The argument goes on. Sam Harris has reacted to my post on profiling.
One line in my article raised a tsunami of contempt for me in liberal and secular circles:
We should profile Muslims, or anyone who looks like he or she could conceivably be Muslim, and we should be honest about it.
Once again, I included myself in this profile—but that did almost nothing to stem the accusations of racism.
He keeps saying that. I don’t know why. The objection isn’t that Sam Harris wants to discriminate against people who don’t look like him, it’s that Sam Harris wants to discriminate against people on the basis of their appearance. The fact that his search criteria are so broad that they include him isn’t a point in his favor, either — it means he favors criteria that produce many false positives.
I really don’t understand why he’s finding that so hard to grasp.
Then he offers an example of how his version of profiling would work. I’ve highlighted a few words that I think are important.
Imagine that you work for the TSA and are executing a hand search of a traveler’s bag. He is a young man in his twenties and seems nervous. You notice that he is carrying a hardcover copy of The Girl with the Dragon Tattoo. You pick up the book and ask him if he likes it. He now appears even more nervous than before. You notice something odd about the book—the dust jacket doesn’t seem to fit. Your remove it and find a different book underneath. How do you feel about this traveler’s demeanor, and the likelihood of his being a terrorist, if the book is:
A. The Qur’an (in Arabic)
B. The Magic Mushroom Grower’s Guide
C. Overcoming Impotence: A Leading Urologist Tells You Everything You Need to Know
D. Dianetics
If you care more about A than B, C, or D, as I think you should, you are guilty of religious profiling (and calling it “behavioral profiling” doesn’t change this fact).
I have no problem at all with that kind of profiling. The matter that raises concerns isn’t that the young man looks Muslim: it’s that he looks nervous, and that he’s hiding something. TSA, please do notice when people’s behavior is peculiar!
But now I’m curious. Change the story a bit. What if the young man were confidently and openly carrying his Qur’an? Should we stop and search everyone with a Muslim holy book?
What if he were nervous, and it was a Christian bible hidden away? Should we then ignore his odd behavior and wave him on to the plane?
What if he were openly carrying that bible? Does proudly carrying a Bible give you a free pass through the screening? He could wear a cross and a flag pin on his lapel, and no longer Look Muslim, I guess.
Harris doesn’t seem to understand that his critics are not saying TSA should be blind and deaf to the people passing through security checkpoints…but that there should be some intelligence behind it, and that the criterion of “looking Muslim” is stupid and useless.
Then he raises a series of strawmen — that we think there is no link between Islam and suicidal terrorism, that we have some pious fantasy of Israeli egalitarianism in their security procedures, and implies that we have no problem with TSA searching toddlers. I said precisely the opposite about Islamic terror tactics; I have no sympathy for Israel’s convergence on fascism; I think most of what TSA does in those security lines is a waste of time and bad security policy.
My fundamental issue with his whole proposal is the shocking innumeracy of it all. Here’s the perfect example: he asks, what percentage of the people who would murder the children boarding a plane, and all who accompany them, are Muslim? And here’s his strange answer.
Some readers might think that this question would be difficult or impossible to answer. Let’s try another, then: What percentage of porn stars are also theoretical physicists? Is this a hard question for which to give a ballpark answer? No. In fact, I would be willing to bet my life that I could get within 10 percentage points of the exact figure without doing any research—and the same holds for the question about using children as bombs on airplanes in the year 2012.
Wow. He gives himself a very broad 10% window for his answer.
So the percentage of porn stars who are also theoretical physicists? I guess 0%. I’m sure I’m within 10% of the correct answer. I won’t go searching porn studios for answers to cosmology questions.
The likelihood of a toddler being used to smuggle a terrorist bomb on board a plane? 0%, again. I’m confident — we aren’t seeing 1 in 5 kids being plucked out of line so the dynamite in their diapers can be thrown away.
But then his peculiar question — what percentage of suicidal terrorists boarding a plane are Muslim — I’d answer with 100% (OK, 90%, so my guess covers a broader range), and I suspect he would, too. But it’s the wrong question. It’s a completely bizarre twisting of what the appropriate question should be. We aren’t screening the guys who look like terrorists at the airport to find out which are Muslim; he wants us to screen the people at the airport who look like Muslims to find the terrorists.
The right question is what percentage of the people who “look Muslim” (whatever that means; Harris hasn’t yet defined it), his screening criterion, are terrorists? And again, the answer is 0%±10%, the same as the percentage of physicist porn stars, or bomb lobbing toddlers. I would agree that just screening Muslims would increase the likelihood of finding a terrorist by some small amount, but it has the problem that I still don’t know what a Muslim looks like, so that pre-selection is going to be awfully leaky, and you’re going to generate such a huge number of false positives that your more rigorous secondary screening is going to get swamped, and you’re going to open the door to even more false negatives as your real terrorists avoid “looking Muslim” in line.
So once again, we sacrifice civil liberties and real effective security for TSA showmanship, as people who “look Muslim” to uniformed low-wage security guards with a GED get thoroughly frisked. I don’t get it. Sam Harris is a scientist; how can he so blithely overlook type I and II errors in a statistical sampling protocol? How can he ignore the ambiguity of his sloppy definition of his primary measure, “looking Muslim”?

