Leggi in ItalianoWhat happens when important decisions have to be made on the basis of information that is scarce, fragmentary or unreliable? The controversy over Iraq’s alleged weapons of mass destruction before the 2003 invasion remains one of the most instructive cases for anyone studying intelligence. It also offers a lesson that reaches well beyond that war.
In October 2002, a National Intelligence Estimate (NIE), bringing together the main assessments of the US intelligence community, concluded that Iraq had resumed its nuclear weapons programme and possessed chemical and biological weapons. Investigations conducted after the invasion did not substantiate those central assessments. How had the intelligence community arrived at conclusions expressed with such confidence?
Part of the answer lies in two official reports that I encourage students to read. The first is the US Senate Select Committee on Intelligence report of 9 July 2004, which examined the intelligence assessments produced before the war. The second is the 31 March 2005 report of the Commission on the Intelligence Capabilities of the United States Regarding Weapons of Mass Destruction. The latter was a presidential commission, not a congressional committee, and paid particular attention to shortcomings in intelligence collection and analysis.
The 2005 Commission made a striking observation: “But they could not analyze data that they did not have.” Behind this simple sentence lies a very concrete problem. US intelligence agencies had collected too little reliable, verifiable information about Iraq’s weapons programmes. Where the sources could not establish what was actually happening, analysts drew on what they knew about the past: Saddam Hussein had developed and concealed prohibited weapons programmes, had used chemical weapons and had obstructed international inspections. These were relevant precedents. They were not, however, sufficient proof that the same stockpiles or active programmes still existed in 2002.
This is where cognitive bias enters the picture. The Senate’s 2004 report explicitly referred to groupthink: a widely shared assumption that Iraq retained active weapons programmes came to shape how evidence was interpreted. Ambiguous information was treated as confirmation, while evidence pointing towards alternative explanations received less attention. Mechanisms intended to challenge prevailing assumptions were not used effectively. A working hypothesis, initially understandable, gradually acquired the status of a presumption that was increasingly difficult to question.
Consider the aluminium tubes that Iraq attempted to procure. Some analysts interpreted them as possible components for centrifuges used to enrich uranium. Yet an alternative explanation existed, supported by technical expertise within the US government: the tubes were more consistent with conventional rocket production. The issue was not whether a dual-use item could conceivably have a nuclear application. It was how much could reasonably be inferred from that item about Iraq’s actual activities and intentions.
The 2005 report identified another particularly revealing analytical error. At certain points, analysts effectively demanded evidence of the absence of weapons rather than seeking positive evidence of their presence. This distinction matters as a question of method. It should not be confused with Iraq’s legal obligations to disarm and cooperate under Security Council resolutions. The question here is how an intelligence service should test a factual hypothesis, not whether a state has fulfilled its international obligations.
The problem became more serious when intelligence assessments reached political decision-makers. The Senate report found that the NIE had not adequately conveyed the uncertainties surrounding its sources and conclusions. The later Chilcot Report of 2016, examining the United Kingdom’s involvement in the war, likewise concluded that the available intelligence had been presented with a degree of certainty that was not justified. Intelligence reliability is, of course, only part of the inquiry into a decision to use armed force. The legality of the use of force must be assessed separately under the United Nations Charter.
A reading exercise for students
I suggest a simple exercise. Take one of the conclusions of the 2002 NIE and reconstruct the reasoning that leads from the information collected to the final assessment. What evidence was actually available? Did it come from a direct source? Was it independently corroborated, or was it already an inference? What alternative explanations were possible? What information might have disproved the original hypothesis? And, above all, did the level of confidence expressed in the conclusion really match the quality of the underlying evidence?
These questions also matter when we study public decision-making, undertake academic research or use artificial intelligence systems. A fluent, detailed and seemingly confident answer may rest on incomplete information, weak sources or assumptions that have never been tested. The quality of the prose is no guarantee of the quality of the knowledge behind it.
I would not draw from the Iraq case the impossible lesson that decisions should be made only when certainty is available. Intelligence assessments inevitably have to be produced under conditions of incomplete information. The more demanding lesson is that we must distinguish what we know, what we infer and what we simply do not know. And we must be willing to communicate those distinctions to decision-makers, even when they would prefer a much more definitive answer.
Further reading:
Commission on the Intelligence Capabilities of the United States Regarding Weapons of Mass Destruction, Report to the President (2005);