Devlin's Angle

May 2002

Randomness at the Airport

During the past two months, I have made 20 domestic flights on commercial airlines. For 7 of those flights, that's just over one in three, I was pulled from the line of passengers waiting to board the plane and subjected to the purportedly "random" additional body and baggage search. (Ironically, the last occasion was when I was about to board my flight back from Washington, D.C. to San Francisco, on my way home from the Board on Mathematical Sciences and their Applications two-day conference on Mathematics and Homeland Security.)

Now, I'll tell you, with data like that, I am sure that, whatever the instructions to the airport security personnel are, those additional searches are not carried out in a random fashion. If they were, then on every flight an average of 80 people would be stopped and searched as they boarded the plane, and no plane would take off on time.

My guess (and my hope) is that the instructions are to check every passenger that fits the very obvious profile of a potential Al Qaeda hijacker, and check sufficient additional passengers to avoid any hint that passengers are being profiled.

Of course, no one is supposed to admit that this is going on, but the authorities would be stupid not to act upon what we know about terrorist risks. In the first rounds of the US War on Terror, we did not after all drop bombs on Omaha, or even many parts of Afghanistan; rather we concentrated our military efforts on those parts of Afghanistan where Al Qaeda terrorists were known or believed to be. Likewise, we should concentrate our necessarily limited airport security efforts on identifiable sections of society where the risk is higher. And that means profiling. In the case of Al Qaeda suicide terrorists - which is after all the real threat - such profiling has unavoidable racial, cultural, and religious components, and that is unfortunate. (It's not just racial and religious, however. Terrorists fit other, behavioral profiles as well.)

Now, I don't like the idea of profiling that has a racial or religious element any more than the next person. Thus, I regard random searches as an arguably necessary nuisance to ordinary airline passengers, a relatively small price to pay in order to avoid overt profiling. (Although I realize that by the same logic we should drop the occasional smart weapon on US cities, just to show we are not singling out any particular group.) And were those additional searches of outside-the-profile passengers truly random, I would have no problem with them. But as my own experience indicates, they are not. (True, I'm a sample of 1, but I fly a lot, so I think the data has significance.)

Now, one of two things could be going on here. First, it is possible that I do in fact fit some profile. I am male, dark-haired, fairly fit and athletic, I always travel light (hand luggage only), and living in California gives me a permanently tanned skin. To be sure, I don't think I look remotely like any of the September 11 hijackers, nor like any other airline terrorist whose picture I have seen. But I cannot rule out the possibility that there is something about me that raises the suspicions of the security personnel. (By the way, I have tried making eye contact with the security personnel, both with and without a smile, and avoiding eye contact altogether. It does not appear to have made any difference. As we shuffle forward toward the gate, the guard inevitably walks up to me anyway.)

On balance however, I don't think any kind of conscious profiling is going on. Rather, I think the issue is a psychological one. The security personnel at the gate do not use a random number generator to select passengers to check. Rather, they stand by the gate and approach people. Now, the last couple of occasions they have chosen me, I've asked them what criteria they used, and both times I was told it was random. I am quite willing to accept that they believed this to be the case. But as anyone who has taught (or taken) an undergraduate course on probability theory will know, humans simply cannot make genuinely random selections, even when choosing abstract objects such as numbers. (Most people are surprised to discover that their attempt to write down a random sequence of numbers is inevitably far from random.) And as numerous studies in sociology and psychology have shown, when it comes to selecting people "at random" from a given population, we are far worse. All kinds of subconscious influences bear on our choice.

I like to think that the reason I keep getting pulled out is that I look like the kind of person who won't cause the security personnel any trouble. We all like to avoid trouble if we can. The security personnel are humans, just like everyone else. Indeed, the majority are immigrants (as am I), whose poor English suggests they are fairly recent immigrants, and moreover the job is a low paying one, so they surely start with a tremendous social disadvantage in having a policing role over obviously affluent, jet-setting professionals with expensive laptop computers and first-class airline tickets. It would be highly surprising if they did not, at least subconsciously, try to minimize the possibility of conflict.

But just think about this. What it means is that the security personnel are concentrating much of their efforts on the people least likely to be terrorists. That's fine - albeit a bit unfair on us mild mannered types - if the "random" searches are purely to hide the profiling that I hope is being done. But if, in addition, there is no profiling, then the system is badly flawed.

If there is no profiling being done, then with the "random" searches concentrating on the "good guys", the present system is just a sham, and a dangerous one at that. On the other hand, if there is profiling, and the "random" searches are just a feelgood addition to feed our admirable need to appear free from prejudice, then by not making them truly random, and allowing them to focus on the "good guys," we miss an excellent opportunity for a genuinely useful second line of defense. Either way, the system is flawed.

I admit that it is possible to cast doubt on much of what I say. Only a systematic scientific survey, carried out using rigorous statistical techniques, could really decide the issue. But as a mathematician I do have a good understanding of randomness, which is a mathematically precise concept. What is, I am sure, clear beyond any doubt, is that those supposedly (and possibly even intendedly) "random" gate searches are anything but. As a consequence, the security benefits that truly random searches could bring are not being taken advantage of.

Moreover, if what I suggest above is correct, then a genuine terrorist could improve his or her chances of taking a weapon onto a plane by lining up just behind two or three of the "safe" passengers (like me) who are likely to be pulled out for extra screening, leaving the screeners fully occupied with harmless passengers while the terrorist walks onto the plane.

A possible solution? In Mexico, when an arriving international passenger gets to the customs control, he or she is asked to press a button, which causes a red or green light to illuminate. If it shows green, the passenger can proceed; if it shows red, a luggage search follows. Now, I dare say that the customs officers can ensure that a red light comes up. (As with security screening, it would be crazy not to make use of human expertise to spot, in this case, potential smugglers.) But I assume that the reason for having the automated system is so that any additional random searches are just that: genuinely random. A similar system could be implemented at the nation's airports. (It could be done by software alone, with a mark on the boarding pass.)

As mathematicians know full well, randomness, when used wisely, is a powerful tool. It could be a tremendous ally in ensuing airline security. But it has to be genuine randomness. With the best will in the world, people cannot make random choices, and if we truly value airline security, we should not ask them to.


Devlin's Angle is updated at the beginning of each month.
Mathematician Keith Devlin ( devlin@csli.stanford.edu) is the Executive Director of the Center for the Study of Language and Information at Stanford University and "The Math Guy" on NPR's Weekend Edition. His latest book is The Math Gene: How Mathematical Thinking Evolved and Why Numbers Are Like Gossip, published by Basic Books.