Conceptual Models and the Cuban Missile Crisis PDF

Conceptual Models and the Cuban Missile Crisis Author(s): Graham T. Allison Source: The American Political Science Review, Vol. 63, No. 3 (Sep., 1969), pp. 689-718 Published by: American Political Science Association Stable URL: http://www.jstor.org/stable/1954423 Accessed: 30/03/2010 09:14 Your use...

Conceptual Models and the Cuban Missile Crisis Author(s): Graham T. Allison Source: The American Political Science Review, Vol. 63, No. 3 (Sep., 1969), pp. 689-718 Published by: American Political Science Association Stable URL: http://www.jstor.org/stable/1954423 Accessed: 30/03/2010 09:14 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=apsa. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. American Political Science Association is collaborating with JSTOR to digitize, preserve and extend access to The American Political Science Review. http://www.jstor.org The Political VOL. LXIII American Science Review SEPTEMBER, 1969 NO. 3 CONCEPTUAL MODELS AND THE CUBAN MISSILE CRISIS* T. ArLLsoN GRAHAM Harvard University The Cuban missile crisis is a seminal event. For thirteen days of October 1962, there was a higher probability that more human lives would end suddenly than ever before in history. Had the worst occurred, the death of 100 million Americans, over 100 million Russians, and millions of Europeans as well would make previous natural calamities and inhumanities appear insignificant. Given the probability of disasterwhich President Kennedy estimated as "between 1 out of 3 and even"-our escape seems awesome.'- This event symbolizes a central, if only partially thinkable, fact about our existence. That such consequences could follow from the choices and actions of national governments obliges students of government as well as participants in governance to think hard about these problems. Improved understanding of this crisis depends in part on more information and more probing analyses of available evidence. To contribute to these efforts is part of the purpose of this study. But here the missile crisis serves primarily as grist for a more general investigation. * A longer version of this paper was presented at the Annual Meeting of the American Political Science Association, September, 1968 (reproduced by the Rand Corporation, P-3919). The paper is part of a larger study, scheduled for publication in 1969 under the title Bureaucracy and Policy: Conceptual Models and the Cuban Missile Crisis. For support in various stages of this work I am indebted to the Institute of Politics in the John F. Kennedy School of Government and the Center for International Affairs, both at Harvard University, the Rand Corporation, and the Council on Foreign Relations. For critical stimulation and advice I am especially grateful to Richard E. Neustadt, Thomas C. Schelling, Andrew W. Marshall, and Elisabeth K. Allison. 'Theodore Sorensen, Kennedy (New York, 1965), p. 705. This study proceeds from the premise that marked improvement in our understanding of such events depends critically on more self-consciousness about what observers bring to the analysis. What each analyst sees and judges to be important is a function not only of the evidence about what happened but also of the "conceptual lenses" through which he looks at the evidence. The principal purpose of this essay is to explore some of the fundamental assumptions and categories employed by analysts in thinking about problems of governmental behavior, especially in foreign and military affairs. The general argument can be summarized in three propositions: 1. Analysts think about problems of foreign and military policy in terms of largely implicit conceptual models that have significant consequences for the content of their thought. Though the present product of foreign policy analysis is neither systematic nor powerful, if one carefully examines explanations produced by analysts, a number of fundamental similarities emerge. Explanations produced by particular analysts display quite regular, predictable features. This predictability suggests a substructure. These regularities reflect an analyst's assumptions about the character of puzzles, the categories in which problems should be considered, the types of evidence that are relevant, and the determinants of occurrences. The first proposition is that clusters of such related assumptions constitute basic frames of reference or conceptual models in terms of which analysts 2In attempting to understand problems of foreign affairs,analysts engage in a number of related, but logically separable enterprises: (a) description, (b) explanation, (c) prediction, (d) evaluation, and (e) recommendation. This essay focuses primarily on explanation (and by implication, prediction). 689 690 THE AMERICAN POLITICAL SCIENCE both ask and answer the questions: What happened? Why did the event happen? What will happen?3 Such assumptions are central to the activities of explanation and prediction, for in attempting to explain a particular event, the analyst cannot simply describe the full state of the world leading up to that event. The logic of explanation requires that he single out the relevant, important determinants of the occurrence.4 Moreover, as the logic of prediction underscores, the analyst must summarize the various determinants as they bear on the event in question. Conceptual models both fix the mesh of the nets that the analyst drags through the material in order to explain a particular action or decision and direct him to cast his net in select ponds, at certain depths, in order to catch the fish he is after. 2. Most analysts explain (and predict) the behavior of national governments in terms of various forms of one basic conceptual model, here entitled the Rational Policy Model (Model I).' In terms of this conceptual model, analysts attempt to understand happenings as the more or less purposive acts of unified national governments. For these analysts, the point of an explanation is to show how the nation or government 2In arguing that explanations proceed in terms of implicit conceptual models, this essay makes no claim that foreign policy analysts have developed any satisfactory, empirically tested theory. In this essay, the use of the term "model" without qualifiers should be read "conceptual scheme." 'For the purpose of this argument we shall accept Carl G. Hempel's characterization of the logic of explanation: an explanation "answers the question, 'Why did the explanadum-phenomenon occur?' by showing that the phenomenon resulted from particular circumstances, specified in Cl, C2, ... Ck, in accordance with laws Lo, L2, . .. Lr. By pointing this out, the argument shows that, given the particular circumstances and the laws in question, the occurrence of the phenomenon was to be expected; and it is in this sense that the explanation enables us to understand why the phenomenon occurred." Aspects of Scientific Explanation (New York, 1965), p. 337. While various patterns of explanation can be distinguished, viz., Ernest Nagel, The Structure of Science: Problems in the Logic of Scientific Explanation, New York, 1961), satisfactory scientific explanations exhibit this basic logic. Consequently prediction is the converse of explanation. Earlier drafts of this argument have aroused heated arguments concerning proper names for these models. To choose names from ordinary language is to court confusion, as well as familiarity. Perhaps it is best to think of these models as I, II, and III. REVIEW VOL. 63 could have chosen the action in question, given the strategic problem that it faced. For example, in confronting the problem posed by the Soviet installation of missiles in Cuba, rational policy model analysts attempt to show how this was a reasonable act from the point of view of the Soviet Union, given Soviet strategic objectives. 3. Two "alternative"conceptual models, here labeled an OrganizationalProcess Model (Model II) and a Bureaucratic Politics Model (Model III) provide a base for improved explanation and prediction. Although the standard frame of reference has proved useful for many purposes, there is powerful evidence that it must be supplemented, if not supplanted, by frames of reference which focus upon the large organizations and political actors involved in the policy process. Model I's implication that important events have important causes, i.e., that monoliths perform large actions for big reasons, must be balanced by an appreciation of the facts (a) that monoliths are black boxes covering various gears and levers in a highly differentiated decision-making structure, and (b) that large acts are the consequences of innumerable and often conflicting smaller actions by individuals at various levels of bureaucratic organizations in the service of a variety of only partially compatible conceptions of national goals, organizational goals, and political objectives. Recent developments in the field of organization theory provide the foundation for the second model. According to this organizational process model, what Model I categorizes as "acts" and "choices" are instead outputs of large organizations functioning according to certain regular patterns of behavior. Faced with the problem of Soviet missiles in Cuba, a Model II analyst identifies the relevant organizations and displays the patterns of organizational behavior from which this action emerged. The third model focuses on the internal politics of a government. Happenings in foreign affairs are understood, according to the bureaucratic politics model, neither as choices nor as outputs. Instead, what happens is categorized as outcomes of various overlapping bargaining games among players arranged hierarchically in the national government. In confronting the problem posed by Soviet missiles in Cuba, a Model III analyst displays the perceptions, motivations, positions, power, and maneuvers of principal players from which the outcome emerged.6 6In strict terms, the "outcomes" which these three models attempt to explain are essentially actions of national governments, i.e., the sum of activities of all individuals employed by a government relevant to an issue. These models focus not on a state of affairs, i.e., a full description of the 1969 CONCEPTUAL MODELS AND THE CUBAN MISSILE CRISIS 691 A central metaphor illuminates differences velopment and support of such a general among these models. Foreign policy has often argument.7 Rather, the sections that follow simbeen compared to moves, sequences of moves, ply sketch each conceptual model, articulate it and games of chess. If one were limited to ob- as an analytic paradigm, and apply it to produce servations on a screen upon which moves in the an explanation. But each model is applied to the chess game were projected without information same event: the U.S. blockade of Cuba during as to how the pieces came to be moved, he would the missile crisis. These "alternative explanaassume-as Model I does-that an individual tions" of the same happening illustrate differchess player was moving the pieces with refer- ences among the models-at work.8 A crisis deence to plans and maneuvers toward the goal of cision, by a small group of men in the context of winning the game. But a pattern of moves can ultimate threat, this is a case of the rational polbe imagined that would lead the serious ob- icy model par excellence. The dimensions and server, after watching several games, to consider factors that Models II and III uncover in this the hypothesis that the chess player was not a case are therefore particularly suggestive. The single individual but rather a loose alliance of concluding section of this paper suggests how semi-independent organizations, each of which the three models may be related and how they moved its set of pieces according to standard can be extended to generate predictions. operating procedures. For example, movement of MODEL I: RATIONAL POLICY separate sets of pieces might proceed in turn, each according to a routine, the king's rook, RATIONAL POLICY MODEL ILLUSTRATED Where is the pinch of the puzzle raised by the bishop, and their pawns repeatedly attacking the opponent according to a fixed plan. Further- New York Times over Soviet deployment of an more, it is conceivable that the pattern of play antiballistic missile system ?9 The question, as the would suggest to an observer that a number of Times states it, concerns the Soviet Union's obdistinct players, with distinct objectives but jective in allocating such large sums of money shared power over the pieces, were determining for this weapon system while at the same time the moves as the resultant of collegial bargain- seeming to pursue a policy of increasing detente. ing. For example, the black rook's move might In former President Johnson's words, "the paracontribute to the loss of a black knight with no dox is that this [Soviet deployment of an anticomparable gain for the black team, but with ballistic missile system] should be happening at a the black rook becoming the principal guardian time when there is abundant evidence that our of the "palace" on that side of the board. mutual antagonism is beginning to ease."10 This The space available does not permit full de- question troubles people primarily because Soviet antiballistic missile deployment, and evidence of world, but upon national decision and implementaSoviet actions towards detente, when juxtaposed tion. This distinction is stated clearly by Harold in our implicit model, produce a question. With and Margaret Sprout, "Environmental Factors on reference to what objective could the Soviet govthe Study of International Politics," in James Rosernment have rationally chosen the simultaneous enau (ed.), International Politics and Foreign pursuit of these two courses of actions? This Policy (Glencoe, Illinois, 1961), p. 116. This requestion arises only when the analyst attempts to striction excludes explanations offered principally structure events as purposive choices of consisin terms of international systems theories. Nevertent actors. theless, this restriction is not severe, since few interesting explanations of occurrences in foreign policy have been produced at that level of analysis. According to David Singer, "The nation state -our primary actor in international relations ... is clearly the traditional focus among Western students and is the one which dominates all of the texts employed in English-speaking colleges and universities." David Singer, "The Level-of-Analysis Problem in International Relations," Klaus Knorr and Sidney Verba (eds.), The International System (Princeton, 1961). Similarly, Richard Brody's review of contemporary trends in the study of international relations finds that "scholars have come increasingly to focus on acts of nations. That is, they all focus on the behavior of nations in some respect. Having an interest in accounting for the behavior of nations in common, the prospects for a common frame of reference are enhanced." 'For further development and support of these arguments see the author's larger study, Bureaucracy and Policy: Conceptual Models and the Cuban Missile Crisis (forthcoming). In its abbreviated form, the argument must, at some points, appear overly stark. The limits of space have forced the omission of many reservations and refinements. 8 Each of the three "case snapshots" displays the work of a conceptual model as it is applied to explain the U.S. blockade of Cuba. But these three cuts are primarily exercises in hypothesis generation rather than hypothesis testing. Especially when separated from the larger study, these accounts may be misleading. The sources for these accounts include the full public record plus a large number of interviews with participants in the crisis. 9New York Times, February 18, 1967. 1OIbid. 692 THE AMERICAN POLITICAL SCIENCE How do analysts attempt to explain the Soviet emplacement of missiles in Cuba? The most widely cited explanation of this occurrence has been produced by two RAND Sovietologists, Arnold Horelick and Myron Rush.1" They conclude that "the introduction of strategic missiles into Cuba was motivated chiefly by the Soviet leaders' desire to overcome . . . the existing large margin of U.S. strategic superiority."12 How do they reach this conclusion? In Sherlock Holmes style, they seize several salient characteristics of this action and use these features as criteria against which to test alternative hypotheses about Soviet objectives. For example, the size of the Soviet deployment, and the simultaneous emplacement of more expensive, more visible intermediate range missiles as well as medium range missiles, it is argued, exclude an explanation of the action in terms of Cuban defensesince that objective could have been secured with a much smaller number of medium range missiles alone. Their explanation presents an argument for one objective that permits interpretation of the details of Soviet behavior as a value-maximizing choice. How do analysts account for the coming of the First World War? According to Hans Morgenthau, "the first World War had its origin exclusively in the fear of a disturbance of the European balance of power.13 In the period preceding World War I, the Triple Alliance precariously balanced the Triple Entente. If either power combination could gain a decisive advantage in the Balkans, it would achieve a decisive advantage in the balance of power. "It was this fear," MIorgenthau asserts, "that motivated Austria in July 1914 to settle its accounts with Serbia once and for all, and that induced Germany to support Austria unconditionally. It was the same fear that brought Russia to the support of Serbia, and France to the support of Russia."'14How is Morgenthau able to resolve this problem so confidently? By imposing on the data a "rational outline."15 The value of this method, according to Morgenthau, is that "it provides for rational discipline in action and creates astounding continuity in foreign policy which makes American, British, or Russian for" Arnold Horelick and Myron Rush, Strategic Power and Soviet Foreign Policy (Chicago, 1965). Based on A. Horelick, "The Cuban Missile Crisis: An Analysis of Soviet Calculations and Behavior," World Politics (April, 1964). 12Horelick and Rush, Strategic Power and Soviet Foreign Policy, p. 154. 13 Hans Morgenthau, Politics Among Nations (3rd ed.; New York, 1960), p. 191. 14Ibid., p. 192. ' Ibid., p. 5. REVIEW VOL. 63 eign policy appear as an intelligent, rational continuum . . . regardless of the different motives, preferences, and intellectual and moral qualities of successive statesmen.'" Stanley Hoffmann's essay, "Restraints and Choices in American Foreign Policy" concentrates, characteristically, on "deep forces": the international system, ideology, and national character-which constitute restraints, limits, and blinders.17 Only secondarily does he consider decisions. But when explaining particular occurrences, though emphasizing relevant constraints, he focuses on the choices of nations. American behavior in Southeast Asia is explained as a reasonable choice of "downgrading this particular alliance (SEATO) in favor of direct U.S. involvement," given the constraint: "one is bound by one's commitments; one is committed by one's mistakes."'8 More frequently, Hoffmann uncovers confusion or contradiction in the nation's choice. For example, U.S. policy towards underdeveloped countries is explained as "schizophrenic.'9 The method employed by Hoffmann in producing these explanations as rational (or irrational) decisions, he terms "imaginative reconstruction." Deterrence is the cardinal problem of the contemporary strategic literature. Thomas Schelling's Strategy of Conflict formulates a number of propositions focused upon the dynamics of deterrence in the nuclear age. One of the major propositions concerns the stability of the balance of terror: in a situation of mutual deterrence, the probability of nuclear war is reduced not by the "balance" (the sheer equality of the situation) but rather by the stability of the balance, i.e., the fact that neither opponent in striking first can destroy the other's ability to strike back.21 How does Schelling support this proposition? Confidence in the contention stems not from an inductive canvass of a large number of previous cases, but rather from two calculations. In a situation of "balance" but vulnerability, there are values for which a rational opponent could choose to strike first, e.g., to destroy enemy capabilities to retaliate. In a 6Ibid., pp. 5-6. Hoffmann, Daedalus (Fall, 1962); reprinted in The State of War (New York, 1965). '8 Ibid., p. 171. 19Ibid., p. 189. Following Robert Maclver; see Stanley Hoffmann, Contemporary Theory in International Relations (Englewood Cliffs, 1960), pp. 178-179. 2 Thomas Schelling, The Strategy of Conflict, (New York, 1960), p. 232. This proposition was formulated earlier by A. Wohlstetter, "The Deli17Stanley cate Balance of Terror,"Foreign Affairs (January, 1959). 1969 CONCEPTUAL MODELS AND THE "stable balance" where no matter who strikes first, each has an assured capability to retaliate with unacceptable damage, no rational agent could choose such a course of action (since that choice is effectively equivalent to choosing mutual homicide). Whereas most contemporary strategic thinking is driven implicitly by the motor upon which this calculation depends, Schelling explicitly recognizes that strategic theory does assume a model. The foundation of a theory of strategy is, he asserts: "the assumption of rational behavior-not just of intelligent behavior, but of behavior motivated by conscious calculation of advantages, calculation that in turn is based on an explicit and internally consistent value system."22 What is striking about these examples from the literature of foreign policy and international relations are the similarities among analysts of various styles when they are called upon to produce explanations. Each assumes that what must be explained is an action, i.e., the realization of some purpose or intention. Each assumes that the actor is the national government. Each assumes that the action is chosen as a calculated response to a strategic problem. For each, explanation consists of showing what goal the government was pursuing in committing the act and how this action was a reasonable choice, given the nation's objectives. This set of assumptions characterizes the rational policy model. The assertion that Model I is the standard frame of reference implies no denial of highly visible differences among the interests of Sovietologists, diplomatic historians, international relations theorists, and strategists. Indeed, in most respects, differences among the work of Hans Morgenthau, Stanley Hoffmann, and Thomas Shelling could not be more pointed. Appreciation of the extent to which each relies predominantly on Model I, however, reveals basic similarities among Morgenthau's method of "rational reenactment," Hoffmann's "imaginative reconstruction," and Schelling's "vicarious problem solving; " family resemblances among Morgenthau's "rational statesman," Hoffmann's "roulette player," and Schelling's "game theorist."23 Most contemporary analysts (as well as laymen) proceed predominantly-albeit most often implicitly-in terms of this model when attempting to explain happenings in foreign affairs. Indeed, that occurrences in foreign affairs are the acts of nations seems so fundamental to think29 2' Schelling, op. cit., p. 4. See Morgenthau, op. cit., p. 5; Hoffmann, Con- temporary Theory, pp. 178-179; Hoffmann, "Roulette in the Cellar," The State of War; Schelling, op. cit. CUBAN MISSILE CRISIS 693 ing about such problems that this underlying model has rarely been recognized: to explain an occurrence in foreign policy simply means to show how the government could have rationally chosen that action.24 These brief examples illustrate five uses of the model. To prove that most analysts think largely in terms of the rational policy model is not possible. In this limited space it is not even possible to illustrate the range of employment of the framework. Rather, my purpose is to convey to the reader a grasp of the model and a challenge: let the reader examine the literature with which he is most familiar and make his judgment. The general characterization can be sharpened by articulating the rational policy model as an "analytic paradigm" in the technical sense developed by Robert K. Merton for sociological analyses.25 Systematic statement of basic assumptions, concepts, and propositions employed by Model I analysts highlights the distinctive thrust of this style of analysis. To articulate a largely implicit framework is of necessity to caricature. But caricature can be instructive. RATIONAL POLICY PARADIGM I. Basic Unit of Analysis: Policy as National Choice Happenings in foreign affairs are conceived as actions chosen by the nation or national ' The larger study examines several exceptions to this generalization. Sidney Verba's excellent essay "Assumptions of Rationality and Non-Rationality in Models of the International System" is less an exception than it is an approachto a somewhat different problem. Verba focuses upon models of rationality and irrationality of individual statesmen: in Knorr and Verba, The International System. ' Robert K. Merton, Social Theory and Social Structures (Revised and Enlarged Edition; New York, 1957), pp. 12-16. Considerably weaker than a satisfactory theoretical model, paradigms nevertheless represent a short step in that direction from looser, implicit conceptual models. Neither the concepts nor the relations among the variables are sufficiently specified to yield propositions deductively. "Paradigmatic Analysis" nevertheless has considerablepromise for clarifying and codifying styles of analysis in political science. Each of the paradigms stated here can be represented rigorously in mathematical terms. For example, Model I lends itself to mathematical formulation along the lines of Herbert Simon's "Behavioral Theory of Rationality," Models of Man (New York, 1957). But this does not solve the most difficult problem of "measurementand estimation." 694 THE AMERICAN POLITICAL government.26 Governments select the action that will maximize strategic goals and objectives. These "solutions" to strategic problems are the fundamental categories in terms of which the analyst perceives what is to be explained. I. Organizing Concepts A. National Actor. The nation or government, conceived as a rational, unitary decisionmaker, is the agent. This actor has one set of specified goals (the equivalent of a consistent utility function), one set of perceived options, and a single estimate of the consequences that follow from each alternative. B. The Problem. Action is chosen in response to the strategic problem which the nation faces. Threats and opportunities arising in the "international strategic market place" move the nation to act. C. Static Selection. The sum of activity of representatives of the government relevant to a problem constitutes what the nation has chosen as its "solution." Thus the action is conceived as a steady-state choice among alternative outcomes (rather than, for example, a large number of partial choices in a dynamic stream). D. Action as Rational Choice. The components include: 1. Goals and Objectives. National security and national interests are the principal categories in which strategic goals are conceived. Nations seek security and a range of further objectives. (Analysts rarely translate strategic goals and objectives into an explicit utility function; nevertheless, analysts do focus on major goals and objectives and trade off side effects in an intuitive fashion.) 2. Options. Various courses of action relevant to a strategic problem provide the spectrum of options. 3. Consequences. Enactment of each alternative course of action will produce a series of "6Though a variant of this model could easily be stochastic, this paradigm is stated in non-probabilistic terms. In contemporary strategy, a stochastic version of this model is sometimes used for predictions; but it is almost impossible to find an explanation of an occurrence in foreign affairs that is consistently probabilistic. Analogies between Model I and the concept of explanation developed by R. G. Collingwood, William Dray, and other "revisionists" among philosophers concerned with the critical philosophy of history are not accidental. For a summary of the "revisionist position" see Maurice Mandelbaum, "Historical Explanation: The Problem of Covering Laws," History and Theory (1960). SCIENCE REVIEW VOL. 63 consequences. The relevant consequences constitute benefits and costs in terms of strategic goals and objectives. 4. Choice. Rational choice is value-maximizing. The rational agent selects the alternative whose consequences rank highest in terms of his goals and objectives. III. Dominant Inference Pattern This paradigm leads analysts to rely on the following pattern of inference: if a nation performed a particular action, that nation must have had ends towards which the action constituted an optimal means. The rational policy model's explanatory power stems from this inference pattern. Puzzlement is relieved by revealing the purposive pattern within which the occurrence can be located as a value-maximizing means. IV. General Propositions The disgrace of political science is the infrequency with which propositions of any generality are formulated and tested. "Paradigmatic analysis" argues for explicitness about the terms in which analysis proceeds, and seriousness about the logic of explanation. Simply to illustrate the kind of propositions on which analysts who employ this model rely, the formulation includes several. The basic assumption of value-maximizing behavior produces propositions central to most explanations. The general principle can be formulated as follows: the likelihood of any particular action results from a combination of the nation's (1) relevant values and objectives, (2) perceived alternative courses of action, (3) estimates of various sets of consequences (which will follow from each alternative), and (4) net valuation of each set of consequences. This yields two propositions. A. An increase in the cost of an alternative, i.e., a reduction in the value of the set of consequences which will follow from that action, or a reduction in the probability of attaining fixed consequences, reduces the likelihood of that alternative being chosen. B. A decrease in the costs of an alternative, i.e., an increase in the value of the set of consequences which will follow from that alternative, or an increase in the probability of attaining fixed consequences, increases the likelihood of that action being chosen.27 2 This model is an analogue of the theory of the rational entrepreneur which has been developed extensively in economic theories of the firm and the consumer. These two propositions specify the "substitution effect." Refinement of this model and 1969 CONCEPTUAL MODELS AND THE V. Specific Propositions A. Deterrence. The likelihood of any particular attack results from the factors specified in the general proposition. Combined with factual assertions, this general proposition yields the propositions of the sub-theory of deterrence. (1) A stable nuclear balance reduces the likelihood of nuclear attack. This proposition is derived from the general proposition plus the asserted fact that a second-strike capability affects the potential attacker's calculations by increasing the likelihood and the costs of one particular set of consequences which might follow from attack-namely, retaliation. (2) A stable nuclear balance increases the probability of limited war. This proposition is derived from the general proposition plus the asserted fact that though increasing the costs of a nuclear exchange, a stable nuclear balance nevertheless produces a more significant reduction in the probability that such consequences would be chosen in response to a limited war. Thus this set of consequences weighs less heavily in the calculus. B. Soviet Force Posture. The Soviet Union chooses its force posture (i.e., its weapons and their deployment) as a value-maximizing means of implementing Soviet strategic objectives and military doctrine. A proposition of this sort underlies Secretary of Defense Laird's inference from the fact of 200 SS-9s (large intercontinental missiles) to the assertion that, "the Soviets are going for a first-strike capability, and there's no question about it."28 VARIANTS OF THE RATIONAL POLICY MODEL This paradigm exhibits the characteristics of the most refined version of the rational model. The modern literature of strategy employs a model of this sort. Problems and pressures in the "international strategic marketplace" yield probabilities of occurrence. The international actor, which could be any national actor, is simply a value-maximizing mechanism for getting from the strategic problem to the logical solution. But the explanations and predictions produced by most analysts of foreign affairs depend primarily on variants of this "pure" model. The point of each is the same: to place the action within a value-maximizing framework, given certain constraints. Nevertheless, it may be helpful to identify several variants, each of which might be exhibited similarly as a paradigm. The first focuses upon the national actor specification of additional general propositions by translating from the economic theory is straightforward. 2 New York Times, March 22, 1969. CUBAN MISSILE CRISIS 695 and his choice in a particular situation, leading analysts to further constrain the goals, alternatives, and consequences considered. Thus, (1) national propensities or personality traits reflected in an "operational code," (2) concern with certain objectives, or (3) special principles of action, narrow the "goals" or "alternatives" or "consequences" of the paradigm. For example, the Soviet deployment of ABMs is sometimes explained by reference to the Soviet's "defense-mindedness." Or a particular Soviet action is explained as an instance of a special rule of action in the Bolshevik operational code.29 A second, related, cluster of variants focuses on the individual leader or leadership group as the actor whose preference function is maximized and whose personal (or group) characteristics are allowed to modify the alternatives, consequences, and rules of choice. Explanations of the U.S. involvement in Vietnam as a natural consequence of the Kennedy-Johnson Administration's axioms of foreign policy rely on this variant. A third, more complex variant of the basic model recognizes the existence of several actors within a government, for example, Hawks and Doves or military and civilians, but attempts to explain (or predict) an occurrence by reference to the objectives of the victorious actor. Thus, for example, some revisionist histories of the Cold War recognize the forces of light and the forces of darkness within the U.S. government, but explain American actions as a result of goals and perceptions of the victorious forces of darkness. Each of these forms of the basic paradigm constitutes a formalization of what analysts typically rely upon implicitly. In the transition from implicit conceptual model to explicit paradigm much of the richness of the best employments of this model has been lost. But the purpose in raising loose, implicit conceptual models to an explicit level is to reveal the basic logic of analysts' activity. Perhaps some of the remaining artificiality that surrounds the statement of the paradigm can be erased by noting a number of the standard additions and modifications employed by analysts who proceed predominantly within the rational policy model. First, in the course of a document, analysts shift from one variant of the basic model to another, occasionally appropriating in an ad hoc fashion aspects of a situation which are logically incompatible with the basic model. Second, in the course of explaining a number of occurrences, analysts sometimes pause over a particular event about which they have a great deal of information and unfold it in such detail that an impression of 29 See Nathan Leites, A Study of Bolshevism (Glencoe, Illinois, 1953). 696 THE AMERICAN POLITICAL SCIENCE randomness is created. Third, having employed other assumptions and categories in deriving an explanation or prediction, analysts will present their product in a neat, convincing rational policy model package. (This accommodation is a favorite of members of the intelligence community whose association with the details of a process is considerable, but who feel that by putting an occurrence in a larger rational framework, it will be more comprehensible to their audience.) Fourth, in attempting to offer an explanationparticularly in cases where a prediction derived from the basic model has failed-the notion of a "mistake" is invoked. Thus, the failure in the prediction of a "missile gap" is written off as a Soviet mistake in not taking advantage of their opportunity. Both these and other modifications permit Model I analysts considerably more variety than the paradigm might suggest. But such accommodations are essentially appendages to the basic logic of these analyses. THE U.S. BLOCKADE OF CUBA: A FIRST CUT30 The U.S. response to the Soviet Union's emplacement of missiles in Cuba must be understood in strategic terms as simple value-maximizing escalation. American nuclear superiority could be counted on to paralyze Soviet nuclear power; Soviet transgression of the nuclear threshold in response to an American use of lower levels of violence would be wildly irrational since it would mean virtual destruction of the Soviet Communist system and Russian nation. American local superiority was overwhelming: it could be initiated at a low level while threatening with high credibility an ascending sequence of steps short of the nuclear threshold. All that was required was for the United States to bring to bear its strategic and local superiority in such a way that American determination to see the missiles removed would be demonstrated, while at the same time allowing Moscow time and room to retreat without humiliation. The naval blackade-euphemistically named a "quarantine" in order to circumvent the niceties of international law-did just that. The U.S. government's selection of the blockade followed this logic. Apprised of the presence of Soviet missiles in Cuba, the President assembled an Executive Committee (ExCom) of the " As stated in the introduction, this "case snapshot" presents, without editorial commentary, a Model I analyst's explanation of the U.S. blockade. The purpose is to illustrate a strong, characteristic rational policy model account. This account is (roughly) consistent with prevailing explanations of these events. REVIEW VOL. 63 National Security Council and directed them to "set aside all other tasks to make a prompt and intense survey of the dangers and all possible courses of action."'3' This group functioned as "fifteen individuals on our own, representing the President and not different departments."32 As one of the participants recalls, "The remarkable aspect of those meetings was a sense of complete equality."33 Most of the time during the week that followed was spent canvassing all the possible tracks and weighing the arguments for and against each. Six major categories of action were considered. 1. Do nothing. U.S. vulnerability to Soviet missiles was no new thing. Since the U.S. already lived under the gun of missiles based in Russia, a Soviet capability to strike from Cuba too made little real difference. The real danger stemmed from the possibility of U.S. over-reaction. The U.S. should announce the Soviet action in a calm, casual manner thereby deflating whatever political capital Khrushchev hoped to make of the missiles. This argument fails on two counts. First, it grossly underestimates the military importance of the Soviet move. Not only would the Soviet Union's missile capability be doubled and the U.S. early warning system outflanked. The Soviet Union would have an opportunity to reverse the strategic balance by further installations, and indeed, in the longer run, to invest in cheaper, shorter-range rather than more expensive longer-range missiles. Second, the political importance of this move was undeniable. The Soviet Union's act challenged the American President's most solemn warning. If the U.S. failed to respond, no American commitment would be credible. 2. Diplomatic pressures. Several forms were considered: an appeal to the U.N. or O.A.S. for an inspection team, a secret approach to Khrushchev, and a direct approach to Khrushchev, perhaps at a summit meeting. The United States would demand that the missiles be removed, but the final settlement might include neutralization of Cuba, U.S. withdrawal from the Guantanamo base, and withdrawal of U.S. Jupiter missiles from Turkey or Italy. Each form of the diplomatic approach had its own drawbacks. To arraign the Soviet Union before the U.N. Security Council held little promise since the Russians could veto any proposed action. While the diplomats argued, the missiles would become operational. To send a secret emissary to Khrushchev demanding that 31Theodore Sorensen, op. cit., p. 675. 32 Ibid., p. 679. a Ibid., p. 679. 1969 CONCEPTUAL MODELS AND THE the missiles be withdrawn would be to pose untenable alternatives. On the one hand, this would invite Khrushchev to seize the diplomatic initiative, perhaps committing himself to strategic retaliation in response to an attack on Cuba. On the other hand, this would tender an ultimatum that no great power could accept. To confront Khrushchev at a summit would guarantee demands for U.S. concessions, and the analogy between U.S. missiles in Turkey and Russian missiles in Cuba could not be erased. But why not trade U.S. Jupiters in Turkey and Italy, which the President had previously ordered withdrawn, for the missiles in Cuba? The U.S. had chosen to withdraw these missiles in order to replace them with superior, less vulnerable Mediterranean Polaris submarines. But the middle of the crisis was no time for concessions. The offer of such a deal might suggest to the Soviets that the West would yield and thus tempt them to demand more. It would certainly confirm European suspicions about American willingness to sacrifice European interests when the chips were down. Finally, the basic issue should be kept clear. As the President stated in reply to Bertrand Russell, "I think your attention might well be directed to the burglars rather than to those who have caught the burglars."34 3. A secret approach to Castro. The crisis provided an opportunity to separate Cuba and Soviet Communism by offering Castro the alternatives, "split or fall." But Soviet troops transported, constructed, guarded, and controlled the missiles. Their removal would thus depend on a Soviet decision. 4. Invasion. The United States could take this occasion not only to remove the missiles but also to rid itself of Castro. A Navy exercise had long been scheduled in which Marines, ferried from Florida in naval vessels, would liberate the imaginary island of Vieques.35 Why not simply shift the point of disembarkment? (The Pentagon's foresight in planning this operation would be an appropriate antidote to the CIA's Bay of Pigs!) Preparations were made for an invasion, but as a last resort. American troops would be forced to confront 20,000 Soviets in the first Cold War case of direct contact between the troops of the super powers. Such brinksmanship courted nuclear disaster, practically guaranteeing an equivalent Soviet move against Berlin. 5. Surgical air strike. The missile sites should 4Elie Abel, The Missile 1966), p. 144. 35Ibid., p. 102. Crisis (New York, CUBAN MISSILE CRISIS 697 be removed by a clean, swift conventionalattack. This was the effective counter-action which the attempted deceptiondeserved.A surgical strike would remove the missiles and thus eliminate both the danger that the missiles might become operationaland the fear that the Soviets would discover the Americandiscovery and act first. The initial attractiveness of this alternative was dulled by several difficulties.First, could the strike really be "surgical"?The Air Force could not guarantee destruction of all the missiles.38Some might be fired during the attack; some might not have been identified.In orderto assure destructionof Soviet and Cuban means of retaliating,what was requiredwas not a surgical but rather a massive attack-of at least 500 sorties. Second, a surprise air attack would of coursekill Russiansat the missilesites. Pressureson the Soviet Union to retaliatewould be so strong that an attack on Berlin or Turkey was highly probable. Third, the key problem with this programwas that of advancewarning. Could the President of the United States, with his memory of Pearl Harbor and his vision of future U.S. responsibility,ordera "PearlHarbor in reverse"?For 175 years, unannouncedSunday morningattacks had been an anathemato our tradition.37 6. Blockade. Indirect military action in the form of a blockadebecame more attractive as the ExCom dissected the other alternatives. An embargoon military shipmentsto Cuba enforced by a naval blockade was not without flaws, however. Could the U.S. blockade Cuba without inviting Soviet reprisal in Berlin? The likely solution to joint blockadeswould be the lifting of both blockades,restoringthe new status quo, and allowing the Soviets additional time to completethe missiles.Second,the possible consequencesof the blockade resembledthe drawbackswhich disqualifiedthe air strike. If Soviet ships did not stop, the United States wouldbe forcedto fire the first shot, inviting retaliation.Third,a blockadewoulddeny the traditional freedomof the seas demandedby several of our close allies and might be held illegal, in violation of the U.N. Charterand international law, unless the United States could obtain a two-thirds vote in the O.A.S. Finally, how ' Sorensen, op. cit., p. 684. Ibid., p. 685. Though this was the formulation of the argument, the facts are not strictly accurate. Our tradition against surprise attack was rather younger than 175 years. For example President Theodore Roosevelt applauded Japan's attack on Russia in 1904. 8 698 THE AMERICAN POLITICAL SCIENCE could a blockade be related to the problem, namely, some 75 missiles on the island of Cuba, approaching operational readiness daily? A blockade offered the Soviets a spectrum of delaying tactics with which to buy time to complete the missile installations. Was a fait accompli not required? In spite of these enormous difficulties the blockade had comparative advantages: (1) It was a middle course between inaction and attack, aggressive enough to communicate firmness of intention, but nevertheless not so precipitous as a strike. (2) It placed on Khrushchev the burden of choice concerning the next step. He could avoid a direct military clash by keeping his ships away. His was the last clear chance. (3) No possible military confrontation could be more acceptable to the U.S. than a naval engagement in the Caribbean. (4) This move permitted the U.S., by flexing its- conventional muscle, to exploit the threat of subsequent non-nuclear steps in each of which the U.S. would have significant superiority. Particular arguments about advantages and disadvantages were powerful. The explanation of the American choice of the blockade lies in a more general principle, however. As President Kennedy stated in drawing the moral of the crisis: Above all, while defending our own vital interests, nuclear powers must avert those confrontations which bring an adversaryto a choice of either a humiliating retreat or a nuclear war. To adopt that kind of course in the nuclear age would be evidence only of the bankruptcy of our policy-of a collective death wish for the world." The blockade was the United States' only real option. MODEL II: ORGANIZATIONAL PROCESS For some purposes, governmental behavior can be usefully summarized as action chosen -by a unitary, rational decisionmaker: centrally controlled, completely informed, and value maximizing. But this simplification must not be allowed to conceal the fact that a "government" consists of a conglomerate of semi-feudal, loosely allied organizations, each with a substantial life of its own. Government leaders do sit formally, and to some extent in fact, on top of this conglomerate. But governments perceive problems through organizational sensors. Governments define alternatives and estimate consequences as organizations process information. Governments act as these organizations enact routines. Government behavior can therefore be 3 New York Times. June, 1963. REVIEW VOL. 63 understood according to a second conceptual model, less as deliberate choices of leaders and more as outputs of large organizations functioning according to standard patterns of behavior. To be responsive to a broad spectrum of problems, governments consist of large organizations among which primary responsibility for particular areas is divided. Each organization attends to a special set of problems and acts in quasi-independence on these problems. But few important problems fall exclusively within the domain of a single organization. Thus government behavior relevant to any important problem reflects the independent output of several organizations, partially coordinated by government leaders. Government leaders can substantially disturb, but not substantially control, the behavior of these organizations. To perform complex routines, the behavior of large numbers of individuals must be coordinated. Coordination requires standard operating procedures: rules according to which things are done. Assured capability for reliable performance of action that depends upon the behavior of hundreds of persons requires established "programs." Indeed, if the eleven members of a football team are to perform adequately on any particular down, each player must not "do what he thinks needs to be done" or "do what the quarterback tells him to do." Rather, each player must perform the maneuvers specified by a previously established play which the quarterback has simply called in this situation. At any given time, a government consists of existing organizations, each with a fixed set of standard operating procedures and programs. The behavior of these organizations-and consequently of the government-relevant to an issue in any particular instance is, therefore, determined primarily by routines established in these organizations prior to that instance. But organizations do change. Learning occurs gradually, over time. Dramatic organizational change occurs in response to major crises. Both learning and change are influenced by existing organizational capabilities. Borrowed from studies of organizations, these loosely formulated propositions amount simply to tendencies. Each must be hedged by modifiers like "other things being equal" and "under certain conditions." In particular instances, tendencies hold-more or less. In specific situations, the relevant question is: more or less? But this is as it should be. For, on the one hand, "organizations" are no more homogeneous a class than "solids." When scientists tried to generalize about "solids,"' they achieved similar results. Solids tend to expand when heated, but some do and some don't. More adequate categorization 1969 CONCEPTUAL MODELS AND THE of the various elements now lumped under the rubric "organizations" is thus required. On the other hand, the behavior of particular organizations seems considerably more complex than the behavior of solids. Additional information about a particular organization is required for further specification of the tendency statements. In spite of these two caveats, the characterization of government action as organizational output differs distinctly from Model I. Attempts to understand problems of foreign affairs in terms of this frame of reference should produce quite different explanations.39 ORGANIZATIONAL PROCESS PARADIGM40 I. Basic Unit of Analysis: Policy as Organizational Output The happenings of international politics are, in three critical senses, outputs of organizational processes. First, the actual occurrences are organizational outputs. For example, Chinese entry into the Korean War-that is, the fact that Chinese soldiers were firing at U.N. soldiers south of the Yalu in 1950-is an organizational action: the action of men who are soldiers in platoons which are in companies, which in turn are in armies, responding as privates to lieutenants who are responsible to captains and so on 'The influence of organizational studies upon the present literature of foreign affairs is minimal. Specialists in international politics are not students of organization theory. Organization theory has only recently begun to study organizations as decisionmakers and has not yet produced behavioral studies of national security organizations from a decision-making perspective. It seems unlikely, however, that these gaps will remain unfilled much longer. Considerable progress has been made in the study of the business firm as an organization. Scholars have begun applying these insights to government organizations, and interest in an organizational perspective is spreading among institutions and individuals concerned with actual government operations. The "decisionmaking" approach represented by Richard Snyder, R. Bruck, and B. Sapin, Foreign Policy Decision-Makinq (Glencoe, Illinois, 1962), incorporates a number of insights from organization theory. " The formulation of this paradigm is indebted both to the orientation and insights of Herbert Simon and to the behavioral model of the firm stated by Richard Cyert and James March, A Behavioral Theory of the Firm (Englewood Cliffs, 1963). Here, however, one is forced to grapple with the less routine, less quantified functions of the less differentiated elements in government organizations. CUBAN MISSILE CRISIS 699 to the commander, moving into Korea, advancing against enemy troops, and firing according to fixed routines of the Chinese Army. Government leaders' decisions trigger organizational routines. Government leaders can trim the edges of this output and exercise some choice in combining outputs. But the mass of behavior is determined by previously established procedures. Second, existing organizational routines for employing present physical capabilities constitute the effective options open to government leaders confronted with any problem. Only the existence of men, equipped and trained as armies and capable of being transported to North Korea, made entry into the Korean War a live option for the Chinese leaders. The fact that fixed programs (equipment, men, and routines which exist at the particular time) exhaust the range of buttons that leaders can push is not always perceived by these leaders. But in every case it is critical for an understanding of what is actually done. Third, organizational outputs structure the situation within the narrow constraints of which leaders must contribute their "decision" concerning an issue. Outputs raise the problem, provide the information, and make the initial moves that color the face of the issue that is turned to the leaders. As Theodore Sorensen has observed: "Presidents rarely, if ever, make decisions-particularly in foreign affairs -in the sense of writing their conclusions on a clean slate . . . The basic decisions, which confine their choices, have all too often been previously made.""' If one understands the structure of the situation and the face of the issue which are determined by the organizational outputs-the formal choice of the leaders is frequently anti-climactic. II. Organizing Concepts A. Organizational Actors. The actor is not a monolithic "nation" or "government" but rather a constellation of loosely allied organizations on top of which government leaders sit. This constellation acts only as component organizations perform routines.42 B. Factored Problems and Fractionated Power. Surveillance of the multiple facets of for4' Theodore Sorensen, "You Get to Walk to Work-," New York Times Magazine, March 19. 1967. 42 Organizations are not monolithic. The proper level of disaggregation depends upon the objectives of a piece of analysis. This paradigm is formulated with reference to the major organizations that constitute the U.S. government. Generalization to the major components of each department and agency should be relatively straightforward. 700 THE AMERICAN POLITICAL SCIENCE eign affairs requires that problems be cut up and parcelledout to various organizations.To avoid paralysis,primarypowermust accompany primaryresponsibility.But if organizationsare permittedto do anything,a large part of what they do will be determinedwithin the organization. Thus each organizationperceivesproblems, processesinformation,and performsa range of actions in quasi-independence(within broad guidelines of national policy). Factored problems and fractionatedpower are two edges of the same sword.Factoringpermitsmore specialized attention to particularfacets of problems than would be possible if government leaders tried to cope with these problemsby themselves. But this additionalattentionmust be paid for in the coin of discretionfor what an organization attends to, and how organizationalresponsesare programmed. C. Parochial Priorities, Perceptions, and Issues. Primary responsibilityfor a narrowset of problems encourages organizationalparochialism. These tendenciesare enhancedby a number of additionalfactors: (1) selective information availableto the organization,(2) recruitmentof personnelinto the organization,(3) tenure of individualsin the organization,(4) small group pressureswithin the organization,and (5) distributionof rewardsby the organization.Clients (e.g., interest groups), governmentallies (e.g., Congressionalcommittees), and extra-national counterparts(e.g., the British Ministry of Defense for the Department of Defense, ISA, or the British Foreign Office for the Department of State, EUR) galvanize this parochialism. Thus organizationsdeveloprelativelystable propensities concerningoperationalpriorities,perceptions,and issues. D. Action as OrganizationalOutput.The preeminent feature of organizationalactivity is its programmedcharacter:the extent to which behaviorin any particularcase is an enactment of preestablishedroutines.In producingoutputs, the activity of each organizationis characterized by: 1. Goals: Constraints Defining Acceptable Performance.The operationalgoals of an organization are seldom revealed by formal mandates. Rather, each organization'soperational goals emergeas a set of constrai