上海交通大学：《当代中国外交》课程教学资源（相关阅读）Geneva, Bandung & Rapprochement_Geneva Conference_Geneva Conference.pdf_大学文库

The Geneva Conference of 1954 A Case of Tacit Deception FRANK C. ZAGARE Department of Political Science Boston University This article uses a game-theoretic model of deception to examine a game played at the Geneva Conference of 1954 by the Western Alliance, the Sino-Soviet bloc and the Vietminh. It argues that if this game were played as a game of complete information, the sophisticated outcome would have been a withdrawal of French forces from Vietnam, followed immediately by an election whose probable winner would have been Ho Chi Minh. For the Western Alliance, especially the United States, this outcome was seen as the least-preferred of the three possible outcomes. However, because the Western Alliance was able to make a false announcement of its preferences, itwas able tacitly to deceive the Soviets, Chinese, and Vietminh into believing that its misrepresentation was its true preference. Thus, it was able to induce its second-most-preferred alternative, the partition of Vietnam, as the (manipulated) sophisticated outcome of the game. Contradictions in a nation's foreign policy pronouncements are sometimes explained in terms of bureaucratic inefficiency (Allison, 1971: ch. 4), the lack of an integrated and coherent policy (Reston, 1955: 62), or shifting or "deteriorating" preferences (Howard, 1971: 148, 199-201). However, such a discrepancy may also indicate that another process is at work, namely, deception. In many situations it may be rational for an actor to deceive another in order to induce a more-preferred outcome.' 1. For a listing of these situations, see Zagare (1977a) and Brams (1977). AUTHOR'S NOTE: I would like to thank Steven J. Brams, Gerald DeMaio, the late Oskar Morgenstern, Richard N. Swift, Maria Zaremba, and Dina A. Zinnes for reading an earlier version of this manuscript and making many helpful suggestions. INTERNATIONAL STUDIES QUARTERLY, Vol. 23 No. 3, September 1979 390-411 o 1979 ISA. 390 This content downloaded on Sun, 27 Jan 2013 21:58:56 PM All use subject to JSTOR Terms and Conditions

Zagare / GENEVA CONFERENCE 1954 391 Contradictory statements by government officials might be a manifestation of such a deceptive strategy. To demonstrate this contention, I will briefly describe a model of deception developed by Brams and Zagare (1977). Then, once a few key game-theoretic terms are operationalized and ascribed empirical meaning, I will show how the model can be used to offer an explanation of the apparent lack of coherence in the Eisenhower Administration's policy toward Southeast Asia in 1954, a policy that one analyst has characterized as wavering "between a point just short of military intervention and a point just short of appeasement" (Reston, 1955: 62). A Brief Exposition of the Deception Model2 Consider a game composed of three players, Plane, Row, and Column, and assume the players must choose from among a set of three alternatives A = {a,, a2, a3}. Let the first alternative, ai, be identified as the status quo. Assume that decisions in this game are a function of the following decision rule: If two or more of the players agree on one of the three alternatives, that alternative is the social choice. If there is no agreement, that is, if all three players disagree, the status quo, ai, prevails. Given this set of alternatives and this decision rule, the threedimensional outcome matrix depicted in Figure I results. Each dimension (plane, row, and column) represents the outcomes associated with the strategy choices available to the players with the corresponding name. In this game, each player has three possible strategies, that is, to pursue one of the three alternatives. This essay assumes that the the choice of the Plane, Row, and Column players associated with the strategy "pursue ai" is the first plane, row, and column respectively, and similarly for the other strategy choices. Hence, the symbol "as" not only stands for an alternative but also for a 2. This section is based on Brams and Zagare (1977, 1979). This content downloaded on Sun, 27 Jan 2013 21:58:56 PM All use subject to JSTOR Terms and Conditions

Zagare GENEVA CONFERENCE 1954 393 Column players results in the outcome al at the intersection of the first plane,first row,and first column. To illustrate the subsequent analysis,assume that the players prefer the outcomes in the order listed: Plane: (a1,a2,a3) Row: (a2,a3,a） Column: (a3,a1,az) How,then,should the players select a strategy that ensures the best possible outcome for themselves?If information is complete, that is,if the players are informed about both the preferences of the other players and the decision rule,a sophisticated strategy is optimal for each player,provided that the other players are also sophisticated (Farquharson,1969). A sophisticated strategy requires each player to eliminate successively his dominated strategies.A strategy is dominated when another strategy available to a player produces at least as good a result for him in every contingency and a better result in one or more contingencies.A strategy which dominates all a player's other strategies is called straightforward.A straight- forward strategy is a player's unconditionally best strategy. In the game outlined above,a emerges as the"sophisticated" outcome,as may easily be demonstrated.From Figure I it can be seen that both Plane and Column have straightforward strategies. For Plane,the choice of his strategy "pursue ar"(the first plane) is unconditionally best since it dominates both of his other two strategies,that is,no matter what choices are made by the other players,the outcomes resulting are either the same as or better than the outcomes resulting from the choice of either of his other two strategies,given his preference scale postulated earlier. Similarly,Column's choice of "pursue a"(the third column)is straightforward-it dominates both his first and second strate- gies. In contrast,Row has no unconditionally best strategy.His second strategy dominates his first but not his third.Therefore, Row's choice of a best strategy depends upon the other two players'choices. ThPM

Zagare / GENEVA CONFERENCE 1954 393 Column players results in the outcome ai at the intersection of the first plane, first row, and first column. To illustrate the subsequent analysis, assume that the players prefer the outcomes in the order listed: Plane: (a,, a2, a3) Row: (a2, a3, al) Column: (a3, a,, a2) How, then, should the players select a strategy that ensures the best possible outcome for themselves? If information iscomplete, that is, if the players are informed about both the preferences of the other players and the decision rule, a sophisticated strategy is optimal for each player, provided that the other players are also sophisticated (Farquharson, 1969). A sophisticated strategy requires each player to eliminate successively his dominated strategies. A strategy is dominated when another strategy available to a player produces at least as good a result for him in every contingency and a better result in one or more contingencies. A strategy which dominates all a player's other strategies is called straightforward. A straightforward strategy is a player's unconditionally best strategy. In the game outlined above, a3 emerges as the "sophisticated" outcome, as may easily be demonstrated. From Figure l it can be seen that both Plane and Column have straightforward strategies. For Plane, the choice of his strategy "pursue ai" (the first plane) is unconditionally best since it dominates both of his other two strategies, that is, no matter what choices are made by the other players, the outcomes resulting are either the same as or better than the outcomes resulting from the choice of either of his other two strategies, given his preference scale postulated earlier. Similarly, Column's choice of "pursue a3" (the third column) is straightforward-it dominates both his first and second strategies. In contrast, Row has no unconditionally best strategy. His second strategy dominates his first but not his third. Therefore, Row's choice of a best strategy depends upon the other two players' choices. This content downloaded on Sun, 27 Jan 2013 21:58:56 PM All use subject to JSTOR Terms and Conditions

Zagare GENEVA CONFERENCE 1954 395 It should be pointed out that this result occurs not because the sophisticated outcome in this example is in any sense"socially preferred."In fact,the configuration of preferences of the three players actually creates a paradox of voting situation and makes no alternative socially preferred,i.e.,majorities are cyclical. If information is complete,Plane has no recourse in this strategically unfavorable position.Sophisticated strategies are optimal when information is complete.However,as Brams and Zagare (1977)have shown,if Plane could conceal his true preferences and somehow announce a false preference order which Row and Column believe,two additional strategies become available.First,after this announcement,Plane could act as if this announcement were his true preferences in his play of the game.This type of deceptive strategy is called tacit deception since the other players cannot detect the deception unless they know the user's true preference order.A second option open to Plane also entails making a false announcement but acting in the play of the game consistently with his true preferences.Since other players can easily detect an action that contradicts the deceiver's announced preference order,this strategy is called revealed deception. To illustrate how these deceptive strategies operate,assume that Plane announces his true preference order to be(a2,a1,a3) instead of(a1,az,a3).If Row and Column believe this (false) announcement,they perceive Plane's second strategy(rather than his first)to be straightforward.Since their preferences remain constant,Column continues to have a straightforward strategy (the third column)but Row does not. After eliminating the(apparent)dominated strategies of Plane and the (actual)dominated strategy of Column from considera- tion,as before,Figure I reduces to Figure 3.This figure is remarkably similar to Figure 2 except that now there is a different outcome (az)associated with Row's second strategy ("pursue a”"). Given Row's preference for az over ar and a3,his rational strategy,if he believes Plane's false announcement,is to"pursue ThPM

Zagare / GENEVA CONFERENCE 1954 395 It should be pointed out that this result occurs not because the sophisticated outcome in this example is in any sense "socially preferred." In fact, the configuration of preferences of the three players actually creates a paradox of voting situation and makes no alternative socially preferred, i.e., majorities are cyclical. If information is complete, Plane has no recourse in this strategically unfavorable position. Sophisticated strategies are optimal when information is complete. However, as Brams and Zagare (1977) have shown, if Plane could conceal his true preferences and somehow announce a false preference order which Row and Column believe, two additional strategies become available. First, after this announcement, Plane could act as if this announcement were his true preferences in his play of the game. This type of deceptive strategy is called tacit deception since the other players cannot detect the deception unless they know the users true preference order. A second option open to Plane also entails making a false announcement but acting in the play of the game consistently with his true preferences. Since other players can easily detect an action that contradicts the deceiver's announced preference order, this strategy is called revealed deception. To illustrate how these deceptive strategies operate, assume that Plane announces his true preference order to be (a2, a,, a3) instead of (a,, a2, a3). If Row and Column believe this (false) announcement, they perceive Plane's second strategy (rather than his first) to be straightforward. Since their preferences remain constant, Column continues to have a straightforward strategy (the third column) but Row does not. After eliminating the (apparent) dominated strategies of Plane and the (actual) dominated strategy of Column from consideration, as before, Figure 1 reduces to Figure 3. This figure is remarkably similar to Figure 2 except that now there is a different outcome (a2) associated with Row's second strategy ("pursue a2"). Given Row's preference for a2 over al and a3, his rational strategy, if he believes Plane's false announcement, is to "pursue This content downloaded on Sun, 27 Jan 2013 21:58:56 PM All use subject to JSTOR Terms and Conditions

Zagare GENEVA CONFERENCE 1954 397 to negotiate a settlement with the Vietminh.The pressures came from three directions:the French left,the Soviet Union,and the Vietminh themselves. In France,dissatisfaction with the war was growing,and the lack of domestic support made it difficult for the Laniel govern- ment to continue its policy of seeking a military solution to the conflict in Indochina.In the Soviet Union,where the new collec- tive leadership dominated by Premier Georgi Malenkov was pursuing a policy of detente with the West,calls for a settlement made it difficult for the French to resist negotiations without risking a serious propaganda defeat (Randle,1969:18).Finally, "peace feelers"from the Vietminh intensified the forces that seemed to be compelling the French to negotiate. THE STRUCTURE OF THE GAME Before 1953,the Franco-Vietminh war was essentially a two- person game.Each side received military aid from its allies,but for the most part the United States,Great Britain,the Soviet Union,and China played passive roles.The prospect of a nego- tiated settlement,however,brought about a more active in- volvement by these other powers. As the game progressed,the players began to cluster into three distinct coalitions:the Western Alliance;the Sino-Soviet bloc; and the single-member coalition,the Democratic Republic of Vietnam (the DRV or Vietminh).The nature and composition of each of these coalitions make it possible to consider each one as a unit in the game that followed. The first important cluster of nations,the Western Alliance, was the coalition that the United States led and dominated and included France,Great Britain,and the quasi-independent State of Vietnam (SVN).American domination rested on the limited ability of the other alliance members to influence the outcome of the negotiations.As Randle (1969:126)has noted,"only the United States...was in a position to make the concessions that could permit a settlement at Geneva."The continuing deteriora- tion of the French military position,symbolized by their defeat at ThPM

Zagare / GENEVA CONFERENCE 1954 397 to negotiate a settlement with the Vietminh. The pressures came from three directions: the French left, the Soviet Union, and the Vietminh themselves. In France, dissatisfaction with the war was growing, and the lack of domestic support made it difficult for the Laniel government to continue its policy of seeking a military solution to the conflict in Indochina. In the Soviet Union, where the new collective leadership dominated by Premier Georgi Malenkov was pursuing a policy of detente with the West, calls for a settlement made it difficult for the French to resist negotiations without risking a serious propaganda defeat (Randle, 1969: 18). Finally, "peace feelers" from the Vietminh intensified the forces that seemed to be compelling the French to negotiate. THE STR UCTURE OF THE GA ME Before 1953, the Franco-Vietminh war was essentially atwoperson game. Each side received military aid from its allies, but for the most part the United States, Great Britain, the Soviet Union, and China played passive roles. The prospect of a negotiated settlement, however, brought about a more active involvement by these other powers. As the game progressed, the players began to cluster into three distinct coalitions: the Western Alliance; the Sino-Soviet bloc; and the single-member coalition, the Democratic Republic of Vietnam (the DRV or Vietminh). The nature and composition of each of these coalitions make it possible to consider each one as a unit in the game that followed. The first important cluster of nations, the Western Alliance, was the coalition that the United States led and dominated and included France, Great Britain, and the quasi-independent State of Vietnam (SVN). American domination rested on the limited ability of the other alliance members to influence the outcome of the negotiations. As Randle (1969: 126) has noted, "only the United States ... was in a position to make the concessions that could permit a settlement at Geneva." The continuing deterioration of the French military position, symbolized by their defeat at This content downloaded on Sun, 27 Jan 2013 21:58:56 PM All use subject to JSTOR Terms and Conditions

398 INTERNATIONAL STUDIES QUARTERLY Dien Bien Phu the day before the Indochinese phase of the Geneva Conference opened,and the mounting domestic criticism of the war,limited French influence both within and outside this coalition (Devillers and Lacouture,1969:108).Because the British had no viable military threat,their influence was also circumscribed at Geneva.Finally,since the Western-supported State of Vietnam was internally weak and totally dependent on American aid,it could not play a substantive role in the negotia- tions. Because of the diminished influence of the French,British,and South Vietnamese,the preferences and actions of the United States can,without distorting this analysis,be viewed as being tantamount to those of the group.Although differences in prefer- ences among the members of this coalition sometimes threatened its cohesion,and sometimes even restricted the coalition's maneuverability,the members ironed out their differences before the critical stage of the conference,making their disagreements irrelevant for this analysis (Devillers and Lacouture,1969:268). Thus,in this essay,the Western Alliance will be viewed as an American-dominated coalition in which the roles played by other members were secondary. The second important player in this game was the coalition resulting from a Sino-Soviet Alliance.Because of the coincidence of Soviet and Chinese interests,discussed below,and the Soviet desire to cooperate with the Chinese in order to enhance its power by demonstrating this alliance,the Soviets and the Chinese acted as a bloc at Geneva(Randle,1969:141). The final coalition in this game was the single entity,the DRV. Vietminh control over a large portion of Vietnam gave them a measure of autonomy at Geneva.Moreover,as will be discussed below,their preferences differed from those of the Soviets and Chinese and marked the DRV as an independent player in this game. THE POSSIBLE OUTCOMES Three alternatives faced the participants at the Geneva Con- ference.The first was a stalemate that would result if the players ThPM rms and Conditions

398 INTERNATIONAL STUDIES QUARTERLY Dien Bien Phu the day before the Indochinese phase of the Geneva Conference opened, and the mounting domestic criticism of the war, limited French influence both within and outside this coalition (Devillers and Lacouture, 1969: 108). Because the British had no viable military threat, their influence was also circumscribed at Geneva. Finally, since the Western-supported State of Vietnam was internally weak and totally dependent on American aid, it could not play a substantive role in the negotiations. Because of the diminished influence of the French, British, and South Vietnamese, the preferences and actions of the United States can, without distorting this analysis, be viewed as being tantamounto those of the group. Although differences in preferences among the members of this coalition sometimes threatened its cohesion, and sometimes even restricted the coalition's maneuverability, the members ironed out their differences before the critical stage of the conference, making their disagreements irrelevant for this analysis (Devillers and Lacouture, 1969: 268). Thus, in this essay, the Western Alliance will be viewed as an American-dominated coalition in which the roles played by other members were secondary. The second important player in this game was the coalition resulting from a Sino-Soviet Alliance. Because of the coincidence of Soviet and Chinese interests, discussed below, and the Soviet desire to cooperate with the Chinese in order to enhance its power by demonstrating this alliance, the Soviets and the Chinese acted as a bloc at Geneva (Randle, 1969: 141). The final coalition in this game was the single entity, the DRV. Vietminh control over a large portion of Vietnam gave them a measure of autonomy at Geneva. Moreover, as will be discussed below, their preferences differed from those of the Soviets and Chinese and marked the DRV as an independent player in this game. THE POSSIBLE OUTCOMES Three alternatives faced the participants at the Geneva Conference. The first was a stalemate that would result if the players This content downloaded on Sun, 27 Jan 2013 21:58:56 PM All use subject to JSTOR Terms and Conditions