The Perils of Analysis: Revisiting Sherman Kent’s Defense of SNIE 85-3-62
Michael Douglas Smith
The crucial lesson of the Cuban analytic experience is that simply being aware of mental traps is not enough.
CIA’s Board of National Estimates (ONE) was criticized for the conclusion its members reached in Special National Intelligence Estimate (SNIE) 85-3-62, published on 19 September 1962, that the Soviets were unlikely to introduce strategic offensive weapons into Cuba. 1 In 1964 Sherman Kent, ONE’s chief from 1952 to 1967, penned a defense of the analytic reasoning and process that produced the flawed judgment. 2
Kent’s article is interesting because he highlighted many of the pitfalls new analysts in the Intelligence Community are now taught to avoid. 3 His defense also indicates that he had most of today’s preferred techniques in mind when the estimate was written. Here I will review the analytic tradecraft Kent set forth in his article, examine the pitfalls the estimate’s drafters fell prey to, and conclude with ideas on what Kent’s essay can still teach analysts. 4
In 2005, the Kent School published a paper looking at common analytic errors identified in CIA critiques of events considered “intelligence failures.” 5 The paper judged that analysts were guilty of
- having a restrictive mind-set;
- engaging in mirror imaging and using a rational actor model;
- engaging in group think;
- employing status-quo thinking;
- exhibiting the paradox of experience;
- being fooled by denial and deception activities; and
- not offering alternative scenarios.
A close examination of Kent’s article shows that the drafters and authorizers of the SNIE did not commit all of these errors and that institutional analytic practices of the period obscured some of the techniques they were accused of omitting.
On one point there is no ambiguity, the estimate incorrectly concluded that the Soviets would not place strategic weapons in Cuba.
We believe that the USSR values its position in Cuba primarily for the political advantages to be derived from it, and consequently that the main purpose of the present military buildup in Cuba is to strengthen the Communist regime there against what the Cubans and the Soviets conceive to be a danger that the US may attempt by one means or another to overthrow it.… At the same time, they evidently recognize that the development of an offensive military base in Cuba might provoke US military intervention and thus defeat their present purpose. 6
In hindsight, we know Soviet leaders did worry about a US invasion, but Nikita Khrushchev calculated that the presence of operational intermediate-range nuclear weapons in Cuba would prevent the United States from acting after the presence of the missiles had become known to US leaders. Thus a key assumption of the drafters was off kilter: Moscow saw a way around the possibility of “provoking US military intervention” that apparently was not considered by the analysts.
This judgment was accompanied by the opinion that
The USSR could derive considerable military advantage from the establishment of Soviet medium and intermediate range ballistic missiles in Cuba, or from the establishment of a Soviet submarine base there. As between these two, the establishment of a submarine base would be the more likely. Either development, however, would be incompatible with Soviet practice to date and with Soviet policy as we presently estimate.
Here we have an alternative hypothesis to the central judgment, but one that is dismissed as a transgression of previous Soviet practice and likely to produce a US reaction Moscow did not want.
Kent acknowledged that “even in the best minds curious derelictions occur,” but he specifically rejected the idea that the analysts went “for the comforting hypothesis, by eschewing the painful.” 7. He noted that the CIA inspector general’s postmortem of the DI’s performance in the crisis identified a mere eight reports out of hundreds coming in on activities in Cuba that “indicated the possible presence of strategic missiles” and that “none of these was available before the crucial estimate was put to bed.”
Kent further argued that photographs available before the SNIE appeared did not provide evidence of missile emplacement and “over and over again it made fools of ground observers by proving their reports inaccurate or wrong.” And he specifically absolved the analysts of “neglect[ing] or wishful misevaluation of evidence because it does not support a preconceived hypothesis.”
In current DI parlance, Kent would have written that his analysts did not fall prey to a rigid mind-set and thus reject a high-impact hypothesis or exhibit an anchoring bias in evaluating information from human intelligence or other sources. Critics, however, have argued that mindset was a problem because analysts did raise the right question but dismissed what turned out to be the right answer because of a scarcity of confirming evidence.
Kent explained how the error occurred: lacking direct evidence before the U-2 photographs of October, the analysts tried to discern “indicators” that pointed to an explanation of what the Soviets were up to. This led them to conclude the buildup was defensive.
Before reaching this judgment, the analysts considered how Moscow might view the idea of using Cuba as a strategic base and applied historical actions to reach their conclusion. This led the analysts to believe the Soviets would be as cautious in 1962 as they had been during earlier Cold War crises and to believe that US outrage at the creation of a communist regime in Cuba was known to Moscow.Kent speculated that hindsight suggested the Soviets may have believed US resolve had weakened after the Bay of Pigs, erection of the Berlin Wall, and the growth of Communist power in Laos; that they saw the strategic value of offensive weapons as outweighing the risks; or that they miscalculated and underestimated the consequences of a resolute US reaction. 8
He then wrote:
[Top of page]
Even in hindsight it is extremely difficult for many of us to follow their inner logic or to blame ourselves for not having thought in parallel with them.
We ask analysts today to avoid a similar misstep by understanding that historical precedent isn’t an infallible guide and to use an analytic tool, such as analysis of competing hypotheses, to see if there is a break in the historical pattern or if a break is of such high impact that the possibility should be conveyed to policymakers.
Today’s analysts have the benefit of more cases in which a foreign leader has acted on a logic alien to that of a “rational” US policymaker. This has sensitized them to the danger of expecting leaders in other countries to act like us (mirror imaging) or calculate the workings of a situation as a US policymaker might.
Kent’s quote also shows the power of an undocumented assumption—that the Soviets understood how angry Americans were over the Castro revolution. There is no indication in Kent’s article that this assumption was ever challenged or subjected to validation. Today we also ask analysts to identify their assumptions just so they can be examined explicitly. And when it seems appropriate analytical papers will list the assumptions or alternative view laid out using different assumptions.
Kent would have been skeptical of the current practice of providing more than one avenue for a policymaker to consider. He argued in the article that a lack of evidence was not an excuse for simply saying this or that may happen, or that the worst case is going to transpire. This, he contended, was of little use to policymakers, and in the instance of presenting the worst case, ran the danger of leading policymakers to stop listening because the analysts “cry wolf” too often.
He also expressed reservations about a common technique used today, the creation of a “red team.” Just tasking a group to try to mimic enemy responses to a situation, argued Kent, did not mean that it would do so successfully. He dismissed the general utility of such efforts and noted that in the case of missiles in Cuba, CIA experts were consulted “as usual.” That they failed “to work out the propositions of an aberrant faction of the [Soviet] leadership,” was not a failure, Kent asserted, because “no estimating process can be expected to divine exactly when the enemy is about to make a dramatically wrong decision. We were not brought up to underestimate our enemies.” He then added:
We could not believe that Khrushchev could make such a mistake.
This opinion is an example of the fallacy of the “rational actor model,” although Kent decried the related mirror-imaging pitfall, when he wrote “that objectivity of judgment about the other man’s probable behavior is the crux of the intelligence business.”
Then Kent stated, “this…suggests that perhaps we do not know some things about Soviet foreign policy decisionmaking that we should.” Kent was oblivious to the possibility that it wasn’t the Soviet decisionmaking process that was opaque and misleading for agency analysts, but the limitations of experts to recognize a radical change in their field.
This we call the “paradox of expertise.” Forty years after Kent made the argument, intelligence analysts are expected to warn, if they can, before an opponent makes a major decision, including a decision that might lead to unusual or unprecedented behavior. There have been too many instances since the Cuban Missile Crisis of leaders choosing paths that wouldn’t seem “rational” to US decisionmakers to do otherwise.
My examination of Kent’s defense leads to two conclusions:
- The analytical process in 1962 pressured analysts to “make a call.”
- The analytic practices of the era had many of the techniques in use today but omitted several current checks and balances specifically designed to avoid analytic pitfalls.
Kent’s words on the uselessness of providing multiple scenarios or worst-case analysis imply that his was then conventional wisdom. They also undoubtedly reflected the desire of most policymakers of the period for such definitive judgments.
Since at least the mid-1990s, however, senior policymakers have increasingly been requiring intelligence analysts to identify and explain plausible scenarios in the estimates they prepare. These are to include those we now label “high impact, low probability” outcomes. On occasion, the outcome may be truly identifiable and a “single outcome prediction” justified, but most of the time the complexity of world affairs precludes such certainty.
Despite acknowledging the pitfall of mirror imaging and the need to “cast yourself in his [the enemy’s] image and see the world through his eyes,” Kent and his colleagues do not appear to have examined their model of a Soviet decision maker, which was essentially a Russian-speaking Western rational actor who made choices with an understanding of US public opinion and pressures on our policymakers.
An “assumption check” would have raised the question of just how well Moscow actually understood US unhappiness with Castro and might have led to Washington’s explicit statement of that feeling to the Soviets. Whether that would have changed the thinking about Khrushchev is debatable, but it could have alerted the analysts to the need to qualify their prediction by presenting this assumption openly in the estimate. 9
There is no suggestion in Kent’s article that Soviet denial and deception activities played any role in misdirecting the analysis. The crucial lesson, therefore, is that simply being aware of our mental traps is not enough. To reduce the potential for analytic errors, some form of analytic structuring technique must be used to overcome cognitive traps. 10
1. “The Military Buildup in Cuba,” SNIE 85-3-62. 19 September 1962. Now declassified and available in several places, including Foreign Relations of the United States, 1961-1962, Volume X, Cuba, 1961-1962. A portion of the estimate and many other documents related to the crisis can be found in CIA Documents on the Cuban Missile Crisis, 1962 (Washington, DC: CIA History Staff, 1992)
2. Kent, Sherman. “A Crucial Estimate Relieved.” Studies in Intelligence 36, no. 5 (Spring, 1964): 111–19. Originally Secret, now declassified and available on cia.gov. Unless noted otherwise, all quotes by Kent are from this article.
3. One element of the Career Analyst Program, a required four-month long training program for new analysts, is a theme addressing the history, mission, and values of CIA and the profession of intelligence analysis. Analysts are required to present briefings to classmates on a variety of historical events that have had a major impact on how the Directorate of Intelligence does its work. The Cuban Missile Crisis is one of those events.
5. A few intelligence officers do not hold that the SNIE was an “intelligence failure.” A future head of the Directorate of Intelligence, Russell Jack Smith, agreed with Kent that the Cuban missile crisis was not a failure of analysis. See Smith’s, The Unknown CIA (Washington, DC: Pergamon-Brassey’s International Defense Publishers, 1989), 155.
7.Kent may have been alluding to the well-documented dissent of the Director of Central Intelligence, John McCone. The DCI believed the Soviets would take the risk of installing offensive missiles in Cuba and said so to a special NSC group on 10 August. He maintained this position despite the opposition of Kent and the Director of Intelligence Ray Cline and the failure of initial photographs to show the weapons. An Agency history of McCone’s directorship highlights this difference of opinion. Published memoirs by some participants also claim McCone wasn’t convincing in his arguments, and one scholarly study goes so far as to assert that the DCI’s “discrepant judgment holds no interesting general lesson for intelligence assessment and hardly seems worth the attention it has received.” See “What Can Intelligence Tell Us About the Cuban Missile Crisis, and What Can the Cuban Missile Crisis Tell Us about Intelligence?” in James G. Blight and David A. Welch (eds.), Intelligence and the Cuban Missile Crisis. London: Frank Cass, 1998, 6.
8. Khrushchev actually viewed the United States as aggressive, not weak, and he was very concerned about losing Cuba to a US invasion. He did see the strategic value of the missiles, but this appears to have been a secondary motivation. See John Lewis Gaddis, We Now Know. Rethinking Cold War History (Oxford: Oxford University Press, 1997), Chapter Nine.
10. McCone and the IC were criticized after the crisis in classified postmortems and newspapers. In November 1962, the President’s Foreign Intelligence Advisory Board (PFIAB) tasked the DCI for an all-source, community-wide examination of collection and analysis relating to the crisis. The result, which characterized IC efforts as on the whole positive, infuriated the board. One member termed McCone’s praise for CIA’s performance a “snow job.” The PFIAB report of 4 February 1963 gave the IC poor marks for its performance before the 14 October 1962 imagery revealed the offensive missiles. The IC’s work after that was given high marks. The Board’s report is in CIA Documents on the Cuban Missile Crisis.