Issues for the US Intelligence Community
Collection and Analysis on Iraq
“The quality of intelligence will be improved only by fundamental changes at the grass roots level.”
Editor’s Note: This is the third in a series of reports supporting the Director of Central Intelligence’s evaluation and critique of intelligence and analysis associated with the war in Iraq. It was prepared on contract by the Kerr Group, a team of former senior intelligence officers. The Group's first report, a classified study, was a documentation of the Intelligence Community's judgments before the war. It characterized the intelligence process, product content, and analytic shortcomings but was not a commentary on the accuracy of those judgments. The second report, also classified, reviewed the intelligence used to support judgments regarding weapons of mass destruction in Iraq. Specifically, it reviewed the reporting used to develop the National Intelligence Estimate “Iraq's Continuing Programs for Weapons of Mass Destruction,” published in October 2002. The third report, which follows, was an unclassified study presented in July 2004. In it, the team presents an assessment of the performance of the Intelligence Community from a broad perspective, focusing on systemic issues that channeled analysts’ evaluations and analyses. Its observations and recommendations continue to have relevance as the Community evolves.
* * *
The Intelligence Community's uneven performance on Iraq from 2002–4 raised significant questions concerning the condition of intelligence collection, analysis, and policy support. The discussion of shortcomings and failures that follows is not meant to imply that all surprises can be prevented by even good intelligence. There are too many targets and too many ways of attacking them for even the best intelligence agencies to discover all threats in time to prevent them from happening. Nonetheless, improving performance requires an acknowledgement of past mistakes and a willingness to change.
This report was prepared at a time of a great rush to reorganize and give the leader of the Intelligence Community new authorities. That probably was a necessary activity. However, to move the organizational boxes and to offer new authorities are not the only answers or perhaps even the best answers. Based on our experience and what we learned during this review, the Group believes that the quality of intelligence will be improved only by fundamental changes at the grass roots level. That is, changes in collection, analysis, the nature of the product, and interaction with policymakers and other customers.
The Intelligence Community itself has made some useful changes and recommended others. Several fixes also have been proposed from outside the Community, which might be helpful but do not address some of the core problems identified by the Group. This report focuses on the question: Does the Community's flawed performance on Iraq represent one-time problems, not to be repeated, or is it symptomatic of deeper problems?
Principal Findings of the Earlier Reports
The central focus of national intelligence reporting and analysis prior to the war was the extent of the Iraqi programs for developing weapons of mass destruction (WMD). The analysis on this issue by the Intelligence Community clearly was wide of the mark. That analysis relied heavily on old information acquired largely before late 1998 and was strongly influenced by untested, long-held assumptions. Moreover, the analytic judgments rested almost solely on technical analysis, which has a natural tendency to put bits and pieces together as evidence of coherent programs and to equate programs to capabilities. As a result the analysis, although understandable and explainable, arrived at conclusions that were seriously flawed, misleading, and even wrong.
Intelligence produced prior to the war on a wide range of other issues accurately addressed such topics as how the war would develop and how Iraqi forces would or would not fight. It also provided perceptive analysis on Iraq's links to al Qa'ida; calculated the impact of the war on oil markets; and accurately forecast the reactions of the ethnic and tribal factions in Iraq. Indeed, intelligence assessments on post-Saddam issues were particularly insightful. These and many other topics were thoroughly examined in a variety of intelligence products that have proven to be largely accurate.
The national intelligence produced on the technical and cultural/political areas, however, remained largely distinct and separate. Little or no attempt was made to examine or explain the impact of each area on the other. Thus, perspective and a comprehensive sense of understanding of the Iraqi target per se were lacking. This independent preparation of intelligence products in these distinct but interrelated areas raises significant questions about how intelligence supports policy. In an ironic twist, the policy community was receptive to technical intelligence (the weapons program), where the analysis was wrong, but apparently paid little attention to intelligence on cultural and political issues (post-Saddam Iraq), where the analysis was right.
With respect to the weapons programs, some critics have argued that the off-the-mark judgments resulted largely from reinforcement of the Community's assumptions by an audience that was predisposed to believe them. This, however, seems to have been less a case of policy reinforcing “helpful” intelligence judgments than a case of policy deliberations deferring to the Community in an area where classified information and technical analysis were seen as giving it unique expertise.
On the other hand, the Intelligence Community's analysis of post-Saddam Iraq rested on little hard information, was informed largely by strong regional and country expertise developed over time, and yet was on the mark. Intelligence projections in this area, however, although largely accurate, had little or no impact on policy deliberations.
The bifurcation of analysis between the technical and the cultural/political in the analytic product and the resulting implications for policy indicates systemic problems in collection and analysis. Equally important, it raises questions about how best to construct intelligence products to effectively and accurately inform policy deliberations.
Any examination of the Intelligence Community must acknowledge the impact of more than 10 years of turmoil that adversely affected all collection and analytic efforts, including those on Iraq. The Intelligence Community was designed to focus on the Soviet Union. It had developed a single-minded rigor and attention to detail that enriched its analysis, particularly with respect to Soviet military issues. The end of the Cold War, however, brought to a close that “stable” bipolar world and left the United States without a principal enemy. Although never perfect, the Intelligence Community's analytic efforts against the Soviet threat were generally insightful and its collection largely effective, reflecting the accumulation of deep understanding developed over many years.
Absent this singular focus, in the post-Cold War environment the Intelligence Community struggled to reestablish its identity and purpose in what had become a world of multiple crises and transient threats. The effort to define its priorities was further complicated as policymakers and others raised questions not only about the role of but even the need for intelligence. Accordingly, intelligence came to be seen as an area where the government could reap resource savings. The resulting cutbacks in collection (technical and HUMINT) and analytic resources had a significant adverse impact on Intelligence Community capabilities.
Nonetheless, during the 1990s the Intelligence Community confronted numerous crises in which to demonstrate the relevance of intelligence analysis to policy deliberations. Regional conflicts, such as the first Gulf war and follow-on sanctions against Iraq, the breakup of Yugoslavia, and emerging threats from North Korea and Iran provided tests for intelligence. The Community's collection and analysis performance over this period, however, was seen as inconsistent and sometimes faulty, leaving important customers still wondering about the relevance of the intelligence input to policy deliberations.
A significant contributor to this uneven performance was, and still is, the Community's tendency to establish single-issue centers and crisis-response task forces. By stripping expertise from regional offices, they diminish the overall ability to provide perspective and context for those issues. The resources seldom get returned to the line offices, which historically have been better equipped to provide complete perspectives on country and regional issues.
Although resources increased marginally over the decade, they were not as robust or focused as the capabilities devoted to the Soviet Union and were seen by the Intelligence Community as inadequate to deal conclusively with the multiplicity of threats. Accordingly, the Community in critical situations has faltered in its analyses and failed to collect pertinent information. This has occurred over a length of time and across crises sufficient in number, quite apart from Iraq, to indicate systemic issues rather than just occasional missteps.
Collection Impeded and Misdirected
Intelligence collection against Iraq fell far short of the mark. The intelligence base for collection and analysis was thin and sketchy. The Intelligence Community had nothing like the richness, density, and detail that it worked hard to develop and became accustomed to having on Soviet issues during the Cold War. To a significant extent this resulted from the reduction over the past decade of the professional collection management cadre capable of integrating HUMINT, imagery, and signals intelligence capabilities into coherent strategies. This development was compounded by the increased separation of collection professionals from the analytic cadre who had been intimately involved in identifying collection gaps, needs, and priorities and developing collection strategies.
Placing these developments in a broader context, however, is important. Iraq was not the only significant intelligence problem facing the Community in the years immediately preceding the war. Counterterrorism and counterproliferation were given higher priority and absorbed much of the clandestine service’s capability and leadership attention. Weapons programs in both North Korea and Iran received higher priority than those in Iraq until late 2002. In Iraq, technical collection priorities emphasized coverage of the Iraqi air defense system in southern Iraq in support of US military operations and prevented collection on other important targets in Iraq.
A number of other factors added to the difficulty of clandestine collection on the Iraq target. The Iraqis took pains to carefully hide their WMD programs. People and operations were protected from US intelligence by a variety of methods, including isolating scientists and technicians involved in the programs and employing effective camouflage, concealment, and deception efforts. The Iraqis had learned well about US intelligence during more than 10 years of confrontation and war.
Nevertheless, collection of information on difficult targets is the core mission of intelligence and in the Iraq case it did not measure up. Many of the more sophisticated clandestine technical collection techniques did not produce results. The Iraq WMD target was given a high priority over more than a decade, even if not the highest. Still, the Intelligence Community did not have conclusive evidence on what the Iraqis were working on, what they had achieved, which programs were ongoing, who was working them, or what the doctrines for use might be. Conversely, the Community saw no evidence that WMD programs were slowed, put on hold, or even nonexistent. Nor did it understand why Saddam's devious and obstructionist behavior continued if, as he claimed, he had no stockpiles of banned weapons.
US intelligence collection strategies contributed to the problem. Looking for information on a particular subject with a preconception of what is needed is almost certain to result in data that reinforces existing assumptions. The Community directed its collection capabilities to filling in what it thought were gaps in information about WMD programs, monitoring progress, looking for new developments in weapons and delivery systems, and identifying efforts to acquire materiel and technology abroad. Based on the hard information collected by US military forces and UN inspectors during and following the first Gulf war, reinforced by subsequent bits of information, the Intelligence Community and the US defense establishment had little doubt that Iraq was continuing development of WMD.
Collection was not focused or conceptually driven to answer questions about the validity of the premise that the WMD programs were continuing apace. This problem is well illustrated by a comprehensive collection support brief describing intelligence needs published by the DCI Center for Weapons Intelligence, Nonproliferation, and Arms Control. It was published contemporaneously with the 2002 National Intelligence Estimate on WMD. The brief describes in great detail the information required to support analysis of Iraq's weapons programs. The intent of the brief was to expose gaps in knowledge about what was believed to be aggressive, ongoing Iraqi weapons programs. The revealed gaps in knowledge were not, however, raised as requirements to address what was not known nor did such gaps raise doubts about prevailing intelligence judgments.
Discussing largely space-based collection systems at an unclassified level is difficult, but a few observations are possible. Despite a wide variety of technical capabilities available to the US, these systems were able to provide accurate information on relatively few critical issues. Monitoring Iraqi reactions to inspections was informative as was reporting on Iraqi acquisition efforts. Technical collection lends itself to monitoring large-scale, widespread targets, a condition not met in the Iraqi case. Analysis of Iraq's WMD programs, therefore, provides an excellent case study for an assessment of the limitations of relying too heavily on technical collection systems with little acknowledgement of the political/cultural context in which such programs exist.
Accordingly, surprisingly little collection was directed against several key issues. Neglected
topics for collection included the social, cultural, and economic impacts on Iraq of nearly 20 years of war and 10 years of sanctions and isolation. Little attention appears to have been paid, for example, to collecting information on the oil-for-food program. Considerable speculation was voiced that several countries and individuals were profiting from this program. Despite the fact that many of the targets for this subject were outside Iraq, it received only sporadic attention.
Although collection itself was a problem, analysts were led to rely on reporting whose sourcing was misleading and even unreliable. In the case of US clandestine reporting, it too often used different descriptions for the same source, leading analysts to believe they had more corroborative information from more sources than was actually the case. In addition, some critical judgments were made on the basis of intelligence provided by foreign intelligence services. Some of those liaison sources were not available to the US, and some key information obtained from liaison proved to be false.
The Intelligence Community knows how to collect secret information, even though in the Iraq situation it did not perform this function well. On the other hand, the acquisition of “softer” intelligence on societal issues, personalities, and elites presents an even greater challenge. This latter information can be found in databases, but they are too often only accessible indirectly and with considerable effort. It may also reside in the minds of groups of people who are accessible but not easily approachable and who do not fall into the category of controlled agents. Although there is a strong argument that the clandestine service should not divert its attention away from collecting “secrets,” information on the stresses and strains of society may be equally, if not more, important. This type of information, however, does not fit with the reward system in the collection world and can be difficult to fully assess and to integrate with other information.
In the case of Iraq, collection strategies were weak and unimaginative and failed to get the richness and density of information required. A careful examination might have addressed the long-neglected question of the value added by the different types of intelligence—e.g., SIGINT and IMINT—relative to the resources devoted to them. Collection on Iraq also was the victim of inadequate funding and too intense competition between top priority targets. Finally, Iraq demonstrates that collection strategies must take into account that the absence of dangerous activity in a targeted country cannot be convincingly demonstrated in the presence of a secretive and devious regime. Or, put differently, collection strategies should recognize the extreme difficulty of requiring such a regime to prove the negative in the face of assumptions that it is dissembling. Overall, the Intelligence Community did not acquit itself well in developing collection strategies on Iraq.
Analysis Adversely Affected
No single act of omission or commission accounts for the inconsistent analytic performance of the Intelligence Community with regard to Iraq. It appears to be the result of decisions made, and not made, since the fall of the Soviet Union, which had an impact on the analytical environment analogous to the effect of the meteor strikes on the dinosaurs. Nothing was the same afterwards. In response to changed priorities and decreased resources, the Intelligence Community's analytic cadre underwent changes in both its organization and its methodological orientation. Perhaps the most significant change was the shift away from long-term, in-depth analysis in favor of more short-term products intended to provide direct support to policy. Done with the best of intentions, this shift seems to have had the result of weakening elements of the analytic discipline and rigor that characterized Intelligence Community products through the Cold War.
The kind of intellectual-capital-intensive analysis that traditionally and effectively preceded policy deliberations was unavailable because of the shift away from research-oriented analytic investments. In reviewing the national intelligence products associated with Iraq, we found that they too often dealt, seriatim, with a broad range of subjects but without extensive cross-reference, and with no attempt to synthesize a broader understanding of Iraq out of the many detailed pieces that were prepared. The absence of such a contextual effort contributed to assessments that failed to recognize the significance of gaps in collection that may have been more evident when viewed from a larger perspective.
The absence of a unifying analysis was also disguised by the rapidity and volume of interactions between intelligence and policy deliberations. Eagerly responsive to quickly developed policy requirements, the quick and assured response gave the appearance of both knowledge and confidence that, in retrospect, was too high.
Of all the methodological elements that contributed, positively and negatively, to the Intelligence Community's performance, the most important seems to be an uncritical acceptance of established positions and assumptions. Gaps in knowledge were left undiscovered or unattended, which to some degree is explainable by the absence of pervasive, intrusive, and effective collection in Iraq. Although many products were appropriately caveated, the growing need to caveat judgments to explain the absence of direct intelligence did not seem to provoke internal review within the Intelligence Community. Indeed, although certain gaps were acknowledged, no product or thread within the intelligence provided called into question the quality of basic assumptions, hastening the conversion of heavily qualified judgments into accepted fact.
As noted earlier, the growing use of centers also contributed to what was at best a problematic result. The Intelligence Community has generally considered centers a useful organizational concept to concentrate analytic and collection capabilities against a carefully defined target set or issue. They also have the effect, however, of drawing resources away from more broadly based organizations. The post-Cold War reductions throughout the Intelligence Community made this a critical but insidious factor. Analysis of Iraq's weapons of mass destruction thus became the purview of technically competent analysts, but as has been described elsewhere, their efforts were not leavened through review by more broadly based colleagues.
Finally, quality control was weakened. The extensive layers of critical management review that traditionally served to ensure both the validity and standing of finished intelligence products seem to have been ineffective in identifying key issues affecting collection and analysis. Allowing for a satisfying sense of voluminous production, and reflecting the approval of receptive consumers, the policy-heavy process provided positive feedback, while the narrowly focused internal architecture lacked the self-awareness that could otherwise have raised serious and timely warnings.
Interaction with the Policy Community
Few issues have engaged greater policymaker interest in intelligence than those concerning Iraq—particularly the questions of weapons of mass destruction and Saddam's links to al-Qa’ida. The demands for intelligence in the months leading up to the war were numerous and intense. The Intelligence Community responded to the overwhelming consumer demand with an ever-increasing stream of analysis— both written and oral. Neither means of communication, however, served the policy community as well as it might have.
In periods of crisis, when demands are high and response time is short, most written intelligence production is in the form of policy-driven memos and briefs and pieces written for daily
publications. The result of this narrowly focused and piecemeal intelligence flow is that it
neither fosters continuity of analysis nor provides a context within which to place seemingly unrelated information. In the case of Iraq, national intelligence did not provide a comprehensive picture of how the country functioned as a whole. The Intelligence Community has made substantial, although sporadic, efforts over the past decade and a half to explore better and more technologically advanced methods of communicating with consumers. The results, however, have been modest at best. The requirement to have background and contextual information available at the policymaker's fingertips in a timely fashion remains unfulfilled.
The policy community was also ill served by the national intelligence estimate (NIE) process. NIEs rarely represent new analysis or bring to bear more expertise than already exists in analytic offices; indeed, drafters of NIEs are usually the same analysts from whose work the NIE is drawn. Little independent knowledge or informed outside opinion is incorporated in estimative products. The preparation of an NIE therefore consists primarily of compiling judgments from previous products and debating points of disagreement. The Iraqi WMD estimate of October 2002 was characterized by all of these weaknesses and more. It was done under an unusually tight time constraint—three weeks—to meet a deadline for congressional debate. And it was the product of three separate drafters, each responsible for independent sections, drawing from a mixed bag of analytic product. Consistent application of analytic or evidentiary standards became next to impossible.
The fundamental question is whether national intelligence estimates add value to the existing body of analytic work. Historically, with few exceptions, NIEs have not carried great weight in policy deliberations, although customers have often used them to promote their own agendas. The time may have come to reassess the value of NIEs and the process used to produce them.
Oral communications have their own set of problems. While direct engagement with the policy community is essential for intelligence to have an impact, too close association with policy deliberations can be troublesome. In the case of Iraq, daily briefings and other contacts at the highest levels undoubtedly influenced policy in ways that went beyond the coordinated analysis contained in the written product. Close and continuing personal contact, unfettered by the formal caveats that usually accompany written production, probably imparted a greater sense of certainty to analytic conclusions than the facts would bear.
Some in the Intelligence Community and elsewhere hold the view that intense policymaker demands in the run-up to the war constituted inappropriate pressure on intelligence analysts. Although viewed in that context as a problem, serious pressure from policymakers almost always accompanies serious issues. The more relevant issue is how the Intelligence Community responded to the climate of policy-level pressure and expectations. Whether or not this climate contributed to the problem of inconsistent analytic performance, however, remains an open question.
The cases of WMD and Iraq's links to al-Qa’ida illustrate two different responses to policy pressure. In the case of al-Qa’ida, the constant stream of questions aimed at finding links between Saddam and the terrorist network caused analysts to take what they termed a “purposely aggressive approach” in conducting exhaustive and repetitive searches for such links. Despite the pressure, however, the Intelligence Community remained firm in its assessment that no operational or collaborative relationship existed. In the case of Iraq's possession of WMD, on the other hand, analytic judgments and policy views were in accord, so that the impact of pressure, if any, was more nuanced and may have been considered reinforcing. Although it is possible that in the absence of strong policy interest, analysts would have been more inclined to examine their underlying assumptions, it is unlikely that such examination would have changed judgments that were longstanding and firmly held.
The intelligence world is one of ambiguity, nuance, and complexity. Dealing with these elements is difficult in the world intelligence serves, where success or failure is the uncomplicated measure by which the Intelligence Community is judged. The controversies over Iraq intelligence can be expressed in the contrast between these two worlds: carefully crafted national intelligence that ultimately failed in its singular mission to accurately inform policy deliberations. This report, the result of over two years of review and consideration, reflects the same contrast. On the one hand, it recognizes the enormous efforts undertaken, the long hours and the intense debate. On the other hand, it describes failures and weaknesses that cannot be ignored or mitigated.
Failures of collection, uncritical analytical assumptions, and inadequate management reviews were the result of years of well-intentioned attempts to do the best job with the resources provided. Decisions were made and their potential risks weighed, but the outcome on important issues proved badly flawed. Recognition of these problems must bring a rapid response.
US intelligence is a robust, highly capable, and thoroughly motivated community that represents an invaluable asset to the nation and its citizens. It must reveal itself as sufficiently mature to both adapt to changing circumstances and counteract the evolutionary processes that have conspired to threaten its reputation and its ability to successfully perform its assigned mission. The alternative is unacceptable and unthinkable.
Richard J. Kerr served as Deputy Director of Central Intelligence; Thomas H. Wolfe as Director of the Office of Near Eastern & South Asian Analysis; Rebecca L. Donegan as Deputy Inspector General; and Aris A. Pappas as Assistant National Intelligence Officer for General Purpose Forces.