Seems a couple of MIT Grad Students have eclipsed the current bunch of 1600 Crew money-spending consultants at the alphabet agencies of DHS and TSA and anywhere else they're spending deficit dollars on security thinking. The grad students have proved that random searches at airports would be more effective at catching bad-guys than the procedures used now.

...two MIT Computer Science graduate students say they have proof that profiling doesn't work. Samidh Chakrabarti, 23, and Aaron Strauss, 22, say that using purely random searches would be more effective at catching terrorists.
...
While the parameters of the system are classified, anyone who is flagged for extra screening knows it as soon as they are pulled aside for special treatment. Chakrabarti and Strauss show, through computer modeling, how the terrorists can easily defeat the system. Put simply, it's all about trial and error. For example, let's say a terrorist cell sends 20 different guys through the airport (with no weapons and no intent to harm), the person who consistently passes through security without extra scrutiny is the best person to send on a destructive mission in the air. The terrorists basically conclude this guy is "profile proof."

On the other hand, according to the MIT grad students, if the airports employed only random searches the terrorists would not be able to practice against the system. The MIT research explains it this way: an average airport has the ability to do extra screening on eight percent of the passengers.
...
That's because with a purely random system, potential terrorists would have no way of knowing ahead of time if they were likely to receive extra screening and they would have no way of practicing against the system.

Doug Laird is a private security consultant and the former Director of Security at Northwest Airlines, who helped craft the CAPPS system. He says that while the paper is compelling, he doesn't want to make the public unnecessarily worried at this point. Laird says Chakrabarti and Strauss make some incorrect assumptions about how the CAPPS system works. He cannot identify which assumptions are incorrect because that information is classified.

Dr. Philip Stark is a Professor of Statistics at UC Berkeley, the top ranked statistics department in the country. Stark reviewed the paper. He had some initial questions about some of the probability equations, but he says even if the equations are slightly off, it would not be by enough to change the conclusions of the paper. He says the model the MIT grad students used is actually very similar to a common aspect of mathematical game theory. The "game" he cites is called "matching pennies."

It works like this: you and I each get a penny and we each decide to put them down as heads or tails. If they match, you keep both, if they don't match, I keep both. Then the question is: what is the best strategy to win the most pennies? All heads? All tails? Two heads then two tails? The problem is that if there is any sort of pattern, your opponent can detect it and counteract it.