Tuesday, December 29, 2009

The TSA Game

Eric Posner explains TSA strategy, once we rule out impossible extremes:
Once you eliminate the implausible corner solutions—the TSA undertakes body cavity searches of every passenger, or security screenings are abolished—a number of consequences follow.

1. The TSA must randomize (play a “mixed strategy,” in game-theoretic parlance). Otherwise, terrorists can predict some of its precautions and evade them. The same principle explains why police vary patrol routes and road blocks. A NYT article today makes clear that the TSA is self-consciously randomizing to keep terrorists off guard.

2. At the social optimum, the number of successful terrorist attacks will be greater than zero. It might be argued that we have had too few successful terrorist attacks over the last few years rather than too many. The question is whether the implicit statistical valuation of life in TSA programs is too high. I suspect that the answer is yes, as is generally the case with airline safety.

3. Profiling is an effective strategy when, as here, terrorists come from a small group of (relatively) easily identifiable people. One suspects that this explains Israel’s success. But profiling places a large portion of the cost of deterrence on a small group, which makes some people morally uneasy.

4. Once the implausible corner-solutions are ruled out, any security policy or threshold will seem arbitrary because you have to draw the line somewhere, which means that it will be easy to point to some permitted activity that is only slightly different from what is forbidden (for example, carrying on 100 ml of liquid rather than 101 ml).

5. As for the “security theater” claim–

a. If ordinary people are fooled into thinking that the TSA is doing more than it is really doing, then at least some potential terrorists will be fooled as well, and so will be deterred from engaging in airplane-terrorism.

b. Ordinary people will also fly more often, which means that one of the goals of terrorists—to terrorize people so that they will pressure their government to make concessions to terrorists—will have failed.

How Chicago was that post? But he's right, though I'll note a tension between numbers 1 and 3. A profiling regime is by definition not random, so terrorists can construct an "anti-profile" to beat the system -- the so-called "carnival booth" (because it encourages terrorists to "step right up and see if they're a winner").

I also think that 5(a) is maybe overstated, as the point is that the facial security measures might fool us average rubes, but not anybody actually intent on breaching airport security. I suppose there might be some aspiring terrorist who is awestruck by the festival of it all, but I have to think that anybody who is intent on breaching TSA security won't give up that easily.

On the other hand, while I don't accept 2 whole-heartedly, I do think it is important to note that pretty much any feasible screening system we come up with will not be fail-proof. Terrorists might get through even if we do everything right. It's really annoying when people act as if pretty well random and unpredictable acts are examples of outrageous bureaucratic incompetence. The TSA isn't the Green Lantern -- it doesn't protect us just by willing it hard enough.

1 comment:

Tim -- tstarks2@gmail.com said...

The TSA has a well-known weakness for the color yellow, however, which arouses suspicion that it is indeed the Green Lantern.