Wednesday, August 23, 2006

Profiling Arabs.

On a similar note to the previous post, note that anti-Arab hysteria is spreading outside the political arena. According to this, an Arabic physician from Winnipeg was escorted off a plane from Denver -- because he was praying (presumably, in Arabic). There's no justification for this sort of obvious racial profiling. I have yet to hear, though, of any official complaint being made by our stalwart Conservative government.

According to this article, the British are being sensibly cynical about the whole exercise. Good points made throughout. However, according to this, the trend is not universal, as two Arabic men were hounded off a plane by racist fellow passengers. (Original article here.)

Thomas Nadelhoffer has an amusing post here, arguing that if we should profile the dangerous kinds of people, we should first be profiling politicians and their kowtowing pundits.

On a serious note, though, am I the only one who remembers Jean Charles de Menezes? How long until the next shooting -- by reason of not being white?


Anonymous said...

As Ann Coulter has expressed so brutally, if Arabs have perpetrated most of the terrorist attacks in the United States, are we not justified in drawing the inductive conclusion that they are more suspicious characters?

What's so wrong about profiling Arabs?

ADHR said...

It depends on how you define "terrorist attacks" and your historical scope. I'm sure that white Europeans have committed more attacks on the US over the past five hundred years than Arabs.

There's three problems. The first is epistemic. The inductive conclusion is broader than the data support. The data show that particular Arabs, in particular social, political, economic, and cultural circumstances, in recent history, have committed attacks on the US. Thus, particular Arabs, in particular social, etc., etc. If you try to generalize much further than that, as with all inductive inferences, you run the grave risk of running rough-shod over too many important distinguishing features. So, inductive inferences, generally, should be as conservative as possible without being vacuous. However, even very conservative inferences can go wrong.

For example, borrowing from Hume, if all I have ever seen are black ravens, I can conclude that all ravens are black. But, if there is such a thing as a white raven, I am simply wrong. The inference is fine, but it's led me astray. Similarly, if all recent attacks on the US have been committed by Arabs, I can conclude that all Arabs are potential attackers. But, if there is one Arab who is not, then I am wrong.

There are also grave moral problems involved: we are judging people not because of particular intentions they have or actions they have done, but on the basis of morally irrelevant features they happen to share with wrong-doers. It's like saying anyone born in New York is a potential terrorist because Timothy McVeigh was born in New York and was a terrorist. Being born in New York is morally irrelevant. Being an Arab, or a Jew, or French, or American is also morally irrelevant. These features do not make you more or less likely to do right or wrong.

Finally, there's a sense of diverting needed resources in a pointless way. (So, this is a prudential concern.) We could, theoretically, track individuals on the basis of a whole range of features. Why focus on race? Because it's expedient. Not because it actually works, but because it's a way to seem to be doing something important. I'd suggest that it would be more sensible to look for features to track which would actually work to prevent further terrorism, rather than looking for obvious features that play well during election season.