So lets take this.
- Everyone getting onto a plane is searched, this should reveal potential weapons and bombs.
- This search is done at a chokepoint.
- Public Transportation systems, in order to be effective, have very few chokepoints, and what chokepoints that exist are high volume.
- We cannot search everyone in a Pubtrans system, because of those chokepoints.
- Any search must then be against a small amount of the total population.
- Assume we can search 10% of the population, this leaves a 90% chance of missing a terrorist.
- If we profile for terrorists, and increase the chances of being searched for individuals matching the profile, we should be able to direct our search, making it more likely that the searched 10% contains a terrorist.
- This decreases the chances of a search for terrorists that don't match the profile.
- The probablity of a successful attack against a Pubtrans system is decreased, but not by much. If we target according to profile, and ensure that 50% of those searched matches the profile, then we only search 5% of everyone else, making a successful attack 95% likely for terrorists that don't match the profile.
The math doesn't work. When you throw in the bones tossed to Civil Liberties, like the fact you can refuse the search and go elsewhere, a terrorist can decline the search and leave, trying again at another location, meaning, that even if they match the profile, and we search half of those that match the profile, then he can make an infinite number of trials with a 50% chance of success at every point. Eventually, he can get through.
Searching doesn't work, at least random searching doesn't. Profiled searches won't either. So what do we do? Well, a good first step would be intelligence that allows us to pinpoint terrorists well before they get to the subway. But we could also invest in non-invasive search techniques, like cadres of bomb sniffing dogs. If the dog smells something, there's now probable cause to search the individual.