Monday, August 29, 2005

ABLE DANGER: Catch-up time

Terry McDermott makes a good point about retrospective memories and high profile events in this LA Times column. It not quite the strong rebuttal to ABLE DANGER that some think. First, he weakens his case by dismissing something he does not understand:

Data mining is a technique in which huge databases are fed into powerful computers that sift them looking for links. It's a technology that holds vast promise, but its main usefulness to date seems to be giving mortgage lenders the ability to find out how much you still owe on your house.

Financial institutions have been doing more than that with data mining for more than 10 years. McDermott should know that if he is going "refute" the ABLE DANGER story. Maybe he should try googling "Capital One" and "strategy" just for a start. Or maybe "data mining" and "banking".

The even bigger problem is in McDermott's assertion that
Atta's academic, immigration, credit, transit and telephone records provide a fairly complete account from the time he left his native Egypt in autumn 1992 to his death. This includes the period during which Able Danger is said to have identified him as a terrorist in the United States. The story those records, and corroborating interviews, tell is that Atta was not in the United States and made almost no contact with the U.S. until June 2000.

Of course, he has to say that. He just invested years of his life writing a book based on the sources who thought they knew all about Atta and his movements based on cell phone records and credit card statements. The whole picture changes if al Qaeda knew how to defeat FBI surveillance and investigations. That is why Hanssen is important; why other security lapses in the FBI are of vital interest.

In my Tailhook post I wrote:

Pretend that you are one of the officers in charge or involved. Your program has promise but it has not caught any terrorists or prevented any attacks. The lawyers think you should shut it down. They are "concerned" that word will get out and ignite a media firestorm over civil liberties, high-tech invasions of privacy, and racial profiling. You suspect that they are right and you also are not confident that one of the lawyers will not leak the details to some obliging reporter.

If the story is leaked, who can you depend on? The brass? Congress? George H. W. Bush's son

This NY Post story suggests that that mindset was at work when ABLE DANGER was shut down:
The private contractors working for the counter-terrorism unit Able Danger lost their jobs in May 2000. The firings following a series of analyses that Pentagon lawyers feared were dangerously close to violating laws banning the military from spying on Americans, sources said.

Captain Ed is right to suggest that pre-9/11 there would have been a bi-partisan frenzy if word of the Rice/China chart had leaked.

I am deeply skeptical of most of what I read about ABLE DANGER. It is not that I think anyone is lying. But the stories we read are coming to us at the end of a game of telephone where each succeeding listener has less knowledge of data mining, statistics, modeling, and probability. Too many journalists are functional innumerates. That is clear, for example, when the NY Post reporter writes that the data-mining group "pinpointed" Rice. Data mining tools do not usually "pinpoint" individuals. The do not have to in order to be successful. Maybe the report's source overstated the accuracy of their work. But I think it is more likely that the reporter used Perry Mason language to describe a probabilistic outcome.

I worry that information is getting lost in transmission. That might explain why Lt. Col. Shaffer has had a hard time making himself understood to non-technical listeners.

No comments: