In response to the July 7 terrorist bombings on London’s tube system, on July 21 police began conducting random searches of bags and packages carried by people on the New York city subway; those who refuse to be searched are not allowed to ride. The New York City Civil Liberties Union is representing plaintiffs in a suit challenging the random search program. I was asked to by the NYCLU to be an expert witness in the trial to assess the effectiveness of such a program – not withstanding any legitimate Fourth Amendment concerns about the right of people to be “secure in their persons, houses, papers, and effects against unreasonable searches and seizures.”
The New York City Police Department claims the subway search program is an effective deterrent against terrorists because it introduces uncertainty and unpredictability that terrorists can’t anticipate. They argue that the mere possibility that a terrorist might encounter a search is sufficient to cause a terrorist to give up on any plans for bombing the subway. Curiously, the NYPD refuses to discuss the numbers that might demonstrate the effectiveness of the search program.
So my job as an analyst is to posit what the effectiveness might be. The New York City subway system has 468 stations, with over 1,000 entrances. On an average weekday, approximately 4.7 million riders enter the city’s subway system. The NYPD concedes that it cannot conduct searches at all of the stations all of the time. According to a survey conducted by the NYCLU of 5,500 subway turnstile entrances from Aug. 25 to Sept. 16, there were checkpoints at 34, or about one-half of one percent.
Clearly, one-half of one percent is not very extensive coverage, and it would be hard to claim that the random search program is effective if the NYCLU survey was representative of the NYPD’s efforts. So let’s assume that the NYPD could do 40 times better and conduct searches at 20 percent of the subway system entrances. Indeed, this would be a fairly significant police effort – 600 officers at 200 entrances (three officers are required to run a checkpoint: one to supervise, one to randomly pick riders to be checked, and one to search bags). Although subway ridership is not uniform throughout the day or across all the stations, we’ll assume that this translates into 20 percent of the riders – 940,000 people – being subject to search.
But not all 940,000 riders will be searched. Instead, checkpoint supervisors decide whom to search based on every nth person in line. So if every 10th person coming through a turnstile is selected, then 94,000 riders will be searched.
So what are the odds that someone riding the subway will be searched? In this particular example, 2 percent. The actual number would change if the assumptions about the number of entrances and frequency of searches changed. But this simple mathematical analysis demonstrates several important points. First, a fairly extensive effort requiring a significant number of officers yields a very low rate of return – which means a terrorist has a very high expectation of success (98 percent in this case), despite the possibility of encountering a checkpoint. Second, to reach double-digit effectiveness (and 10 percent is still not highly effective and, therefore, not likely to have significant deterrent value) the NYPD would have to expand its effort fivefold to search 470,000 people.
Yet even if the NYPD could somehow find the manpower to conduct 470,000 searches, a strange feature of the subway search program essentially reduces its effectiveness to near zero. The way the program is constructed, anyone can walk away from a checkpoint – even after they are selected for search – without any action taken against them. Moreover, people don’t even need to get to the turnstile to know that a search is being conducted because they are given advance notice that there is a checkpoint at an entrance via signs and public address announcements.
So what are the odds that a would-be terrorist would willingly subject himself to search? Near zero. Instead, he would simply walk away from an entrance that has a checkpoint and find another entrance that doesn’t have one. The NYCLU survey suggests the vast majority of entrances don’t have checkpoints, which means the probability of finding a way into the subway system without being detected is extremely high – bordering on nearly 100 percent – despite any unpredictability or uncertainty of encountering a checkpoint.
I know the NYPD takes its job of protecting the citizens of New York seriously, so one has to wonder why their container search program would have such a gaping loophole. The answer lies in the fact that, apparently, this and other details of the program were designed by a lawyer in the department’s legal bureau, not their counterterrorism division.
Nonetheless, the head of the counterterrorism division claims that the random searches are an effective deterrent (although the fact that the NYPD refuses to disclose any data on the program suggests that maybe their own numbers would produce results very similar to or worse than my hypothetical example). The NYPD argues that random security checks on the Brooklyn Bridge caused Iyman Faris to decide that “the weather is too hot” and call off a plot to bomb the bridge. But comparing the Brooklyn Bridge to the New York City subway system is mixing apples with oranges. The Brooklyn Bridge plot involved cutting the suspension cables that hold up the bridge. So placing security at either end of the bridge (which is what the police department did) would be an effective way to prevent access to the cables and thus deter a terrorist. But unlike the Brooklyn Bridge, the New York City subway system has 1,000 access points, not two. So placing security at only a few access points doesn’t prevent entry via another location and is not an effective deterrent – especially if a terrorist knows he can easily evade the security and find a way to penetrate the system.
Even more incomprehensible is that the NYPD argues that access to interrogation of the likes of Khalid Sheikh Mohammed somehow gives them confidence that the random subway search program is an effective deterrent. Yet by their own admission, the terrorists who pose a threat are smart, plan thoroughly, engage in extensive surveillance, and are motivated to succeed. It seems that the NYPD believes that a smart terrorist would behave only in the way they predict in order to make the program effective and that adaptive behavior is not possible. But we know al-Qaeda learns and is adaptive. And you only need common sense – not access to interrogations and other classified information – to figure out that the barriers posed by the random subway search program are very low and easily circumvented.
In many ways, the NYPD’s container search program looks like pre-9/11 airline security. It is designed to be effective if – and only if – the terrorists act exactly as the NYPD predicts they will act. But there’s no reason to believe that terrorists will behave in such a manner. For example, the NYPD is rightly concerned about a terrorist attack during the rush hours, so that’s when they’re more likely to be conducting searches. But because the subway system is open 24/7 and there are no restrictions on how long a person can remain in the system, a terrorist who wants blow up a train during rush hour can enter the system before rush hour when there is likely to be less (or zero) security in place and ride around the city until the appointed time to detonate a bomb. How hard is that to figure out? And would that not fit the definition of a smart terrorist who plans thoroughly and is motivated to succeed?
What 9/11 demonstrated is that a-Qaeda and its ilk will think creatively. If there are relatively easy ways to exploit weaknesses in security to achieve success, that security will not pose an insurmountable barrier. Indeed, all of the 9/11 hijackers had to walk through airport security and did so believing that they would succeed. If we are dealing with smart terrorists – as the NYPD asserts – then we need to be smarter. On that count, New York City’s random subway search program fails to make the grade.
