July 08, 2004

Why Method is Important

In the next few days the 911 Commission is going to issue it's report, which will point to some rather significant failures in the intelligence community. But since the Commission isn't really equipped to deal with, or even conceptualize, complex or wicked problems, it'll simply point a few fingers, get a few people fired, and generally not make any important changes. Well, I'm just guessing from some stuff I heard Senator Levin talking about on CSPAN an hour ago.

We don't solve complex problems, because our entire system is geared toward solving simple problems... and if a problem isn't simple we just toss simple solutions at it like darts at a dartboard, because that's what we do.

So Senator Levin sees the problem as the Bush administration having failed to communicate that, for instance, the CIA "doubted" that the meeting between Atta and the Chief of Iraqi Intelligence took place in Prague. They criticize the CIA scathingly, yet they believe we can just accept the CIA's "doubts" about such a meeting at face value.

Does the CIA know, of a certainty, where Atta was at the time of the alledged meeting? If not, why not? What was their basis for doubt? More importantly, if we're more concerned with Type II errors, as we ought to be in this case, then the issue isn't the CIA's doubt, but why they still kept the possibility that such a meeting did take place open. And more importantly why that was the right and correct thing to do, for the Administration. (Levin seems to think it was the wrong thing to do, but can't express a coherent reason for such a presumption.) In other words a prudent person ought to ask: "Have you any proof that the meeting did not take place? Proof that rises to the level of the proof that's now being demanded of those who say the metting did take place?"

Because that's the only way we could have made a valid and appropriate assumption that there was no such meeting.

Let me repeat that, for emphasis: Only if we can prove that the meeting did not take place to a virtual certainty would it have been appropriate to assume that it did not. Otherwise the appropriate assumption was that the meeting did take place. (There is a body of evidence pointing to such a meeting. It's simply that the evidence isn't conclusive, or has other explanations.)

Now, of course, we can assume that it didn't take place... but only because the risks associated with a Type II error (concluding that there was no Al Qaeda/Saddam collaboration, when there was) have been reduced to almost zero. And those risks have been reduced because we acted on the assumption that there was a collaborative relationship, even though we knew there was no certaintly about that.

So why doesn't Levin understand this?

Beats me.

Posted by Demosophist at July 8, 2004 11:47 PM | TrackBack
Comments