Why Intelligence Fails: Lessons from the Iranian Revolution and the Iraq War
Robert Jervis, Ithaca, NY: Cornell University Press, 2010. 238 pp, endnotes and index.
Can intelligence failure be avoided? Robert Jervis begins his study of two well-known cases, the 1979 Iranian Revolution and the 2003 Iraq War, by noting that the question is more complicated than it may first appear. The most common understanding is that “intelligence failure” occurs when, as Jervis puts it, “there is a mismatch between the estimates and what later information reveals.” But intelligence has no crystal ball, and no one should be surprised that assessments of things that are hidden and projections about the future sometimes miss the mark. In this sense, intelligence failures are indeed inevitable, whatever steps might be taken to try to avoid them. A more interesting question is whether analysts succeed or fail in making the most of information available to them. In two case studies, Jervis identifies key reasons why analysis fell short while also demonstrating that the most common explanations for these failures are wrong. His conclusion in both cases is that if analysts had done their best, i.e., “succeeded,” they would have reached many of the same judgments, albeit with a reduced degree of certainty.
Jervis’s study of why the CIA failed to anticipate the revolution that deposed the shah of Iran was written in 1979 and only recently declassified. Despite the intervening years, its insights remain fresh and relevant to today’s intelligence challenges. The fundamental reason for the failure, according to Jervis, was that judgments were based mostly on their inherent plausibility and alternative possibilities were not seriously considered. The shah had defied previous predictions of his demise and was expected to do so again. Analysts didn't understand the nature of the opposition, particularly the religious dimension—which was dismissed as an anachronism. CIA believed that the shah would crack down if his rule was threatened, apparently not taking into account that this expectation was at odds with US advice that he should continue to pursue democracy and reform. Most important, analysts did not recognize that this key belief was not “disconfirmable”—that is, it could not be shown to be false until the shah had already been deposed.
Jervis’s Iraq study is less comprehensive and acknowledges some missing pieces, but he finds the basic mechanism of failure to be similar to that in the Iran case. Analysts had developed plausible inferences about what was happening in Iraq that guided their interpretation of the relatively few specific bits of information that were available. It made no sense that Saddam Hussein would continue to obstruct inspections and risk a US attack if he had nothing to hide. This general presumption, rather than the specific evidence being reported, was the basis for the judgment that Iraq had WMD. Analysts assumed that they were seeing only a small portion of Iraq’s effort because of Saddam’s well-developed program of denial and deception. As in the case of Iran, they did not take into account that there was no way to determine if this core belief was true or false.
Jervis does not discount or excuse the specific errors of analysis and sourcing that received most of the attention in the official postmortems of the Iraq failure. However, he notes that critics invariably leave the impression that had these mistakes been avoided, the Intelligence Community could have reached the correct judgments about Iraq’s weapons. In fact, given the information available, the least damning verdict that might have been offered was that there was no solid evidence of continuing programs. Any claim that Saddam had ended his WMD programs would have been seen as highly implausible, even if there was evidence to support it. As Jervis notes, critics do not wish to acknowledge this because there is a presumption that “bad outcomes are explained by bad processes.” It is more comforting to believe that if the right reforms and organizational changes are made, future failures can be avoided.
This is not to say that the IC could not do a better job. Jervis’s main criticism is the failure to apply what he calls “social science methods,” which might be thought of more generally as critical thinking skills. Analysts tend to look for (and find) what they expect to see. They do not think enough about the potential significance of things that are not seen (“dogs that do not bark”). Most important, they do not make an effort to consciously articulate the beliefs that guide their thinking and consider what evidence should be available if they were true, or what it would take to disprove them. Facts do not speak for themselves but inevitably are seen in a framework of understanding and belief—whether that framework is recognized or not. Analysts rarely think about that contextual framework or what it would take to make them change their views.
The perils of such thinking traps are not a new concern to intelligence analysts. Indeed, Jervis begins his book with a quotation from Sherman Kent, one of the founding fathers of the profession, who observed that intelligence officers are supposed to be distinguished from others by their “training in the techniques of guarding against their own intellectual frailties.” However, as Jervis also notes, many aspects of routine practice and culture in the IC do not encourage attention to this problem. Intelligence products tend to focus on the latest events, reporting the facts with little reflection or interpretation. Conclusions are too often merely assertions without explanation or support beyond their inherent plausibility. Although it has all the necessary raw materials, the IC has never developed an effective peer review process for analytic production. “Coordination” tends to focus on superficial language changes rather than a serious examination and debate about fundamental premises.
In the aftermath of post-9/11 and Iraq war critiques, the IC has placed renewed emphasis on enhancing collaboration and improving the quality of analysis. In accordance with the Intelligence Reform and Terrorism Prevention Act of 2004, analysts are applying new guidelines designed to improve characterization of sourcing, clarify assumptions, and encourage consideration of alternative possibilities. Jervis does not assess the merits of these initiatives specifically, but he clearly believes that the prospects for improvement are limited by the fundamentally intractable nature of the problem. He suggests that better analysis requires a robust examination of how judgments are reached and a sharp focus on underlying factors that are often overlooked. Why do specific judgments seem plausible and are there alternative possibilities? Could the information advanced in support of a particular thesis be explained by other factors? Are we misunderstanding the impact of political and historical factors unique to the issue or region? He recommends supplementing this program of self-scrutiny with substantively focused peer review and extensive study of a range of historical cases.
Even as Jervis explains the challenge of overcoming congenital intelligence limitations, he also warns that better analysis in the sense he suggests might not be particularly welcomed by consumers. By their nature, decision makers need to have conviction and are focused on selling and implementing their policies. Intelligence analysis that gives more scope to alternative interpretations of the evidence is not likely to be well received. Jervis offers a colorful quote from John Maynard Keynes to illustrate the point: “There is nothing a Government hates more than to be well-informed; for it makes the process of arriving at decisions much more complicated and difficult.” Perhaps the best contribution intelligence can offer, Jervis suggests, is a nuanced evaluation of alternative possibilities and the key factors at work. Ideally, this could raise the level of understanding and debate before policymakers make decisions. At the same time, however, they are unlikely to pay attention unless they are already seized with the issue, so there is a narrow window for such inputs.
There is much more of value to the intelligence professional in this concise but densely packed volume, including a discussion of the complexities of politicization, specific insights on other historical cases of interest, and detailed endnotes that constitute a survey of relevant literature. It is essential reading that gets beyond the conventional wisdom about intelligence failure and provides nuanced insight into what Jervis describes as the “insoluble dilemmas of intelligence and policymaking.”
[Top of page]
All statements of fact, opinion, or analysis expressed in this article are those of the author. Nothing in the article should be construed as asserting or implying US government endorsement of an article’s factual statements and interpretations.