Monday, July 25, 2005
Terrorism, Deterrence, and Searching on the Subway
Dave Hoffman (law, Temple) over at the Conglomerate blog, has written a very thoughtful retort to a recent post of mine (cross-posted at PrawfsBlawg and Balkinization) regarding the searching of baggage on NYC subways. I argued that:
It is another big waste of money and time, as well as a needless invasion of civil liberties -- all for a cosmetic security benefit. There are 4.5 million passengers each day on the NYC subways. What good could a few random checks do? The odds of the police finding the terrorist with a bomb this way are about as good as the odds of being hit by lightning. I doubt it will have much of a deterrent effect either.
Dave argues that I’m “underestimating the effect of random searches on public safety.” I am not much of an expert in behavioral law and economics and deterrence, so I’m straying far away from my turf. But I’ll try my best to defend the territory I’ve wandered onto.
Dave argues: “Terrorists are notoriously risk averse - they obey the law punctiliously until they attack. Thus, even a relatively minor risk of being caught will act as a very large deterrent, forcing terrorists to find other paths.” Really? Since when are terrorists risk adverse? Anybody willing to blow oneself up in the process of committing his or her crime is hardly risk adverse. And I just don’t see the relevance of the fact that terrorists obey the law prior to engaging in terrorism. However, suppose Dave were correct and the terrorists would “find other paths.” There are so many other targets that are even more dangerous and damaging. So we close off one target but leave the rest unguarded. Are we really any safer?
Dave also contends that if the police vary the number of searches, it will make it “hard for terrorists to intelligently evaluate the likelihood of being stopped on any given trip.” But unless the number of searches were really great in proportion to the number of subway riders (4.5 million each day), the variation would be relatively small. Wouldn’t the terrorists think that if three or four of them tried to bomb the subway on a given day, probably all (or at least most) would get through without being searched?
Dave argues that “there is a good argument that terrorists, subject to human behavioral tics, are likely to vastly overestimate the likelihood of being caught and therefore be more deterred than rational terrorists (what a contradiction in terms that is!) would be.” But Dave forgets that many terrorists are different from ordinary criminals in that terrorists are often on a suicide mission. They care about getting caught only because their mission might fail, not because of any potential legal sanction that might be imposed. If Dave is right, why on 9-11 did the terrorists try to use planes? Why not try some other means of terrorism? After all, planes involve a lot of security whereas other targets don’t. Wouldn’t the “risk-adverse” terrorist who might overestimate being caught attempt something else? Why did they go to flight school and expose themselves at many points to being detected when they could have tried something different? I’m certainly no expert on terrorist behavior, but I’m not very convinced by Dave’s theory.
Dave says: “Will terrorists then move on to other targets of opportunity? Probably. But forcing them to do so would be a victory.” I’m not so sure. This depends upon what the other targets are. Is it a victory to stop a terrorist from bombing a subway car and killing 40 people so that the terrorist decides instead to blow up a building or mall killing thousands?
Dave says I speak of two kinds of costs – law enforcement costs and civil liberties costs. For law enforcement costs, Dave argues that I neglect the other law enforcement benefits such as catching drug use and guns. True, the searching might help the police enforce other criminal laws, but I worry that the “special justifications” for fighting terrorism will then be used as a way to conduct general policing. The issue is whether we want ordinary crime policed at the same degree of invasiveness as terrorism.
The civil liberties costs are high, which Dave admits. There are also other costs as well, such as inconvenience and hassle, something which New Yorkers don’t like very much. Frankly, I wonder how long New Yorkers will be willing to put up with these searches.
Now, of course, let’s assume that the searches are not done using some kind of racial profiling – that they truly are random. If they’re not, then we need to address the profiling issue, which involves another cost Dave isn’t accounting for.
Finally, I’m a bit confused by Dave’s example. He uses a model of 1000 random searches per day, and calculates a “7% chance of being searched over the course of a year of weekdays.” I don’t know enough to say whether his rough calculations are correct, but I question the basic underlying assumption. Why look over the course of a year? Doesn’t this assume that a person rides the subway each and every day? Are terrorists likely to ride regularly each day and always be transporting materials for their plot on the subway each time?
Dave makes his arguments with humility, admitting that many of his points are made based on assumptions and models of behavior that he’s not entirely sure are correct. My arguments are made with a similar humility. I’m speculating a lot and am resting on a number of assumptions too.
There is one argument Dave doesn’t raise against my position that is worth thinking about. He might contend that even if I’m right that the searching provides mostly a “cosmetic” benefit, is there still a benefit worth considering? If a cosmetic but ineffective security measure makes people feel better, doesn’t making people feel better have value? So if Security Measure X is much less effective and more costly than Security Measure Y, but X makes people feel much better, to what extent should this attribute weigh in the balance? But even if we can placate people based on false perceptions, should we?
TrackBack URL for this entry:
Listed below are links to weblogs that reference Terrorism, Deterrence, and Searching on the Subway:
Brief responses to your thoughtful post:
"Anybody willing to blow oneself up in the process of committing his or her crime is hardly risk adverse."
Well, sure. But risk aversion isn't an on-off life switch; terrorists can be very risk averse in terms of avoiding potential loss of their ability to execute their long-term goals (they are surely less likely to, for example, speed, buy drugs, shoplift, etc., than ordinary folks or even criminals). These risk preferences are what differentiate terrorists from, say, sociopaths.
"If Dave is right, why on 9-11 did the terrorists try to use planes?"
Because it was a relatively cheap way to make a very visible point. Subway attacks, which hurt not only those directly attacked but also millions of fellow commuters for long periods of time after the incident, are quite similar. That is: if you force the terrorists to bomb a specific location, instead of a transportation medium, you've gained something in a terrible calculus. On your logic, most of our anti-terrorist prevention activities would seem to be a waste of money because they merely shift risk from one point to another. Not so: the goal should be to protect high-vulnerability/high-impact targets first.
"I don’t know enough to say whether his rough calculations are correct, but I question the basic underlying assumption. Why look over the course of a year? Doesn’t this assume that a person rides the subway each and every day? Are terrorists likely to ride regularly each day and always be transporting materials for their plot on the subway each time?"
Sure, the 7% risk is high given usage patterns, but the point was that it wasn't quite so impossible to imagine being searched as the low daily odds initially suggest. And my general point was that even if the odds look low (from an ex ante perspective), decision making under uncertainty is likely to be quite risk averse. Even assuming the terrorist cell has to make 5 prepatory rides, that is five shots at what they would perceive to be a catastrophic risk of discovery.
Posted by: Dave Hoffman | Jul 25, 2005 5:05:23 PM
Even assuming the terrorist cell has to make 5 prepatory rides, that is five shots at what they would perceive to be a catastrophic risk of discovery
What's catastropic about it? That's what I don't understand. If they get searched when they're "casing" the place, presumably they won't have bombs on them when the search happens, so no harm. If they get searched carrying a bomb, all they have to do is set the bomb off as soon as the cop says "hey you, open up your bag" and they achieve almost all of their goal. So what's the risk to the terrorist?
Posted by: Paul Gowder | Jul 25, 2005 5:28:42 PM
There are many possible outcomes to a search.
1. Ts are searched when casing the target, with no incriminating pre-bomb devices on them. This results, most of the time, in no police action. Ts are made nervous - and might change their future plans because of vivid recent search experience (think about how rarely you return to a place you've gotten food poisoning, even though you realize that it is mostly random).
2. Ts are searched when casing the target or simply when travelling from place to place, and they have bomb parts (non-explosive) on them. This results in police action some of the time (function of police sensitivity/willingness to bear risk of wrongful arrest suit/likelihood of ethnic profiling). Same deterrence outcomes as #1 + we get the bonus of occasionally actually arresting someone and busting the cell.
Obviously we want to be in categories 1 & 2. And we want the Ts to overreact to the risk of 1 or 2 occuring. We can convince Ts that 1 & 2 are likely outcomes by using techniques detailed in my original post.
But what if 3, you scenario arises:
3. Ts are searched with a bomb on them. Some of the time - and this is the good bit from a public policy perspective - the police either stop the bombing entirely or limit its effectiveness (consider bombing #2 in London last week as an example of this). Ts get nervous; they don't place the device in the right place; they explode it before or after rush hour. From Ts' ex ante perspective, this is actually a catastrophic outcome: less spectacular death = less martydom, right? They too are deciding how to spend scare resources (lives). We can make it more expensive to spend them in the subway, as opposed to elsewhere.
One other issue to think about on the privacy end of the scale is the potential privacy losses that would accompany another devastating terrorist strike in New York. Sure, this argument sweeps far; I'm well aware of EP, racial profiling, and like objections. But it seems to me that if the cops can create a shadow of deterrence over the subway system by expending relatively little privacy cost, they ought to do so. Is this privacy cost too high? Again, I think you have to rely on the answer of actual New Yorkers - and one way to do so is to have the program announced in an election year.
Posted by: Dave Hoffman | Jul 28, 2005 12:03:03 AM
Dave: some brief comments...
On scenario 1, I fail to see why the terrorists would be scared off. They read the newspaper too, they know that random searches are happening, so why would they freak out if it happens? If I'm about to get on the freeway and speed, and I see a traffic cop going the other direction, it doesn't deter me from speeding because I already knew traffic cops were out there. I mean, sure, there's salience effects, but presumably the risk of being searched is already salient to them given the big hooh-hah over the policy and all, plus the fact that they'll see cops searching others. How much more salient will it get because it happens to them?
For scenario 2, I find it entirely too optimistic to think that terrorists will go where, again, they know they're subject to search, carrying incriminating materials that they can't use. I wish they were so stupid!
On scenario, 3, I don't know that we know enough about their calculus. How much of a deterrent is "maybe we'll have to blow it up in non-rush-hour" to them? And can't they minimze those risks, by, say, only carrying around bombs when they think there's a reasonable probability that they'll be able to use them effectively on a moment's notice? How many bombers get caught BEFORE they bomb, anyway? Not many!
Posted by: Paul Gowder | Jul 28, 2005 12:40:23 AM