Jeffrey LewisIntelligence Analysis and Bayesian Analysis

I am writing a piece for Arms Control Today on foreign counterspace systems. The mandate was to summarize foreign counterspace capabilities to accompany an article being written by Michael Krepon.

The more I look, the more I realize that the intelligence community on this issue is all method, no evidence.

Most people reason by taking a prior judgment (lets call it a prior, for short) and updating the prior as new information becomes available. This is why, on a date, a first impression determines whether you get laid. The relationship between prior beliefs and new information was formally stated by a mathematically inclined Presbyterian minister in Britain, Thomas Bayes (1702-1761).

This approach, which has advantages, also creates problems for intelligence judgments. Bruce Blair, in The Logic of Intelligence Hype and Blindness (December 2003), demonstrates that two strong prior beliefs (a terrorist attack is unlikely, Saddam Hussein lusts after WMD) resulted in two recent intelligence failures that you may have read about.

At best, Blair notes, “Bayesian math points to a fairly slow learning curve” among US policymakers. At worst, I would suggest, Bayesian math leaves some priors essentially impervious to revision for some policymakers and analysts.

What does this mean for intelligence estimates of counterspace capabilities? Take this statement by Donald Rumsfeld:

As we become increasingly dependent on space for communications, situational
awareness, positioning, navigation and timing, space will necessarily become an
area we have to defend. Adversaries are likely to develop ground-based lasers,
space jamming and “killer” micro-satellites to attack U.S. space assets.

This paragraph reveals a very strong prior belief that other states must be pursuing counterspace capabilities because of our vulnerability–no evidence, all theory. Actually, very strong doesn/’t convey the centrality of the assumption to Rumsfeld/’s worldview. Here is how The Commission on US National Security Space Management and Organization, which Rumsfeld chaired until his nomination as Secretary of Defense, put the point even more strongly in its consensus report: “We know from history that every medium–air, land and sea–has seen conflict. Reality indicates that space will be no different.”

All of history is a strong prior, on par with Mom loves me and The voices told me to do it. Moving Rummy et al off the Chinese counterspace obsession will require a complete revision of their interpretation of all of recorded human history. How many years do you think that will take?

Some hint is provided by the recent brouhaha over the Chinese “parasite” microsatellite. The Department of Defense claims China “has reportedly begun testing an anti-satellite, that is an ASAT system,” a claim that Gregory Kulacki and David Wright tracec to a single Chinese military enthusiast in Anhui province.

Calling out Rumsfeld like this in a meeting or public debate would shred credibility and undermine everything he ever said after that. He could tell a story about his grandaughter/’s fifth birthday party and people would fact check the name of the clown. In other words, “Sayonara, ethos!” But logos is stubborn. (Diamond Don is s.o.l. on the pathos front).

Will the revelation that the story is bogus result in a fundamental change of Pentagon attitudes?

Well, when confronted with evidence, Flex Plexico, a Pentagon spokesman, responded with a classic summary of Bayesian analysis: He called the UCS report “additional information” that “our analysts will evaluate fully and weigh against the other information they have.”

In other words, no.