Read an Excerpt
Cooperation and Discord in U.S.-Soviet Arms Control
By Steve Weber PRINCETON UNIVERSITY PRESS
Copyright © 1991 Princeton University Press
All rights reserved.
ISBN: 978-0-691-07837-3
CHAPTER 1
Introduction
This book is a study of U.S.-Soviet efforts to cooperate in the limitation of strategic nuclear weapons systems. Current theory in international relations provides a powerful analysis of the many impediments to cooperation between states, but it does not yet offer an adequate explanation of why those impediments are sometimes overcome. Arms control cooperation between two adversarial superpowers would be a particularly difficult case to explain. Why have the United States and the Soviet Union achieved cooperative agreements with the goal of enhancing their mutual security in certain arms control issues, but not in others?
This is an important question, not only for international relations theorists but also for political leaders concerned with superpower arms control and the broader U.S.-Soviet relationship in which it is embedded. The success of cooperative arms control efforts has become, for better or for worse, a crucial indicator of larger trends in U.S.-Soviet relations. There is clearly some connection between the two. The Limited Test Ban Treaty concluded in 1963 was an important part of the thaw in relations that followed the Cuban Missile Crisis of 1962. The SALT I accords, signed by President Richard Nixon and General Secretary Leonid Brezhnev in May 1972, ushered in and became a symbol of the detente of the 1970s. The "new detente" of the late 1980s was similarly accompanied by the December 1987 signing of a treaty abolishing the superpowers' intermediate-range nuclear forces.
Arms control may be important in and of itself, apart from its connection to broad trends in U.S.-Soviet relations. Ideally, superpower arms control should help both states to reduce the costs of their continuing military competition. It should also serve to reduce the risks of war, and to limit damage to both sides should war somehow occur. Politicians, political scientists, and strategic analysts disagree strongly over whether U.S.-Soviet arms control has achieved any of these objectives. Nor is it obvious that arms control has, on balance, contributed toward a more constructive and mutually beneficial superpower relationship. In the course of this study, I will consider these questions and set forth an alternative perspective on the variegated history of U.S.-Soviet security cooperation. I will also prescribe an alternative approach to the practice of superpower arms control, an approach that I argue has greater potential to maximize the limited contribution that cooperation can make to the security interests and obligations of the United States and the Soviet Union. These analyses and prescriptions follow directly from the findings on the central question of this book: why have the superpowers achieved cooperation in certain arms control issues but not in others?
I seek an answer to this important question by developing a theory of international cooperation that starts from, but goes beyond a formal, game-theoretic approach. In his landmark 1984 work, Robert Axelrod set forth an elegantly simple formal model of the evolution of cooperation in a Prisoner's Dilemma game. This book is an attempt to explore and extend the applicability of Axelrod's model to the problem of explaining cooperation in U.S.-Soviet arms control. I will examine three historical cases—attempts to limit antiballistic missile systems, multiple independently targetable reentry vehicles, and antisatellite weapons (ABM, MIRV, and ASAT)—and I argue that each case fulfills the initial conditions for cooperation identified by Axelrod's model. The formal theory points toward a cooperative outcome in each case. But the actual outcomes do not match the prediction. Cooperation emerges in ABM but not in MIRV, and in ASAT a partially cooperative arrangement deteriorates over time.
Several conclusions are possible. Axelrod's theory could be "wrong" in and of itself. Or it could be valid, and simply not applicable to these cases. Neither of these conclusions is correct. I argue instead that the game-theoretic approach to the study of cooperation is incomplete. Axelrod's theory yields important insights about the evolution of cooperation, but it is not by itself sufficiently discriminating to explain the variance in outcomes among these three important cases. How can we do better?
One reason why the theory is incomplete is that it does not include an adequate analysis of strategy. Strategy, in game theory, is simply a decision rule—an algorithm that players use to select a course of action in a game. Axelrod shows that the simple strategy of "Tit for Tat," which cooperates on the first round and thereafter does whatever the adversary did on the previous round, can under certain conditions elicit cooperative outcomes in the Prisoner's Dilemma. But strategy in international politics is different. Strategy is the combination of political, military, and diplomatic measures a state employs to promote desired outcomes in its interactions with other states. States face a difficult challenge in developing a strategy that will elicit cooperation from a rival state. Tit for Tat may be an effective strategy in game theory but it does not translate directly into real world behavior. States can employ many different types of strategies based on reciprocity that are broadly consistent with the game-theoretic concept of Tit for Tat. And while Axelrod's "players" can change their behavior and adapt their strategies easily, neither is true of states. In U.S.-Soviet security relations, strategies shift only at critical junctures; once established, they are difficult and costly to modify.
I start with the hypothesis that the specifics of strategy make a difference. In other words, differences among strategies based on reciprocity can explain an important part of the variance in empirical outcomes that is left unexplained by the formal theory. I argue that the particular strategy of reciprocity that a state chooses at a critical juncture is a central determinant of whether cooperation does or does not occur. The basic reason for this is that influencing another state's behavior through a strategy of reciprocity is a more complex process than is captured in game-theoretic models. My hypothesis calls for an explanation of two processes: why states choose the particular strategies they do; and how different strategies bring about changes in the behavior of both states that may or may not produce cooperation.
Using the case studies as an inductive guide, I construct a typology of strategies based on reciprocity that the superpowers have used in their arms control relationship. It turns out that the range of strategies that will successfully promote cooperation in U.S.-Soviet arms control is severely circumscribed to a small subset of those that would be consistent with Tit for Tat. I will make the case that strategies based on one specific ideal type—what I call "enhanced contingent restraint"—will most successfully promote cooperation in U.S.-Soviet security relations. I explore in three case studies the conditions under which a strategy of enhanced contingent restraint can be implemented and how it operates to produce a cooperative outcome. I also consider why other types of reciprocal strategies fail.
Could we have gotten to the same point through formal theory? In principle, the answer may be "possibly"; in practice, the answer is no. The issues of interest here—how states conceive, develop, implement, and respond to strategies of reciprocity—are not well-suited for analysis by formal game-theoretic models. And because they are issues of the first importance for theorists and policymakers, I argue that empirically based research is for many purposes a more promising way to build theories of cooperation than is the further development of highly complex formal models.
This book is in no sense a definitive "test" of Axelrod's theory. Quite the contrary: since the theory does not yet explain the variance in outcomes among important cases that fit its assumptions and initial conditions, it cannot as it stands be tested against historical evidence. The purpose of this book is more circumscribed: to extend Axelrod's work in an empirical direction that generates new insights about cooperation and U.S.-Soviet arms control, while providing an important agenda for further research.
The Problem of Cooperation
Cooperation occurs in international relations when states adjust their policies in a coordinated way, such that each state's efforts to pursue its interests facilitate rather than hinder the efforts of other states to pursue their own interests. Cooperation is not always seen as a "problem" per se. It is common to hear that we live in a world of increasing interdependence, where global issues and shared interests create strong demands for international cooperation. Even security, the basic province of sovereignty, has been invaded by interdependence. Because nuclear weapons render the superpowers utterly and equally vulnerable, cooperation in limiting the shared threat to the survival of humanity has become an "imperative." A group of authors, sometimes lumped under the rubric "liberal interdependence" thinkers, argue that though it may take some time to work out the details, the United States and the Soviet Union will learn to cooperate in arms control because they must do so in order to survive.
This view is not acceptable to proponents of the more traditional, realist perspective on international relations. Realists justly criticize some of their more optimistic colleagues as being naive about power and conflict, and for underestimating the incentives for states to compete even when all stand to lose in absolute terms. But this critique can be carried too far. Joseph Grieco, among others, extends the realist view to suggest that "cooperation theory" is fundamentally wrong-headed because it adopts a liberal interdependence perspective on international relations and understates the importance of conflict.
This is a mistake. Cooperation theory, as I develop it here, starts from the fundamental premises of realism. I assume that the international system remains fundamentally anarchic. Without an international authority capable of making and enforcing rules, there are no inherent limits on the pursuit of sovereign interests by individual states. That extends to the use of force at any time. The consequence of anarchy is that each state must observe the principle of self-help and rely primarily on its own capabilities to ensure security and well-being.
Anarchy also places a premium on the "relative" position of states, as opposed to the absolute gains that may be realized from cooperation. Cooperation may make everyone better-off, but one state will be wary if it helps another state more than itself, since the other may then "use its disproportionate gain ... to damage or destroy" the state. In a realist world, states with conflicting interests are strongly inclined to eschew potential gains from cooperation with potential rivals.
The point is that "imperatives" do not always occur under anarchy, and the most compelling shared interests and the strongest of incentives will not always produce cooperation. Even under the most auspicious circumstances, cooperation is risky. The potential for exploitation is high, and the penalty for misplaced trust can be damaging or even catastrophic. Prudent political leaders are wary of conciliatory gestures, and they take seriously the worst-case interpretation of an adversary's intentions.
The "security dilemma" is an important result. States that are engaged in military preparations, even if intended for defense, will tend to provoke responses from other states, who must always fear the threat arising from any state's enhanced capabilities, regardless of intentions. In some situations, the incentives to exploit another state or to act preemptively in the face of possible exploitation may be compelling. And if these structural factors were not by themselves sufficient to hamper security cooperation between two competing superpowers, the difficulties are enhanced by the general hostility and distrust that has characterized U.S.-Soviet relations over time. The fact that each superpower's ideology long identified the other as the major threat to international peace and progress does not make cooperation any easier.
Yet few relationships between states are characterized solely by conflicting interests. U.S. and Soviet leaders have at various times perceived a number of interests in common, including interests relating to their own most fundamental security concerns. Some common interests can be achieved through unilateral actions that states can take on their own—for example, deterrence based on mutual assured destruction (MAD) can be had if both states independently deploy large arsenals of survivable nuclear weapons suitable for retaliatory attacks against cities. However, states in this position may wish to coordinate their behaviors so they can achieve the same result at lesser cost. Sometimes the incentives can be great. If Washington and Moscow agree not to deploy counterforce weapons that threaten each other's retaliatory capabilities, the problem of guaranteeing MAD can be solved more easily, less expensively and with greater confidence. There are other instances where common interests can only be realized through coordinated action. Either of these conditions can spawn a demand for cooperation, even between long-term adversaries.
In the starkest of realist worlds, this demand for cooperation would go unmet. But the real world is not so stark: despite the obstacles, the scope and magnitude of international cooperation have sometimes been quite extensive. International relations theory needs to account for this cooperation in some way, because it affects the character of life in the international system for all states.
Realism, by itself, has trouble explaining some forms of cooperation. A realist view can accomodate limited cooperation that occurs when states' short-term or "myopic" self-interests are identical: U.S.-Soviet cooperation that is directly tied to the superpowers' overwhelming interest in avoiding a catastrophic nuclear war is not a puzzle for realism. But superpower cooperation in arms control has at times gone beyond shared myopic self-interest. In 1972, Nixon and Brezhnev signed a treaty that severely limited the deployment of antiballistic missile systems. During the 1960s and much of the 1970s, the two sides refrained from developing and deploying antisatellite weapons. More recently, the superpowers agreed to remove a large number of medium-range missiles from Europe; and in 1990 they are on the verge of accepting substantial reductions in their strategic arsenals. None of these actions was necessary to prevent nuclear war. All involved the possibilities of cheating and relative gains. Realism can accommodate this kind of cooperation only uncomfortably. The logic of realism leads us to expect that these more elaborate forms of security cooperation will be short-lived, tenuous, and insignificant in the evolution of relations between the superpowers.
These are a few "successes" of arms control. There have also been many failures. Washington and Moscow failed to place meaningful constraints on MIRV warheads in the early 1970s, when the arguments for doing so were at least as convincing as those favoring restraint in ABM. And although the character of U.S.-Soviet relations has changed dramatically at the beginning of the 1990s, the potential for additional costly failures in arms control remains.
It is an important weakness in the realist argument that it cannot explain why cooperation develops in some of these instances but not in others. Cooperation is not inconsistent with realism, but this is a much weaker statement than to say that cooperation is explained. Success or failure matters, and sometimes it matters greatly. Cooperation has not always been short-lived and tenuous. It has had tremendous impact on the evolution of U.S.-Soviet relations. Realist theory as it stands addresses the constraints on security cooperation, but it has little to say about how states sometimes overcome those constraints, and even less to say about the consequences of their having done so.
Cooperation theory does not deny the constraints. It does not replace realism. It is an addition to realism. It elaborates and supplements the theory to provide a more precise explanation: to address why, how, and under what circumstances cooperation will emerge between states in an anarchic environment. It should also deal with the consequences of cooperation.
(Continues...)
Excerpted from Cooperation and Discord in U.S.-Soviet Arms Control by Steve Weber. Copyright © 1991 Princeton University Press. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.