Download Tit-for-tat and the Negotiation of Nuclear Arms Control and more Study notes Negotiation in PDF only on Docsity!
Tit-for-tat and the Negotiation
of Nuclear Arms Control
George Bunn* and Rodger A. Payne**
Robert Axelrod's The Evolution Of Cooperation has been widely ac- claimed in the few years since its publication.^1 Given its promise for promoting cooperation in 'prisoner's dilemma games' (PDGs), such praise is unsurprising. The purpose of this paper is to consider whether the tit-for-tat (TFT) strategy Axelrod recommends for achieving an evolution of cooperation in PD situations has application to the negotia- tion of arms control treaties. The structure of many important social and political problems resem- bles the PDG interaction (see the Appendix for a brief description of PDGs). For instance, many competitive bilateral situations - including arms races - are said to share the logic of the PDG. Two states in an arms race have incentives to continue the competition, although they will be worse off if they escalate the race than if they agree to stop it.^2 Each state fears it will suffer a 'sucker' payoff should it practice restraint while the opponent continues escalating the competition. Like- wise, each is tempted to seek unilateral advantage when the opponent practices restraint. In all PD situations, individually rational decisions can lead to inferior group outcomes. Searching for ways that players can overcome the dilemma and receive high payoffs, social scientists have thoroughly examined various strategies that can be employed in the PDG. Axelrod, among others, recognized that players have an incentive to cooperate over repeated plays of the game.^3 In an iterated PDG (a so-called 'supergame'), players eventually discover that non-cooperative behavior yields poor payoffs for all. They also learn that mutual cooperation provides the greatest rewards over time. Axelrod finds that players in a PD supergame should adopt a tit-for-tat (TFT) strategy so long as a few basic pre-requisites are met. Assuming players are rational egoists who do not greatly discount the value of future payoffs, all players should make moves based on an opponent's previous move.^4 TFT can provide the large payoffs promised by coopera- tion; yet, at the same time, players retain the option to defect when faced with an uncooperative opponent. Using a computer tournament,
- Member in Residence, C.I.S.A.C., Stanford University. ** C.I.S.A.C., Stanford University.
208 ARMS CONTROL
Axelrod discovered that TFT performs better than virtually any other strategy in a PD supergame between two players. Tit-for-tat is robust and stable. Over time, it can 'invade' a world of non-TFT strategies and prove successful. In this way, cooperation can evolve in almost any PD situation.
TFT IN THE LITERATURE Because of TFT's apparent ability to yield cooperative results when competition seems rational, many international relations scholars have applied Axelrod's findings to their own work. Even Axelrod says that states in a bilateral relationship can escape the temptations of the prisoner's dilemma. Rivals locked into an ongoing competition can learn to cooperate using a tit-for-tat strategy. Axelrod cites specific examples from international relations history where TFT-like cooperative behav- ior might explain the course of events. For instance, he devotes a chapter to the evolution of cooperation between opposing forces in the trenches of the First World War.^5 Taking his argument several steps further, some scholars argue that competition in the U.S.-Soviet relationship can be eased via the appli- cation of tit-for-tat strategies. A few have even argued that TFT is sometimes employed by the superpowers. For instance, Nye says that the U.S.-Soviet relationship reflected in the current 'Non-Proliferation Treaty loosely represents a situation that game theorists call iterated prisoner's dilemma. Each has learned that cheating and taking short-run advantage could lead to similar behavior by the other and loss of their common interest in slowing the spread of nuclear weapons.'^6 Nye is referring not to the 'offer and acceptance' bargaining which produced the text of the NPT, but to the trial-and-error, action-reaction process which convinced each superpower that such a treaty and other related agreements and practices which comprise the non-proliferation regime were in its interests. The offer-and-counteroffer-or-acceptance bargaining which produces a specific agreement is often difficult to distinguish from the action and reaction which brings participants to the bargaining table, and which influences them later to implement what they have negotiated. As Alexander L. George suggests, however, such distinctions are essential for proper consideration of the usefulness of game theory strategies such as TFT in the negotiation of U.S.-Soviet treaties and the building of U.S.-Soviet regimes.^7 The difference is relatively clear. In bargaining, two players come to agreement on what they will do before they act. In action-reaction, the two players act independently without talking - although each knows the other's prior moves.
210 ARMS CONTROL
side's 'cooperative impulses' to find an agreement leaving both better off, and its 'competitive impulses' to get more than an opponent from the situation. Neither Kennedy nor Khrushchev could gain desired results without acceding to a ban on atmospheric testing in his own country. Thus, each faced a 'negotiator's dilemma' not unlike the prisoner's dilemma.
- Second, the action and reaction which often precedes serious bargain- ing may be better modeled as employment of TFT strategy in a PDG, than as offer-counteroffer bargaining at a negotiating table. The payoff structure states seem to perceive in pre-negotiation action-reaction may resemble a PDG, and the cooperative strategies they use to begin nego- tiations share many characteristics with TFT. This notion is explained below in the section on 'nice' characteristics.
- Next, the claimed benefits of arms control justify a thorough examina- tion of alternative means of achieving it. A large number of scholars now assume that PDGs accurately model arms races, and a growing number believe TFT induces cooperation in PDGs, even between states.^10 For our purposes, we have assumed that PDG accurately models the arms race in order to examine whether TFT strategy can contribute to arms control negotiation. If the prisoner's dilemma game accurately models the competition of arms races, perhaps the tit-for-tat strategy can suggest how to slow them. After all, TFT can apparently induce an evolution of cooperation in many otherwise competitive environments. Some important scholars have explicitly called for an application of TFT to arms control bargaining. Duncan Luce argues forcefully that TFT should be applied to superpower arms control: 'For the arms race, it may be that Mr. Axelrod's teaching approach is our best hope, provided decision makers on both sides... adopt explicitly his Tit for Tat strategy.'^11
- Fourth, an examination of tit-for-tat seems worthwhile because empirical studies of past arms control negotiations find that TFT-like reciprocity often describes superpower behavior in this situation. Interestingly, studies testing (and supporting) two divergent theories of negotiation behavior commonly find that TFT-like reciprocity occurs when parties reach agreement. Both are described here. Frequently, studies of nuclear arms negotiations seem to assume that the first of these, the concession/convergence bargaining model, is a good 'fit'. This model suggests that each side starts negotiating from a position different from the opponent. Agreement is reached by exchanging concession for concession until the positions converge.
THE NEGOTIATION OF NUCLEAR ARMS CONTROL 211
Empirical studies of the concessions and retractions in nuclear test ban negotiations and SALT (Strategic Arms Limitation Talks) conclude that they were characterized much of the time by reciprocity in concessions, or 'various forms of tit for tat'.^12 It should be noted, however, that the test ban studies were based almost entirely on the unsuccessful attempts to achieve a comprehensive treaty banning all tests, rather than the successful negotiations which led to the LTBT.^13 Despite an assumption of concession/convergence from the begin- ning, some studies of SALT concessions and retractions show generaliz- ed reciprocity, though not exactly Axelrod's TFT. Studies by Jensen, and by Stoll and McAndrew, weigh and analyse the moves which produced four agreements: the SALT I Interim Agreement, the Anti-Ballistic Missile (ABM) Treaty, the 1974 Vladivostok formula agreement, and the SALT II Treaty. For the first two agreements, Stoll and McAndrew found TFT-like cooperative reciprocity in that concessions begat conces- sions. However, this cooperation was usually more like tit-for-two-tats, or even tit-for-three-tats. For the latter two agreements, they found inverse reciprocity - concessions begat retractions more often than concessions. Overall, the correlation between a single concession and a responding concession (which would be quite like TFT) was not strong. However, a trend of concessions or a group of concessions on one side seemed to make a difference to the other - though it did not always produce cooperation. Despite this mixed evidence, Stoll and McAndrew conclude that 'U.S.-Soviet interactions were characterized more often by cooperative reciprocity (various forms of tit for tat) than by inverse reciprocity (various forms of exploitation)'.^14
In the literature concerned with international negotiations, 'formula' approaches are contrasted to concession/convergence theories. Scho- lars subscribing to the formula approach believe that negotiators usually strive to find an initial formula for agreement which will link issues to meet interests sought by both sides. This is especially typical in the beginning stages, when many outstanding issues exist.^15 William Zartman's analyses of case studies and interviews with international negotiators suggest that formula/detail models better describe international negotiation than do concession/convergence models. When only one important quantifiable issue was on the table - a wage rate, for instance - concessions and convergence were obvious, but always in reference to an accepted standard or formula (like the cost of living). With many issues on the table, and not all of them quantifiable on the same scale, the use of a formula to link them was even more evident. While not rejecting concession/convergence models, especially for negotiating details after a formula has been chosen, Zartman found the formula/details approach dominant in international negotiations.^16
THE NEGOTIATION OF NUCLEAR ARMS CONTROL 213
to offer the best application of TFT strategy to arms control.^19 Propon- ents foresee mutual restraint which occurs without formal agreement through action and reaction. It is therefore closer to the TFT strategy than the offer and counteroffer of treaty negotiation. Scholars (prin- cipally Schelling) and policymakers (including former ACDA Director Adelman) have seriously considered tacit arms control.^20 Nonetheless, we do not place much emphasis on this kind of restraint.^21
'Mutual example' restraints are unlikely to yield much more than tem- porary halts in weapons testing and the like. Formally negotiated arms control treaties have come to wield considerable symbolic significance. Meaningful reductions seem quite improbable in the absence of a formal treaty. Only certain types of restraint, perhaps involving new areas of the arms race - such as the mutual reluctance to test submarine missiles in depressed trajectory flights, or ASAT test moratoria - seem likely to 'evolve' tacitly.^22 Today's pressing issues, like warhead limits, de- MIRVing, and deep reductions, would probably require formal treaties.
The United States' legal framework also makes eventual congres- sional approval of agreements to limit or reduce arms a necessity.^23 Certain types of arms control are illegal in the absence of treaties or congressional legislation. Further, verification problems cast doubt on the desirability of tacit arms control agreements. What constitutes a violation of an agreement that has not been formalized? How can tacit agreements survive 'minor' defections? The formal process at least has a Standing Consultative Commission (SCC) to institutionalize punishment and forgiveness. Thus, we focus on whether policymakers seeking arms control treaties can or should adopt TFT strategy to initiate or conduct negotiations, although some of the discussion applies to tacit agreement. Our analysis covers two primary areas of interest. Initially, we look at the four key substantive characteristics of TFT strategy identified by Axelrod which help make it a useful action strategy. It is clear, nice, reciprocal (or retaliatory) and forgiving. We ask how readily nuclear arms bargaining can share these characteristics. Next, a few process questions are con- sidered. We analyse the potential problems faced by the bureaucracy, the Congress, and the President in employing TFT, especially given public opinion. Throughout the argument, both theoretical and practical objections are raised.^24
214 ARMS CONTROL
THE FOUR CHARACTERISTICS OF TFT
Clear First, Axelrod argues that the tit-for-tat strategy is unambiguous and that such clarity is a strength. Simply, it allows ready recognition by the players. Opponents expect both cooperative and uncooperative strategies to be reciprocated because their moves are recognizable. However, clarity is actually highly problematic given the difficulty of defining bargaining moves in a TFT context of strategic choice. What offers reflect cooperation or defection in a negotiating forum? Cooperation might mean acceptance of offers, although a TFT strategist could argue that it also included some counteroffers, or even willingness to continue talking without reaching agreement. Defection could per- haps include rejection of the other side's offers, withdrawn offers, or introduction of unreasonable offers. However, as will be argued below, defection would probably also be symbolized by actions outside the negotiating forum. At the most basic level, players may not be able to interpret even what entails a move or response. There is no easy solution to this problem of definition. To test whether TFT would remain robust when clarity is imperfect, Axelrod included a tournament round in which there was a one per cent chance of misperception by one player of the other's intention to cooperate or defect. TFT remained the best strategy in this round, though it 'got into a lot of trouble when a single misunderstanding led to a long echo of alternating retaliations'.^25 Suppose, however, the chance of misperception is 25 per cent, or more.^26 In the real world of superpower relations, clarity may be no better. One cannot look simply at the offer and counteroffer at a bargaining table to determine whether to respond cooperatively or to defect. For example, intelligence failures, whether of commission (analysis) or omission (detection) can lead to misperceptions. The status of the Soviet phased array radar in Krasnoyarsk during the ongoing strategic arms reduction talks (START) illustrates this problem. What does this radar symbolize? Should the U.S. have ignored it and concentrated on the cooperative signals it received at the bargaining table? The Soviets denied the ABM tracking capabilities of the radar and apparently agreed to delay construction to allow for negotiations. Thus, though no 'defection' may have been intended, non-cooperative signals were received by the U.S., which formally charged the Soviets with an ABM Treaty violation. The data about other alleged instances of Soviet cheating are even more ambiguous and reflect the continuing
216 ARMS CONTROL
has noted on pork-barrel legislation. Simply stated, broad packages offer something for everyone, reducing incentives to defect.^27 However, despite some talk of linkage, arms control policymakers do not always exhibit a willingness to trade votes. For instance, Secretary of State George Shultz readily linked Soviet human rights behavior to arms control, noting that 'Soviet abuses can only jeopardize efforts to make progress in all areas of U.S.-Soviet relations'. However, Shultz refused to accept Soviet linkage demands. Despite 'their efforts to link a possible INF [intermediate nuclear force] agreement with our acceptance of their position on SDI,' he said, 'we don't see any reason why these issues should be linked.'^28 Another clarity problem concerns the role of superpower commu- nication. Both the U.S. and Soviet Union do some of their arms control bargaining outside the negotiating rooms - seeking a pub- lic relations effect - which can render interpretation of real offers problematic. Moreover, persuasive communication can be used to alter payoff structures - often by attempts to link or 'de-link' issues. Some research suggests that seemingly slight changes in the way choices are 'framed' can greatly affect outcomes.^29 In short, through both verbal and written discourse, the superpowers often transmit threats, bargains, 'policy positions', and intended actions; yet, superpower communication can also serve propaganda or public relations goals.^30 The nuclear age can be considered a rhetorical age, but what rhetoric is important and what is not? Leaders can find it nearly impossible to distinguish among intentions, hyperbole, and actions. The U.S. also has difficulty making its positions clear because of its pluralistic government. The executive and legislative branches, for instance, are often in disagreement about the course of U.S. arms policy. Sometimes it seems as if these two policy actors share the role of the 'U.S. player' versus the usually more unified 'Soviet player'.^31 Although not a bargaining example, different opinions on the fate of SALT II illustrate the magnitude of disagreement. Just in this decade, the Reagan Administration denounced SALT II as 'fatally flawed' (despite voluntarily complying with its provisions for about five years), while the Congress voted for bills to force U.S. compliance. Likewise, the Administration condemned Soviet anti-satellite (ASAT) devices and called for a U.S. counterpart, even as Congress voted to deny funds for testing the U.S. system, and numerous members of Congress called for negotiations. These examples illustrate the problem of achieving clarity within the relatively narrow realm of nuclear arms control. In short, even if most of the political differences between the U.S. and the Soviet Union can be isolated from arms bargaining, ambiguity is still possible, if not likely.
THE NEGOTIATION OF NUCLEAR ARMS CONTROL 217
Nice TFT is said to be nice because a player employing this strategy cooper- ates in the initial move. According to Axelrod's computer PDG compe- tition results, nice strategies consistently scored higher than did those beginning with unfriendly moves, during iterated play. However, in the context of the nuclear arms race, nice strategies in Axelrod's sense are impossible to implement. The opportunity to make the initial nice move is long past for either state. For our purposes, we will assume that a state beginning a TFT strategy via cooperative moves is 'nice'. Still, we would argue that such a strategy may be impossible to implement and some observers would say nice strategies are dangerous to national security. The practical problem is the easiest to demonstrate, but not nec- essarily the simplest to overcome. The U.S. and the Soviet Union have been locked into competition for over forty years. Therefore, a history of defections must be considered when approaching nuclear arms bargaining, whether tacit or formal.^32 Indeed, although instances of cooperation are sprinkled throughout the history, the periods of great tension have been longer than the periods of detente. Theorists might argue that greater cooperation will eventually evolve as opponents learn the folly of their ways, but practice in this instance suggests scepticism. Can credible cooperative plays be implemented during uncoopera- tive periods? Would the opponent exploit them, or choose instead to cooperate? Decision makers attempting to integrate an explicit TFT strategy into policy would face questions like these. Some theorists might argue that several consecutive cooperative plays following a period of defections would convince an opponent of one's sincerity and lead to mutual cooperation.^33 Alternatively, others might call for a GRIT strategy (graduated reduction in tensions) to initiate cooperation, based on carefully planned and announced conciliatory moves.^34 However, any decision maker would be faced with a dilemma. If cooperative plays are insignificant, the opponent can readily dismiss them as propaganda. The fact that cooperative moves have been sprinkled throughout the arms race compounds the problem because few would believe that the current moves are any more indicative of cooperation. If the cooperative plays are significant, the opponent can exploit them. Neither fate seems particularly promising. The risk - and promise - of cooperative plays following long periods of hostility are illustrated by President Kennedy's efforts to achieve a nuclear test ban treaty. Clearly he faced an uphill battle. After a brief thaw in U.S.-Soviet relations produced the first serious test ban nego- tiations near the end of the Eisenhower Administration, Khrushchev
THE NEGOTIATION OF NUCLEAR ARMS CONTROL 219
the Cuban Missile Crisis] - in the early stages of a dispute and then switches to conciliation'.^38 Lessons learned from analyses of games, experiments, and real life case studies collected by Patchen suggest the utility of such a strategy for beginning the cooperation which can lead to serious bargaining in some circumstances. Finally, the initiation of cooperative arms bargaining ploys might be perceived as dangerous and naive by foreign policy actors in the U.S, particularly during some future period of tension. Many decision mak- ers, for instance, have long advocated bargaining chip theories in nuclear arms negotiating situations. The Reagan Administration's strategy has explicitly been to bargain with the Soviets from a position of strength. Weapons are developed or deployed before they are bargained away. Given the claimed success in the area of intermediate nuclear forces, this strategy will probably gain in popularity among foreign policy elites. It is not difficult to imagine negotiators resisting attempts to adopt nice strategies by arguing that their bargaining positions would be undercut.
Reciprocal Axelrod states that another strength of the TFT strategy is its promise of reciprocal punishment. Retaliation is required in-kind for any defection, thus the provocable threat of defection serves to induce cooperation. Again, however, the achievement of true reciprocity is problematic in the real world of nuclear arms control bargaining. The vast agendas of the superpowers complicate reciprocal bargaining. Just as diverse interests confound clarity, they make reactions difficult to interpret. Which 'bad' actions merit punishment? What defections would be recognized as punishment by the opponent? In a complex world, these are questions that are not readily answered. For instance, how should the Soviets have interpreted the Reagan Administration's reaction to their downing of Korean Air Lines flight 007 in 1983? Few specific sanctions were enacted and the President soon delivered a fairly conciliatory speech at the United Nations calling for arms control. Yet the Administration initially employed very hostile rhetoric, calling the incident a massacre of innocent civilians.
This example is not atypical. Policymakers usually rely on rhetoric to explain their retaliatory moves, although verbal descriptions are inherently imprecise. For instance, the Carter administration said that it would not accept the status quo in Cuba upon discovery of the Soviet brigade there, but little more than one month later it basically did accept the status quo, despite making claims to the contrary. Appropriate reciprocal moves might be difficult to determine even if the context is confined to nuclear arms. Limiting the appraisal to the offers and counteroffers at the bargaining table would not stop
220 ARMS CONTROL
the inevitable linkages with other issues. How, for example, should a state respond to cheating? Should all ongoing negotiations be can- celled? Should the opponent cheat as well? If so, should its cheating be 'proportional' to the first cheating, as the Reagan Administration suggested in the case of SALT II? Obviously, each of these questions invokes a series of difficult problems, not all of them directly related to arms bargaining. If cheating is merely taken to the SCC, it might not be resolved satisfactorily for some time. Yet, reciprocal cheating might lead to the collapse of all arms control, as could withdrawal from ongoing negotiations. Thus cheating, though dealing with treaty implementation, clearly affects what happens at the bargaining table. Even Axelrod recognizes that retaliation must not be too great, lest it lead to an unending echo of defections.^39 Finally, the U.S. might find it politically impossible to implement TFT strategies because of the timing demands. Axelrod says that some defections must be answered immediately, and other scholars have also noted the importance of prompt retaliatory responses.^40 Unfortunately, neither the bureaucracy nor the Congress of the U.S. are especially well known for their ability to respond quickly to foreign policy demands. Even a president can fail to act quickly in the face of apparent defection, perhaps delayed by inadequate information. For instance, SALT II suffered during Carter's long delay before responding to the Soviet brigade in Cuba. Discovery was revealed in August, but Carter did not deliver his speech on this issue until October.
Forgiving Finally, Axelrod argues that the TFT strategy is useful because it is forgiving. Players need not find themselves locked into defection even though payoff temptations may occasionally result in uncooperative plays. Nuclear arms bargainers are typically not very forgiving. Defections are exploited politically by the opponent. For example, when the U.S. refuses to concede on a Soviet bargaining point, it is likely to be attacked publicly as the obstacle to agreement. Beyond the bargaining table, cases of cheating are often highly politicized. In the U.S., the existence of third parties - public opinion, Congress, allies - provides ample opportunities to exploit defections and makes forgiveness more difficult. The Reagan administration continually pointed to the Krasnoyarsk radar, along with a few other alleged Soviet arms control violations, to demonstrate Soviet non-compliance. In turn, it used this cheating as an argument for larger U.S. defense budgets and justified some potentially uncooperative positions in ongoing negotiations by highlighting these violations. For instance, an oft-repeated U.S. demand for on-site inspection techniques
222 ARMS CONTROL
Because formal arms control processes are the most likely, the general public, the bureaucracy, and the Congress, as well as the president, take on important roles. In short, the political significance of a treaty (as opposed to merely tacit restraint) virtually assures the active partici- pation of all the actors. Yet, TFT cooperative strategies cannot readily emerge out of a system that features a somewhat 'schizophrenic' public, an entrenched bureaucracy, an independent Congress, and a strong president. Axelrod himself recognized that 'bureaucratic politics' was not incorporated in his PDG tournaments and that political leaders with strong self interest, especially lame ducks, can easily choose to 'seek private goals rather than maintain a pattern of cooperation'.^42
Public Opinion Obviously, members of the general public do not formally participate in the arms control negotiation process. Nonetheless, as discussed above, popular opinion can have a significant impact on the fate of treaties by influencing the negotiation and ratification processes. Put simply, Congress and the President alike take public opinion into account when planning arms control policy. Because public opinion can vary greatly on these issues, however, a TFT negotiating strategy would be difficult to implement. Public opinion certainly seems 'schizophrenic' on arms control issues.^43 For the most part, this is because diverse members of the general population hold widely different views on the superpower arms competition. Many seem to believe, for instance, that the control of nuclear weapons is a very high priority. They may advocate unilateral U.S. restraint even if the Soviets do not reciprocate. During the peak of the anti-nuclear weapons movement in the early 1980s, 45 per cent of the public supported a nuclear freeze even if it was to be a unilateral U.S. effort.^44 This portion of the population apparently felt that the accumu- lation of arms was more likely to lead to disaster than would unilateral Soviet arms increases. Those who hold this view would be generally un- supportive of a TFT strategy. If their views were implemented into policy, the U.S. would probably adopt cooperative strategies even in the face of Soviet defections. This would endanger reciprocity. Still other views on arms issues are at the other extreme of the scale. A significant portion of the public consistently supports increased defense spending because of fear of the Soviet Union, even at great cost. For instance, opinion polls taken after the Reykjavik summit showed that most people support continued SDI funding even if this action leads to deployment and assures that no arms control agreement can be achieved.^45 Many of these members of the general population believed President Reagan when he referred to the USSR as the 'evil
THE NEGOTIATION OF NUCLEAR ARMS CONTROL 223
empire', and consequently do not support cooperative measures. Their views are generally inconsistent with TFT strategies. If these ideas were implemented into policy, the U.S. would probably adopt defection strategies even in the face of Soviet cooperation. Policy would be neither nice nor forgiving. Of course, there is a range of public opinion about arms control falling between these two extremes. However, some views carry more weight when they are upheld by organized interest groups like the Committee on the Present Danger or the Council for a Livable World. U.S. arms control policy is influenced when political leaders pander to one or more of the particular views. Undoubtedly, public protests against the MX slowed the system's deployment, and theoretically had an impact on U.S. arms negotiation positions. President Reagan argued against MX limits by saying that the Soviet Union would not cooperate on arms control if it knew that the U.S. would unilaterally cooperate. Similarly, public outrage at the Soviet invasion of Afghanistan and troop presence in Cuba during 1979 influenced the fate of SALT II and ultimately slowed all arms bargaining for years. Some leaders attempt to be 'all things to all people' and design strategies that baffle Soviet leaders. These unclear approaches are no better than those that tilt toward special interests. Seemingly, President Carter's ill-fated 1977 arms control proposal was of this sort. Carter tried to appease conservatives (especially Senator Jackson) and liber- als alike by designing a new deep reductions arms control proposal. The Soviets were shocked and confused by the plans and immediately rejected them.^46
The Bureaucracy
When it comes to arms control negotiations, the U.S. does not bargain as an individual rational actor would, but is instead a victim of bureaucratic politics. American negotiators are typically messengers, not movers. Their instructions are usually developed through a bureaucratic wrangle which often gives key agencies, senators, and allies an effective veto over them.^47 In terms familiar to rational choice theorists, this unanimity rule makes interest aggregation a substantial problem.^48 For example, it is the conventional wisdom of experienced arms control negotiators that no significant arms control treaty can get a two-thirds vote in the Senate if it is opposed by the Joint Chiefs of Staff. Thus, the Chiefs, among others, have a very powerful voice in this in-house negotiation.^49
The overall result is that American negotiators often begin with an agency-consensus position which would give up little on the U.S. side to gain a great concession from the Soviets. This negotiating stance might be called a 'lowest common denominator'. Bargainers
THE NEGOTIATION OF NUCLEAR ARMS CONTROL 225
easily see development of new weapons as defections. Moreover, either reciprocity or forgiveness is likely to be jeopardized in this process if any Soviet actions during the ratification debate, or even during negotiation, anger senators. Failure to respond would potentially signal weakness to conservative senators and simultaneously threaten reciprocity. Strong response might pacify senators, but endanger forgiveness.
The President
The president and his appointed advisors obviously have a strong stake in arms control negotiations. However, it is not clear that their position makes TFT practical. First, since the president cannot unilaterally make arms control trea- ties, clarity is always uncertain. This has already been explained in the above sections dealing with the bureaucracy and Congress. Second, even a president in control of the bureaucracy could easily adopt a strategy that does not reflect TFT. In fact, there are often great temptations to reject cooperative reciprocal strategies. Competing pol- itical interests - especially public opinion - can lead any president away from adopting TFT. Domestic political incentives can push the president and his immediate advisors to an uncooperative negotiating stance. The U. S. polity has been described as schizophrenic in its attitudes towards the Soviet Union and arms control. Members fear (and distrust) the Soviets, but also usually support the control of nuclear weapons through agreements. As Anthony Downs explained in his classic argument, these kinds of competing political forces invite politicians to take median positions while trying to outflank all political opposition.^52 Downs said that politicians tend to react to voter attitudes in much the same way that resource allocations in the economy respond to changes in consumer demand. Thus, whenever public fear of the Soviet Union is thought to be more important than pressure for arms control, political leaders seek to reflect that sentiment, thereby outmaneuvering opponents, but rejecting cooperative arms control strategies. TFT cannot succeed if it is held hostage to political whims of this sort.
Next, presidents can have alternative preference structures that would prevent their application of TFT strategies. For instance, some may desire U.S. strategic superiority, if such a goal is attainable. The Reagan Administration's early bargaining appeared to have been based on this approach.^53 In game theoretic terms, this strategy would make the U.S. a difference maximizer, rather than a utility maximizer.^54 An advantaged position is sought even if it means a suboptimal outcome. As potentially measured by an interval payoff schedule after an iterated PDG, a 10 (U.S.) to 5 (Soviet) arms race payoff is preferred to a
226 ARMS CONTROL
greater - although more equitable - outcome obtained through arms control bargaining, perhaps utilizing TFT strategies (such as 14 to 13). The employment of other non-utility driven bargaining schemes could conceivably make the implementation of TFT difficult, if not impos- sible. For instance, an egalitarian bargainer seeks equal outcomes, even sometimes at the expense of his own utility. If a president seeks any equivalence, rather than the best long-run U.S. payoffs, skewed outcomes may result. Carter's SALT II negotiating result could be called egalitarian. For many critics, Carter accepted some equal numerical arms limits for the sake of obtaining agreement, although the ceilings were not necessarily in the best interests of the U.S.5 5 Neither difference maximizing nor egalitarian strategies are very compatible with TFT. A basic reason players employ TFT strategies in PDG situations is that they are trying to maximize their own payoffs in iterated play. However, those who hold alternative preference structures reject the logic of rational egoism. In standard PDG formats, two rational egoists prefer a cooperate/cooperate payoff over iterated play, but an egalitarian seeks any result approximating equality (including potentially a defect/defect payoff), and a difference maximizer seeks a 'temptation' outcome.^56 Thus, a president with one of these alternative preference structures would conduct arms control negotiations very differently than would a rational egoist employing TFT.
IMPLICATIONS
This essay has argued that TFT strategies cannot be easily imple- mented in nuclear arms control negotiations. Both theoretical and practical objections were raised. The purpose has not been to dismiss Axelrod's 'evolution of cooperation' idea. Rather, we have attempted to demonstrate the difficulty faced by arms control negotiators attempt- ing to implement such plans. Unfortunately, interesting theoretical constructions sometimes cannot be readily applied to practice. This section discusses a few implications of our argument for policymakers and scholars.
Policymakers
We agree with many analysts that the iterated PDG is suggestive though imperfect in modeling the superpower arms race.^57 Additionally, the implementation of some U.S.-Soviet arms control arrangements - such as the non-proliferation regime - resembles employment of TFT strategy in iterated PDGs. Moreover, assuming sufficient common interest to support negotiation of a U.S.-Soviet arms control agreement, and a relationship permitting joint action, techniques for eliciting cooperation
