeneous are at
a somewhat higher risk of renewed war. But Table a.3 shows that, even taking
all of this into account, countries that experience coups are about 20 percent
more likely to return to war than identical postwar regimes that successfully
manage intra-coalition politics.
The theory and the empirics of the cases suggest that armed actors had
incentives to build new institutions quickly to attract foreign wealth and
manage distributional politics among each other. But what kind of institutions?
Are there any patterns to the kinds of institutions that emerge and endure after
civil war? And do global patterns match with the story of patronage politics
buttressed by strong, personality-based party networks to monitor defections,
described in Chapter 5?
Yes, they do, though with the caveat that the answer depends critically
on what time period is being studied. Military regimes can certainly manage
distributional politics. These regimes, however, had their heyday in the 1970s,
against the backdrop of Cold War competition, and are relatively rare today.
For the purposes of exposition, postwar regime types were coded based on
what form of government was sustained for the five-year period after the
peace process commenced. Mechanically, I added a number of columns to the
Doyle and Sambanis (2006) replication dataset of all civil war settlements since
1945, based on a coding scheme developed by Magaloni (2008), who classifies
different kinds of dictatorships based on their “launching organization” and
the number of viable political parties. For simplicity in interpretation, binary
variables were identified for Military Dictatorships (23 percent of the sample),
Hegemonic Party Regimes (23 percent of the sample), Single Party Regimes
204 Case Selection and External Validity

(11 percent of the sample), and “true” Democracies (11 percent of the sample).
For the purposes of this statistical analysis, these categories are exclusive.9
Control variables were included for characteristics that might affect postwar
institutions (e.g., wealth, oil, the degree of fragmentation caused by the war,
and a few other common control variables).10
During the Cold War, coup leaders could seize power and then extort the
two great powers for recognition simultaneously – threatening, implicitly or
explicitly, to offer their territory and assets to the geopolitical competitor.
Marxist ideology provided a moral and scientific justification for single-party
dictatorship. With the end of the Cold War, Hegemonic Party systems – similar
to the caricature of Chicago politics, where one party wins predictably, but
voters are permitted to waste their vote on opposition parties if they choose –
come into prominence. UN peacekeepers were deployed more often, increasing
the prominence of the residual Transition category (a government assisted by a
transnational presence).
Some authoritarian regimes seem to be systematically better “coup-proofed”
than others, as the next set of regressions show. Each of the seven models is a
simple logit estimator with clustered standard errors by region, with Coup,
a binary indicator of a leader’s irregular exit in the first five years after the
initiation of the peace process, as the dependent variable. Fully institutionalized
democracies are the most coup-proof, as we see in Models 4, 5, and 7. The
11 percent of regimes that manage to emerge from civil war with consolidated
democratic institutions and the 9 percent of countries that received large
peace-enforcement missions were relatively successful at avoiding coups.11 In
the aftermath of civil conflict, regime investments in party infrastructure – with
elections used to legitimate the regime and then distribute patronage – inoculate
regimes against coups. Hegemonic Party regimes appear unusually resilient,
as do single party regimes. Military dictatorships are vulnerable. Table a.5
displays the substantive interpretation of the coefficients via simulation for

9 A Hegemonic Party regime is a regime that allows multiparty competition, but where the
incumbent dominates the state apparatus. A Single Party regime is one in which there is
only one legal party (overwhelmingly communist regimes). A Military regime is one in which
authority ultimately rests within the armed forces. Democracies are defined as countries that
achieve a Polity score of 6 or higher any time in the five years after the peace process.
10 Rather than fully replicate the full Magaoloni structure here, I also included a scaled polity
variable that runs from 0 to 20, which provides a rough measure of the distinction between
“soft” and “hard” authoritarian nondemocracies.
11 There is substantial crossover between these two categories, and a glance at the these
twenty-three cases suggests that they were not analogous to the highly fragmented political
reconstruction projects in Georgia and Tajikistan. Rather, they tend to be peripheral
secessionist insurgencies in states where democratic institutions were strong in the country’s
“core” (e.g., India, the Philippines) or countries where combatants came to the table
peace was underwritten UN peacekeeping and international peace guarantees (e.g., Cyprus,
Mozambique, South Africa, El Salvador).
 table a.4. Which Post-War Institutions Are Most Coup-Prone?

(1) (2) (3) (4) (5) (6) (7)

Hegemonic Party −2.158** −2.248** −2.468** −2.294** −2.437** −2.407** −3.33.9**
(0.251) (0.092) (0.248) (0.214) (0.251) (0.337) (0.375)
Military Regime 0.88.9* 0.854++ 0.734 0.179 0.674 0.819 −0.784
(0.458) (0.521) (0.615) (0.604) (0.558) (0.601) (0.966)
Single Party Regime −0.217 −0.630 −1.195 −0.756 −0.639 −2.224++
(1.028) (1.134) (1.002) (1.124) (1.136) (1.189)
Number of Warring Factions 0.165** 0.207** 0.188** 0.248** 0.1.95** 0.188**
(0.054) (0.049) (0.027) (0.038) (0.045) (0.033)
GDP per Capita (Start of War) −0.370* −0.376* −0.371++ −0.335 −0.364
(0.192) (0.156) (0.199) (0.230) (0.279)
Postwar Democratization =! (if =0) =! (if =0)
(Polity Score 6+) [18 [18
dropped] dropped]
Strong UN Mission =! (if =0) =! (if =0)
[12 [5
dropped] dropped]
Oil Exporter −0.374 −0.372
(1.020) (1.098)
Polity Score Scaled −0.049
(0−20) (0.071)
Constant −1.368* −1.827** −1.275* −0.655 −1.260++ −1.258* 0.507
(0.459) (0.421) (0.584) (0.450) (0.615) (0.582) (0.651)
Pseudo r2 (0.109) (0.122) (0.162) (0.184) (0.163) (0.164) (0.226)
N 138 134 134 116 122 134 93
Logit regressions with clustered standard errors by region, with “Coup within five years of the start of the Peace Process” as the dependent variable.
205

Robust standard errors are in parentheses. ++ p < 0.1; *p < 0.05; ** p < 0.01.
206 Case Selection and External Validity

table a.5. Institutions after Civil War: Practical Coup-Proofing (Simulations)

Proposed All at All at

Variable Mean Value Variation Mean (%) Median (%)
Coup (G&G) (DV) 0.20 0–1 15 27
(Baseline) (Baseline)
Hegemonic Party 0.23 0–1 −11 −23
Military Regime 0.23 0–1 +8 +6
Single Party Regime 0.11 0–1 −4 −8
Warring Factions 3.23 2–4 +4 +7
GDP Per Capita 1.79 0.59–2.34 −9 −12

Model 2, the most straightforward to interpret.12 Results are interpreted in

terms of their individual impact on the probability that a government will
suffer a coup, holding all other variables at either their mean or median value.
Postwar authoritarian institutions matter, and building parties after civil wars
is a common strategy for keeping mass violence from breaking out again.
The inference I draw from these trends is that Georgian and Tajik actors
engineered the same kinds of outcomes that can be observed in other states since
the end of the Cold War, only faster. Though the experiences of these two states
exposes processes of civil war termination that have resonance in other settings,
I leave rigorous cross-national tests to future work. Following King, Keohane,
and Verba (1994), most of my inferences come from “increasing my n” within
countries rather than across countries – comparing the life experiences of many
soldiers, and the consolidation strategies of many warlords, while holding the
real estate constant. It would be a simple matter to assemble anecdotes from
around the globe that echo the logic of the model. I prefer to let data sourced
from the Georgian and Tajik streets tell their own story.

12 Obviously, this model pools the twenty-three cases with either consolidated democracies or
strong UN Enforcement Missions (which both predict the outcome perfectly and are thus
unsuitable for simulations in CLARIFY). The substantive effects of this pooling are that the
“baseline” probability of a country suffering a coup is lower than it would be if these cases
were excluded. As previously, the change in outcome values for binary variables is displayed
for switching the independent variable to 1; for continuous variables the change comes from
switching the independent variable from the 25th percentile to the 75th percentile.
Appendix B

Mathematical Proofs

This appendix presents formal propositions to supplement the presentation in

Chapter 2. To recap:

• Each of n warlords chooses simultaneously whether to battle for total

dominance of a state (“Fight”), or to support the candidacy of a civilian
president (“Install”). Call the full set of warlords W. Call the total lootable
resources in the country v. If the total number of warlords supporting a
president is less than a commonly known stability threshold s, the outcome
is war. In war, each warlord who played “Fight” gets nv − c and each
warlord who played “Install” gets nv − c − w.
• Define the number of warlords who play “Install” as k. Call this subset
of k warlords W P . If k ≥ s, a president is installed. Warlords who played
“Fight” in the first stage get a reservation value r. The game ends for these
warlords.
• The lootable resources of the state v increase to v∗ . A figurehead president
P is installed. He selects l warlords, such that k ≥ l ≥ s. Call this subset of l
members W L . P proposes a distribution of v∗ among these l warlords and
himself, x = (xi , xj , . . . xq , xP ). Warlords in W P but not in W L get a payoff
of zero. The game ends for these warlords.
• Each warlord in W L observes his distribution and either “Accepts” or
“Coups” the president. Playing “coup” imposes costs c on a warlord, but
this warlord will succeed in installing himself as president (claiming v∗ )
with some probability p. The “Coup” payoff, then, is pv∗ − c. If s or more
warlords play “Accept,” the president’s distribution x is implemented. If
fewer than s warlords play “Accept,” the president receives zero and each
warlord receives his coup payoff. The game ends.

The appropriate solution concept for this game is a subgame perfect Nash
equilibrium (SPNE), in which no player can promise or threaten actions

207
208 Mathematical Proofs

that he would not take if presented with the option. To simplify matters
and highlight essentials, my analysis focuses on simple strategies in which a
warlord conditions second-stage actions only on whether or not a government
formed, without reference to particular composition of the coalition.1 Even
with this constraint, the game contains many such equilibria. Once we eliminate
strategies that are weakly dominated – in other words, once we restrict analysis
to equilibria sustained by strategies by which players could do no worse, but
possibly improve their welfare, regardless of strategies chosen by other players
– it turns out that only a few of these equilibria are important for analysis.

Proposition 1. This game has a subgame perfect Nash equilibrium in which no

player plays a weakly dominated strategy in which all warlords choose “Fight”
in the first period. Call this a State Failure equilibrium.

Proof. One has only to consider a defection by a single warlord in the first
stage. Consider the simple case where n = 3 and s = 2. If i knows that both
of the others will choose “Fight,” second-stage payoffs will not be realized. A
comparison between nv − c and nv − c − w makes “Fight” the best reply. Any
strategies off the equilibrium path can be chosen, but “All Fight” will still
remain an equilibrium.

This essentially captures the situation described in Chapter 3: many warlords

fight, each hoping to control the capital. This is an inefficient equilibrium for
two reasons. First, war is costly. Fighting destroys productive assets, leaving v
smaller for whatever warlord succeeds in the military contest. Second, fighting
in the first stage foregoes the wealth associated with international recognition
(v < v∗ ). Nevertheless, if other warlords are planning to try to violently seize
the statehouse, any warlord i can only make himself worse off by not taking
part in the scramble. As in Rousseau’s stag hunt and Jervis’s security dilemma,
gains from cooperation do not easily overcome incentives for defection when
trust is low and stakes are high.
We now move to analysis of the final subgame, with the assumption that
warlords will backward-induct strategies in the first “installation” stage of the
game based on expectations of what will unfold in the second “consolidation”
stage. Because warlords have the last move in the game, it is intuitive that they

1 It is reasonable to object that this is not how the game would play out. In a two-stage game,
warlords should be able to condition their strategies in the second stage on what becomes known
about the outcome of the first stage. In actual play, there are good reasons to expect that the
content of the coalition membership should matter. Warlord i should be able to identity of the
warlords who joined the government. His strategy should allow him to “Coup” if warlord j is
part of W P and “Agree” if j is not part of W P . In this more complex and realistic setting the
set of outcomes for the first round is an n-tuple containing 2n possibilities in the (pure) strategy
space and kaleidoscopic complexity as various actors attempt to “out think” each other. This is
certainly what happens when I play this game in class with my students.
Mathematical Proofs 209

will be well positioned to extort the president. Yet, as has been shown in many
contexts, an early mover with proposal power in a bargaining game can extract
substantial advantages from the ability to limit her strategic opponents’ choice
sets.2

Proposition 2. In the final subgame (starting with P’s proposal), there is always
a subgame perfect Nash equilibrium where P distributes x such that xi = pv∗ −c
to each of s warlords in W L .

Proof. First consider defection by a single warlord i who has been offered
xi = pv∗ − c. By changing strategies to “Coup,” i will only receive pv∗ − c,
which he is already getting as xi . Next, consider whether P, who can keep for
himself what he does not transfer, can improve his welfare by changing his
distribution. P knows he will receive 0 in the event of a successful coup by any
warlord i, or if fewer than s warlords play “Accept.” Looking down the game
tree, he knows he must devise transfer schemes that induce exactly s warlords
to play “Accept.” Because P gets to keep for himself whatever part of v∗ he does
not distribute in the form of xP , he loses utility if he transfers any more than
the minimum necessary. Consider the most constraining case for collusion,
where s = n. In this example, each warlord must be included in W L , and if a
single warlord plays “Coup” P will receive zero. P can pay each of n warlords
xi = pv∗ − c and keep a positive transfer xP = nc for himself (the rents from
sparing all warlords the cost of fighting). Because neither the president nor any
of the warlords can change strategies and improve their welfare, this is a SPNE.

Every warlord i ∈ W P 3 W L will receive zero. At the time that a warlord

chooses to play “Install” or “Fight” he understands that P, if installed, only
needs the loyalty of s warlords to pass the stability threshold,. Though there
is nothing in the structure of the game that requires that the president select
exactly s warlords to be in W L , as P gets to keep what he does not distribute, P
will have the ability and incentive to offer exactly s warlords exactly what they
would receive in a coup, and no more. P cannot credibly commit to paying
all k warlords who install him if k > s.

Proposition 3. In the final subgame, any outcome in which P is removed in a

coup, or any outcome in which the sum of P’s proposed transfer to warlords
x1 +x2 . . . xn exceeds s(pv∗ −c), requires the play of weakly dominated strategies
on the equilibrium path, off the equilibrium path, or both.

Proof. Because the president P gets to keep v∗ − l(pv∗ − c) for himself, his
payoff is strictly decreasing in l. P should want to include exactly s warlords
in l, which is the minimum necessary to keep himself in power. If c ≥ pv∗ ,

2 See Schelling (1960) and Osborne (2004) generally, but especially Ferejohn (1986).
210 Mathematical Proofs

then the president is no longer incentivized to keep the coalition small, but
can no longer credibly commit to any transfer of wealth to any warlord. As
shown in Proposition 2, P should always be able to stay in power through
some correctly-calculated allocation x. For a warlord i to be induced to play
“Accept,” he must be transferred xi ≥ pv∗ − c.

If warlords understand the broad contours of the game, in other words,

if they understand that the president will pay exactly s warlords their coup
value, understand that they are interchangeable, and they choose strategies
accordingly, then in the installation stage, they are essentially gambling when
they install a figurehead president. This gamble can be justified. It is possible
to support a subgame perfect Nash equilibrium in which every warlord opts to
join the state. I will refer to this as a full incorporation equilibrium.

Proposition 4. Subgame perfect Nash equilibria can exist in which (a) no

warlord plays a weakly dominated strategy and (b) all warlords play “Install”
in the first period. Call these Full Incorporation equilibria.

Proof. For “Install” to be a best reply, it must be true that r ≤ ( ks )(pv∗ − c).
This is true for every warlord i if r ≤ ( ns )(pv∗ − c). Many distributions of x by
P are supportable equilibria, but in each distribution the president will select s
warlords and transfer each of them pv∗ − c. There are n!s! SPNEs of this sort. For
example, if W={A, B, C, D, E, F}, and s = 5, there are six different distributions
of x that are six different SPNEs – one where each of A, B, C, D, E, and F is
transferred xi = 0, while the other five are transferred xi = pv∗ − c and P keeps
xp = v∗ − 5(pv∗ − c) for himself. If n = 6 and s = 4, there are thirty different
SPNEs. In each of these, four warlords receive xi = pv∗ − c, 2 warlords receive
zero, and the president retains xp = v∗ − 4(pv∗ − c). When “Install” is chosen,
this distribution x is unknown.

As a general principle, each warlord i should compare his life opportunities

in W P to his reservation value r. Because s is static and warlords are symmetric,
i’s utility is strictly decreasing in k. Every warlord who plays “Install” worsens
the W L lottery odds for every warlord in W P . Still, if r is low, v∗ is high, and
s is high, that is, the lottery odds are good, it is possible to support situations
where every warlord strictly prefers joining the state to staying outside of it.

Proposition 5. Subgame perfect Nash equilibria can exist in which no warlord

plays a weakly dominated strategy where some k ≥ s warlords play “Install”
and join W P , but at least one warlord plays “Fight” in the first period and
remains outside of the consolidating state. Call these Partial Incorporation
equilibria.
Mathematical Proofs 211

Proof. There can exist a k0 such that r ≤ ( ks0 )(pv∗ − c) but r > ( k0 +1
s
)(pv∗ − c).
If one more warlord were to enter, it would no longer pay (in expectation) for
any of them to enter.3 In this setting, k0 is approximately equal to ( sr )(pv∗ − c).

As a simple extension: Consider a more realistic model that introduces

initial heterogeneity in the warlords’ value for the outside option if the state
consolidates, but they opt to stay out of the state. Instead of having all warlords
symmetric in r, assume ri is not the same for all i ∈ W, and that they cannot
(at low cost) mimic the characteristics of the “high reservation wage” types to
get the best deal. In that case, F(.) represents the cumulative distribution of ri ,
meaning that F(z) is the share of warlords with ri ≤ z and nF(z) is the number
of warlords with ri ≤ z. The equilibrium is determined by a cutpoint value r∗
s
 
∗
r = (pv∗ − c)
nF(r∗ )
Note that the left-hand side of the equation is increasing in r while the
right-hand side is decreasing. This means that the equilibrium will be unique.
All warlords with ri > r∗ will stay out, and those with ri ≤ r∗ will enter. All
the comparative statics on s, v∗ , and the average r value from above still hold
in this richer setting. But now the game produces a natural result in which
warlords who stay out have an idiosyncratic feature – such as safe refuge across
an interstate border or sustained military support from a great power – that
gives them better payoffs outside the consolidating state.

3 Reaching this arrangement, where some stay in and others stay out, still requires that the
warlords solve a coordination problem among themselves. To understand why this is so, imagine
a symmetric mixed strategy equilibrium, where all the warlords are identical, and each warlord
plays “Install” with an identical equilibrium probability. In this case, where the question of
whether the stability threshold s is passed is resolved probabilistically, the players may or may
not reach consolidation (enough to select P) in equilibrium.
Appendix C

Ninety-Seven Anonymous Warlords

The purpose of this appendix is to provide details about the data on Tajik
warlords who joined the state and to expand on the brief presentation in
Chapter 5.
With the aid of research teams based in Bishkek and Dushanbe, I revisited
the Small Arms Survey’s secondary source materials. My aim was to identify
additional characteristics of each field commander’s private army and resolve
a number of inconsistencies in the application of coding categories. The
data collection effort, conceived in the spring of 2006, began as an attempt
to systematically collect information on recruitment techniques, control of
resources, political connections to groups in the capital, financial support, and
characteristics of command and control for each warlord who fought in the
Tajik civil war. Additional interviews with area specialists, military and embassy
professionals, and journalists filled in the gaps left by interviews with former
combatants. All final coding decisions were my own. None of my research
assistants retain project materials.
What is a warlord, exactly? The coding rules used admitted an additional
field commander to the dataset if (1) it was possible to find at least
three secondary or two primary sources that confirmed that an individual
actually existed (e.g., that the newly discovered commander was not simply
a pseudonym or nom de guerre), and (2) at least one source suggested that the
warlord could call on the services of at least twenty-five men through channels
other than the official state hierarchy (e.g., being a general or police colonel did
not automatically lead to inclusion in the dataset).1 The dataset likely includes
at least one individual who may have had a bigger public presence than he

1 The twenty-five threshold is admittedly arbitrary. Though most of the selective violence during
the consolidation phase was perpetrated (or threatened) by a small core of trusted killers kept
on close retainer, to put on a public face and “lobby,” it was important to be able to organize
rallies and occasionally stage large-scale violent demonstrations (see Chapter 4).

213
214 Ninety-Seven Anonymous Warlords

actually commanded in terms of street recruits, but because their names exist in
the historical record it was decided at the time that excluding them would have
been arbitrary. Individuals who remained completely outside of the political
consolidation scramble – apolitical criminals and drug dealers who kept their
heads down and never entered militia politics – are also excluded from these
data. These ninety-seven warlords represent what I believe to be the universe
of independent Tajik militia leaders as best as they could be identified.2 The
unit of analysis in the regression analysis that follows is the individual field
commander. The period of observation is 1992–2006.
The post-incorporation longevity of warlords in my dataset diverged
dramatically. A few were still affiliated with the regime at the end of the
study period. Many more were purged, jailed, or killed. The model, which
treats field commanders as symmetric and indistinguishable, makes no attempt
to explain this variation in side payoffs or tenure in the ruling coalition.3
Indeed, if it were possible to compare post-incorporation political fortunes,
the model’s prediction would be that, conditional on joining the state, all
warlords should be on equal footing when it comes to keeping their position
and avoiding a purge. This seems a poor fit with the narrative data presented
in the book. Ethnic minority warlords were easily marginalized, and strong

2 There are two caveats. First, There were thirteen warlords for whom it was not possible to
determine a precise place of birth. In these cases, two additional criteria were employed. For
eight of the observations, interviews with former combatants made it possible to determine the
area where the field commander’s men were primarily recruited. For the remaining five we coded
based on the region of the country where the warlord was known to carry out operations, on the
assumption that his men were probably relying on nearby family for shelter and aid. This criteria
may would have the effect of slightly inflating the percentage of Gharmis in the dataset. There
were five warlords where birthplace data existed but was contradictory, and we had no trusted
source to settle the issue. Five warlords were coded as having multiple birthplace regions. These
ambiguities mostly relate to the misuse of the term “Gharmi” to refer either to someone from
the region of Gharm or or a collective slur against the tens of thousands of families that were
forcibly relocated moved from the mountains (“Gharm”) to the cotton-producing lowlands as
part of Soviet population transfer policy in the 1940s, 50s, and 60s. In the summary statistics in
Table 4.3 each warlord is counted only once for each area; in the statistical analysis duplicate
birthplaces are “double-counted,” meaning the effects of the additional dummy variables are
incorporated into the Weibull simulation results. Family membership also posed a problem. In
more than two cases, brothers were identified as warlords operating in tandem. This created
a problem. Should these men should be treated as independent observations, or a single one?
It seems unlikely that they would turn on each other, and in practice the research team never
found an example of a particular pair of brothers shooting at each other. Yet if we choose to
treat them as a pooled observation, should we do the same for warlords who are cousins? Or
warlords from the same clan-regional faction? To avoid a difficult mapping project, a decision
was made to treat all warlords as independent observations unless they were immediate familial
relations. Individuals who were literally brothers were treated as single observation.
3 If the warlords themselves could have predicted ex ante which characteristics would maximize
their odds of successful coalition survival, I assume they would have adopted those character-
istics. Though clan/regional affiliation and place of birth are fixed for an individual, strategic
marriages allow flexibility even for these “immutable” characteristics.
Ninety-Seven Anonymous Warlords 215

Warlords Purged from Tajik State 1992–2006

Number of Warlords Purged/Demoted by Home Region and Year

0
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Kulobi Hissori
Pamiri Gharmi

figure c.1. The dependent variable: purges over time.

warlords seem to have done better than weak warlords. Statistical rigor is
the antidote to anecdote, however. A dataset of Tajik warlord biographies
allows for a transparent evaluation of the model’s most controversial starting
assumption.
The dependent variable used in this analysis is the number of years that
a warlord was permitted to take part in the bonanza of postwar state
corruption before he was dismissed or disappeared. Figure c.1 provides a
basic visualization of these purges, grayscale-coded by warlord home region.
Warlords were removed from the state apparatus at a rate of approximately
three per year. Every purge could be interpreted as a failure by a warlord to
correctly anticipate the value of acting to install a president – either strategic
misplay, or (as this book argues) a bet that simply did not pay out. Figure c.2
shows the Kaplan–Meier survival estimate, with “failure” defined as leaving
the state. For most field commanders, the arrangement that initially convinced
them to join the state was void within seven or eight years.
Survival analysis can be used to assess if regional patterns reported in
Chapter Five (Table 5.2) are spurious. Multivariate Weibull analysis displays
transparently which individuals were more or less likely to politically survive
successive rounds of coalition formation (with implications for physical
survival). The reported coefficients are expressed as the multiple by which
a field commander’s expected time in the state will change when the factor
is present. The independent variables in the dataset are either individual
characteristics of warlords or characteristics of their army. A binary UTO
216 Ninety-Seven Anonymous Warlords

How Long Before Warlords Are Purged from State?

Kaplan–Meier Survival Estimate
100%
Probability a Warlord Remains

75%
in the Tajik State

50%

25%

0%
0 5 10 15
Years Already Spent in the Tajik State

95% CI Survivor Function

figure c.2. Time trends: almost all warlords get purged sooner or later.

variable was created to capture the master cleavage of the civil war, coded
“1” if a warlord joined the state as part of the 1997 United Tajik Opposition
(UTO) amnesty agreement. As militias fought with similar weapons technology
and deployed light infantry tactics, an estimate of Troop Strength (the total
number of soldiers under a warlord’s command at the time of integration) was
triangulated through a variety of sources. As an additional indicator of relative
warlord power, a binary variable was constructed to identify commanders
that had careers in the Soviet army, the MVD (Ministerstvo Vnutrennikh Del)
(police), or KGB (security service). The Soviet Security variable was intended to
serve as a proxy for leadership and tactical training beyond the standard single
tour of conscripted service.4 A few trends emerge clearly from these data.

4 Although the secondary literature notes that these commanders had little training, it is still
striking how few of the field commanders had any real military expertise before the fighting
broke out. Despite the fact that one would think military and police professionals would
be perfectly positioned to take advantage of the breakdown of order, only thirteen of the
ninety-seven warlords identified by the coding project had previous experience in the Red
Army or the KGB, and eleven of the fifty-eight analyzed in the subsample of “joiners” came
out of the security structures. The armies that these men led were no larger on average than
the armies of their counterparts who lacked their institutional advantages. One explanation
for this might be that the Russian army stayed carefully neutral in the Tajik Civil War until a
clear military victor emerged and it could play kingmaker, and career military officers within
the army unlikely to be identified as “warlords” precisely because they had the resources to
remain aloof from consolidation politics and keep their names out of the newspapers. Another
potential explanation for the unusual trend in the data is that training in the security services
 table c.1. Which Warlords Are Purged Quickly? Which Warlords Survive?

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7

Hissori/Uzbek 0.361** 0.422** 0.412** 0.416** 0.387** 0.375** 0.315**
(0.097) (0.107) (0.108) (0.109) (0.112) (0.95) (0.106)
Gorno-Badakshon (Pamiri) 1.349 1.583++ 1.641++ 1.569++ 1.468 1.681* 1.530
(0.364) (0.408) (0.450) (0.408) (0.416) (0.427) (0.467)
Soviet Military/KGB 2.122** 2.125** 2.103** 2.159** 2.246** 2.286**
(0.621) (0.619) (0.620) (0.640) (0.644) (0.670)
Political Opposition (UTO) 0.914 0.872
(0.221) (0.224)
Gharmi (from Gharm region) 0.939 0.879
(0.245) (0.267)
Kulobi 0.835 0.736
(0.216) (0.223)
Strength/Troop Estimate 1.000 1.000
(0.000) (0.000)
p 1.469 1.567 1.575 1.569 1.556 1.626 1.627
SE (p) (0192) (0.201) (0.203) (0.201) (0.199) (0.214) (0.213)
N 58 58 58 58 58 58 58
N(ended) 42 42 42 42 42 42 42
Weibull regression with length of time serving in the state, or in cooperative arrangement with state officials (measured in years) as
the dependent variable. Coefficients in the table report the estimated multiplicative effect of a one-unit change in the independent
variable on the average duration that a warlord will remain in the state; e.g., 0.294 means that a one-unit change is associated with
a reduction in the mean length of time in the state by a factor of 3. Standard errors in parentheses. ++ p < 0.1; ∗ p < 0.05; ∗∗ p < 0.01.
217
218 Ninety-Seven Anonymous Warlords

The central trend of note is that most warlords were, statistically speaking,
equally likely to be purged in a given year. This is consistent with the theoretical
assumption that they were functionally indistinguishable coalition members.
Winners of the civil war were no more likely to survive successive rounds
of coalition reshuffling than losers. Whether one considers the civil war as
being fundamentally a matter of ideology (“UTO Islamists vs. PFT Rump
Communists”) or regional politics (“Kulob, Hissor, and Khojand vs. Gharm
and Badakhshon”), the master cleavage of the war provided no predictive
power in terms of which warlords were most likely to endure as coalition
members.
Warlord military power – measured by the number of troops at the time
of incorporation – had no statistically significant impact on warlord tenure
in the state. This suggests two things. First, the same “multiple equilibria”
problem that makes coalition formation games difficult for political scientists
to productively analyze from a distance made the consolidation process difficult
to navigate for strategic actors on the ground. Given the opportunity, a large
number of weak warlords could (and did) substitute for a single strong warlord
in securing domestic order.
Ethnicity matters. Minority warlords faced different opportunities and
constraints compared to warlords from the national group when navigat-
ing the contours of coalition politics. Pamiri warlords – who held out
against Rakhmonov’s consolidation project in the impenetrable mountains of
Badakhshon – tend to last a long time in the state once they joined it, about
50 percent longer than average. Hissori warlords, and other groups who had
historically traced their patronage to Tashkent, were by contrast easy targets
for purges once geopolitical frictions emerged between Russia and Uzbekistan.
They exit the ruling coalition about three times as fast as the average
warlord. Though it is unlikely that these trends could have been predicted
ex ante with either existing models of civil war settlement or expert area
knowledge, these trends contradict the model’s assumption of simple warlord
interchangeability.
Commanders who had prior careers in the Soviet security services emerged
as unusually successful survivors, lasting about twice as long as average. As
of 2006, five of these eleven warlords still held official state positions, putting
their survival rate far above average. My interpretation of this trend is that
access to clandestine networks of former Soviet security officials provided these
men advantages in identifying lucrative niches in the Tajik shadow state as
it consolidated. The president turned on warlords from his home region at
a rate that is virtually indistinguishable from the average. In this light, the
“Kulobization” of the state apparatus under Rakhmonov represents a general

allowed unusual social mobility – languages, contacts, and access to foreign currency – which
allowed these men to flee the state with their families between 1988 and1992.
Ninety-Seven Anonymous Warlords 219

victory for Kulobi apparatchiks at the expense of the men who fought to put
the civilian apparatus into power. But it is also interesting to note that the
subrepublic, local informal networks that mattered so much in mobilization –
clan, family, avlod, and kolkhoz ties – were less robust predictors of political
survival over time than the informal networks of security personnel.
The primary inference that I draw from these regression results is that the
ruling coalition changed over time. Warlords simply could not predict how the
coalition would change fifteen years into the future. Though there are certainly
omitted variables in this analysis, I am confident that there is nothing that
distant scholars could code ex post that would have been known to the actors
ex ante.5
Certain area specialists, who have no interest in external validity or
generalization beyond the Tajik case, might find it useful to lower the
microscope and add more granular data to the analysis. Modern social science
methods could shrink the error term on predictions of longevity in the Tajik
state even further. Software for mapping social networks or forensic accounting
technologies could be brought to bear. Drug routes and ministry positions could
be coded. When I began this data collection process, I had vague ambitions to
do all of these things. After all, I had chosen my research question, in part, out of
a desire to revisit an understudied period of history and bring new facts to light.
But as I describe in Chapter 1, at some point along the way I became uneasy
with the task to which I had set myself. Social memory is a complicated and
contingent thing. I did not trust all of the data I was collecting. I also began
to fear that my research team was taking risks by asking too many sensitive
questions. I remember the exact moment that it dawned on me that the process
of collecting systematic information on individual people, in the way that I
was doing it, was difficult to distinguish from putting together a targeting list.
As I slowly internalized the implications of this, I decided that continuing to
add either columns or rows to the dataset exposed all of my research assistants
to charges of treason or espionage. Drawing a map of these social networks
could easily have unanticipated effects, potentially exposing my host family in
Tajikistan or my research team to violence. To limit my own liability, and set
an example for future scholarship in this vein, replication data are stripped of
names and identifiers.

5 For additional visualizations of the trends described earlier, see Driscoll (2012).
References

Abduallev, Kamoludin, and Catherine Barnes. 2001. “The Politics of Compromise –

The Tajikistan Peace Process.” Conciliation Resources, London. Available on
http://www.c-r.org/accord/tajikistan
Akbarzadeh, Shahram. 1996. “Why Did Nationalism Fail in Tajikistan?” Europe-Asia
Studies 48(7):1105–1129.
Akbarzadeh, Shahram. 2001. Abdullajanov and the “Third Force.” Conciliation
Resources, chapter 7, Retrieved from http://www.c–r.org/our–work/accord/
tajikistan/third–force.php.
Akhmedov, Said. 1998. Tajikistan II: The Regional Conflict in Confessional and
International Context. In Conflicting Loyalties and the State in Post-Soviet Russia, ed.
Michael Waller, Bruno Coppiters, and Alexei Malashenko. Frank Cass, pp. 171–186.
Akiner, Shirin. 2001. Tajikistan: Disintegration or Reconciliation. The Royal Institute
of International Affairs, London.
Akiner, Shirin. 2005. “Violence in Andijan, 13 May 2005: An Independent Assessment.”
Silk Road Paper, Central Asia-Caucasus Institute Silk Road Studies Program.
Albini, Joseph, R. E. Rogers, and Victor Shabalin. 1997. “Russian Organized Crime: Its
History, Structure, and Function.” Trends in Organized Crime 3(1):39–40.
Allison, Graham, and Robert Blackwill. 1991. “America’s Stake in the Soviet Future.”
Foreign Affairs 70(3): 77–97.
Anchabadze, Jurij. 1999. History: The Modern Period. In The Abkhazians, ed. George
Hewitt. Curzon Press, pp. 132–146.
Anderson, Benedict. 1991. Imagined Communities: Reflections on the Origin and
Spread of Nationalism. Verso.
Anderson, Jon Lee. 2007. “Inside the Surge: The American Military Finds New
Allies, but at What Cost?” The New Yorker. Available on http://www.newyorker.
com/magazine/2007/11/19/inside-the-surge, last accessed November 19, 2014.
Areshidze, Irakly. 2007. Democracy and Autocracy in Eurasia: Georgia in Transition.
Michigan State University Press.
Atkin, Muriel. 1997a. Tajikistan: Reform, Reaction, and Civil War. In New States, New
Politics: Building Post-Soviet Nations, ed. Ian Bremmer and Ray Taras. Cambridge
University Press, pp. 604–634.

221
222 References

Atkin, Muriel. 1997b. Thwarted Democratization in Tajikistan. In Conflict, Cleavage,

and Change in Central Asia and the Caucasus, ed. B. Parrott and K. Dawisham.
Cambridge University Press, pp. 277–311.
Atkin, Muriel. 2002. Tajikistan: A President and His Rivals. In Power and Change in
Central Asia, ed. Sally N. Cummings. Routledge, pp. 97–130.
Aunger, Robert. 1995. “On Ethnography: Storytelling or Science?” Current Anthropol-
ogy 36(1):97–130.
Auten, Brian. 1996. “Tajikistan Today.” Studies in Conflict and Terrorism 19:199–212.
Aves, Jonathan. 1993. Underdevelopment and the New Georgian Nationalist Move-
ment. In In a Collapsing Empire: Underdevelopment, Ethnic Conflicts, and
Nationalisms in the Soviet Union, ed. Mark Buttino. Feltrinelli Editore Milano,
pp. 225–236.
Aves, Jonathan. 1996. Georgia: From Chaos to Stability? The Royal Institute of
International Affairs, London.
Ayres, Sabra. 2014. “Amid Staggering Destruction, Eastern Ukraine Looks to Rebuild.”
Al Jazeera America. Available on http://america.aljazeera.com/articles/2014/9/28/
east-ukraine-reconstruction.html, last accessed September 28, 2014.
Babak, Vladimir, Demian Vaisman, and Aryeh Wasserman. 2004. Republic of Tajikistan.
In Political Organization in Central Asia and Azerbaijan: Sources and Documents,
ed. Vladimir Babak, Demian Vaisman, and Aryeh Wasserman. Cummings Center for
Russian and Eastern European Studies. Frank Cass Publishers.
Balta, E. 2007. Military Success, State Capacity, and Internal War-Making in Russian
and Turkey. PhD thesis, City University of New York.
Barkey, Karen. 1994. Bandits and Bureaucrats. Cornell University Press.
Bates, Robert. 2001. Prosperity and Violence: The Political Economy of Development.
W. W. Norton and Company.
Bates, Robert H. 1984. Market and States in Tropical Africa. University of California
Press.
Bates, Robert H., Avner Greif, Margaret Levi, J. L. Rosenthal, and Barry Weingast.
1998. Analytic Narratives. Princeton University Press.
Bates, Robert H., Avner Greif, Margaret Levi, J. L. Rosenthal, and Barry Weingast.
2000. “The Analytic Narrative Project.” The American Political Science Review
94(3):696–702.
BBC. 2014. “Russia’s Lavrov Says Time for a ’Reset 2.0’ in US ties.” BBC News. Avail-
able on http://www.bbc.com/news/world-europe-29401840, last accessed September
28, 2014.
Beissinger, Mark. 2013. “The Semblance of Democratic Revolution: Coalitions in
Ukraine’s Orange Revolution.” American Political Science Review 107(3):1–19.
Beissinger, Mark R. 2002. Nationalist Mobilization and the Collapse of the Soviet State:
A Tidal Approach to the Study of Nationalism. Cambridge University Press.
Berman, Eli. 2009. Radical, Religious and Violent: The New Economics of Terrorism.
The MIT Press.
Birnbaum, Michael. 2014. “Western Sanctions for Ukraine Conflict Hurt Russian
Economy, Open up Rifts.” Washington Post. Available on http://www.bbc.com/news/
world-europe-30436200
Blasi, J., M. Kroumova, and D. Kruse, eds. 1997. Kremlin Capitalism: Privatizing the
Russian Economy. Cornell University Press.
References 223

Blattman, Christopher and Eduard Miguel. 2009. “Civil War.” NBER Working Paper
No. w14801. March. National Bureau of Economic Research.
Blauvelt, Timothy. 2013. “Endurance of the Soviet Imperial Tongue: The Russian
Language in Contemporary Georgia.” Central Asian Survey 32(2):189–209.
Bornet, Jean-Marc. 1998. The International Committee of the Red Cross and the
Conflict in Tajikistan. In Tajikistan: The Trials of Independence, ed. Shirin Akiner
Mohammad-Reza Djalili, and Frederic Grare. Curzon Press, pp. 219–228.
Bowen, John, and Roger Petersen. 1999. Critical Comparisons. In Critical Comparisons
in Politics and Culture, ed. John Bowen and Roger Petersen. Cambridge University
Press, pp. 1–20.
Brass, Paul R. 1997. Theft of an Idol: Text and Context in the Representation of
Collective Violence. Princeton University Press.
Brenninkmeijer, Olivier A. J. 1998. International Concern for Tajikistan: UN and OSCE
Efforts to Promote Peace-Building and Democratisation. In Tajikistan: The Trials of
Independence, ed. Shirin Akiner Mohammad-Reza Djalili and Frederic Grare. Curzon
Press, pp. 180–216.
Brown, Archie. 2007. Seven Years that Changed the World. Oxford University Press.
Brown, Bess A. 1998. The Civil War in Tajikistan, 1992–1993. In Tajikistan: The Trials
of Independence, ed. Shirin Akiner Mohammad-Reza Djalili and Frederic Grare.
Curzon Press, pp. 86–96.
Brown, Michael E. 1997. The Causes of Internal Conflicts. MIT Press, pp. 1–31.
Buford, B. 1993. Among the Thugs. Vintage Press.
Bukharbaeva, Galima. 2005. “Testimony: Statement of Galima Bukharbaeva,
Correspondent—Institute for War and Peace Reporting.” Helsinki Commission
Hearing.
Bunce, Valerie. 1999. Subversive Institutions: The Design and Destruction of Socialism
and the State. Cambridge University Press.
Bushkov V., and Mikulskii, D. 1995. Anatomiya grazhdanskoi voiny v Tadzhikistane
(ethno-sotsial’nye protessy I politeskaya bor’ba, 1992–1995. Moscow.
Capote, Truman. 1965. In Cold Blood. First ed. Vintage International.
Carothers, Thomas, ed. 2006. Promoting Rule of Law Abroad: In Search of Knowledge.
Carnegie Endowment for International Peace, Washington, DC.
Centlivres Pierre, and Micheline Centlivres-Demont. 1998. Tajikistan and Afghanistan:
the Ethnic Groups on Either Side of the Border. In Tajikistan: The Trials of
Independence, ed. Shirin Akiner Mohammad-Reza Djalili, and Frederic Grare.
Curzon Press, pp. 3–13.
Chwe, Michael Suk-Young. 1998. “Culture, Circles, and Commercials: Publicity,
Common Knowledge, and Social Coordination.” Rationality and Society 10(1):
47–75.
Collier, P. H., and A. Hoeffler. 2004. “Greed and Grievance in Civil War.” Oxford
Economic Papers 56:563–595.
Collins, K. 2006. Clan Politics and Regime Transition in Central Asia. Cambridge
University Press.
Cooley, Alexander. 2012. Great Games, Local Rules: The New Great Power Contest in
Central Asia. Oxford University Press.
Cornell, Svante E. 2001. Small Nations and Great Powers: A Study of Ethnopolitical
Conflict in the Caucasus. Curzon Press.
224 References

Cornell, Svante E. 2002. “Autonomy as a Source of Conflict: Caucasian Conflicts in

Theoretical Perspective.” World Politics 54(2):245–276.
Cornell, Svante E. 2003. The Growing Threat of Transnational Crime. In The South
Caucasus: A Challenge for the E.U., ed. Dov Lynch. Chaillot Papers, No. 65.
Cunningham, David E. 2006. “Veto Players and Civil War Duration.” American Journal
of Political Science 50(4):875–892.
Daly, John C. K. 2005. “The Andijan Disturbances and Their Implications.”
Central Asia-Caucasus Analyst 29. Available on http://www.cacianalyst.org/
publications/analytical-articles/item/10049-analytical-articles-caci-analyst-2005-6-
29-art-10049.html, last accessed 29 June, 2005.
Daniel, E. Valentine. 1996. Charred Lullabies. Princeton University Press.
Darchiashvili, D. 1997a. Georgia: The Search for Security. Caucasus Working Papers,
CISAC, Stanford.
Darchiashvili, David. 1997b. “Georgia: A Hostage to Arms.” Saferworld Arms and
Security Programme, London.
Darchiashvili, David. 2005. Georgian Defense Policy and Military Reform. In Statehood
and Security: Georgia after the Rose Revolution, ed. Bruno Coppieters and
Robert Legvold. MIT Press, pp. 117–154.
Darchiashvili, David, and Ghia Nodia. 2003. The Weak State Syndrome and
Corruption. In Building Democracy in Georgia: Power Structures, the Weak State
Syndrome, and Corruption in Georgia. International Institute for Democracy and
Electoral Assistance.
Darden, Keith. 2008. “Graft and Governance: Corruption as an Informal Mechanism
of State Control.” Politics and Society 36(1):35–60.
Davenport, Christian. 2007. State Repression and the Domestic Democratic Peace.
Cambridge Studies in Comparative Politics. Cambridge University Press.
De Figueiredo, R., and B. Weingast. 1999. The Rationality of Fear. In Civil Wars,
Insecurity, and Intervention, ed. B. Walter and J. Snyder. Columbia University
Press.
de Waal, Thomas. 2003. Black Garden: Armenia and Azerbaijan through Peace and
War. New York University Press.
de Waal, Thomas. 2005. Georgia and Its Distant Neighbors. In Statehood and Security:
Georgia after the Rose Revolution, ed. Bruno Coppieters and Robert Legvold.
MIT Press, pp. 307–338.
Demetriou, S. 2002. “Politics from the Barrel of a Gun: Small Arms Proliferation and
Conflict in the Republic of Georgia.” Small Arms Survey, Geneva.
Denber, Rachel, Barnett Rubin, and Jeri Laber. 1993. “Human Rights in Tajikistan in
the Wake of Civil War.” Human Rights Watch (Organization), Helsinki Watch, New
York.
Derluguian, G. M. 2005. Bourdieu’s Secret Admirer in the Caucasus: A World-System
Biography. University of Chicago Press.
Devdariani, Jaba. 2005. Georgia and Russia: The Troubled Road to Accommodation.
In Statehood and Security: Georgia After the Rose Revolution, ed. Bruno Coppieters
and Robert Legvold. MIT Press, pp. 153–204.
Dixit, Avinash K. 2004. Lawlessness and Economics: Alternative Modes of Governance.
The Gorman Lectures in Economics. Princeton University Press.
References 225

Donetsk People’s Republic Plans to Switch to Ruble. 2014. Retrived from http://www.
kyivpost.com/content/ukraine/donetsk-peoples-republic-plans-to-switch-to-ruble-
not-ready-for-own-currency-366278.html
Doyle, M., and Sambanis, N., 2006. Making War and Building Peace. Princeton
University Press.
Driscoll, Jesse. 2012. “Commitment Problems or Bidding Wars? Rebel Fragmentation
as Peace Building.” Journal of Conflict Resolution 56(1):118–149.
Driscoll, Jesse, and Timothy Blauvelt. 2012. Translation Bottlenecks and Truth
Hierarchies: A Language Experiment in Post-War Georgia. In Association for the
Study of Nationalities Annual Meeting. Association for the Study of Nationalities,
Columbia University, New York.
Dudoignon, Stephane A. 1998. Political Parties and Forces in Tajiksistan, 1989–1993.
In Tajikistan: The Trials of Independence, ed. Shirin Akiner Mohammad-Reza Djalili
and Frederic Grare. Curzon Press, pp. 52–85.
Easterly, William. 2001. The Elusive Quest for Growth: Economists’ Adventures and
Misadventures in the Tropics. MIT Press.
Ekedahl, Carolyn McGiffert, and Melvin A. Goodman. 1997. The Wars of Eduard
Shevardnadze. Pennsylvania State University Press.
Ellis, S. 1999. The Mask of Anarchy: The Destruction of Liberia and the Religious
Dimension of an African Civil War. C. Hurst.
Elster, Jon. 1989. Nuts and Bolts for the Social Sciences. Cambridge University Press.
Elster, Jon. 2000. “Rational Choice History: A Case of Excessive Ambition.” American
Political Science Review 94(3):685–695.
Emerson, Robert M., Rachel I. Fretz, and Linda L. Shaw. 1995. Writing Ethnographic
Field Notes. University of Chicago Press.
Enloe, Cynthia H. 1980. Ethnic Soldiers. University of Georgia Press.
Espiritu, Yen Le. 1996. Racial Killing or Barroom Brawl? Multiple Explanations for the
Killing of Vincent Chin. In Riots and Pogroms, ed. Paul Brass. New York University
Press, pp. 221–234.
Fairbanks, C. H., Jr. 1995. “The Postcommunist Wars.” Journal of Democracy 6:18–34.
Fairbanks, C. H. 2002. Weak States and Private Armies. In Beyond State Crisis?:
Post-Colonial Africa and Post-Soviet Eurasia in Comparative Perspective, ed. M.
Beissinger and C. Young. Woodrow Wilson Press, pp. 129–159.
Fairbanks, Charles. 1996. Party and Ideology in the Former U.S.S.R. In Left, Right
and Center: Party and Ideology after the Cold War, ed. Richard Zinman and Jerry
Weinberger. Cornell University Press.
Fairbanks, Charles. 2004. “Georgia’s Rose Revolution.” Journal of Democracy,
110–124.
Fairbanks, Charles. 2012. “Georgia’s Prison Rape Scandal – And What It Says
About the Rose Revolution.” The Atlantic, Available on http://www.theatlantic.com/
international/archive/2012/09/georgias-prison-rape-scandal-and-what-it-says-about-
the-rose-revolution/262720/, last accessed September 24, 2012.
Fanon, Franz. 1968. The Wretched of the Earth. Grove Press.
Fearon, J. 2004. “Why Do Some Civil Wars Last So Much Longer Than Others?”
Journal of Peace Research 41(3):275–301.
Fearon, J. D. 2007. “Iraq’s Civil War.” Foreign Affairs 82(2):2–16.
226 References

Fearon, James. 1997. Deliberation as Discussion. In Deliberative Democracy, ed. Jon

Elster. Cambridge University Press, pp. 44–68.
Fearon, James. 2003. “Ethnic and Cultural Diversity by Country.” Journal of Economic
Growth 8(2):195–222.
Fearon, James, and David Laitin. 1996. “Explaining Interethnic Cooperation.”
American Political Science Review 90(4):715–735.
Fearon, James, and David Laitin. 2000. “Violence and the Social Construction of Ethnic
Identity.” International Organization 54(4):845–877.
Fearon, James, and David Laitin. 2003. “Ethnicity, Insurgency, and Civil War.”
American Political Science Review 97(1):75–90.
Fearon, James, and David Laitin. 2004. “Neotrusteeship and the Problem of Weak
States.” International Security 28(4):5–43.
Fearon, James, and David Laitin. 2007. “Civil War Termination.” Prepared for the 2007
Annual Meetings of the American Political Science Association.
Felter, Joseph. 2005. Taking Guns to a Knife Fight: Counterinsurgency in the
Philippines. Unpublished doctoral dissertation, Stanford University.
Ferejohn, J. 1991. Rationality and Interpretation: Parliamentary Elections in Early
Stuart England. In The Economic Approach to Politics: A Critical Reassessment of
the Theory of Rational Action, ed. Kristen Renwick Monroe. HarperCollins.
Ferejohn, John. 1986. “Incumbent Performance and Electoral Control.” Public Choice
30(Fall):5–25.
Ferejohn, John. 2004. Internal and External Explanation. In Problems and Methods in
the Study of Politics, ed. Ian Shapiro. Yale University Press, pp. 144–166.
Ferguson, James. 2005. “Seeing Like an Oil Company: Space, Security, and Global
Capital in Neoliberal Africa.” American Anthropologist 107(3):377–382.
Filkins, Dexter, and Carlotta Gall. 2010. “Taliban Leader in Secret Talks Was
an Impostor.” New York Times. November 22, 2010, page A1. Available on
http://www.nytimes.com/2010/11/23/world/asia/23kabul.html?pagewanted=all
Fortna, Page. 2008. Does Peacekeeping Work? Princeton University Press.
Gagnon, V. P. 1995. “Ethnic Nationalism and International Conflict: The Case of
Serbia.” International Security 19(3):130–166.
Gambetta, D. 1996. The Sicilian Mafia: The Business of Private Protection. Harvard
University Press.
Gambetta, Diego. 2002. Corruption: An Analytical Map. In Political Corruption in
Transition, ed. Andras Sajo and Stephen Kotkin. Central European University Press,
pp. 33–56.
Geertz, Clifford. 1977. The Interpretation of Cultures. Basic Books.
Girod, Desha. 2012. “Effective Foreign Aid Following Civil War: The Nonstrategic-
Desperation Hypothesis.” The American Journal of Political Science 56(1):188–201.
Girod, Desha and Jennifer Tobin, “Take the Money and Run: The Determinants of
Compliance with Aid Agreements,” International Organization (forthcoming).
Giustozzi, Antonio. 2009. Bureaucratic Facade and Political Realities of Disarmament
and Demobilization in Afghanistan. In Reintegrating Armed Groups after Conflict,
ed. Mats Berdal and David Ucko. Routledge Studies in Intervention and Statebuilding,
pp. 67–88.
Gleason, Gregory. 1997. Central Asian States: Discovering Independence. Westview
Press.
References 227

Goemans, H. E., K. S. Gleditsch, and G. Chiozza. 2009. “Introducing Archigos: A

Dataset of Political Leaders.” Journal of Peace Research 46: 269–283.
Goldenberg, Suzanne. 1994. Pride of Small Nations: The Caucasus and Post-Soviet
Disorder. Zed Books.
Goldgeier, James, and Michael McFaul. 2003. Power and Purpose. Oxford University
Press.
Goldsmith, Aurthur A. 2008. “Making the World Safe for Partial Democracy?
Questioning the Premises of Democracy Promotion.” International Security 33(2):
120–147.
Goldstone, Jack A. 1994. “Is Revolution Individually Rational? Groups and
Individuals in Revolutionary Collective Action.” Rationality and Society 6(1):
139–166.
Gorvin, Ian. 1998. The Human Rights Situation in Tajikistan (1992–1993). In
Tajikistan: The Trials of Independence, ed. Shirin Akiner Mohammad-Reza Djalili
and Frederic Grare. Curzon Press, pp. 229–236.
Gourevitch, Peter. 1978. “The Second Image Reversed: The International Sources of
Domestic Politics.” International Organization 32(4):881–912.
Gretsky, S. 1995. Civil War in Tajikistan: Causes, Development, and Prospects for
Peace. In Central Asia: Conflict, Resolution, and Change, ed. Roald Sagdeev and
Susan Eisenhower. Center for Post-Soviet Studies, The Eisenhower Institute, MD.
http://www.amazon.com/Central-Asia-Conflict-Resolution-Change/dp/0967023319
Gretsky, Sergei. 1997. Russia’s Foreign Policy Toward Central Asia. Carnegie Endow-
ment for International Peace. Moscow.
Grossman, Dave. 1996. On Killing. Back Bay Books.
Grossman, H. I. 1999. “Kleptocracy and Revolutions.” Oxford Economic Papers
51:267–283.
Grotz, Florian. 2001. Tajikistan. In Elections in Asia and the Pacific: A Data Handbook,
Vol. 1; Middle East, Central Asia and South Asia, ed. Florian Grotz, Dieter Nohlen,
and Christof Hartmann. Oxford University Press.
Haas, Peter M. 1992. “Introduction: Epistemic Communities and International Policy
Coordination.” International Organization 46(1):1–35.
Handelman, S. 1995. Comrade Criminal: Russia’s New Mafiya. Yale University Press.
Harris, Colette. 2004. Control and Subversion. Pluto Press.
Heathershaw, John. 2007. “Peacebuilding as Practice: Discources from Post-Conflict
Tajikistan.” International Peacekeeping 14(2):219–236.
Heathershaw, John. 2008. “Seeing Like the International Community: How Peacebuild-
ing Failed (and Survived) in Tajikistan.” Journal of Intervention and Statebuilding
2(3):329–351.
Heathershaw, John. 2009. Post-Conflict Tajikistan: The Politics of Peacebuilding and
the Emergence of Legitimate Order. Routledge.
Helpman, Elhanan. 2004. The Mystery of Economic Growth. Belknap Press of Harvard
University Press.
Herman, Edward S., and Noam Chomsky. 1987. Manufacturing Consent. Pantheon.
Herr, Michael. 1968. Dispatches. ed. Vintage International.
Hobbes, Thomas. 1651. Leviathan. 1991 reprint ed. Cambridge University Press.
Horsman, S. 1999. “Uzbekistan’s Involvement in the Tajik Civil War 1992–1997:
Demostic Considerations.” Central Asian Survey 18(1):37–48.
228 References

ICG. 2004. Tajikistan’s Politics: Confrontation or Consolidation? Technical Report.

International Crisis Group. Bishkek/Brussels.
ICG. 2005. “The Curse of Cotton: Central Asia’s Destructive Monoculture.” Asia
Report (93). Bishkek/Brussels, pages i-50 http://www.crisisgroup.org/en/regions/
asia/central-asia/093-the-curse-of-cotton-central-asias-destructive-monoculture
aspx.
IDEA. 2011. “IDEA: International Institute for Democracy and Electoral Assitance.
Voter Turnout for Tajikistan.” Retrieved from http://www.idea.int/vt/countryview.
cfm?CountryCode=TJ
Ilkhamov, Alisher. 2006. “The Phenomenology of ‘Akromiya’: Separating Fact from
Fiction.” China and Eurasia Forum Quarterly 4(39):39–48.
Jackall, Robert. 2005. Street Stories: The World of Police Detectives. Harvard University
Press.
Jackson, Robert H. 1990. Quasi-States: Sovereignty, International Relations, and the
Third World. Cambridge University Press.
Jalali, Ali A. 2006. “The Future of Afghanistan.” Parameters Spring 36(1):4–19.
Jankowski, Martin S. 1992. Islands in the Street: Gangs and American Urban Society.
University of California Press.
Jones, Stephen. 1997. Georgia: The Trauma of Statehood. In New States, New Politics:
Building Post-Soviet Nations, ed. Ian Bremmer and Ian Taras. Cambridge University
Press, pp. 505–546.
Jones, Stephen. 2006. “The Rose Revolution: A Revolution Without Revolutionaries?”
Cambridge Review of International Affairs 19(1):33–48.
Jones, Stephen F. 1996. Adventurers or Commanders? Civil-Military Relations in
Georgia Since Independence. In Civil-Military Relations in the Soviet and Yugoslav
Successor States, ed. Daniel Zirker and Constantine P. Danopoulos. Westview
Press.
Jones-Luong, Pauline. 2002. Institutional Change and Political Continuity in Post-Soviet
Central Asia: Power, Perceptions, and Pacts. Cambridge University Press.
Jonson, Lena, and Shirin Akiner. 1998. Tajik War: A Challenge to Russian Policy. Royal
Institute of International Affairs, London.
Kaldor, M. A. 1999. New Wars: Organized Violence in a Global Era. Polity Press.
Kalyvas, S. 2003. “The Ontology of “Political Violence”: Action and Identity in Civil
Wars.” Perspectives on Politics 1(3):475.
Kalyvas, S. 2004. “The Urban Bias in Research on Civil Wars.” Security Studies
13(3):160.
Kalyvas, S. 2006. The Logic of Violence in Civil War. Cambridge Series in Comparative
Politics. Cambridge University Press.
Kalyvas, S., and M. Kocher. 2007. “How “Free” Is Free Riding in Civil Wars? Violence,
Insurgency, and the Collective Action Problem.” World Politics 59:177–216.
Kalyvas, Stathis. 2007. Civil Wars. In Oxford Handbook of Political Science, ed. Carles
Boix and Susan Stokes. Oxford University Press, pp. 416–434.
Kalyvas, Stathis, and Laia Balcells. 2011. “International System and Technologies of
Rebellion: How the End of the Cold War Shaped Internal Conflict.” American Political
Science Review 104(3):415–429.
Katz, Jack. 1988. Seductions of Crime: Moral and Sensual Attractions in Doing Evil.
Basic Books.
References 229

Kaufmann, Chaim. 1996. “Possible and Impossible Solutions to Ethnic Civil Wars.”
International Security 20(4):136–175.
Keegan, J. 1994. The History of Warfare. Vintage International.
Keen, David. 1998. “The Economic Functions of Violence in Civil Wars.” Vol. 320:
Adelphi Paper. (Oxford University Press for International Institute for Strategic
Studies).
Kendzior, Sarah. 2006. “Inventing Akromiya: The Role of Uzbek Propagandists in the
Andijon Massacare.” Demokratizatsiya: The Journal of Post-Soviet Democratization
14(4):545–562.
Kennan, George F. 1997. “A Fateful Error.” New York Times, February 5.
King, Charles. 1997. “Ending Civil Wars.” Adelphi Paper (Oxford University Press for
International Institute for Strategic Studies).
King, Charles. 2000. “The Benefits of Ethnic War: Understanding Eurasia’s Unrecog-
nized States.” World Politics 53:524–552.
King, Charles. 2001a. “The Myth of Ethnic Warfare.” Foreign Affairs November/
December:n/a.
King, Charles. 2001b. “Potemkin Democracy.” The National Interest Summer.
King, Charles. 2004a. “The Micropolitics of Social Violence.” World Politics
56(3):431–455.
King, Charles. 2004b. “A Rose among Thorns: Georgia Makes Good.” Foreign Affairs
83(2):13–18.
King, Charles. 2008a. “The Five-Day War.” Foreign Affairs. November/December
2008 Issue; available on http://www.foreignaffairs.com/articles/64602/charles-king/
the-five-day-war
King, Charles. 2008b. “Salon Radio: Professor Charles King on Russia/Georgia.” Salon
Radio. Retrieved from http://www.salon.com/2008/08/11/king_7/
King, Gary, Robert Keohane, and Sidney Verba. 1994. Designing Social Inquiry.
Princeton University Press.
Kotkin, Stephen. 2002. “Trashcanistan: A Tour through Wreckage of the Soviet
Empire.” The New Republic Magazine, 226(14):26–38.
Kotkin, Stephen. 2008. Armageddon Averted. Oxford University Press.
Kramer, Andrew. 2014. “Ukraine Picks Motley Group to Exchange for Prisoners.” New
York Times, September 24, A4
Krasner, Stephen. 2001. Sovereignty: Organized Hypocrisy. Princeton University Press.
Kubicek, Paul. 1999–2000. “Russia’s Foreign Policy and the West.” Political Science
Quarterly 114(4):547–568.
Kuran, Timur. 1991. “Now Out of Never: The Element of Surprise in the Eastern
European Revolution of 1989.” World Politics 44:7–48.
Kuzmin, A. I. 1997. Grazhdanskaya Voina v Tadzhikisstane: Istoki I Perspektivy. TsMI,
MGIMO, Moscow.
Lacina, Bethany. 2006. “Explaining the Severity of Civil Wars.” Journal of Conflict
Resolution 50(2):276–289.
Lacina, Bethany. 2014. “India’s Stabilizing Segment States.” Ethnopolitics 13(1):13–27.
Laitin, David. 1977. Politics, Language, and Thought: The Somali Experience.
University of Chicago Press.
Laitin, David D. 1994. “The Tower of Babel as a Coordination Game: Political
Linguistics in Ghana.” American Political Science Review 88(3):622–634.
230 References

Laitin, David D. 1998. Identity in Formation. Cornell University Press.

Laitin, David D. 1999a. National Revivals and Violence. In Critical Comparisons in
Politics and Culture, ed. Roget Petersen and John Bowen. Cambridge University Press.
Laitin, David D. 1999b. Somalia: Civil War and International Intervention. In Civil
Wars, Insecurity, and Intervention, ed. Jack Snyder Barbara F. Walter. Columbia
University Press, pp. 146–180.
Laitin, David D. 2001. “Secessionist Rebellion in the Former Soviet Union.” Compara-
tive Political Studies 34(8):839–861.
Laitin, David D. 2006. “Caucasuan Crucibles.” New Left Review 38.
Laitin, David D., and Rogers Brubaker. 1998. “Ethnic and Nationalist Violence.”
Annual Review of Sociology 24:423–452.
Larsson, Robert L. 2004. “The Enemy Within: Russia’s Military Withdrawal from
Georgia.” Journal of Slavic Military Studies 17(3):405–424.
LeBon, Gustave. 1895. The Crowd: A Study of the Popular Mind. Second – 1982 reprint
ed. Cherokee Publishing Company.
Leeuw, Charles Van Der. 1999. Storm Over the Caucsus in the Wake of Independence.
Curzon Press.
Levitsky, Steven and Lucan Way. 2005. “International Linkage and Democratization.”
Journal of Democracy 16:20–34.
Lezhnev, Sasha. 2006. Crafting Peace: Strategies to Deal with Warlords in Collapsing
States. Lexington Books.
Lichbach, Mark I. 1995. The Rebel’s Dilemma. University of Michigan Press.
Licklider, R. 1995. “The Consequences of Negotiated Settlements in Civil Wars,
1945–1993.” The American Political Science Review 89(3):681–690.
Luttwak, Edward. 1999. “Give War a Chance.” Foreign Affairs 78(4):36–44.
Lyall, Jason. 2010. “Are Coethnics More Effective Counterinsurgents? Evidence from
the Second Chechen War.” American Political Science Review 104(1):1–20.
MacFarlane, Neil. 1999. Realism and Russian Strategy after the Collapse of the USSR.
In Unipolar Politics, ed. Ethan B. Kapstein and Michael Mastanduno. Columbia
University Press, pp. 218–260.
MacFarquhar, Neil and Andrew Kramer. 2014. “Praising Rebels, Putin Toughens Tone
on Ukraine.” New York Times, August 29, 2014, page A1.
Mackinder, Halford. 1942. Democratic Ideals and Reality: A Study in the Politics of
Reconstruction. Classic ed. National Defense University Press.
MacPhee, Craig. 2004. “Expert on Georgia Explores ‘Can Saakashvili Outdo She-
vardnadze?’.” Newswise. January 5, 2004. Available on http://www.newswise.
com/articles/expert-on-georgia-explores-can-saakashvili-outdo-shevardnadze
Magaloni, Beatriz. 2008. “Credible Power-Sharing and the Longevity of Authoritarian
Rule.” Comparative Political Studies 41:715–741.
Mardin, Serif. 1978. “Youth and Violence in Turkey.” Archives of European Sociology
XIX:229–254.
Markowitz, Lawrence. 2011. “Unlootable Resources and State Security Institutions in
Tajikistan and Uzbekistan.” Comparative Political Studies 44(2):156–183.
Markowitz, Lawrence. 2013. State Erosion: Unlootable Resources and Unruly Elites in
Central Asia. Cornell University Press.
Mars, Gerald, and Yochanan Altman. 1983. “The Cultural Bases of Soviet Georgia’s
Second Economy.” Soviet Studies XXXV(4):546–560.
References 231

Marten, Kimberly. 2012. Warlords: Strong-Arm Brokers in Weak States. Cornell Studies
in Security Affairs. Cornell University Press.
Martin, Terry. 2001. The Affirmative Action Empire: Nations and Nationalism in the
Soviet Union, 1923–1939. Cornell University Press.
Mason, T. David, and Dale A. Krane. 1989. “The Political Economy of Death Squads:
Toward a Theory of the Impact of State-Sanctioned Terror.” International Studies
Quarterly 33(2):175–198.
Matveeva, Anna. 2009. “The Perils of Emerging Statehood: Civil War and State
Reconstruction in Tajikistan.” CSRC Working Papers, 2(46):1–57.
McCormick, Gordon, Stephen Horton, and Lauren A. Harrison. 2007. “Things
Fall Apart: The Endgame Dynamic of Internal Wars.” Third World Quarterly
28(2):321–367.
McFaul, Michael. 2001. Russia’s Unfinished Revolution. Cornell University Press.
McFaul, Michael. 2002. “The Fourth Wave of Democracy and Dictatorship: Noncoop-
erative Transitions in the Postcommunist World.” World Politics 54(2):212–244.
McFaul, Michael. 2014. “Michael McFaul on Vladimir Putin and Russia.” Retrieved
from https://www.youtube.com/watch?v=iHgp9fLUzpE
Miller, Eric. 2004. “Smelling the Roses: Eduard Shevardnadze’s End and Georgia’s
Future.” Problems of Post-Communism, 51(2):12–21.
Mirzoev, Tokhir. 2002. Post-Soviet Corruption Outburst in Post-Conflict Tajikistan.
In Political Corruption in Transition, ed. Andras Sajo and Stephen Kotkin. Central
European University Press, pp. 353–382.
Mitchell, Lincoln. 2004. “Georgia’s Rose Revolution.” Current History 103(675):
342–348.
Mueller, John. 2000. “The Banality of ’Ethnic War’.” International Security 25(1):
42–70.
Mueller, John. 2003. “Policing the Remnants of War.” Journal of Peace Research 40(5):
507–518.
Mueller, John. 2004. The Remnants of War. Cornell University Press.
Naimark, Norman. 2002. Fires of Hatred: Ethnic Cleansing in Twentieth-Century
Europe. Harvard University Press.
Nakaya, Sumie. 2009. State Reconstruction in Tajikistan. PhD thesis, City University of
New York.
Nassim Jawad, and Shahrbanou Tadjbakhsh. 1995. Tajikistan: A Forgotten Civil War.
Minority Rights Group. London.
Neumann, Iver, and Sergei Solodovnik. 1996. The Case of Tajikistan. In Peacekeeping
and the Role of Russia in Eurasia. Westview Press, pp. 83–102.
Nodia, G. 2002. Putting the State Back Together in Post-Soviet Georgia. In Beyond State
Crisis?: Post-Colonial Africa and Post-Soviet Eurasia in Comparative Perspective, ed.
M. Beissinger and C. Young. Woodrow Wilson Press.
Nodia, Ghia. 1996. “How Different Are Postcommunist Transitions?” Journal of
Democracy 7(4):15–29.
Nodia, Ghia. 1998a. The Conflict in Abkhazia: National Projects and Political
Circumstances. In Georgians and Abkhazians: The Search for a Peace Settlement,
ed. Bruno Coppieters, Ghia Nodia, and Yury Anchabadze. BIOst.
Nodia, Ghia. 1998b. “Causes and Visions of Conflict in Abkhazia.” Working Paper:
Berkeley Program in Soviet and Post-Soviet Studies.
232 References

North, Douglass C. 2006. Understanding the Process of Economic Change. Princeton

Economic History of the Western World. Princeton University Press.
North, Douglass C., John J. Wallis, and Barry R. Weingast. 2009. Violence and
Social Orders: A Conceptual Framework for Interpreting Recorded Human History.
Cambridge University Press.
Nourzhanov, K. 2005. “Saviors of the Nation or Robber Barons? Warlord Politics in
Tajikistan.” Central Asian Survey 24(2):109–130.
Nourzhanov, Kirill, and Christian Bleur. 2013. Tajikistan: A Political and Social History.
Australia: Australian National University E-Press.
Nunn, Samuel. 1991. “Soviet Defense Conversion and Demilitarization.” Congressional
Hearings.
O’Ballance, Edgar. 1997. Wars in the Caucasus, 1990–1995. New York University
Press.
Olcott, Martha Brill. 1994. Emerging Political Elites. In The New Geopolitics of Central
Asia and Its Borderlands, ed. Ali Banuazizi and Myron Weiner. Indiana University
Press. Warsaw.
Olcott, Martha Brill. 1997. Central Asia’s New States. United States Institute of Peace
Press.
Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of
Groups. Harvard University Press.
O’Prey, Kevin. 1996. Keeping the Peace in the Borderlands of Russia. In UN
Peacekeeping, American Politics, and the Uncivil Wars of the 1990s, ed. William J.
Durch. Stimpson Center Press. Warsaw.
Orwell, George. 1952. Homage to Catalonia. Harcourt Brace and Company.
Osborne, Martin J. 2004. An Introduction to Game Theory. Oxford University Press.
OSCE. 2005. “Preliminary Findings on the Events in Andijan, Uzbekistan, 13
May 2005.” OSCE Office for Democratic Institutions and Human Rights,
pp. 44–67.
Panfilov, Oleg. 2003. Tajikistan: Journalists in Civil War (1992–1997). Russian ed.
Goryaie Tochki.
Papava, Vladimer. 2006. “The Political Economy of Georgia’s Rose Revolution.” Orbis,
50(4): 657-667. Available on http://www.sciencedirect.com/science/article/pii/S00
30438706000779
Papava, Vladimer, and Vepkhia Chocheli. 2003. Financial Globalization and
Post-Communist Georgia: Global Exchange Rate Instability and Its Implications
for Georgia. iUniverse.
Paxson, Margaret. 2005. Solovyovo: The Story of Memory in a Russian Village.
Woodrow Wilson Press.
Petersen, Roger. 2001. Resistence and Rebellion: Lessons from Eastern Europe.
Cambridge University Press.
Petersen, Roger. 2002. Understanding Ethnic Violence: Fear, Hatred, and Resentment
in Twentieth-Century Eastern Europe. Cambridge University Press.
Petersen, Roger. 2011. Western Intervention in the Balkans: The Strategic Use of
Emotion. Cambridge Series in Comparative Politics. Cambridge University Press.
Pevehouse, Jon C. 2005. Democracy from Above? Regional Organizations and
Democratization. Cambridge University Press.
Pileggi, Nicholas. 1985. Wise Guy. Pocket Books.
References 233

Pirseyedi, Bobi. 2000. “The Small Arms Problem in Central Asia: Features and
Implications.” United Nations Institute for Disarmament Research (UNIDIR).
Popkin, S. 1979. The Rational Peasant: The Political Economy of Rural Society in
Vietnam. University of California Press.
Popkov, Viktor. 1998. Soviet Abkhazia 1989: A Personal Account. In The Abkhazians,
ed. George Hewitt. Curzon Press, pp. 102–131.
Posen, Barry. 1993. “The Security Dilemma and Ethnic Conflict.” Survival 35(1):27–47.
“Putin’s Approval Rating Soars to 87%, Poll Say”. 2014. Retrieved from http://www.
themoscowtimes.com/article/504691.html
Putnam, Robert. 1988. “Diplomacy and Domestic Politics: The Logic of Two-Level
Games.” International Organization 42(3):427–460.
Rakowska-Harmstone, Teresa. 1970. Russia and Nationalism in Central Asia: The Case
of Tadzhikistan. Johns Hopkins University Press.
Rashid, Ahmed. 2002. Jihad: The Rise of Militant Islam in Central Asia. Yale University
Press.
Rashid, Ahmed. 2008. Descent into Chaos. Viking Press.
Rashidov, Akbar. 1993. Family and Tribal Structures and Social Conflicts in Soviet
Central Asia. In In a Collapsing Empire: Undervelopment, Ethnic Conflicts, and
Nationalisms in the Soviet Union, ed. Mark Buttino. Feltrinelli Editore Milano,
pp. 291–299.
Ratiani, Shorena. 2004. “Georgia: Corruption Crackdown Makes Waves.” IWPR
Caucasus Reporting Service, no. 231.
Remnick, David. 1994. Lenin’s Tomb: The Last Days of the Soviet Empire. Vintage
International.
Remnick, David. 2014. “Watching the Eclipse.” New Yorker, August 11, 2014. Available
on http://www.newyorker.com/magazine/2014/08/11/watching-eclipse
Reno, William. 1998. Warlord Politics in African States. Rienner.
Reno, William. 2002. Mafiya Troubles, Warlord Crises. In Beyond State Crisis?:
Post-Colonial Africa and Post-Soviet Eurasia in Comparative Perspective, ed. Mark R.
Beissinger and Crawford Young. Woodrow Wilson Press, pp. 105–127.
Roeder, Philip G. 1993. Red Sunset: The Failure of Soviet Politics. Princeton University
Press.
Roeder, Philip G. 2007. Where Nation-States Come From: Institutional Change in the
Age of Nationalism. Princeton University Press.
Roth, Andrew. 2014a. “A Separatist Militia in Ukraine with Russian Fighters Holds a
Key.” New York Times. June 5, 2014, page A12.
Roth, Andrew. 2014b. “Ukraine Faces Struggle to Gain Control of Militias, Including
Those on Its Side.” New York Times. June 5, 2014, page A12.
Roth, Andrew, and Sabrina Tavernise. 2014. “Russians Revealed Among Ukraine
Fighters.” New York Times. June 5, 2014, page A12.
Roy, Olivier. 2000. The New Central Asia: Geopolitics and the Birth of Nations.
New York University Press.
Rubin, Barnett R. 1993. “The Fragmentation of Tajikistan.” Survival 35(4):71–79.
Rubin, Barnett R. 1998. Russian Hegemony and State Breakdown in the Periphery:
Causes and Consequences of the Civil War in Tajikistan. In Post-Soviet Political
Order: Conflict and State Building, ed. Jack Snyder Barnett and R. Rubin. Routledge,
pp. 128–161.
234 References

Rucker, Philip. 2014. “Hillary Clinton Says Putin’s Actions are Like ‘what Hitler
did Back in the ’30s’ ”. Retrieved from http://www.washingtonpost.com/blogs/po
st-politics/wp/2014/03/05/hillary-clinton-says-putins-action-are-like-what-hitler-did
-back-in-the-30s/
Rumer, Boris. 1989. Soviet Central Asia: A Tragic Experiment. Unwin Hyman.
Sacco, Joe. 2003. The Fixer. First ed. Drawn and Quarterly.
Sambanis, Nicholas. 2002. “A Review of Recent Advances and Future Directions in the
Quantitative Literature of Civil War.” Defense and Peace Economics 13(3):215–243.
Satz, Debra, and John Ferejohn. 1994. “Rational Choice and Social Theory.” Journal of
Philosophy 92:71–87.
Schelling, Thomas. 1960. A Strategy of Conflict. Harvard University Press.
Schenkann, Nate. 2014. “Band of Outsiders: How Sanctions Will Strengthen
Putin’s Regional Clout.” Foreign Affairs, August 27, 2014. Available on
http://www.foreignaffairs.com/articles/141941/nate-schenkkan/band-of-outsiders
Scheper-Hughes, Nancy, and Philippe Bourgois, eds. 2004. Violence in War and Peace:
An Anthology. Blackwell Readers in Anthropology.
Scott, James. 1976. The Moral Economy of the Peasant. Yale University Press.
Shabad G., and Llera F. 1994. Political Violence in a Democratic State: Basque
Terrorism in Spain. In Terrorism in Context, ed. Martha Crenshaw. Pennsylvania
State University Press.
Shapiro, Jacob. 2013. The Terrorist’s Dilemma: Managing Violent Covert Organiza-
tions. Princeton University Press.
Shelley, Louise I. 2004. “Organized Crime in the Former Soviet Union: The Distinctive-
ness of Georgia.” TRACCC (Transnational Crime and Corruption Center), Tbilisi,
Georgia.
Shleifer, Andrei. 2005. A Normal Country: Russia after Communism. Harvard
University Press.
Simis, Konstantin. 1982. USSR: The Corrupt Society: The Secret World of Soviet
Capitalism. Simon and Schuster.
Slade, Gavin. 2007. “The Threat of the Thief: Who Has Normative Influence in
Georgian Society?” Global Crime 8(2):172–179.
Slider, Darrell. 1997. Democratization in Georgia. In Conflict, Cleavage, and Change
in Central Asia and the Caucasus, ed. Karen Kawisha and Bruce Parrot. Cambridge
University Press.
Smale, Alison, and Andrew Roth. 2014. “Ukraine Says That Militants Won the East.”
New York Times. April 30, 2014, page A1.
Solnick, Steven L. 1998. Stealing the State: Control and Collapse in Soviet Institutions.
Harvard University Press.
Stedman, Stephen. 1997. “Spoiler Problems in Peace Processes.” International Security
22(2):5–53.
Stedman, Stephen. 2003. “Where Do Peacekeepers Go?” International Studies Review
5(4):37–54.
Stinchcombe, Arthur. 1968. Constructing Social Theories. Harcourt, Brace and World.
Suny, Ronald Grigor. 1995. Elite Tranformation in Late-Soviet and Post-Soviet
Transcaucasia, or What Happens When the Rulling Class Can’t Rule? In Patterns
in Post-Soviet Leadership, ed. Robert C. Tucker and Timothy J. Colton. Harvard
University Russian Research Center, pp. 141–165.
References 235

Tadjbakhsh, S. 1993. “The Tajik Spring of 1992.” Central Asia Monitor 2:21-29.
Tajima, Yuhki. 2014. The Institutional Origins of Communal Violence: Indonesia’s
Transition from Authoritarian Rule. Cambridge University Press.
Tarzi, Amin. 2005. “Disarmament in Afghanistan – Which Militias and What
Weapons?” Radio Free Europe/Radio Liberty (RFE/RL).
Tavernise, Sabrina, and David Herszenhorn. 2014. “Patchwork Makeup of Rebels
Fighting Ukraine Makes Peace Talks Elusive.” New York Times. July 9, A4
Thelen, Kathleen. 1999. “Historical Institutionalism in Comparative Politics.” Annual
Review of Political Science 2:23–49.
Tilly, Charles. 1985. War Marking and State Making as Organized Crime. In Bringing
the State Back In, ed. Theda Skocpol, Peter B. Evans, and Dietrich Rueschemeyer.
Cambridge University Press, pp. 169–191.
Tishkov, Valery. 1995. “"Don’t Kill Me, I’m a Kyrgyz!” Anthropological Analysis of
Violence in the Osh Ethnic Conflict.” Journal of Peace Research 32(2):133–149.
Tishkov, Valery. 1997. Ethnicity, Nationalism, and Conflict in and after the Soviet
Union: The Mind Aflame. SAGE.
Toft, Monica. 2010. Securing the Peace: the Durable Settlement of Civil Wars. Princeton
University Press.
Toft, Monica Duffy. 2005. The Geography of Ethnic Violence: Identity, Interests, and
the Indivisibility of Territory. Princeton University Press.
Torjensen, S. 2005. “Tajikistan’s Road to Stability: Reduction in Small Arms Prolifera-
tion and Remaining Challenges.” Small Arms Survey, Geneva.
Torjensen, Stina, and S. Neil MacFarlane. 2007. “R Before D: The Case of Post Conflict
Reintegration in Tajikistan.” Conflict, Security, and Development 7(2):311–332.
Treisman, Daniel. 1999. After the Deluge: Regional Crises and Political Consolidation
in Russia. University of Michigan Press.
Trenin, Dmitri. 1995. Russian Peacemaking in Georgia. In Crisis Management in the
CIS: Whither Russia?, ed. Hans-Georg Ehrhart, Anna Kreikenmeyer, and Andrei
Zagorski. Nomos Verlagsgesellschaft.
Tudoroiu, Theodor. 2007. “Rose, Orange, and Tulip: The Failed Post-Soviet Revolu-
tions.” Communist and Post-Communist Studies 40:315–342.
Tullock, G. 1971. “The Paradox of Revolution.” Public Choice 11(1):
89–99.
Ucko, David. 2009. Militias, Tribes, and Insurgents: The Challenge of Political
Reintegration in Iraq. In Reintegrating Armed Groups after Conflict, ed. Mats Berdal
and David Ucko. Routledge Studies in Intervention and Statebuilding, pp. 89–118.
Van Atta, Don. 2008. “The Failure of Land Reform in Tajikistan.” Paper prepared for
the 13th Annual World Convention of the Association for the Study of Nationalities.
Van Evera, Stephen. 1994. “Hypotheses on Nationalism and War.” International
Security 18(4):5–39.
Varese, Federico. 2001. The Russian Mafia: Private Protection in a New Market
Economy. Oxford University Press.
Venkatesh, Sudhir. 2007. Off the Books: The Underground Economy of the Urban Poor.
Harvard University Press.
Venkatesh, Sudhir. 2008. Gang Leader for a Day. Penguin Press.
Vigil, James Diego. 1988. Barrio Gangs: Street Life and Identity in Southern California.
University of Texas Press.
236 References

Volkov, V. 2001. Who Is Strong When the State Is Weak. In Beyond State Cri-
sis?: Post-Colonial Africa and Post-Soviet Eurasia in Comparative Perspective, ed.
M. Beissinger and C. Young. Woodrow Wilson Press, pp. 81–103.
Volkov, Vadim. 1999. “Violent Entrepreneurship in Post-Communist Russia.”
Europe-Asia Studies 51(5):741–754.
Wagner, R. Harrison. 2007. War and the State: The Theory of International Politics.
University of Michigan Press.
Walter, Barbara. 1997. “The Critical Barrier to Civil War Settlement.” International
Organization 51:335–364.
Walter, Barbara F. 2002. Committing to Peace: The Successful Settlement of Civil Wars.
Princeton University Press.
Walter, Barbara F. 2009. “Bargaining Failures and Civil War.” Annual Review of Political
Science Spring. 12:243–261.
Wantchekon, Leonard. 2004. “The Paradox of ”Warlord” Democracy: A Theoretical
Investigation.” The American Political Science Review 98(1):17–33.
Warren, Camber, “Not by the Sword Alone: Soft Power, Mass Media, and the
Production of State Sovereignty,” International Organization (forthcoming).
Weber, Max. 1953. From Max Weber: Essays in Sociology. Oxford University Press.
Weinstein, Jeremy. 2006. Inside Rebellion: The Political Economy of Rebel Organiza-
tions. Cambridge University Press.
Wheatley, Jonathan. 2005. Georgia: From National Awakening to the Rose Revolu-
tion – Delayed Transition in the Former Soviet Union. Ashgate.
Whitlock, Monica. 2005. The Land Beyond the River. St. Martin’s Press.
Whyte, William Foot. 1982. Interviewing in Field Research. In Field Research: A
Sourcebook and Field Manual, ed. Robert G. Burgess. George Allen and Unwin.
Wood, Elisabeth. 2003. Insurgent Collective Action and Civil War in El Salvador.
Cambridge University Press.
Wood, Elisabeth. 2007. Field Methods. In The Oxford Handbook of Comparative
Politics, ed. Carles Boix and Susan Stokes. Oxford University Press.
Wright, Joseph, and Matthew Winters. 2010. “The Politics of Effective Foreign Aid.”
Annual Review of Political Science 13:61–80.
Zuercher, Christoph. 2005. The Post-Soviet Wars: Rebellion, Ethnic Conflict, and
Nationhood in the Caucasus. New York University Press.
Zverev, Alexei. 1996. Ethnic Conflicts in the Caucasus. In Contested Borders in the
Caucasus, ed. Bruno Coppieters. VUB University Press, pp. 13–71.
Zviagelskaya, Irina. 1998. The Tajik Conflict: Problems of Regulation. In
Tajikistan: The Trials of Independence, ed. Shirin Akiner Mohammad-Reza Djalili
and Frederic Grare. Curzon Press, pp. 161–179.
Index

Abashidze, Aslan 114, 137, 142, 161, 168 n88 Clinton, William 9 n18, 133
Abdullojonov, Abdumalik 86–87, 92, 119, collective action 28, 35 n7, 53–54, 129–130
152–153 Collins, Kathleen 72 n71, 147–148 n51
Afghanistan Commonwealth of Independent States (CIS)
American invasion 39, 80, 132, 148, 172 9–10, 86, 92 n16, 127
civil war in 80 community defense militias 76, 79, 106
narcotics and illicit economies in 143 n41, corruption 24, 126, 144–145, 161, 172
159, 171 as incentive for warlords to join state 34, 36,
analytic narrative 16–17, 26 38–9, 139–43
anarchy, political economy of 1–3, 96 n24, counterinsurgency
100–101, 112 n53, 196 generally 15
Andizjan 21 Russian approach 43, 65, 80, 93
assurance games 196 n43 two-player models of 52, 191–193
Attrition coups
decisive victory in civil war 81 n105, 84
by warlords 35, 52, 69, 91, 201–203
among militias 106–110, 114–115, 121,
in Georgia 118–119
194–195
in Tajikistan 158
autocracy
presidential buy-off 35–37, 43, 52
risk of conflict in 193–194
credible commitments
from Presidents 6, 120, 139–140, 143, 153,
Baker, James 133 169
Boimatov, Ibdullo 108 n45, 153–155, 164 from warlords 57, 100, 102, 104
to ethnic minorities 63–64, 104
capital city, strategic importance criminal behaviors
of capture 26, 28, 30–33, 37, 43, 55, 76, 85, drug trafficking 14, 97–98, 132, 143 n41,
94–97, 114 148, 158, 159, 171–172
cessation of conflict among militias 81, 95, 101, 114, 120–121
by decisive victory 3–4, 81 n105, 84, war crimes 79 n94, 83, 177 n5
200 n3 criminal networks
postmodern imperialism 6–7, 10 as a source of human capital for militias
role of credible commitments 6 49, 87
third party guarantors 4–6 Dushanbe mafia 86
checkpoints See roadblocks Russian transit networks 48, 59, 64

237
238 Index

criminal networks (cont.) maximization of state value (v*) 33, 41, 43,
vori v zkone 64, 69, 105 n37 85–86, 92, 126–133, 138, 144,
cultural differences, as spoiler for cooperation 166–172, 181, 186–189, 203–204, 208
42–43 president, management of warlord coalition
34–36, 60–61, 67, 75 n82, 85, 91, 95,
97, 115, 117, 124, 129–131, 134, 147,
Darchiashvili, David 67
151–152, 162–164, 171, 174–175, 213,
Darden, Keith 145
219
democratization 14 n25
president, payments to warlords 1, 24,
Derluguian, Georgi 21, 46–47 n4
35–37, 43, 60–61, 75, 88, 90, 93–94,
desertion 15, 19, 82
106, 126, 160, 174, 176, 186, 201
deterrence 4, 60, 114–117, 195
reservation value (r) 37, 39–40, 90, 93–94,
devolution of autonomy 165, 191
128, 130, 132, 135, 139, 148, 155, 181,
differential effects of war on rural and urban
186, 188, 208–209
populations 94–95, 112
stability threshold (s) 32–33, 39–40, 85,
Dostum, Abdul Rashid 80, 93 n17, 154
119, 126, 131–132, 144, 147, 149,
Doyle and Sambanis civil conflict dataset 14,
151, 159, 162, 186, 188,
201
208–209
Drug routes
war economy payout 33, 37, 41, 68
cigarettes 14, 97–98, 157
warlords, coups after installation of regime
heroin 132, 143 n41, 148, 171–172
35, 52, 69, 118–119, 207
duration of civil conflict 2–3, 198–200
warlords, installation of puppet regime 1,
11–12, 14–15, 28–29, 31, 44, 85–89,
economic development and insecurity 138 92, 94–96, 102, 116, 123, 174, 189
Elster, John 26 n31
ethnic irredentism 62, 80, 151 n57 warlords, integration into the state 1, 10, 12,
ethnic politics 28–29, 31, 34, 39–41, 52–53, 95–98,
in Georgia 62–64, 127, 142 107, 110, 115, 130–131, 135, 138–141,
security dilemma 78–79 144, 156–162, 176, 181
in Tajikistan 72–73, 77–79, 81, 83, 104, warlords, manipulation of coalition by
164–165, 218 presidents 34–36, 60–61, 67, 75 n82,
in Ukraine 183 85, 91, 95, 97, 115, 117, 124, 129–131,
Eurasian Economic Union (EEU) 182, 190 134, 147, 151–152, 162–164, 171,
174–175, 213, 219
warlords, post-installation competition
Fearon, James duration of conflict dataset
between 86, 89, 91, 114–120, 159, 174
198–199
warlords, payoffs from regime 1, 24, 35–37,
43, 60–61, 75, 88, 90, 93–94, 106, 126,
Game Theoretic Concepts 160, 174, 176, 186, 201
coalition formation 32, 38, 52, 85–89, 102 warlords, posturing and demonstration
equilibrium, subgame perfect (SPNE) 129, violence 91, 96–97, 99–103, 106–111,
207-8 114, 121, 185
sucker costs 32 warlords, relative armed strength 60–61, 86,
two player game, criticism 52, 191–193 90, 163, 215–216, 218
Game of Predator Collusion warlords, replaceable coalition members 193
cost of fighting 32, 188 Gamsakhurdia, Zviad 60, 66–70, 82–83,
equilibrium, full incorporation 37–39, 41, 88–91, 117, 126–127, 161
145, 169–170, 178–179 Zviadist insurgency 27, 82, 89, 127 n6, 134,
equilibrium, partial incorporation 37–39, 136
41–43, 178–179, 186, 188, 190 Gorno–Badakshon Autonomous Oblast
equilibrium, state failure 37–39, 41, 187, (GBAO) 73–77, 81, 89–90, 150 n55,
196, 208 158–159, 163-165, 171–172, 218
Index 239

Georgia Khudoiberiyev, Mahmud 108 n45, 115, 151,

Abkhazia 24–25, 52 n14, 60–65, 68–70, 89, 153–158, 164
94–95, 117 n57, 120, 133, 136–137, Khudonazarov, Davlat 74
186–187 Kitovani, Tengiz 51, 67–69, 83, 88–89, 115
Adjara 43, 81–82, 94, 136–137 n54, 139, 142–143, 192
Armenian enclaves 63, 70, 73, 126 n6, 136, Komitet gosudarstvennoy bezopasnosti (KGB)
142 68, 84, 128, 153, 159, 216–217
civil war 136–137 Kuvvatov, Sulton 158
coups 118–119 Kyrgyzstan 18 n33, 46, 127, 147 n51
ethnic politics in 62–64, 127, 142 Osh 46, 235
Gali 137–138
geography 8, 23
language, political mobilization of 63 n39, 183
Georgian Dream Party 171
Led Zepplin 101
membership in NATO 168–169, 178–179
Levada Center 190
Mingrelia 62–63, 82, 127 n6, 138, 142
Liberal Interventions 4–5, 10
Ministry of the Interior 66–68, 88, 98,
local feuds 76
115–116, 118, 127, 131, 138–144
nationalism 64, 169
post-Soviet history 131–145 major powers
Rose Revolution 132–133, 168–171 demand for local order 1, 6, 40, 93, 127,
South Ossetia 39, 60–61, 66, 81–82, 94, 148–149, 172 n95, 195
136–137, 161, 169, 176, 178–179, extortion by warlords 1, 10, 12–14, 93
191 maximizing value of state capture 33, 41,
Sukhumi 69, 83, 117, 135, 179, 191 43, 85–86, 92, 126–133, 138, 144,
Tbilisi 58, 62, 67–69, 83, 95–96 166–172, 181, 186–189, 203–204, 208
Tbilisi, militias in 58, 60–61, 95, Makhkamov, Kakhar 73–74
107–108 Margiani, Avantandil 142
United National Movement 170 Massoud, Ahmed Shah 39, 74, 80
Giorgadze, Igor 118, 134, 144 mediation 4–7, 10
Gorbachev, Mikhail 92, 133 methodology
Gulua, Gregory 108 case selection 22–23, 198–200
external validity 201–203
field experience 16–23
Hobbes, Thomas 1, 3 human subjects and safety 20–23, 65 n48,
219
Ioseliani, Dzhaba 50–51, 65–69, 88, 115 n54, mathematical proofs 207–211
118–119, 123, 135, 139, 142–145, 192 sampling and bias 17–23
Iran 9, 80, 93, 132–133, 149, 176, 178 translation and note taking in
Iskanderov, Akbarshah 75, 80, 126 interviews 21
Islam, political mobilization of 62 n14, 73, 79, unreliable responses 20, 65 n48, 180, 193
148, 151 n57, 172 use of formal models and analytical
Ivanishvili, Bidzina 180 narratives 16–17, 26
Mirzoev, Rustam 86–87
Mirzoyev, Ghaffor 87, 156, 158
Kabardino-Balkaria 46 Mkherdrioni (Knights of Georgia) 17, 50–51,
Karakashvili, Georgi 139 65–69, 94, 103, 113, 115, 161
Karimov, Islam 83–84, 149, 151 n57 conflict with Shevarnadze 135
Kasimov, Shurob 139 fractionalization 118
Kekelidze, Nika 108 waning power 105–110, 139–144
Kenjayev, Safarali 74–75, 89, 146 n48, 157, monitoring 4–5, 10
164–165 mujahideen 47, 93, 149, 157
Khachishvili, Temur 88 Mullen, Mark 161
240 Index

Nabiev, Rakhmon 60, 74–75, 83, 89 Presidents

Nadibaidze, Vardiko 134 credible commitment problem 6, 120,
nationalism 47, 51, 61, 64 139–140, 143, 153, 169
natural state (Tajikistan as a) 52, 59 n28, developing independent constituencies
171 125–126
Naujavano, Mamadaez 84 dissolution of Parliament 59
neo-trusteeship See post-modern imperialism independence from warlords 28–29, 35–37,
Ninidze, Teto 110 135, 143, 158, 162–163, 168
Nodia, Ghia 59–61, 134, 161 n79 installation as puppet 1, 11–12, 14–15,
nomenklatura 47–48, 59, 71, 98 28–29, 31, 44, 85–89, 92, 94–96, 102,
non-governmental organizations 116, 123, 174, 189 n31
aid provided through 5 n11, 43, 166 management of coalition 34–36, 60–61, 67,
exploitation by warlords 1, 5 n11, 12, 24, 75 n82, 85, 91, 95, 97, 115, 117, 124,
43, 93 129–131, 134, 147, 151–152, 162–164,
maximizing value of state capture 33, 41, 171, 174–175, 213, 219
43, 85–86, 92, 126–133, 138, 144, payments to warlords 1, 24, 35–37, 43,
166–172, 181, 186–189, 203–204, 60–61, 75, 88, 90, 93–94, 106, 126,
208 160, 174, 176, 186, 201
organizational culture 2, 116, 175, personalist rule 44, 135
196 replaceability of warlord partners 193
Nuri, Said Abdullo 155 two-level games 125–126, 130–132, 154,
174
use of media 125 n4
Open Access Society primordialism (as a theory of civil conflict) 25,
Doorstep Conditions 171 42–43, 48, 50, 55
organized crime 48–49, 59, 64, 69, 86–87, privatization 11, 24, 31, 34, 41, 123, 130–131,
105 n37 140, 174
Orwell, George 172 n97, 196 property rights 52, 116
protection rackets 51, 109, 121
Pacacia, Otar 142 Putin, Vladimir 183–192
Pamiri militias
fate after independence 73–77, 81, 104, Rahkmonov, Emomali 29, 81, 87, 89–92, 104,
163–165 106, 108, 119–128, 132, 146–159,
geography and strategy 71 n66, 104, 127, 162–166, 171, 173, 218–219
132 realism 2–4, 10, 175
political participation 89, 150 n55, recruitment
158–159, 163–165, 171–172, 218 as a collective action problem 53–54
resistance 90 credible commitments 102
Papava, Vladimir 134 n16, 140 family networks 106
participation, models of 26 n16 ideological justifications 55, 65, 76–77, 103
patronage 24, 47, 57, 149–150, 157, importance of trust among recruits 103
164 motivations 19, 55–58
police reform 12–13, 67, 131, 145–146 promises of future spoils and protections 57,
absence of 90, 142 n37, 155, 161 99–100
sacking of warlords 162–163, 168 sociopathy 53–54, 100, 103, 113, 184
political parties 135, 149–152, 204 triggering and sustaining mechanisms
Polity Index 24, 204–205 55 n21
Popular Front 59–60, 81, 83, 89, 94, 104–108, use of selective incentives 54–56, 100
110, 120, 122, 146–147, 149–151, roadblocks, tactical use 56, 109
154 n66, 156–158, 162–165, 171 Rodeli, Alexander 119, 141, 178–179
Poroshenko, Petro 186, 192 Roy, Olivier 71 n67, 76, 164 n84
post-modern imperialism 6–7, 10 Rubin, Barnett 90
Index 241

Russia Tajikistan
anti-Russian rhetoric 144 avlod and kinship 49, 55 n22, 72 n69,
counterinsurgency tactics 43, 65, 80–82, 93 77–78, 83, 87, 99, 128, 146, 158
intervention in Georgia 65, 81–82, 84, civil war deaths 70
118–119, 126–127, 134–136, 178 cotton 52 n14, 61–62, 71–72, 78–79, 147
intervention in Tajikistan 80, 82, 84, 127, n51, 150, 214 n2
132, 149–152, 178 coups 158
intervention in Ukraine 182–189 Dushanbe 22 n40, 54, 70 n64, 73–81,
post-modern imperialism 7, 9, 43 89–91, 95–96, 106, 115–16, 146,
relations with Afghanistan 127 152–159
spheres of influence 9, 134, 137, 182, 189 ethnic politics in 72–73, 77–79, 81, 83, 104,
treaties with satellite states 136, 149 164–165, 218
201st Motorized Division 61, 79, 82–83, 93 family and social structure 72 n69, 147
world reaction to 9, 134, 184, 189 n31 Ferghana Valley 25, 46, 74, 127
geography 8, 23, 94, 158
Gorno-Badakshon 22 n40, 74, 78,
Saakashvili, Mikheil 65 n49, 161 n78,
80 n100, 119, 149, 159, 164–165,
168–170, 179
217–218
Safarov, Sangak 51, 77–80, 86–89, 118,
151–152 Hissori 89, 106, 110, 119–120, 146–151,
159, 163–164, 218
Saidov, Faizali 108, 152
Saliev, Rauf 86–87 Islamic parties 73, 79
Salimov, Yakub 86–87, 158 Khojand 9, 72–75, 80, 86, 89, 93, 106,
secession movements 119–120, 146–154, 164, 218
in Georgia 137 kollkhoz (collective farms) 53, 61, 70 n64,
in Tajikistan 80 n101 71–72, 76–78, 81 n103, 219
in Ukraine 191 Kulob 83, 89–90, 104 n35, 108 n45,
security dilemma 78–79 117–122, 146–155, 158, 163–165,
security forces, reform See police reform 217–219
Segura, Tengiz 69–70, 88 Kurgon-Tubbe 58, 78–80, 119, 151
settlement of conflict Lakai 106, 108, 146 n48, 164
by decisive victory 3–4, 81 n105, 84, 200 n3 Leninobod 72, 75, 89, 117, 151–153
post-modern imperialism 6–7, 10 Ministry of the Interior 74, 80, 86, 98, 110,
role of credible commitments 6 115, 148
third party guarantors 4–6 nationalism 74
Shevardnadze, Eduard Popular Front 59–60, 81, 83, 89, 94,
end of political career 168–171 104–108, 110, 120, 122, 146–147,
as independent president 29, 115–148, 149–151, 154 n66, 156–158, 162–165,
158–161, 173–174, 181, 192 171
return to Georgian politics 68, 87–104 post-Soviet history 70–81, 146–159
Shishlyannikov, Alexander 127–128 strategic institutional reform 150
side-switching 12, 35, 55 n22, 58, 96, 98–99, Tamunindze, Kakha 116
111, 114, 154 taxation
Small Arms Survey 162, 213, 224, 235 as a fungible income stream 14, 34, 53, 69,
social capital, role of 33, 50–54, 92, 96, 100, 98, 107, 110, 121
102, 110–114, 149–150, 180 under militia rule 14, 69, 101, 107, 109,
sociopathy 116, 121, 138, 140, 161
among recruits 53–54, 100, 103, 113, 184 terrorist organizations, control of 1, 6, 40, 93,
among respondents 20, 65 n48, 180, 193 127, 148–149, 172 n95, 195
stag hunt 32–33 theft of civilian property 59, 69, 81, 83, 101,
state failure 92, 96 n24 116
sub-proletariat mobilization 1, 47, 50 Turajonzoda, Hoji Akbar 156
242 Index

Ukraine exit from coalition 90, 99

Crimea 184–190 fight or join calculus 10, 40, 207
ethnic politics in 183 installation of puppet president 1, 11–12,
Maidan Square 182–186, 190 14–15, 28–29, 31, 44, 85–89, 92,
membership in NATO/EU 188–189 94–96, 102, 116, 123, 174, 189 n31
nationalism 187 integration into state and regulatory capture
Orange Revolution 192 1, 10, 12, 28–29, 31, 34, 39–41, 52–53,
policy prescriptions 187–190 95–98, 107, 110, 115, 130–131, 135,
Russian intervention 182–189 138–141, 144, 156–162, 176, 181
United Nations manipulation of coalition by President
peacekeeping 2, 5, 7, 177 34–36, 60–61, 67, 75 n82, 85, 91, 95,
Russian influence 9 97, 115, 117, 124, 129–131, 134, 147,
United Tajik Opposition 90, 94, 129 n10, 149, 151–152, 162–164, 171, 174–175, 213,
152, 226 219
Usmon, Davlat 157 origins 48, 50, 53, 68
USSR outside of coalition 33, 37, 41, 68, 114,
civil conflict after collapse 2, 46–50, 71, 198 117–118, 120, 146 n49, 158, 161
disposition of Soviet assets upon collapse payoffs from regime 1, 24, 35–37, 43,
49–50 60–61, 75, 88, 90, 93–94, 106, 126,
use of Soviet institutions by successor states 160, 174, 176, 186, 201
4, 48, 60, 170 post-installation competition between 86,
Uzbekistan 19 n33, 86, 93, 117–120, 127 89, 91, 114–120, 159, 174
posturing and demonstration violence 91,
vote buying 31, 142, 189 96–97, 99–103, 106–111, 114, 121,
185
predictors of success 111–113, 162–163,
war economy 33, 37, 41, 68 215–216, 218
warlords
as producers and distributors of public
brokerage 50–51, 64, 69, 142
goods 102
capture by state 1, 10, 12, 28–29, 31, 34,
rehabilitation of reputation 192
39–41, 52–53, 95–98, 107, 110, 115,
upstart 99, 102, 107
130–131, 135, 138–141, 144, 156–162,
use of incitement rhetoric 51, 60–61, 66,
176, 181
77–79, 81
capture of third parties 1, 5 n11, 6, 10,
Weberian state 175–176
12–14, 24, 40, 93, 127, 129, 142,
White Eagle militia 17, 68, 138, 160
148–149, 157, 172 n95, 181, 187–189,
White Falcon militia 139
195, 203–204
weapons of mass destruction and terrorism 6,
collective action 29, 130
9 n18, 193, 195
coups by 35, 52, 69, 91, 201–203
credible commitment problem 57, 100, 102,
104 Yanukovich, Victor 182, 185–186
definition of 213–214 Yeltsin, Boris 9, 71, 93, 127, 134, 152–153
distrust among 44
effective control of foot soldiers 109,
138–139 Zakharchenko, Alexander 192
establishment of war economy 14, 51, 56, Zhvania, Zurab 143, 145
69, 101, 107, 109, 116, 121, 138, 140, Zioyev, Mirzo 146 n49, 157, 162 n82
161 Zukhurov, Saidamir 87, 128, 158
Other Books in the Series (continued from page iii)

Carles Boix, Political Order and Inequality: Their Foundations and Their
Consequences for Human Welfare
Carles Boix, Political Parties, Growth, and Equality: Conservative and
Social Democratic Economic Strategies in the World Economy
Catherine Boone, Merchant Capital and the Roots of State Power in
Senegal, 1930–1985
Catherine Boone, Political Topographies of the African State: Territorial
Authority and Institutional Change
Catherine Boone, Property and Political Order in Africa: Land Rights and
the Structure of Politics
Michael Bratton, Robert Mattes, and E. Gyimah-Boadi, Public Opinion,
Democracy, and Market Reform in Africa
Michael Bratton and Nicolas van de Walle, Democratic Experiments in
Africa: Regime Transitions in Comparative Perspective
Valerie Bunce, Leaving Socialism and Leaving the State: The End of
Yugoslavia, the Soviet Union, and Czechoslovakia
Daniele Caramani, The Nationalization of Politics: The Formation of
National Electorates and Party Systems in Europe
John M. Carey, Legislative Voting and Accountability
Kanchan Chandra, Why Ethnic Parties Succeed: Patronage and Ethnic
Headcounts in India
Eric C. C. Chang, Mark Andreas Kayser, Drew A. Linzer, and Ronald
Rogowski, Electoral Systems and the Balance of Consumer-Producer
Power
José Antonio Cheibub, Presidentialism, Parliamentarism, and Democracy
Ruth Berins Collier, Paths toward Democracy: The Working Class and
Elites in Western Europe and South America
Pepper D. Culpepper, Quiet Politics and Business Power: Corporate Control
in Europe and Japan
Rafaela M. Dancygier, Immigration and Conflict in Europe
Christian Davenport, State Repression and the Domestic Democratic Peace
Donatella dellaPorta, Social Movements, Political Violence, and the State
Alberto Diaz-Cayeros, Federalism, Fiscal Authority, and Centralization in
Latin America
Thad Dunning, Crude Democracy: Natural Resource Wealth and Political
Regimes
Gerald Easter, Reconstructing the State: Personal Networks and Elite
Identity
Margarita Estevez-Abe, Welfare and Capitalism in Postwar Japan: Party,
Bureaucracy, and Business
Henry Farrell, The Political Economy of Trust: Institutions, Interests, and
Inter-Firm Cooperation in Italy and Germany
Karen E. Ferree, Framing the Race in South Africa: The Political Origins of
Racial Census Elections
M. Steven Fish, Democracy Derailed in Russia: The Failure of Open Politics
Robert F. Franzese, Macroeconomic Policies of Developed Democracies
Roberto Franzosi, The Puzzle of Strikes: Class and State Strategies in
Postwar Italy
Timothy Frye, Building States and Markets After Communism: The Perils
of Polarized Democracy
Geoffrey Garrett, Partisan Politics in the Global Economy
Scott Gehlbach, Representation through Taxation: Revenue, Politics, and
Development in Postcommunist States
Edward L. Gibson, Boundary Control: Subnational Authoritarianism in
Federal Democracies
Jane R. Gingrich, Making Markets in the Welfare State: The Politics of
Varying Market Reforms
Miriam Golden, Heroic Defeats: The Politics of Job Loss
Jeff Goodwin, No Other Way Out: States and Revolutionary Movements
Merilee Serrill Grindle, Changing the State
Anna Grzymala-Busse, Rebuilding Leviathan: Party Competition and State
Exploitation in Post-Communist Democracies
Anna Grzymala-Busse, Redeeming the Communist Past: The Regeneration
of Communist Parties in East Central Europe
Frances Hagopian, Traditional Politics and Regime Change in Brazil
Henry E. Hale, The Foundations of Ethnic Politics: Separatism of States
and Nations in Eurasia and the World
Mark Hallerberg, Rolf Ranier Strauch, and Jürgen von Hagen, Fiscal
Governance in Europe
Stephen E. Hanson, Post-Imperial Democracies: Ideology and Party For-
mation in Third Republic France, Weimar Germany, and Post-Soviet
Russia
Silja Häusermann, The Politics of Welfare State Reform in Continental
Europe: Modernization in Hard Times
Michael Hechter, Alien Rule
Gretchen Helmke, Courts Under Constraints: Judges, Generals, and
Presidents in Argentina
Yoshiko Herrera, Imagined Economies: The Sources of Russian
Regionalism
J. Rogers Hollingsworth and Robert Boyer, eds., Contemporary Capitalism:
The Embeddedness of Institutions
John D. Huber and Charles R. Shipan, Deliberate Discretion? The
Institutional Foundations of Bureaucratic Autonomy
Ellen Immergut, Health Politics: Interests and Institutions in Western
Europe
Torben Iversen, Capitalism, Democracy, and Welfare
Torben Iversen, Contested Economic Institutions
Torben Iversen, Jonas Pontusson, and David Soskice, eds., Unions,
Employers, and Central Banks: Macroeconomic Coordination and
Institutional Change in Social Market Economies
Thomas Janoski and Alexander M. Hicks, eds., The Comparative Political
Economy of the Welfare State
Joseph Jupille, Procedural Politics: Issues, Influence, and Institutional
Choice in the European Union
Stathis Kalyvas, The Logic of Violence in Civil War
David C. Kang, Crony Capitalism: Corruption and Capitalism in South
Korea and the Philippines
Stephen B. Kaplan, Globalization and Austerity Politics in Latin America
Junko Kato, Regressive Taxation and the Welfare State
Orit Kedar, Voting for Policy, Not Parties: How Voters Compensate for
Power Sharing
Robert O. Keohane and Helen B. Milner, eds., Internationalization and
Domestic Politics
Herbert Kitschelt, The Transformation of European Social Democracy
Herbert Kitschelt, Kirk A. Hawkins, Juan Pablo Luna, Guillermo Rosas,
and Elizabeth J. Zechmeister, Latin American Party Systems
Herbert Kitschelt, Peter Lange, Gary Marks, and John D. Stephens, eds.,
Continuity and Change in Contemporary Capitalism
Herbert Kitschelt, Zdenka Mansfeldova, Radek Markowski, and Gabor
Toka, Post-Communist Party Systems
David Knoke, Franz Urban Pappi, Jeffrey Broadbent, and Yutaka Tsuji-
naka, eds., Comparing Policy Networks
Allan Kornberg and Harold D. Clarke, Citizens and Community: Political
Support in a Representative Democracy
Amie Kreppel, The European Parliament and the Supranational Party
System
David D. Laitin, Language Repertoires and State Construction in Africa
Fabrice E. Lehoucq and Ivan Molina, Stuffing the Ballot Box: Fraud,
Electoral Reform, and Democratization in Costa Rica
Mark Irving Lichbach and Alan S. Zuckerman, eds., Comparative Politics:
Rationality, Culture, and Structure, 2nd edition
Evan Lieberman, Race and Regionalism in the Politics of Taxation in Brazil
and South Africa
Richard M. Locke, Promoting Labor Standards in a Global Economy: The
Promise and Limits of Private Power
Pauline Jones Luong, Institutional Change and Political Continuity in
Post-Soviet Central Asia
Pauline Jones Luong and Erika Weinthal, Oil Is Not a Curse: Ownership
Structure and Institutions in Soviet Successor States
Julia Lynch, Age in the Welfare State: The Origins of Social Spending on
Pensioners, Workers, and Children
Lauren M. MacLean, Informal Institutions and Citizenship in Rural Africa:
Risk and Reciprocity in Ghana and Côte d’Ivoire
Beatriz Magaloni, Voting for Autocracy: Hegemonic Party Survival and Its
Demise in Mexico
James Mahoney, Colonialism and Postcolonial Development: Spanish
America in Comparative Perspective
James Mahoney and Dietrich Rueschemeyer, eds., Comparative Historical
Analysis in the Social Sciences
Scott Mainwaring and Matthew Soberg Shugart, eds., Presidentialism and
Democracy in Latin America
Isabela Mares, From Open Secrets to Secret Voting: Democratic Electoral
Reforms and Voter Autonomy
Isabela Mares, The Politics of Social Risk: Business and Welfare State
Development
Isabela Mares, Taxation, Wage Bargaining, and Unemployment
Cathie Jo Martin and Duane Swank, The Political Construction of Business
Interests: Coordination, Growth, and Equality
Anthony W. Marx, Making Race, Making Nations: A Comparison of South
Africa, the United States, and Brazil
Doug McAdam, John McCarthy, and Mayer Zald, eds., Comparative
Perspectives on Social Movements
Bonnie M. Meguid, Party Competition between Unequals: Strategies and
Electoral Fortunes in Western Europe
Joel S. Migdal, State in Society: Studying How States and Societies
Constitute One Another
Joel S. Migdal, Atul Kohli, and Vivienne Shue, eds., State Power and Social
Forces: Domination and Transformation in the Third World
Scott Morgenstern and Benito Nacif, eds., Legislative Politics in Latin
America
Kevin M. Morrison, Nontaxation and Representation: The Fiscal Founda-
tions of Political Stability
Layna Mosley, Global Capital and National Governments
Layna Mosley, Labor Rights and Multinational Production
Wolfgang C. Müller and KaareStrøm, Policy, Office, or Votes?
Maria Victoria Murillo, Labor Unions, Partisan Coalitions, and Market
Reforms in Latin America
Maria Victoria Murillo, Political Competition, Partisanship, and Policy
Making in Latin American Public Utilities
Monika Nalepa, Skeletons in the Closet: Transitional Justice in
Post-Communist Europe
Ton Notermans, Money, Markets, and the State: Social Democratic
Economic Policies since 1918
Eleonora Pasotti, Political Branding in Cities: The Decline of Machine
Politics in Bogotá, Naples, and Chicago
Aníbal Pérez-Liñán, Presidential Impeachment and the New Political
Instability in Latin America
Roger D. Petersen, Understanding Ethnic Violence: Fear, Hatred, and
Resentment in Twentieth-Century Eastern Europe
Roger D. Petersen, Western Intervention in the Balkans: The Strategic Use
of Emotion in Conflict
Simona Piattoni, ed., Clientelism, Interests, and Democratic Representation
Paul Pierson, Dismantling the Welfare State? Reagan, Thatcher, and the
Politics of Retrenchment
Marino Regini, Uncertain Boundaries: The Social and Political Construc-
tion of European Economies
Kenneth M. Roberts, Changing Course in Latin America: Party Systems in
the Neoliberal Era
Marc Howard Ross, Cultural Contestation in Ethnic Conflict
Ben Ross Schneider, Hierarchical Capitalism in Latin America: Business,
Labor, and the Challenges of Equitable Development
Lyle Scruggs, Sustaining Abundance: Environmental Performance in Indus-
trial Democracies
Jefferey M. Sellers, Governing from Below: Urban Regions and the Global
Economy
Yossi Shain and Juan Linz, eds., Interim Governments and Democratic
Transitions
Beverly Silver, Forces of Labor: Workers’ Movements and Globalization
since 1870
Theda Skocpol, Social Revolutions in the Modern World
Dan Slater, Ordering Power: Contentious Politics and Authoritarian
Leviathans in Southeast Asia
Regina Smyth, Candidate Strategies and Electoral Competition in the
Russian Federation: Democracy Without Foundation
Richard Snyder, Politics after Neoliberalism: Reregulation in Mexico
David Stark and László Bruszt, Postsocialist Pathways: Transforming
Politics and Property in East Central Europe
Sven Steinmo, The Evolution of Modern States: Sweden, Japan, and the
United States
Sven Steinmo, Kathleen Thelen, and Frank Longstreth, eds., Structuring
Politics: Historical Institutionalism in Comparative Analysis
Susan C. Stokes, Mandates and Democracy: Neoliberalism by Surprise in
Latin America
Susan C. Stokes, ed., Public Support for Market Reforms in New
Democracies
Susan C. Stokes, Thad Hall, Marcelo Nazareno, and Valeria Brusco,
Brokers, Voters, and Clientelism: The Puzzle of Distributive Politics
Duane Swank, Global Capital, Political Institutions, and Policy Change in
Developed Welfare States
Sidney Tarrow, Power in Movement: Social Movements and Contentious
Politics, Revised and Updated 3rd Edition
Tariq Thachil, Elite Parties, Poor Voters: How Social Services Win Votes in
India
Kathleen Thelen, How Institutions Evolve: The Political Economy of Skills
in Germany, Britain, the United States, and Japan
Kathleen Thelen, Varieties of Liberalization and the New Politics of Social
Solidarity
Charles Tilly, Trust and Rule
Daniel Treisman, The Architecture of Government: Rethinking Political
Decentralization
Guillermo Trejo, Popular Movements in Autocracies: Religion, Repression,
and Indigenous Collective Action in Mexico
Lily Lee Tsai, Accountability without Democracy: How Solidary Groups
Provide Public Goods in Rural China
Joshua Tucker, Regional Economic Voting: Russia, Poland, Hungary,
Slovakia, and the Czech Republic, 1990–1999
Ashutosh Varshney, Democracy, Development, and the Countryside
Jeremy M. Weinstein, Inside Rebellion: The Politics of Insurgent Violence
Stephen I. Wilkinson, Votes and Violence: Electoral Competition and
Ethnic Riots in India
Jason Wittenberg, Crucibles of Political Loyalty: Church Institutions and
Electoral Continuity in Hungary
Elisabeth J. Wood, Forging Democracy from Below: Insurgent Transitions
in South Africa and El Salvador
Elisabeth J. Wood, Insurgent Collective Action and Civil War in El Salvador