A provocative op-ed entitled “America’s Intelligence Denial in Iran” in the Wall Street Journal on Wednesday made a startling claim about the intelligence community. Fred Fleitz, a longtime employee of the intelligence community, wrote that analysts are “unwilling to conduct a proper assessment of the Iranian nuclear issue” due to political considerations, and that the latest National Intelligence Estimate on Iran was “poorly written [with] little improvement over the 2007 version.”
Mr. Fleitz is onto something: NIEs in general are terribly written. After all, they are written by committee, subject to laborious review sessions by the 16 agencies of the intelligence community. They are important documents for the insight they bring into the intelligence process, but it is a mistake to consider them authoritative. A good NIE is filled with qualifications, caveats and warnings about the quality of the information it contains, since it is an aggregation of many agencies’ views.
The 2002 NIE on Iraq famously stated (pdf) that Saddam Hussein was “reconstituting its nuclear program” — a claim that in retrospect is almost laughably false. That NIE lent a false sense of certainty about Iraq’s nuclear programs. In hindsight, it demonstrated how poor assumptions and an over-reliance on confirmatory reviewers led policymakers to make resoundingly poor foreign policy decisions.
That wasn’t the first time. In 1995 an NIE claimed North Korea was 15 years away from developing a ballistic missile capable of hitting the United States. In response, some members of Congress complained the NIE was too politicized and organized an independent panel — led by former Director of Central Intelligence and current Secretary of Defense Robert Gates — to review its assumptions. That panel, while pointing out that NIEs rely too much on weak assumptions and uncritical reviewers, largely confirmed the original analysis.
In July 1998, the so-called Rumsfeld Commission, convened by House Republicans and chaired by Donald Rumsfeld, refuted the Gates Panel and claimed North Korea’s missile program posed a bigger threat than the intelligence community would admit. Two months later, North Korea launched a missile over Japan — a feat few believed it capable of at the time. While that lent credence to the Rumsfeld Commission, Gates’ bigger argument about NIE’s being compromised by poor reviewing practices and sloppy research didn’t substantially change how they were written.
So when Fleitz argues that the intelligence community is downplaying the threat from Iran thanks in part to biased reviewers and hazy assumptions, he might be onto something. NIEs have, in the past, badly misjudged the threats they were meant to identify. The problem with his argument, however, is that it is shortsighted. For example, he notes that the intelligence community is gun shy after the fiasco resulting from the intelligence community’s overzealous assessments of Saddam Hussein’s Iraq. He’s right, but I’d argue that this is actually a good thing: The intelligence community was rightly condemned for the blunders that led the country into the Iraq war. They should be more cautious in future assessments and mindful of how their assessments might be appropriated (or misappropriated) by policymakers.
Fleitz’s complaint about liberals in the intelligence community trying to undermine U.S. policymakers has a historical precedent in national security studies. In his newest book, “Fixing the Facts: National Security and the Politics of Intelligence,” Joshua Rovner discusses the controversial history of NIEs. He details various efforts on the part of some conservative policymakers to cast the intelligence community as liberals seeking to downplay threats to the U.S. In the mid-1970s, he explains, Albert Wohlstetter, a professor at the University of Chicago who focused on strategic security issues, started openly accusing the CIA of deliberately underestimating Soviet military capability (here are some of his papers from this area). In response, the CIA, led by future President George H.W. Bush created “Team B” in 1976. “Team B” amplified Wohlstetter’s claim about the Soviets and laid the groundwork for the arms race of the ’80s.
Once the Cold War was over, the same dynamic of conservative analysts and scholars accusing the intelligence community of underestimating threats to the U.S. played out over North Korea, Iraq and Iran. Then in 2002, with a conservative president in the White House, the intelligence community wrote a much more alarmist NIE about Iraq’s nuclear capabilities. This NIE helped sell the invasion of Iraq to the public (even though public versions excluded (pdf) much of the hedging language and caveats of the classified NIE, which created a misleading picture of what the government really knew before the invasion). Now, the old Wohlstetter-esque complaint, that the intelligence community is too cautious because of its liberal bias, is being filed over Iran.
It’s difficult to locate the offending language in the 2007 Iran NIE. It discussed two separate issues: enrichment and weaponization. The 2007 NIE stated that Iran had an ongoing enrichment program that would start producing weapons-grade uranium by about 2009 or so, but that a program to weaponize that uranium was much further off. That is more or less what has played out: the latest outrage over Iran’s nuclear program is over its increased enrichment capability. There is a widespread assumption that this new enrichment will make weaponization easier, but is not itself a weaponization program: Iran is not building nuclear weapons, it’s just making a lot of uranium.
We don’t know what is contained in the 2011 NIE, which is still classified. And it is there that Fleitz has a stronger argument about how the intelligence community seeks out review of its products. If, as he alleges, the reviewers were chosen for their politics and their sympathetic leanings toward the report’s conclusions, then that is a serious problem.
Confirmation bias is a critical issue in intelligence analysis. I’ve worked in four intelligence programs since 2006, and all of my bosses save one were openly conservative (the other was very liberal). I never felt their politics affected their integrity as analysts in any way. What did affect our analysis, however, was more basic analytic tradecraft: vetting sources, piecing together bits of information that tell a coherent story and identifying trends that might affect U.S. policy. It is easy to get lost seeking evidence to support a conclusion you’ve already decided upon.
Intelligence analysis is challenging work, and analysts are often expected to predict the outcomes of extremely complex processes and events. Given the recent scars of the Iraq war that are still fresh for many in the intelligence community, we shouldn’t judge them too much for being cautious. But that doesn’t mean we should excuse poor judgment in reviewing their analysis, either.