Ermarth, Fritz W. Book reviews of Essence of Decision: Explaining the Cuban Missile Crisis by Graham T. Allison and Victims of Group Think by Irving L. Janus
CIA HISTORICAL REVIEW PROGRAM
>RELEASE IN FULL
2 JULY 96
INTELLIGENCE IN RECENT PUBLIC LITERATURE
ESSENCE OF DECISION: EXPLAINING THE CUBAN MISSILE CRISIS. By Graham T. Allison. (Little, Brown, and Company, Boston, 1971.)
VICTIMS OF GROUPTHINK. By Irving L. Janis. (Houghton, Mifflin Company, Boston, 1972.)
Both of these books are about "decision making," a trend of contemporary social and political science that, in Stanley Hoffman's acerbic regard, involves digging around in other people's waste baskets to see who did or said what to whom. Both treat, from different vantage points, the problem of how governments make decisions or shape their behavior on the gravest matters of national security. One is exclusively on the Cuban missile crisis; the other devotes much attention to it. Both have vital lessons for intelligence. Allison's book is about how to understand the decisions and actions of governments; its main aim is to present tools that can improve that understanding. Allison thus speaks directly to the tasks of analysis in most of the intelligence community's production components. Janis, although in search of understanding and better analysis, is really after improvements in policy decision making itself; but his insights into policy decision making have a useful bearing on what we may call intelligence decision making.
Reviewing Allison's book at this time presents the reviewer with a dilemma. Those who have not come into contact with it, now some three years after its publication, must either be little interested in its subject or have no time for reading. They would best be served by a fairly extensive survey of its contents. Those in the intelligence community most directly interested in its message, however, have in all likelihood read it carefully and have already been influenced professionally by it. What they need would be more in the nature of a status report. Neither task can be adequately met in a short review. The dilemma is sharpened by the fact that this reviewer has a strong stake in the popularity of the Allisonian view but remains at heart somewhat ambivalent as to its value.
The Essence of Allison lies in three approaches to understanding government behavior, three conceptual models, as he calls them. The models are composed of the assumptions we use, the questions we ask, the information we seek, the vocabulary we employ; and they shape the answers we get.
Model I is the Rational Actor. It states or, more correctly, implicitly assumes that governments are akin to rational individuals who have values (or cost/benefit calculations), purposes, and an instrumental command of tactics. They establish aims, gather and assess information, weigh risks, then choose and implement a plan of action as an exceedingly sensible man would buy a car or play a hand of poker. If the Rational Actor fails or gets in trouble, it is because he lacked the necessary information, miscalculated, or was lacking in rationality. The noun-verb combinations of this Model are straightforward and familiar: "The USSR seeks ...", "Moscow has apparently decided ...", "The Politburo believes ...", "The Russians are now going to ..." The subject may be plural; but the notion is singular and the action conscious and purposeful. Most important, the all-pervasive assumption is that of a fully reasoned correlation of ends and means, and complete self-control on the part of the actor, the government in question.
Allison's purpose is to challenge the Rational Actor Model "on its home ground" — that of deep international crisis where reason and self-control are at a premium — and trim it down to size as a tool for understanding government behavior. He starts essentially from the realization of any attentive newspaper reader that governments are not really Rational Actors. A government is an assortment of disparate institutions, each with its own preoccupations and habits. Further, a government is an arena in which groups and individuals compete for power and influence. These characteristics of government are as important in shaping government behavior as are rational calculation and purpose, perhaps more so. From them he derives two alternative models to complement the perspective of the Rational Actor.
Model II is called the Organizational Process Model. It is concerned with the role of standard operating procedure of governmental entities in the aggregate behavior of the government they make up. Any member of any organization can understand the power of Model 11. Many actions of the organization take place, not because they are sensible or some powerful influence wants it that way, but because that is just the way things are done. Large organizations have to have standard operating procedures to handle important and complex matters or they will lapse into complete paralysis. Moreover, government organizations are created to handle enduring, repetitive missions; they cannot develop new strategies or operational repertoires from scratch in each new instance. Thus, they are usually called into action to do something more or less as they've always done it, and you get the standard operating procedure with minor variations.
Model III is also quite congenial to a layman's view of reality; it is the Governmental or Bureaucratic Politics Model. When you put people into an organization, or little organizations into bigger organizations, you have Politics. People and organizations — "players" in Allisonian terms — compete for status or influence, or perhaps to avoid influence and the risks that go with it. This means struggle, factionalism, even duplicity on the part of the players, the antithesis of what the Rational Actor is supposed to stand for.
The backbone of Allison's book is a series of chapters in which he first introduces the logical or theoretical machinery of his three models along with a précis of their academic antecedents, and then methodically applies them to the history of the Cuban missile crisis to see what they explain about the behavior of the Soviet and American governments in that harrowing event. Both the theoretical and applied chapters are rich in value and thoroughly worth reading. Because he is supplied with an abundance of data, it is the American side of the story, not surprisingly, that shows Models II and III to best effect. In a familiar and poignant episode we see a human confrontation between Model I in the form of Secretary McNamara and CNO Admiral Anderson representing Models II and III. The Secretary wants to know how the CNO will implement the quarantine to see that it will conform with the carefully calculated strategy of the Administration (Model I). The CNO cites the Manual of Naval Regulations (Model II) and suggests that the matter be left to the Navy (Model III). To Model I's way of thinking, Model II or "how John Paul Jones would have handled it," much less Model III or "leave it to the Navy," just wasn't good enough.
Unfortunately for us, Dr. Allison's tour de force falls short precisely where we are most interested — in explaining Soviet behavior. At the outset, Allison poses several key questions about the crisis that he feels have not been satisfactorily answered, two of them about Soviet behavior: Why did the Soviets try the Cuban missile gambit? Why did they pull out of it? In the end, his effort to apply Models II and III are forced and contrived, despite a treatment that is factually largely accurate and carefully done. The main problem, of course, is data. As has been argued persuasively by Messrs. Johnson, Steinbrunner, and Horelick in a study commissioned by CIA (The Study of Soviet Foreign Policy: A Review of Decision-Theory Related Approaches, the Rand Corporation, forthcoming), Models II and III, along with other approaches focused on the inner workings of governments, are voracious consumers of detailed information. When that information is lacking, the models do not work well. They turn into largely speculative excursions, worthy of pursuit and inspiring to the imagination, but devoid of reliable explanatory, much less predictive, power. In applying Model III to Soviet behavior in the crisis, Allison explores the role of Khrushchev, his conflicts with other members of the leaderships, and the possible impact of those conflicts on Soviet decisions. Incidentally, while he treats these matters fairly well, he is largely dependent on the research of academic and official analysts who, without the aid of his models, were hot on the scent of Kremlin conflict even during the crisis. But in the end, the theoretical apparatus of Model III does not fill the gaps of absent information.
Allison's efforts to apply Model II to Soviet behavior focus on asserted conflicts between what the Soviets may be presumed to have been seeking in putting the missiles in Cuba and the way they actually went about it. In essence, Allison claims that the Soviets "blew it" because their standard operating procedures for deploying the missiles and associated defenses revealed the move either too early, before the missiles were operational, or too late, when it was very difficult to pull back. In Allison's view, the Soviet authorities in charge, namely the Soviet Rocket Forces (SRF), set about deploying missiles as they always had, in nice identifiable sites, mindless of the need to orchestrate with Soviet diplomacy what the Americans learned and when they learned it.
If "A" for effort is ever warranted, surely it is here. But the result is not quite convincing. As we all know, the SRF can be quite secretive when it wants to be. But Soviet military and political decision makers alike were surely aware that they could start the Cuban move in secret, but they could not keep it a secret from the U.S. Government very long. Why did they think they could go ahead with it when it was discovered in the face of the kind of political pressure that Senator Keating and others were placing on the Kennedy Administration? The reviewer is indebted to Mr. W. P. Southard of CIA and to Mr. Anatoliy Gromyko, currently of the Soviet Ministry of Foreign Affairs and son of the Soviet Foreign Minister, for a plausible answer to this riddle. Mr. Southard suggests that the Soviets, knowing they could not keep the secret long into actual deployment, believed that they could count on the U.S. Government to keep it a secret from the public and its political opponents, a not uncharacteristic Soviet expectation repeatedly disappointed, and that the Kennedy Administration was in fact signalling its willingness to let the missiles be deployed. In the weeks preceding the outbreak of the crisis, Kennedy was saying in effect that he knew what was going on in Cuba, that offensive missiles were not there, and "were it otherwise, the gravest issues would arise" — this when it was already "otherwise" and the Soviets thought we knew it. From this they may have concluded that the U.S. would acquiesce in the missile move as long as the Soviets kept it from public view, as would the U.S.
The foregoing may seem farfetched but essentially no more so than Allison's assumption that the Soviet government, not particularly given to light-handed management, would allow the trickiest undertaking since Alamogordo to run on unexamined standard operating procedures. The Southard thesis has gained novel support from the junior Gromyko who, in an article for a Soviet book on international crises, argued that prior to the crisis, Kennedy did not directly challenge the Soviets as to what was going on, and that they were as surprised as the U.S. public when Kennedy threw down the gauntlet in his TV address. Admittedly there must be in this an element of post hoc rationalization on behalf of Gromyko senior; but it is not therefore a false view of Soviet perceptions. Which of the two theses fares best under Occam's razor may be left to the reader and future historians. The point is that Model 11 facts can be made to work just as nicely in a Model I explanation.
In fact, although Allison sets out to challenge him on his own ground, the Rational Actor remains standing astride the history of the Cuban missile crisis like the jolly Green Giant. Allison has a clear polemical interest in deprecating the power of the Rational Actor Model. Among its offshoots, he accounts the various sub-models called deterrence theory, strategic calculus, or missile power. These he finds inadequate to explain Soviet behavior. But to this reviewer the strategic power approach offers about as good an explanation of Allison's key questions — why they started and stopped the missile gambit — as any available.
Despite the cruciality of strategic issues in the crisis, the Soviet view of these issues preceding and during the crisis has never been fully sorted out in public discussions. From the position of Moscow, or Khrushchev, or the Soviet General Staff, the strategic situation in early 1962 must have appeared positively horrendous. After much ballyhoo, the missile gap had collapsed in one speech by Roswell Gilpatrick; the Soviets had only a few dozen soft and very slow-reacting ICBMs; a small, very vulnerable bomber force; and a rag-tag assortment of missile submarines that the U.S. Navy had under constant trail. Meanwhile, the U.S. had about 100 Atlas and Titan ICBMs by mid-year 1962, a formidable force of 1,500 heavy and medium bombers, and 96 operational Polaris SLBMs. Moreover, it had been toying since McNamara came into the Pentagon with a counterforce doctrine that looked fearfully like a theory of preventive war to the Soviets. This was enunciated in McNamara's Ann Arbor address which not only pronounced counterforce, but implied a U.S. expectation that Soviet retaliation with any small surviving strategic force could be deterred. And finally, Minuteman was coming into the force at what the Soviets must have found a mind-boggling rate; Penkovsky's contribution from the SRF Bulletin of summer 1961 indicates that the Soviets saw this with chilling clarity. In short, the Soviets faced a near future of woeful strategic vulnerability; they knew it and knew we knew it.
What could they do? They could try to change the political relationship. This is what they did after the missile crisis convinced them they had no choice. But in the summer of 1962 Khrushchev, both for Model I and Model III reasons, was not yet willing to scuttle his past tactics of confrontation. The Cuban missile gambit was a cheap and daring way to fix the problem temporarily. Much is made of the probable political value that successful deployment of missiles to Cuba would have had. But it would also have had direct and tangible value in enhancing Soviet deterrence against a surprise attack. For it could have created the kind of synergistic relationship not unlike that existing between U.S. Minuteman ICBMs and bombers today. Facing deployed missiles in Cuba, U.S. strike planners would have to choose between launching a missile attack against the USSR simultaneously and an attack on the Cuban bases or timing a missile attack on the USSR and an attack on Cuba such that they arrived at the same time. In the first tactic, the strike on Cuba would arrive first, and rudimentary communications would allow the Soviets to launch USSR-based systems on warning. In the second, warning of a ballistic missile attack on the USSR might allow some of the Cuban missiles to get off before they were destroyed. In practice, none of this would have worked very well, but simply complicating U.S. operational problems was a plus. And the Cuban missiles could have substantially increased the megatonnage targeted on the U.S. in a preemptive strike.
When the U.S. finally made it clear it would not stand for this, the Soviets had no choice but to back off, for the very same reasons they initiated the missile venture: they were too vulnerable. The sole remaining mystery in this line of reasoning is this: If Khrushchev was so impressed by actual and impending U.S. strength that he would try such a desperate move, how could he believe the U.S. would let him get away with it? Again, the simplest explanation may be the best. After the Bay of Pigs and the Vienna summit, Khrushchev thought he could psych Kennedy out. By inference from the Gromyko essay, he continued to think so until very late in the game. Nobody said the Rational Actor of Model I couldn't make mistakes.
Now where does all this leave us? Several useful lessons emerge from facing the analytical challenges that Allison presents to intelligence analysis:
- A deliberate quest for different perspectives and approaches to explain government behavior is definitely useful, because of the questions raised if not for the answers found.
- Almost all "facts" can be treated in several different ways; and a corollary, there can be facts without intelligence, but not intelligence without facts.
- It is extremely important to be explicit about assumptions and the distinction between logical inference and speculation. Speculation should be promoted but not confused with inference.
- Finally, on really important matters, it is unfortunately the charge of official intelligence to be more than insightful; it must be right. This requires carrying the methodological excursions of scholars forward to a synthesis that they rarely achieve. When faced with an urgent intelligence problem, it will not do to report that we see a bit of Models I, II, and III plus a few others we might invent. They all have to be put together and the best explanation with the most predictive power derived.
How this is done is indirectly a concern of Janis' Victims of Groupthink, provocatively subtitled "A psychological study of foreign policy decisions and fiascoes." The message of this book is simply conveyed: Why do individually wise, able, informed, and dedicated foreign policy decision makers sometimes make some absolutely disastrous decisions and at other times do fairly well? Part of the answer, according to Janis, lies in the pernicious influence of Groupthink. When this syndrome is present and strong, there is bound to be trouble; when absent or controlled, things will turn out better. Groupthink is the purely internal pressure for consensus that is generated by the social dynamics of small, cohesive, deliberative groups of people. It includes the pressure to "get along and go along," the tendency of action groups to idealize their image of themselves and demonize their image of adversaries, the incentives to get difficult things over with, and resistance to scrutiny of biases and assumptions that will challenge the group's cohesion and self-image.
Janis traces the deleterious effect of Groupthink through four modern episodes of national security policy: the Bay of Pigs, U.S. operations in North Korea, Pearl Harbor, and Vietnam. By way of counterpoint to these fiascoes, in his view, he offers examples in which success attended the control or suppression of Groupthink: the Cuban Missile Crisis, and the formulation of the Marshall Plan. The author is particularly attentive to the lessons of the Bay of Pigs and the Cuban _Missile Crisis because they were such starkly contrasting performances by essentially the same group of decision makers.
The Bay of Pigs episode stands for Janis as "a perfect fiasco," a failure of collective reason so dramatic as to stagger the imagination. Why did it happen? Why did sensible people drift so uncritically into so wrong-headed an operation? In reviewing the published histories, he finds that key figures in the Kennedy Administration went along even though they felt and expressed reservations which on their face were profound but still were glossed over or ignored. A whole set of wrong assumptions was bought, from the military viability of the plan to the prospects for an anti-Castro revolt. Janis cites four so-called official explanations for this episode employed after the fact by analysts and participants: The Administration had to act on the plan for political reasons; the Administration was new and inexperienced; operational secrecy kept needed expertise out of the deliberations; doubting decision makers did not want to damage their reputations by casting doubt. The author finds all of these wanting for various reasons and goes on to assay how the real villain was Groupthink, the pressure for consensus. It took the guise of an "illusion of invulnerability," and "illusion of unanimity," "suppression of personal doubts," "self-appointed mindguards" and other forms. Few who read Janis' book will attend their next meeting without sensing the demonic presence of Groupthink.
Top level decision making during the Cuban Missile Crisis was a success because it consciously fought the Groupthink phenomenon. It encouraged dissent and repeated review of judgments. It took place in a changing organizational context that obstructed the establishment of set patterns of authority and influence. The President kept out of group deliberations so as not to intimidate subordinates. Fatiguing as it was, decisions were allowed to be reopened. Janis offers in his historical and concluding chapters some interesting views on where and why Groupthink arises. He is not very successful in telling us why in a few happy instances it does not arise. With regard to Cuba in 1962, he suggests, almost as an aside, that the threat of nuclear war might have had something to do with it.
From Janis' perspective there are some similarities between policy decision making and intelligence analysis, also a kind of decision making in that it involves a weighing of evidence and then a decision on what judgment to put forward. The lessons derived are also similar: The most important one is to make sure that assumptions are made as explicit as possible and scrutinized with the same rigor as the evidence. The value of this lesson stands out in official reviews of the intelligence community's performance prior to the outbreak of the October Middle East War.
Unfortunately, there are a great variety of "think" syndromes that impair intelligence analysis. Perhaps worst of all is Nothink when the day-to-day hassle prevents recognition of impending problems and thought about them until they are blazing hot. Then there are Bossthink and Bureauthink, hierarchical derivatives of Groupthink, whereby analysts almost unconsciously assimilate the views of superiors and their organizations and drive evidence to fit them. One encounters "I've-seen-it-all-before"-think, a peculiar peril of the experienced analyst in a profession that is obliged to appear, if not be, omniscient. Another one could be called Lobbythink. This occurs when some preferred policy position is being pushed in the guise of intelligence analysis. When the intelligence is clearly linked organizationally with a policy making institution, this syndrome is readily detectable. But it may appear in the most pristine garb of "objectivity" or in a determination to see some objectionable point of view counterbalanced.
Finding a remedy for all these potential maladies is happily not the objective of a book review. But if they are genuine problems, surely part of the solution lies in reflection, an enterprise for which the intelligence profession must allot a good deal of time. A few hours with Allison and Janis couldn't hurt.
Fritz W. Ermarth