Monday, December 04, 2006

Profiling - what's wrong with it?

Profiling seems to be getting a bad rap these days, whether it's racial profiling (getting pulled over for "driving while black", for example) or other forms of profiling, such as suspicions of anyone doing anything overtly Islamic, such as the recent detention of several Muslim clerics from a US Airways flight. (Note that I am deliberately using simply "profiling" rather than the more commonly used "racial profiling" because profiling can go beyond simple race, as the example above illustrates.)

Is profiling inherently problematic? The answer, surprisingly, I think is "no." From a civil-rights perspective, I think there is nothing wrong with profiling per se, provided that it meets one critical test: is it effective. Most of the time, it isn't, and this is where profiling inevitably crosses the line from inconvenience to personal infringement for the person being profiled.

Consider two examples to prove my point about effectiveness. In the first, there is a report of a shooting in a predominantly black neighborhood. If the police are out looking for suspects and focus exclusively on blacks, that IMO is "bad" profiling - you're casting unfair and undue suspicion upon someone because of skin color (seems like an infringement on their rights to me), and you're also ineffective because you are needlessly cutting off leads that could be fruitful.

On the other hand, suppose in the same situation there are eyewitness reports that the shooter was black. In this case, I don't think anybody would think it controversial to focus on black suspects; frankly, I doubt anyone would actually call this "profiling" in this situation.

Over the past few years, I think we've seen a lot of hysteria over profiling with regard to terrorism. The imams mentioned above are not the first example of Muslims being removed from airplanes because of middle-eastern looks or because they were overtly Muslim. I hope I'm not saying anything scandalous in saying that this is pretty bad profiling. After all, I'm not aware that we've ever actually caught a terrorist this way. On the other hand, the sort of profiling that has actually stopped terrorists has been looking for people who, say, try to light their shoes on fire on an airplane.

Much of this hysteria I think arises from a mistaken understanding of statistics and probability. For example, it is a fact that the most deadly of terrorist attacks (9/11) and the vast majority of terrorist attacks around the world over the past several years have been committed by Muslims. Sad fact, but pretty obviously true. So the odds that a given terrorist is Muslim are much higher than the odds that the terrorist is, say, Catholic. However, we aren't looking at random terrorists, we're looking at random people. And even if the odds that a random Muslim is a terrorist are twice that of a random non-Muslim, you're still looking at minuscule odds in either case. Stated another way, with over a billion Muslims in the world, if there are even 1 million terrorists among them (which I'm sure there are not), that's still only a 0.1% chance that any given Muslim is a terrorist.

So suspecting a given Muslim of being a terrorist simply because of his religion is a lot like trying to solve a rape case by detaining all of the men in the city. You haven't really narrowed the problem in any meaningful way, and you're violating the rights of a lot of people in the process.

This isn't just an academic point - efficiency in security is actually critical. Trying to achieve good security without efficiency has two very dangerous side effects: (a) it overwhelms the process that you are trying to keep secure, and (b) all the effort expended on pointless security procedures distracts resources from genuinely useful security work, thus creating security holes or lapses that can be exploited by people who don't simply look like the stereotype of the bad guys but who actually are the bad guys.

Removal of shoes and bans on liquids at the airport both offer examples of the latter, and as far as I can tell neither of these time-consuming, arduous, and expensive procedures has stopped a terrorist. Alert passengers on a plane stopped the shoe bomber, and good intelligence stopped the liquids-bombers.

Frankly, I've yet to hear of examples of profiling that are actually effective in practice. And that is what makes the inconvenience and infringement that results so unacceptable.

No comments: