Why strategic intelligence analysis has limited influence on American foreign policy

Transcription

1 Intelligence and National Security ISSN: (Print) (Online) Journal homepage: Why strategic intelligence analysis has limited influence on American foreign policy Stephen Marrin To cite this article: Stephen Marrin (2017) Why strategic intelligence analysis has limited influence on American foreign policy, Intelligence and National Security, 32:6, , DOI: / To link to this article: Published online: 04 Jan Submit your article to this journal Article views: 4321 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at Download by: [ ] Date: 26 December 2017, At: 16:57

2 Intelligence and National Security, 2017 VOL. 32, NO. 6, ARTICLE Why strategic intelligence analysis has limited influence on American foreign policy Stephen Marrin ABSTRACT Why does strategic intelligence analysis have limited influence on American foreign policy? Intelligence analysis is frequently disregarded, this paper contends, because it is a duplicated step in the decision-making process and supplements but does not supplant policy assessment. Many intelligence analyses will confirm policy assessments and be redundant or if the assessments are different policy-makers will choose their own interpretations over those of intelligence analysts. The findings of this paper provide scholars with important insights into the limits of intelligence analysis in the foreign policy process as well as recommendations for increasing its positive impact on policy. Strategic intelligence analysis frequently does not influence the creation and implementation of American foreign policy. This is contrary to an expectation that intelligence analysis should matter in the making of foreign policy and national security. Better intelligence analysis should lead to better policy. Ideally, independent and objective intelligence analysts provide the information and assessments that decisionmakers use to make the best possible decisions. Yet strategic intelligence analysis appears to have had limited influence on national security decision-making. As former senior intelligence official and Temple University historian Richard Immerman suggests, in terms of foreign policy instances of the formulation of national security policy, or grand strategy, hinging on intelligence collection and analysis are few and far between if they exist at all. 1 Former National Intelligence Officer Paul Pillar has also made a similar argument on the irrelevance of intelligence to great decisions. 2 Their conclusions are not new, however. For decades, there have been questions about the degree to which strategic intelligence influences policy. In 1953, Roger Hilsman, then a professor at Columbia University and later Director of State Department s Bureau of Intelligence and Research, observed that decisionmakers frequently made decisions without asking for or using intelligence assessments. 3 In 1960 Benno Wasserman said that despite its voluminous labours, intelligence does not influence policy greatly because decisionmakers frequently do not read or accept the intelligence analysis they are provided with. 4 In 1966 Sherman Kent, the most important foundational figure in the development of American intelligence analysis, also pointed out that no matter how right we are (policymakers) will upon occasion disregard the thrust of our findings. 5 In the decades since, many others have referenced the frequent irrelevance of strategic intelligence analysis to national security decision-making. 6 Intelligence analysts can become frustrated when their analysis does not have the influence or impact that they think it should. In 1953, Roger Hilsman observed that an intelligence analyst is not only frustrated when he decides that policy is made without benefit of the information supplied by CONTACT Stephen Marrin marrinsp@jmu.edu This article was originally published with errors. This version has been amended. Please see Corrigendum ( ) Informa UK Limited, trading as Taylor & Francis Group

3 726 S. MARRIN intelligence, but bewildered and indignant as well. 7 Hilsman quotes one frustrated official who said that contrary to the way things should work, policy was made without intelligence or was only supplemented by intelligence. Intelligence people always had to analyze what had already happened, or merely to give support for policy decisions that were already made. Intelligence did nothing but hack work and research. In practice, the thing was all backwards. 8 Strategic intelligence analysis has had less influence on American foreign policy than many would expect, and the literature on American foreign policy contains the frustration of intelligence analysts whose work was not accepted by policymakers: Samuel Adams on the order of battle assessments during the Vietnam War; Melvin Goodman on Cold War assessments regarding the Soviet Union; and Paul Pillar on his assessments before the Iraq War. 9 Forgotten by scholars, in 1974 Ray Cline then the recently retired Director of the US State Department s Bureau of Intelligence and Research eviscerated both Henry Kissinger and the Nixon Administration for their failure to rely on intelligence analysis in their decision-making. 10 Obviously frustrated and disappointed, Cline titled the article Policy without Intelligence. Taken in the aggregate, these incidents indicate that there is an unresolved dilemma that lies at the core of the American intelligence and foreign policy process. Strategic intelligence analysis, which many think should be the foundation for decision, is frequently not relied upon in the making of decisions. Yet many scholars, practitioners, and even the general public persist in retaining this belief that intelligence analysis should play a central role in the making of American foreign policy. To suggest that strategic intelligence analysis frequently does not influence policy is not to argue that strategic intelligence analysis never influences policy. In fact, strategic intelligence analysis has influenced policy for the better in some situations. 11 But it happens less frequently than many would expect, and seems to happen even less frequently on the most important national security decisions. So why does strategic intelligence analysis have limited influence on American foreign policy? Competing explanations Explanations for the lack of intelligence analysis influence on policy suggests that it is due to the combination of political commitments and cognitive biases at the highest levels of national policy-making which hinder or otherwise prevent intelligence analysis from influencing policy. Scholars referencing the limited influence of intelligence analysis on policy rely upon these explanations, including Richard Immerman, Robert Jervis, Joshua Rovner, and Paul Pillar. 12 There is some evidence to support the contention that politics and cognitive bias affect some foreign policy decisions. As former CIA officer Harold Ford explains, policy-making is not based on orderly reason; instead all kinds of forces go into their making of policy, not excluding timidity, ambition, hubris, misunderstandings, budgetary ploys, and regard for how this or that policy will play in Peoria. 13 The CIA itself said as much when it observed that national estimates have played only a modest role in the formulation of policy because national security policy is not directly driven by facts, analyses, and judgments, and knowledge does not lead to action in any simple way. What also come into play are competing bureaucratic pressures, a wide range of domestic considerations, the concepts, perceptions and stereotypes held by the policy makers themselves, and a host of essentially political inputs. 14 But the predominant explanation for policy-maker disregard of intelligence analysis is insufficient because it cannot account for the extent to which it arises as a problem across space, time, and different subject matter domains. The limited influence that intelligence analysis has had on decisions is noticeable over an extended period of time in many different countries, as well as in other fields such as law enforcement and business, the prevailing explanation which is essentially an idiosyncratic one dependent on the prevailing mindset of the policy-maker does not seem sufficient to account for the scope of the problem. 15 So why does intelligence analysis frequently not matter in all these different contexts?

4 INTELLIGENCE AND NATIONAL SECURITY 727 This paper contends that the primary problem is not with the policymakers who disregard intelligence analysis, but rather with how scholars have been conceptualizing the intersection of intelligence analysis and decision-making. Scholars appear to be assuming that intelligence analysis should influence policy and then explain why decisionmakers were not influenced by the intelligence analysis. But why should intelligence analysis influence policy-making in the first place? Instead of asking why do decisionmakers not accept intelligence analysis? better questions to ask are how is strategic intelligence analysis used by decisionmakers? and why should decisionmakers accept intelligence analysis as better than their own analysis? Answers to these questions show that the privileging of intelligence analysis in the policy process which exists in the current problem specification is based on normative expectations derived from idealized conceptions of the intelligence and policy process. In the end, intelligence analysis will inevitably have limited influence on policy-making because it is a duplicative step in the policy process. Policymakers are analysts too Intelligence analysis is frequently disregarded because the analysis of a situation is a duplicated step in the decision-making process and supplements but does not supplant policy assessment. Yet this is not the impression most scholars have of the intelligence and policy process. The prevailing conceptualization of the relationship between intelligence and policy is best illustrated using the intelligence cycle, which is a simplified heuristic to describe the different steps in intelligence production. The intelligence cycle starts with information requirements levied on intelligence collection capabilities, the collection and processing of the raw intelligence and transmission of this material to analysts who decipher its meaning and relay their understandings to the decisionmakers, who levy additional requirements. While there is some criticism of the intelligence cycle for being an over-simplification of the process, it is the predominant framework used to understand and explain intelligence production. For that reason it is frequently used as the foundation for the Standard Model of the intersection of intelligence analysis and decision-making which is dominant in the intelligence and security studies literature. 16 The intelligence cycle differentiates between intelligence and decision-making functions, implying that intelligence organizations collect and analyze intelligence, and decisionmakers make decisions based on the information and assessments provided. The reason intelligence and policy functions are separated, in this ideal relationship, is to ensure that decisionmakers get the objective, unbiased information and interpretation they need to make the best possible decisions. As Paul Pillar observed, the proper relationship between intelligence gathering and policy-making sharply separates the two functions. The intelligence community collects information, evaluates its credibility, and combines it with other information to help make sense of situations abroad that could affect U.S. interests. 17 Decisionmakers then use this intelligence analysis in the making of their decisions. The role of intelligence analysis is conceived either as a way of reducing uncertainty, or helping to ensure that power is applied more effectively and efficiently. 18 For those employing normative expectations derived from this idealized conception of the relationship, when decisions are made that disregard intelligence analysis in favor of that from other sources, including ideological or partisan ones, controversy and criticism erupts as it is taken as indicative of a corrupted decision-making process. In 1949, Sherman Kent implied as much when he suggested that decisionmakers should not disregard intelligence analysis because if they do they would be turning their backs on the two instruments by which western man has, since Aristotle, steadily enlarged his horizon of knowledge the instruments of reason and scientific method. 19 This expectation that policymakers should use intelligence analysis can be found in more contemporary evaluations that identify policy-maker neglect of intelligence analysis as one of the primary pathologies of intelligence-policy. 20 But a different conceptualization of the intelligence and policy process emphasizes the ways in which each process overlaps the other. Ernest May and Philip Zelikow developed a model for evaluating the intersection of intelligence analysis and policy-making which summarizes decision-making

5 728 S. MARRIN as the process of evaluating international events and issues that have the potential to affect national interests (what s going on?), the analysis and evaluation of this information (so what?), and the ability to create and implement effective policies (what is to be done?). 21 In this three-step process, intelligence organizations perform a delegated function, by either finding out what is going on (collection and analysis), or determining what it means (analysis answering the so what question), or both. Contrary to conventional portrayals, decisionmakers do not just make decisions after receiving finished intelligence analysis. Instead, they engage in all three steps, focusing their attentions on what to do about the situation at hand. This reconceptualization of the intelligence and policy process shows that the analytic function is duplicated in both domains. If intelligence analysis were the primary locus of analysis and assessment, then policy-maker disregard of intelligence analysis would be potentially significant. But the assessment function is duplicated in both the intelligence and policy domains. As Sherman Kent observed, there is no universal law which obliges policy, plans, and operations to accept and use (intelligence analysis) findings. 22 Over 50 years later, former CIA officer Jack Davis observed that policymakers can reach, publicize, and act upon their own judgments which may different from those of intelligence agencies precisely because intelligence analysis is an input to and not a substitute for policy analysis. 23 Why should policymakers base their decisions on intelligence analysis and assessment in the first place? The normative expectation that intelligence analysis should be accepted by decisionmakers is based assumptions that intelligence analysis more accurate because analysts have more and better information and are more expert than policymakers, and that accuracy or truth can be achieved through intelligence agencies ability to accumulate, aggregate, and expertly interpret information independent of policy preferences. But on closer examination, each of these assumptions has problems. Policymakers also have information and expertise Intelligence analysts do not necessarily have more or better information than decisionmakers do. For example, decisionmakers have access to raw intelligence in fact, according to former senior CIA manager Martin Petersen, the same stuff intelligence analysts are reading 24 as well as the finished intelligence analysis. In 1989 Robert Gates suggested that the creation of the White House Situation Room in the early 1960s had severely degraded the influence of current intelligence analysis for the President in a way that had yet to be fully appreciated. 25 More than 20 years later, decision-maker access to raw intelligence continues to marginalize the contributions of intelligence analysts. In addition, policymakers acquire a lot of information through non-intelligence channels including friends, broader networks, and other contacts in government, academia and business. 26 Sometimes these other sources provide information that is more accurate than that provided by intelligence agencies. Very little of the information acquired by policy officials is shared with the intelligence community. As a result, on strategic issues it can be very difficult for the intelligence analyst to provide decisionmakers with better information than they already have access to. Also, if intelligence analysts were more expert than policymakers, then that might also provide sufficient justification for arguing that they produce more accurate assessments on a regular basis. But on a closer examination does not hold up to scrutiny either. While the literature sometimes portrays decisionmakers as political generalists and the intelligence analysts as knowledgeable experts, with the inference being that the generalist should defer to the knowledge and expertise of the experts, the reality is that strategic national security policymakers frequently have considerable knowledge and expertise of their own and they trust their own judgments. Robert Jervis has said that one reason why policy-makers paid little attention (to intelligence analysis) was that they thought their own general knowledge of the situation provided an adequate basis for their decisions. As one senior State Department official put it, A policy-maker usually has some expertise of his or her own, after all. I use the intelligence community as a resource for factual information, but I don t need it for opinions. I have my own. 27

6 INTELLIGENCE AND NATIONAL SECURITY 729 Or, as three CIA officers put it, policy people regard themselves as having certain expertise and ample sophistication. They are accustomed to interpreting, analyzing, and projecting, as well as planning, deciding, and operating. Intelligence that attempts to do for them what they believe they can accomplish competently themselves generally is less well received. 28 In fact, policymakers sometimes have more expertise than intelligence analysts. As former CIA analyst Richard Russell says, it is not uncommon for policy makers to be more expert than most CIA managers and analysts. 29 Or as former CIA analyst Harold Ford says, senior policy-makers often know much more about given foreign questions than do the experts preparing intelligence for them. 30 As a result, as Robert Gates has observed, on more than a few occasions, policymakers have analyzed or forecast developments better than intelligence analysts. 31 Why should policymakers accept the judgments of intelligence analysts when they conflict with their own? Before one can criticize decisionmakers for rejecting intelligence analysis one has to establish that they should have deferred to intelligence analysts judgments in the first place. But that case has not yet been made. Policymakers can produce good analyses different from intelligence analysts If accuracy or truth could be achieved through independent accumulation and aggregation of information, then that might also suggest that better information combined with more knowledge leads to more accurate assessment. But this assumption, too, has difficulty withstanding scrutiny. The embedded assumption here is based on the portrayal of intelligence assessment as primarily inductive in nature, where facts lead to judgment in a simple and demonstrable way. The primary source of this empirical, inductive analytic model is Sherman Kent himself. Kent s underlying assumption was that there was one and only one correct interpretation to every set of data, and that the job of the intelligence analyst was to interpret the information available in order to come up with that one correct interpretation. 32 The problem with Kent s assumption is that intelligence assessments are not derived solely from the data; facts do not speak for themselves. They have to be interpreted, and that requires some form of conceptual framework to organize the information and derive inferences from it. 33 This is important because it means that intelligence analysts can arrive at different answers to the same question. As Richards Heuer, one of the CIA s foremost analytic methodologists, has observed, research in cognitive psychology indicates that analysts do not fit information together into pictures the way someone putting together a puzzle might do. Instead, analysts typically form a picture first and then select the pieces to fit. 34 Under these conditions, different interpretations are possible depending at least as much upon the conceptual framework employed to analyze the data as it does upon the data itself. 35 This also explains how decisionmakers can arrive at assessments that are different from those of intelligence analysts. In the case of policymakers, the deductive frameworks used to interpret the significance of events are frequently driven by world views or perspectives that are grounded in a particular political philosophy. Policymakers use these as filters accept or reject the intelligence assessments they are provided with. Some observers appear to believe that politics or ideology should not shape policy interpretations, or that such interpretations are less legitimate than those made by intelligence analysts. But policymakers are inherently political decisionmakers, responsible to the people through elections. If one believes in democracy as a form of government, one also believes that the democratic impulse as operationalized in the political process should be incorporated into the decision-making process as well. One could make the argument that political predispositions and perspectives should be the driving force behind the assessment of national security situations, to include the intelligence collected about them. Senior CIA analyst Fulton Armstrong acknowledges this directly when he says that the policymakers and the political processes that empower them determine what is in the national interest and as a result our system encourages a political competition to define problems as well as solutions. 36 Political preferences and values implicitly shape the intelligence analysts evaluation of the problem as well, and this can limit the degree to which intelligence analysis is incorporated into policy. In

7 730 S. MARRIN 1949, Yale University professor Willmoore Kendall argued that the functional differentiation between intelligence analysts and decisionmakers would mean that the analysts would become mere research assistants to the George Kennans precisely because the independent and objective role would prevent them from being able to provide decisionmakers with the kinds of knowledge that determines the pictures they have in their heads of the world to which their decisions relate. 37 Kendall s conclusions are similar to Richard Betts later observation that the strict separation between intelligence analysis and decision-making may be an unrealistic application of pure science norms to areas where fact and value are inseparable and judgment must preside. 38 Intelligence analysis increases uncertainty and complicates policy-making An additional problem for policy receptivity to or acceptance of intelligence analysis is that, in the context of duplicated assessment, intelligence analysis increases uncertainty and complicates policy-making. Contrary to portrayals of intelligence analysis as reducing uncertainty, Robert Jervis has suggested that strategic intelligence analysis may actually increase decision-maker uncertainty. 39 Increasing decision-maker uncertainty may, in fact, be what the founders of the US intelligence community wanted intelligence analysis to accomplish in the first place. In Sherman Kent s 1949 argument for the creation of strategic intelligence analysis as a separate career field, he acknowledged that the acquisition and interpretation of information could be done by policy staffs, and that policymakers would prefer this. 40 But the risk of relying on policy staff for evaluating intelligence was that they would, to use a term developed decades later, provide intelligence to please that was consistent with the biases and preferences of the decision-maker. To prevent this outcome, Kent argued strongly that intelligence analysis should be independent of policy. The purpose of independent intelligence analysis is to provide the inconvenient facts and unwanted interpretations which challenge reigning assumptions, presumptions, and preferences. As Kent put it, the job of intelligence analysis is to see that the doers are generally well-informed; its job is to stand behind them with the book opened at the right page, to call their attention to the stubborn fact they may be neglecting. 41 Or, according to Roger Hilsman, by having one man collect facts without thinking of policy and another use the facts to make policy, one at least guarantees that the policy man will have to face the unpleasant facts that do not support his policy. Thus the policy man who has become wedded to an incorrect policy can be faced with the unpleasant facts which he has ignored and be made to see the one right solution. 42 In other words, the purpose of independent intelligence analysis is to increase decision-maker uncertainty rather than reduce it. The implicit assumption was that policymakers would or should adjust policy for the better when confronted with intelligence assessments that challenged their interpretations and policies. This appears to be what Sherman Kent expected to happen when he argued that strategic intelligence analysts should be as independent from policy as possible. Sherman Kent did not say that decisionmakers actually use these inconvenient facts and unwanted interpretations in their deliberative and decision-making processes; instead, he said that they should. In a particularly strong rhetorical flourish for 1949, he says that policymakers should not reject their intelligence analysis or else they risk the failures of Hitler by turning their backs on reason and the scientific method. 43 But history has shown that on receiving intelligence analysis that is inconvenient or unwanted, policymakers are more likely to reject or ignore the intelligence analysis than change policy. The literature does not provide any evidence indicating that intelligence analysis influences policy-maker judgment when provided with unwelcome analyses. Instead, it indicates that they will disregard or ignore the analysis in favor of other assessments. As Robert Gates points out, in his experience having supported five presidential administrations, the usual response of a policy-maker to intelligence with which he disagrees or which he finds unpalatable is to ignore it. 44 Gates does not say they ignore it sometimes. Nor does he say they ignore it frequently. Instead, he says they ignore it as a matter of course; that ignoring inconvenient intelligence analysis is the usual state of affairs.

8 INTELLIGENCE AND NATIONAL SECURITY 731 Some might argue that policy-maker rejection of intelligence analysis is evidence of a corrupted or at least inefficient or ineffective decision-making process. But this conclusion would not be justified as there is not yet any evidence that policy assessment in general is any less accurate than intelligence analysis. Examples of policymakers making more correct assessments of situations can be found in their rejection of inaccurate warning. Intelligence analysts in general irrespective of the more specialized dedicated warning function have an incentive to provide warning of harm to national security resulting from the downside risk or negative consequence of both action and inaction. Policymakers frequently and appropriately discount or disregard these warnings, as acting on them would lead to expensive policy failures. As Erik Dahl says, In some cases intelligence might be wrong, and the result of policy-maker receptivity could be the failure of policy or even disaster. Mark Lowenthal has pointed out that in some cases military leaders do well by failing to pay attention to intelligence (that is, by being unreceptive). 45 This kind of correct rejection of inaccurate intelligence analysis is only infrequently captured in the literature, yet provides a significant counterpoint to the charge that ignoring intelligence analysis produces policy failure. While specific instances can be identified where policy-maker disregard of intelligence analysis led to policy failures, there are many other instances in which policymakers rejected intelligence analysis warnings which if acted upon would have led to other kinds of policy failures. The rejection of the inconvenient fact or unwanted interpretation in and of itself is not evidence of a flawed decision-making process. How policymakers use intelligence analysis Taken all together, this means that intelligence analyses frequently does not matter in terms of policy analyses. Viewed from the perspective of decisionmakers, intelligence analysis serves as a check on their judgment; if it confirms what they already think of a situation, then it is both confirmatory and redundant. 46 But if the intelligence analysis provides a new interpretation or assessment, then the decisionmakers must decide whether to use that different interpretation or assessment as a way to revise their own analysis. The policymakers subsequent receptivity to this different interpretation or assessment can be highly variable. 47 Frequently policymakers will choose their own interpretation over that of intelligence analysts for any of a wide variety of legitimate reasons, making the intelligence analysis appear as if it were ignored. In only a subset of cases does the conventional explanation apply; that intelligence analysis is rejected due to inappropriate policy-maker commitments to policy preferences and biases. In contrast to those who might view policy-maker disregard of intelligence analysis as one of the pathologies of intelligence-policy relations, this duplication suggests that it really may be the normal state of affairs in foreign policy decision-making. When the evaluative and assessment process is deficient at the level of policy, intelligence analysis can make a positive contribution to policy. When policymakers might have misunderstood something that they did not have the time to assess, intelligence analysis can make a positive contribution to policy. When policy assessment is shaped by motivated bias, intelligence analysis may come to a different answer. But what matters here is not the actual analysis itself, but the differential between the intelligence analysis and the assessment or assessments made by the decisionmakers themselves. The place to start understanding the impact that intelligence analysis has on a decision is not with the intelligence available or the analysis of it, but rather decision-maker understandings of what is happening, why it is happening, and what they are planning to do about it. Explaining historical cases of intelligence and foreign policy Strategic intelligence analysis has a role to play in national security decision-making, but it is a much more limited role than the conventional explanation implies. The reason for this much more limited role is because legitimate differences of interpretation frequently lead decisionmakers discount the

9 732 S. MARRIN evaluation embedded in intelligence assessments. This is borne out through study and evaluation of historical cases. Persian Gulf War ( ) In the case of the Persian Gulf War, strategic intelligence analysis had limited influence mostly because it was consistently incorrect and disregarded in favor of more accurate policy analysis. Specifically, intelligence analysis first incorrectly assured policymakers that Iraq would not invade Kuwait via a 1989 NIE which said that Iraq would not attack any of its neighbors for two to three years. 48 Then, while warning of Iraqi troops massing on the Kuwaiti border during the summer of 1990, intelligence assessments incorrectly suggested that the Iraqi presence was an effort to put pressure on Kuwait for negotiating purposes and that they would not invade. 49 Just before the invasion, the National Intelligence Officer for Warning Charlie Allen warned that an invasion might be imminent, but this warning was not provided early enough to implement policies which would deter or prevent it. 50 Even after the Iraqi invasion of Kuwait, US intelligence analysts believed that war was not forthcoming. Instead, in late 1990 they suggested that Iraq would withdraw from Kuwait rather than engage with invading US forces. 51 None of this intelligence analysis had much of an influence on American foreign policy. In this case, Paul Wolfowitz the Undersecretary of Defense for Policy during the Gulf War has said that the December 1989 NIE on Saddam Hussein s intentions did not influence policy one way or the other. 52 Instead, he says that by October or November 1989 policymakers were already debating how to respond to a possible Iraqi invasion of Kuwait and Saudi Arabia. The reason they did this was not based on intelligence assessments, but instead as part of a strategic planning process with a five-to-ten year time frame and recognizing the reasonable range of uncertainty about what could happen. Later intelligence analyses were subsequently disregarded in favor of the analyses produced in the policy planning process as well. The reason intelligence analysis did not have influence here is not because of cognitive bias or political bias, but because the intelligence assessments were redundant. In this case, Wolfowitz suggests that intelligence agencies do not have any intrinsic advantage in terms of their ability to assess this kind of uncertainty. As he put it in terms of the 1989 NIE, it would be totally unreasonable to expect an NIE at that time to have predicted what (Saddam Hussein) was going to do. I do not think Saddam himself knew what he was going to do. It is clear that most of the leaders of the Arab world, who had a very large stake in predicting what he was going to do, had no sense of what was coming. 53 Given this level of uncertainty and the significant consequences involved, Wolfowitz suggested that you would have to conclude this (the Iraqi invasion of Kuwait) was a serious possibility. After all, we were dealing in a world where ten percent possibility was something we might plan for and, in fact, we did. That it turned out to be a hundred percent possible was not foreseen. 54 What Wolfowitz is suggesting is that policymakers had already considered the prospect of Iraqi invasion of Kuwait and were taking it seriously prior to intelligence assessments which said such an eventuality was unlikely. While an intelligence analyst or an observer at the time at the time might have attributed this policy-maker disregard of intelligence analysis to cognitive biases expecting or perhaps even desiring war with Iraq, history has subsequently shown that different interpretations are possible from the same set of information and that sometimes international actors are more threatening than intelligence analysts expect them to be. The duplication of analysis in both the intelligence analysis and policy domains led to a differential in assessment. The policymakers chose to act on their own more alarmist interpretation, which in this case appears to have been the more accurate one. This may be a case where the independent, objective intelligence assessments were incorrect relative to assessments that policymakers either acquired from other sources or contingency plans they were developing for themselves.

10 INTELLIGENCE AND NATIONAL SECURITY 733 Yugoslavia (1990) Another case of strategic intelligence analysis that did not influence policy was the 1990 NIE that predicted Yugoslavia would collapse into violence and ultimately fall apart. 55 As Gregory Treverton has said, this NIE predicted Yugoslavia s tragedy with a prescience that is awe inspiring. (The NIE) concluded that Yugoslavia s breakup was inevitable. The breakup would be violent, and the conflict might expand to spill into adjacent regions. this analysis actually predicted that future. 56 But, as Treverton goes on to say, Yet so far as I can tell, the document had no effect (on policy). None. A 2003 case study written by Thomas Shreeve for the Intelligence Community s Case Method Program comes to the same conclusion. 57 A standard interpretation of this failure of accurate intelligence analysis to influence decision would attribute this to the cognitive preconceptions of policymakers. Shreeve implies this is the case when he says that the NIE s assessment was also fundamentally inconsistent with what US policymakers wanted to happen in the former Yugoslavia. 58 As Treverton suggests, the senior levels of the Bush Administration s foreign policy machine were preoccupied with other issues, so being told of one more disaster lurking in the wings was unwelcome information. It was news they did not want to hear. Hoping that Yugoslavia might somehow stay together was convenient. 59 The implication is that they rejected the warning due to a combination of cognitive biases and political predispositions. But a different interpretation is possible based on the duplication of analysis in both intelligence and policy domains. As Treverton suggests, some policy officials recall agreeing that any breakup of Yugoslavia was bound to be violent, indeed extremely violent. Thus, for them the estimate s foreboding was on the mark but not news. 60 A senior member of the National Security Council staff agrees with this; that the NIE wasn t news for those of us who had been following developments there. It was not a surprise. 61 It was not a surprise because they had come to much the same conclusions. State Department reporting was already highlighting how nationalistic forces were putting pressure on the political stability of the country, and leading to pressures to fracture. 62 The reason it appears that the intelligence analysis did not matter was because the assessment it contained was redundant. Intelligence analysts do not like it when their assessments fail to produce changes in policy. Shreeve quotes a former Deputy National Intelligence Officer for Eastern Europe criticizing the policy response to the warning. As he said, The United States simply stopped caring about Yugoslavia If Yugoslavia had fallen apart without bloodshed, we would have seen no US interest at all. It is only atrocities that catch our attention. 63 But as John Gannon points out, it is important to remember that the break-up was not in fact inevitable. In retrospect, analysts may not appreciate that their certainty was not entirely justified. The negative outcome the violent dissolution of Yugoslavia happened not because policymakers ignored the intelligence analysis or failed to appreciate that it might happen, but rather because they failed to implement policies which prevented that outcome. Why did they fail? Perhaps, as Shreeve points out, their time, attention, energy and resources were being devoted to the Middle East and addressing the Iraqi invasion of Kuwait. 64 Or perhaps, as a result of the end of the Cold War, Yugoslavia s strategic importance was much less than it used to be. Or perhaps policy officials thought there was nothing they could do, so did little. Shreeve quotes a US government official as saying that policymakers believed there was nothing they could do. Brent Scowcroft, the National Security Advisor at the time, said much the same thing. 65 The US Ambassador to Yugoslavia found this to be frustrating and tried to prevent the NIE from becoming a self-fulfilling prophecy by pushing for other policy outcomes which in the end failed as well. Contrary to the prevailing explanation, this NIE was not disregarded due to policy-maker commitments or cognitive biases. Instead, its conclusions were the same as those of other policy assessments, making it redundant in the decision-making process. That some conclude it was ignored because it did not produce a change in policy is based on an inappropriate normative expectation that intelligence analysis will be both a precursor to and foundation for policy. In this case, policy failed not because it

11 734 S. MARRIN ignored the intelligence analysis but rather because the policymakers could not come up with a policy which would prevent the negative outcome. Terrorist threat before 9/11 attacks Another case where strategic intelligence analysis did not significantly influence or change policy for similar reasons was strategic assessment of the threat from Al Qaida before the 11 September 2001 terrorist attacks. In 1995 and 1997 two NIEs had been produced on the terrorist threat, but neither of these analytic products appears to have had any impact on decision-making. The 9/11 Commission Report said that the 1995 NIE warned that this danger would increase over the next several years and predicted future terrorist attacks against the United States and in the United States. 66 This NIE said that the greatest threat stemmed from the network of Islamist groups that had formed during the jihad against the Soviet Union in Afghanistan and that they were looking at targets in both Washington DC and New York City with civil aviation as a vulnerable and attractive target. 67 There are no indications that this NIE led to any changes to counterterrorist policy. Nor is there any evidence that the NIE produced in 1997 The Foreign Terrorist Threat in the US: Revisiting Our 1995 Estimate had any impact on decision-making either. 68 The reason these NIEs appear not to have had much influence on policy was not because they were disregarded or ignored by policymakers, but rather because the Clinton Administration already appreciated the threat posed by al Qaeda and was in the process of implementing policies to ramp up counterterrorism efforts. By the time the 1995 NIE was produced, decision-makers were already sensitized to the threat of terrorism and had decided to address the terrorist threat through Presidential Decision Directive 39 (PDD-39) US Policy on Counterterrorism which, according to the 9/11 Commission Report, said that the United States should deter, defeat and respond vigorously to all terrorist attacks on our territory and against our citizens. 69 Unfortunately, these counterterrorism policies were ineffective to the point that officers at all levels questioned the effectiveness of the most active strategy the policymakers were employing to defeat the terrorist enemy. 70 Even Richard Clarke, the strongest proponent of more robust counterterrorist policy, admits that his policy advice, even if it had been accepted immediately (in 2000) and turned into action, would not have prevented 9/ This statement from Clarke is significant, demonstrating that the national security decision-maker most responsible for coordinating counterterrorist policy before 9/11 did not believe that even the most robust of planned policies were adequate as a way of addressing the threat. This failure of the Clinton Administration to respond effectively to the terrorist threat is not a result of either inadequate intelligence analysis or disregard of intelligence analysis, but rather because constraints on policy prevented them from being able to act effectively. As a general observation, Michael Herman a former chairman of the British Joint Intelligence Committee, and professor at Oxford University has said, Even if intelligence points to what should be done it may be impossible to do it; the best warnings of attack may lead to no action if there is nothing that can be done. 72 This can help explain the failure of pre-9/11 counterterrorism policies. As Joshua Rovner and Austin Long suggest, elected officials from both parties were unwilling to take the unpalatable but necessary steps that, with hindsight, might have stopped bin Laden. This is the real failure of September A surface-level evaluation consistent with the conventional explanation for policy disregard of intelligence analysis could attribute the Clinton administration s failure to be receptive to intelligence analysis warning them of the threat from al Qaeda to a hope that current counterterrorism policies were effective and succeeding. As with the Yugloslavia case, such a belief would have been convenient, and consistent with both policy preferences and cognitive biases. But the reality is more complicated. Policy analysis of the threat of terrorism led to the initiation of more robust counterterrorism policies, while intelligence analysis (the NIEs) which continued to highlight the threat provided assessments which were redundant. Similar to the Yugoslavia case, the primary policy failure associated with the 2001 attacks

12 INTELLIGENCE AND NATIONAL SECURITY 735 was not due to a failure of intelligence analysis or policy disregard of intelligence analysis, but rather because policymakers did not implement policies which would have prevented the negative outcome. Iraq War (2003) The Iraq War also demonstrates the limited influence that strategic intelligence analysis can have on American foreign policy. For example, strategic intelligence analysis which had limited influence on policy include the 2002 NIE on Iraq s weapons of mass destruction and the two 2003 NIC assessments on the likelihood of postwar problems. 74 That the 2002 NIE on Iraqi WMD failed to influence policy should not be a surprise; it was not requested by the Bush Administration, but rather by Congress. While it stimulated debate in the halls of Congress, there is no evidence that the 2002 NIE had any impact on the process of making the decision to go to war in Iraq. As Mark Lowenthal the Assistant Director of Central Intelligence for Analysis and Production and Vice Chairman of the NIC at the time the NIE was produced has said, this NIE had virtually no effect on anyone s decision about whether to go to war. 75 The 2002 NIE assessment that Iraq possessed WMD was not influential because it was consistent with intelligence assessments that had been produced over the past 10 years. That this assessment was communicated with greater certainty and confidence than may have been justified is a problem, but even a more nuanced, accurate assessment would have concluded that Iraq probably possessed WMD. 76 The reason this NIE did not have any obvious influence on decision-making is because its conclusions were already known and incorporated into policy planning. As a result, their findings were consistent with the analyses of policymakers, thus making their findings redundant. As for the two 2003 national intelligence assessments, Paul Pillar has criticized the Bush Administration for its failure to incorporate pessimistic assessments of post-war outcomes from two 2003 National Intelligence Council products into its deliberative process. As he pointed out, the Bush Administration went to war without requesting and evidently without being influenced by any strategic-level intelligence assessments on any aspect of Iraq. 77 Per the conventional explanation for the lack of influence of intelligence analysis, perhaps the national security leadership chose to ignore these warnings due to a desire not to consider the possibility of a negative outcome along combined with its commitment to a policy of military invasion. Alternatively, perhaps these 2003 intelligence assessments did not influence policy because, while they warned that potentially serious problems could develop in Iraq as a result of the invasion, these kinds of warnings were incorporated into and addressed in the policy planning process. As President Bush has said, efforts were taken to prepare for the postwar period and provide security after Saddam, but at the same time we were aware of our limitations. 78 That the Bush Administration decided to invade despite the warnings from intelligence analysts does not necessarily indicate that the warnings were ignored. Instead, as occurred in the case of Yugoslavia and counterterrorism policies before 9/11, efforts were made to address the problems but those efforts were not effective in preventing the negative outcome. This limited influence of strategic intelligence analysis on policy has received a lot of criticism because, as previously established, for those who possess normative expectations that policymakers should rely on intelligence analysis in their decision-making processes, when decisions are made that disregard intelligence analysis it can be perceived as indicative of a corrupted decision-making process. But rather than attribute the lack of influence of intelligence analysis to policy-maker cognitive bias and political predisposition, a much simpler explanation is that the Bush Administration decided that the risk of inaction had a greater chance of hurting national interest than the possible costs of action taken to protect national security. In retrospect this decision may have been mistaken given the combination of inaccurate intelligence on Iraqi possession of WMD and the post-war problems that resulted. That policymakers chose to rely on what in retrospect appear to have been less accurate analyses over more accurate analyses is tragic but not necessarily indicative of a corrupted decision-making process.