CHAPTER 1

THE WORK OF INTELLIGENCE

THIS CHAPTER HAS FOUR SECTIONS. The first section makes the case that intelligenceis a social problem, a recognition that has significant implicationsfor the work of the CIA. The second section introduces the theoretical viewpoint,social constructivism, and explains why it is well suited to investigatethe CIA's work. In sum, this is because intelligence work happens not merelyin the minds of individual analysts but in a distinctive community, the CIA.This section also spends time illuminating the details of exactly what is meantby "intelligence work," especially "intelligence analysis," to demonstrate itsessentially social nature. The third section introduces a crucial distinction betweentwo types of strategic surprises, secrets and mysteries. The fourth andfinal section introduces the intelligence cycle, a model that we use to examinethe impact of the CIA's identity and culture on its work.

THE SOCIAL FOUNDATIONS OF INTELLIGENCE

Explicit recognition of the social nature of intelligence analysis has emergedonly in the last few years. In the following pages, however, we examine theactual process of intelligence analysis in detail and expose it as an almost entirelysocial process and therefore one well suited to a social constructivistexamination. Time spent laboring over the social nature of intelligence analysisin this section illuminates an activity that those outside the world of intelligencehave difficulty picturing precisely. A close look at the actual processesof analysis here also introduces documentary material that Chapter 2 drawson to elucidate the social mechanisms that create and maintain the agency'sidentity.

Anecdotal accounts of both intelligence analysis and of specific strategicsurprises have always contained accounts of social interactions, but scholarsand practitioners have explicitly recognized the essentially social nature ofintelligence analysis only in the last few years. The literature targeting improvedanalysis has usually consisted either of collections of practical analytictechniques for the individual analyst (essentially, what an individual "shoulddo") or descriptions of the various psychological traps to which individualanalysts are prone (essentially, what an individual "should not do"). One canobserve this social void in both CIA publications about intelligence analysisand in external sources.

The slighting of the essentially social basis of U.S. intelligence analysisbegan at its birth. Sherman Kent, in Strategic Intelligence for American WorldPolicy, describes a seven-step process of intelligence analysis. None of Kent'sanalytical steps overtly recognizes the social nature of analysis. Quite the contrary:Step One of Kent's process of analysis reads, "1. The appearance of aproblem requiring the attention of a strategic intelligence staff." Note a peculiarthing about this step: The problem to be analyzed simply "appears"—theanalyst and the agency as a whole are unproblematically presented by the exogenousenvironment with this problem; they do not participate in its definitionor creation.

This uncritical, deus ex machina introduction of a discrete intelligenceproblem is even more peculiar considering Step Two of Kent's process:"2. Analysis of this problem to discover which facets of it are of actual importanceto the U.S., and which of several lines of approach are most likelyto be useful to its governmental consumers." Clearly, Kent is describing anessentially social process as unproblematically as if intelligence issues wereatomic particles.

For the readers of his book, Kent's positivistic approach is not a surprise.In the preceding paragraphs (by the man, one may note, called "the godfatherof National Intelligence Estimates," after whom the CIA's school for analystsis named, and whose "Principles of Intelligence Analysis" analysts still use intraining), Kent says:

A medieval philosopher would have been content to get his truth by extrapolatingfrom Holy Writ, an African chieftain by consultation with his witchdoctor, or a mystic like Hitler from communion with his intuitive self. But weinsist, and have insisted for generations, that truth is to be approached, if notattained, through research guided by a systematic method. In the social scienceswhich largely constitute the subject matter of strategic intelligence, thereis such a method. It is much like the method of the physical sciences. It is notthe same method but it is a method none the less.

Kent then elucidates in a footnote the qualification to this naked positivismmade in the final sentence quoted above: namely, that in the social sciencethere is "enormous difficulty" in "running controlled and repetitive experiments."This idea, while true, does not reveal any appreciation by Kent forthe distinction between natural and social facts or any insight into the socialnature of analysis.

One might object that Kent's book is a 1950s relic. As far as its attitudes tosocial facts are concerned, it is not. To offer but one example, Abram Shulskyand Gary Schmitt's Silent Warfare: Understanding the World of Intelligence, a2002 book still widely respected among analysts and used in many courses onintelligence, says:

Analysis refers to the process of transforming bits and pieces of informationthat are collected in whatever fashion into something that is usable by policymakers and military commanders. The result, or "intelligence product," cantake the form of short memorandums, elaborate formal reports, briefings, orany other means of presenting information.

Silent Warfare then goes on to describe cryptanalysis, telemetry analysis,photo interpretation, and the production of scientific and technical intelligence,military intelligence, political intelligence, and economic and (even)"social" intelligence (sic) without addressing the social aspects of the analyticalprocess. The closest that the authors come to acknowledging that the analyticalprocess is a social process is through such asides as, "In some cases,such as the production of economic and political intelligence, the techniques[of analysis] are not distinguishable from those of the corresponding socialsciences" (p. 52). Such asides hardly go to the heart of the epistemologicalproblems raised by the approach described in the preceding paragraphs.

In the same way, one of the CIA's attempts to improve analysis, the oft-citedvolume Richards Heuer's The Psychology of Intelligence Analysis pullstogether articles written in the Directorate of Intelligence's in-house journal,written between 1978 and 1986. Here, too, the focus is almost entirely on theinternal cognitive challenges to the individual analyst, at one point comparingthe analyst to a "chess master." The analyst exists in splendid, endogenousisolation, handed discrete, exogenous "problems" from on high. The analystworks in isolation: The second sentence of chapter 1 of Heuer's magnum opusbegins with "Intelligence analysis is fundamentally a mental process ..."

The single (partial) exception to that generalization is in itself revealing. Inthe final section, "Improving Intelligence Analysis," Heuer acknowledges theneed for CIA personnel to have "exposure to alternative mind-sets." He writes:

The realities of bureaucratic life produce strong pressures for conformity.Management needs to make a conscious effort to ensure that well-reasonedcompeting views have the opportunity to surface within the IntelligenceCommunity. Analysts need to enjoy a sense of security, so that partially developednew ideas may be expressed and bounced off others as sounding boardswith minimal fear of criticism for deviating from established orthodoxy ...[Management should promote] the kinds of activities that confront analystswith alternative perspectives—consultation with outside experts, analyticaldebates, competitive analysis, devil's advocates, gaming, and interdisciplinarybrainstorming.

These measures seem like an implicit acknowledgment that intelligenceanalysis is a social activity. One must realize, however, that Heuer's remarksaim to aid individual analysts to keep an "open mind." Observe in this passagethat management needs to make this effort and that only "well-reasoned"competing views should have the opportunity to surface to challenge "orthodoxy."Observe too that it is the "sense of security" of the individual analystthat needs nurturing and that it is individual analysts who need "confronting"with "alternative viewpoints." One can conclude, therefore, that the essentiallydynamic, social aspects of analysis are ignored, and recommendationsare made to improve the analytical performance of individual CIA "chessplayers," who are conceived in isolation from the chess board, pieces, or rules.

In the same manner, in Anticipating Surprise: Analysis for Strategic Warning,a 1970s manual for training analysts at the CIA (called "mandatory readingfor intelligence analysts whose job it was to forecast threats to the UnitedStates" during the Cold War) also largely ignores the social nature of intelligenceanalysis. In it, the social nature of the analytical process gets a nod butlittle more, and the focus remains on individual judgments made (seemingly)in a social void.

A change comes in Robert Clark's Intelligence Analysis: A Target-CentricApproach, published in 2004. This work is mostly a toolbox of analytical techniqueswritten by a veteran CIA analyst and executive in the Directorate ofIntelligence. It does not raise larger issues of problem formation and definitionin analysis, but at least it dwells on the activity's social aspects: Three sectionsclearly address the "ideal" analyst's social attributes, or "interpersonal skills."These are the "analyst as team player," the "analyst as advocate," and the "analystas communicator." The description offered of ideal individual analysts, forexample, states that:

They are persuasive. They enjoy interacting with people and teaching othershow the analytical game is played. They choose their words with care, andwhen they speak, customers listen and respect their opinions. They are highlyregarded by their peers and can organize and work with a team on analysis.But they have the courage to stand alone in their judgments. They are good,and they know it. Their self-confidence, like that of the Israeli intelligenceanalyst who spotted the oncoming Yom Kippur attack, tends to perturb theirsuperiors.

Note that this description ends with these social traits in an analyst underscoredas a factor in preventing a strategic surprise! Clark goes on to state:"The process of getting an answer, especially on complex intelligence problems,is fundamentally a social one." Unfortunately, however, the hypothesisis not pursued further!

The following year (2005), in the beginning of his pathbreaking anthropologicalstudy of intelligence analysis, Dr. Rob Johnston (a Director of CentralIntelligence postdoctoral research fellow at the CIA's Center for the Study ofIntelligence), finally defined intelligence analysis in a social manner: "Intelligenceanalysis is the application of individual and collective cognitive methodsto weigh data and test hypotheses within a secret socio-cultural context."He did so while introducing the results of a two-year study to "investigateanalytic culture, methodology, error and failure within the Intelligence Community,"in the course of which he conducted 489 interviews, "direct participantobservations," and focus groups. Johnston concludes his introduction tothe analysis process thus: "My work during this study convinced me of theimportance of making explicit something that is not well described in the literature,namely, the very interactive, dynamic, and social nature of intelligenceanalysis."

Johnston's volume offers abundant evidence to confirm the essentiallysocial nature not just of intelligence work generally but of intelligence analysisat the CIA. In fact, Johnston says, "Despite the seemingly private andpsychological nature of analysis as defined in the literature, what I found wasa great deal of informal, yet purposeful collaboration during which individualsbegan to make sense of raw data by negotiating meaning among the historicalrecord, their peers, and their supervisors." He then offers even more detail bydescribing a typical description of the analytic process in the words of a CIAanalyst:

When a request comes in from a consumer to answer some question, the firstthing I do is read up on the analytic line. [I] check the previous publicationsand the data. Then, I read through the question again and find where there arelinks to previous products. When I think I have an answer, I get together withmy group and ask them what they think. We talk about it for a while and cometo some consensus on its meaning and the best way to answer the consumer'squestions. I write it up, pass it around here, and send it out for review.

This description neatly brings us to further evidence that intelligence analysisis essentially social: the "review process." The review process—so key tothe CIA's analytical work—is clearly social, not merely individual and cerebral.Both for that reason and because of its centrality to the CIA's work as awhole, the review process is worth understanding in detail.

Evidence of the Social Nature of Analysis:The Review Process

Martin Petersen opens his 2005 Studies in Intelligence article, "Making theAnalytic Review Process Work," with the words, "If there is a first principlein producing written intelligence, it is that finished intelligence is a corporateproduct, not a personal one." This article provides rich fodder for a socialconstructivist analysis of strategic surprise because it further exposes the socialnature of the CIA's analytical work.

Petersen begins by reminding his CIA audience that the review process inintelligence analysis is not mere bureaucratic pettifogging, and it is more thanediting: "Editing is NOT review. Editing is a mechanical task that should beaccomplished by the first-level reviewer or by a staff. Review is about thinking,about questioning evidence and judgments. It focuses on the soundness ofthe analytic points that are being made and the quality of the supporting evidence."In this view, "review" in intelligence might resemble review of a physicsproblem by a more experienced physicist. After an analysis is finished, forexample, "The drafter's supervisor is almost always the first-level reviewer";this supervisor "bears the greatest responsibility—after the author—for thesubstantive accuracy of the piece." Use of the words supervisor and responsibilityimplies a culture recognizes that hierarchy, but so far it remains at leastdebatable whether review is purely social.

A second level of review of the analysis then occurs, and this level of reviewis more clearly social: We not only have further mentions of responsibility,but a relationship—closeness to the policy maker—is cited as a virtue (thegreater significance of which is explored in Chapter 2). In addition, Petersencontinues, one of the key questions that the second-level reviewer must askis about consistency: "Is this piece consistent not only with previous work onthis topic but also with other analysis being done in the issue group?" Such aquestion foregrounds the social nature of intelligence analysis, as it stronglyimplies that consistency with other, past analysis is a screen through which ananalytical piece must pass to get to the next level. Natural sciences, however,recognize neither "arguments from authority" nor a majority consensus tosettle disputes—those are usually used to settle social, not scientific questions.Next, Petersen explains the third level of review:

The third-level review should be done by the office director or the staff of asenior officer in the organization. On a particularly sensitive piece, both mayweigh in ... Like earlier reviews, the third level needs to ponder core tradecraftquestions: is it clear what is known and not known and what the level of confidenceis? What assumptions underpin the analysis? And does the piece addresspolicymaker concerns? ... is it consistent with other work being done in the organization?... The third-level reviewer should focus most on whether the rightquestions have been asked and what the key variables are.