29 October 2012
Flawed estimates of Iraq’s supposed chemical, biological and nuclear weapons were the result of a collapse in intelligence and policy community relations, argues Joshua Rovner. He attributes this collapse to the partisan political pressure that built up prior to the 2003 invasion of Iraq – a buildup which prompted intelligence agencies to transform worst-case scenarios into most-likely estimates.
By Joshua Rovner for Centre for International Policy Studies (CIPS)
Editor's note: Some of the West’s most high-profile intelligence agencies began the 21st century inauspiciously. Not only were they accused of failing to anticipate an upsurge in transnational terrorism, there were also fears that their relationships with their political task masters had been seriously damaged in the aftermath of the 2003 invasion of Iraq. This latter fear, however, was not totally justified. Both the intelligence agencies and the political elites they served were quick to acknowledge that they both needed to adjust to 21st century realities.
As part of our week-long focus on how intelligence agencies are adapting to a new range of security threats, today we present an overview of the problems these organizations confront and some of the adjustments they have made (or are trying to make) to address them. We begin with an article written by Joshua Rovner that traces the breakdown of relations between US policymakers and their intelligence agencies in 2002-2003. The case is of course an infamous one – it turned on the critical question of whether Iraq’s chemical, biological and nuclear weapons facilities existed or not. While relations were “poisoned by politicization” prior to the invasion of Iraq, Rovner contends that efforts to improve the dialogue between both parties have been slowly progressing in a positive direction.
Our second article assesses some of the practical measures that have been introduced in the aftermath of the US intelligence agencies’ failure to predict the 9/11 and other terrorist attacks. Not only have governments poured extra resources into intelligence operations, they have also endowed their agencies with enhanced legal powers. Many intelligence organizations have also been reformed in order to encourage greater intra- and inter-agency information sharing. The Center for Security Studies’ (CSS) Prem Mahadevan describes some of these changes in detail and measures the costs and benefits associated with the widespread reform of Western intelligence organs.
We know a lot about intelligence failure. Scholars, journalists, and government commissions have described many reasons why intelligence agencies fail to provide accurate estimates of looming strategic threats, and why they fail to provide warning of impending disasters. One reason intelligence analysts make mistakes is that they usually work with partial and dubious information. Another reason is that analysts are human, prone to the same psychological problems as anyone else. Individuals are not terribly good at interpreting new information that is inconsistent with their beliefs or expectations, and this makes intelligence analysts vulnerable to surprise if they downplay data that falls outside their worldview. Teamwork can be an antidote to these psychological biases, but it can also lead to groupthink. Bureaucratic politics and organizational pitfalls are further sources of failure. Intelligence agencies are bureaucracies, and like all bureaucracies they are interested in wealth, autonomy, and prestige. Such incentives may encourage them to deliver estimates that tend to support their own bureaucratic preferences, even if those conclusions are inaccurate. Finally, a failure to coordinate the activities of multiple intelligence agencies can prevent accurate assessments based on shared information.
But suppose intelligence agencies overcome all of these problems. Imagine a scenario in which they are not hamstrung by the psychological biases of individual analysts, the perils of groupthink, or the bureaucratic and organizational roadblocks to cooperation. Imagine they gain access to a treasure trove of relevant and reliable information, and they interpret that raw data through a thoroughly rational and coordinated inter-agency process. Imagine that at the end of this process they deliver the perfect intelligence product: an accurate, timely, comprehensive, and reliable strategic estimate about a critically important policy problem. Even then they are not finished. Indeed, they will not have succeeded unless they can find a receptive consumer, because even the best intelligence is irrelevant if policymakers are hostile or indifferent. Intelligence is less important than intelligence-policy relations.
Healthy intelligence-policy relations are characterized by candor. Intelligence officials have the freedom to deliver discomfiting conclusions to policymakers without fear of reprisal. It is not easy being the bearer of bad news, especially on high-stakes issues. Ideally, intelligence officials can expect that their conclusions will be received at face value and will stand a reasonable chance of informing the policy debate. This is not to say that they will be automatically accepted or that policy will move in lockstep with intelligence judgments. It is only to say that intelligence officials can be objective and forthright without fear of being punished for their views. At the same time, policymakers will have the freedom to offer constructive criticism of intelligence without being accused of politicization. Leaders may reasonably criticize intelligence for being tardy with key judgments on pressing issues; for presenting estimates in ways that encumber rather than facilitate the decision-making process; or for asking the wrong questions. In some cases they may rightly conclude that intelligence estimates are the product of sloppy analysis based on unreliable information. These are all legitimate complaints and policymakers should not fear that they will be accused of politicization for making them.
Some amount of friction is normal in intelligence-policy relations. Policymakers are not always receptive to bad news, and they often respond badly. Intelligence officials are not always receptive to criticism, and they often suspect policy meddling. There are a number of other sources of friction. Policy generalists demand timely answers and clear judgments so they can make decisions on short notice; intelligence specialists may shy away from unequivocal conclusions without having much time for consideration. Policymakers tend to be self-confident and action-oriented; intelligence analysts tend to be more reserved and conscious about the complexity of issues and the uncertainty of international politics. For these reasons and others, intelligence-policy friction is the norm.
A little friction is not a bad thing, because too much harmony will encourage overconfidence and discourage reassessment. But there are some instances in which normal friction boils over. Extraordinary deviations from this baseline are what I call the pathologies of intelligencepolicy relations. Two are particularly important. The first is neglect, which describes cases in which policymakers ignore intelligence out of hand or only pay attention to estimates that support their expectations or beliefs. Intelligence officials can also contribute to neglect by distancing themselves from policy circles out of fear of being corrupted. Extreme neglect is a serious problem because it removes a potentially important source of information and insight from the policy process. Intelligence also functions as a check against wishful thinking. If nothing else, neglect leads to waste as intelligence agencies continue to collect information and churn out estimates that no one will ever read.
The second major pathology of intelligence-policy relations is politicization, defined as the manipulation of estimates to reflect policy preferences. Policymakers can manipulate intelligence directly through crude arm-twisting, or they can do it indirectly by sending subtle signals to intelligence officials about what they expect to hear. Intelligence officials can also politicize intelligence by allowing their own policy preferences to color their analyses.
Politicization does enormous damage. Politicized intelligence estimates tend to reach confident conclusions even when the underlying data is thin. This is a recipe for dramatic errors. Worse, politicization discourages reassessment—even if better information becomes available—because that would constitute an admission of earlier failure. Politicization also does lasting harm to the quality of intelligence-policy relations. Episodes of politicization reinforce mutual negative stereotypes: policymakers increasingly believe that intelligence officials are bureaucratic enemies, and intelligence officials increasingly believe that policymakers are all bullies. These suspicions can persist for years after the fact.
Leaders ignore intelligence for simple reasons. They might indulge the psychological need for cognitive consistency and ignore intelligence that cuts against their beliefs. Or they might just be unsatisfied with the quality of intelligence and choose to trust their own sources and their own instincts. Whatever the reason, leaders are under no procedural or legal obligation to listen to intelligence. Politicization is somewhat more puzzling, because public revelations of meddling can have serious political consequences. If they don’t like intelligence, they can ignore it. Why would leaders ever take the risk of a scandal by trying to change it? The answer has to do with domestic politics. Leaders politicize intelligence when they make strong public commitments in the face of substantial domestic opposition. In these cases, leaders have strong incentives to present an image of consensus support from across the national security establishment. Choreographing a coterie of military officers and diplomats helps strengthen policymakers’ arguments in front of skeptical audiences. Intelligence officials are particularly important to the consensus because intelligence agencies control secret information. The imprimatur of intelligence allows policymakers to claim that they are acting on the basis of the best possible information and thus deserve the benefit of the doubt. Intelligence is a very persuasive advocacy vehicle, and policymakers may be tempted to exploit the aura of secrecy in order to overcome domestic opposition.
There is no need to manipulate intelligence that is already consistent with policy preferences. But controversial issues are usually open to multiple interpretations—that is why they are controversial— and intelligence is likely to reflect this uncertainty and disagreement. While admitting uncertainty might be intellectually honest, it is intolerable for policymakers who are trying to win domestic battles. They need intelligence to offer clear support for their positions, not pusillanimous and hedging judgments. So instead of accepting intelligence at face value, they will pressure it to deliver strong conclusions that align with policymakers’ public rhetoric.
This process explains much about the intelligencepolicy debacle before the war in Iraq.3 U.S. intelligence before the summer of 2002 usually assumed that Saddam Hussein still harbored dreams of an arsenal of chemical, biological, and nuclear weapons. At the same time, most estimates found that Iraq had few if any existing chemical and biological weapons, and they concluded that Iraq’s nuclear program would have to be rebuilt. Intelligence analysts were very candid about the poor quality of information coming from Iraq, especially after the departure of UN weapons inspectors in 1998. And they were forthright about different interpretations within the intelligence community. While nobody was naive about Saddam Hussein, intelligence analysts were careful not to claim too much about what was going on in Iraq.
All that began to change in the summer of 2002. Earlier in the year, the Bush administration had begun signaling that it was committed to regime change in Iraq—by force if necessary. President Bush had reason to believe that this would not arouse much public opposition, given his extraordinarily high approval ratings after the 9/11 attacks. In the summer, however, Senate Democrats began questioning the rationale for going to war with Iraq. Democrats were given political cover by prominent conservatives like Brent Scowcroft, the former National Security Advisor for President George H.W. Bush, who published a widely read anti-war op-ed in the Wall Street Journal.
Suddenly the president faced serious domestic opposition to his strong public commitment to regime change in Iraq. At this point, the administration began pressuring the intelligence community to exaggerate the threat of Iraq’s supposed unconventional weapons program. It subsequently used that intelligence to bolster the case for war.
Months of politicization had large effects on intelligence. By the fall, the intelligence community was painting a much more ominous portrait of Iraqi capabilities. In October, the community released a National Intelligence Estimate (NIE) that increased the amount of suspected Iraqi chemical agent to as much as 500 tons, including mustard gas, sarin gas, cyclosarin and the nerve agent VX. No previous estimate had claimed that Iraq had more than 100 tons in storage. The NIE also provided the first definitive communitywide claim Iraq possessed any biological weapons, and it warned that Saddam’s operatives could disperse them though a frightening array of “bombs, missiles, aerial sprayers, and covert operatives.”4 Finally, the estimate concluded that international sanctions were fraying so badly that Iraq might be able to acquire a nuclear weapons capability by 2010, if not sooner. None of this was based on new information. Instead, the stark change in the content and tone of intelligence occurred because policy pressure encouraged analysts to transform their worst-case scenarios into mostlikely estimates.
Politicization also inhibited reassessment after UN and IAEA inspectors returned to the country in late 2002. Inspectors scoured the country in the months before the war began in March 2003, conducting hundreds of site visits and no-notice inspections in order evade Iraqi concealment. The inspectors also brought advanced technologies like ground penetrating radar in order to assess suspected underground storage facilities. They found nothing. While Iraq had produced some ballistic missiles in excess of allowed range limits, inspectors found no sign of a reconstituted chemical, biological, or nuclear weapons program. What they found was not consistent with intelligence estimates from the previous year. But their reports had no effect on U.S. intelligence agencies, and there was no serious effort to reassess Iraqi capabilities. Under pressure from policymakers, intelligence officials had published bold estimates of the Iraqi threat, and they were in no mood to reconsider.
Intelligence-policy relations suffered in the aftermath of the war. Intelligence officials blamed policymakers for inflating Iraqi capabilities, without questioning their own flawed assumptions about Iraqi intentions. Policymakers criticized the community for overselling the quality of intelligence, without questioning their own role in exaggerating the estimate. Anecdotal evidence suggests that the atmosphere has improved somewhat over the past few years, but mutual suspicion and doubt still lingers. As in past cases of politicization, it will take a long time to return relations to a state of normal friction.
Joshua Rovner is Associate Professor of Strategy and Policy at the US Naval War College. He is the author of the recent book, Fixing the Facts: National Security and the Politics of Intelligence (Cornell University Press, 2011).
"Intelligence, Policy, and the War in Iraq" was originally published as CIPS Policy Brief No 18, June 2012. Republished with permission.