3 ms·
Schneier On Security: The Trouble With Airport Profiling
- maeon3 15y agoLets suppose there are going to be 10 events in the next 5 years where passengers try to take down airplanes from the interior. Also suppose there are going to be 800 million people per year boarding planes. Any system that identifies the 10 people among the (800 million * 5) will have to have "ten nines" worth of accuracy: 99.99999999% in order for the system to have any value. These systems do not exist, and almost provibly cannot exist. We are back to square one: brute force, check each passenger for every possible threat.
- B-Con 15y agoThe most important argument he mentioned against profiling, IMO, is that attackers can change to evade profiles. In the case of terrorism, there are about 1 billion Muslims and only about 300 million Middle Easterners. It's trivial for radical Muslims to recruit non-Middle Easterners, and they do. Profiling on Middle Easterners sets you up to only catch 1/3 of the potential attackers at best, and that assumes they don't pick their recruits to not look suspicious. The type of thinking that thinks that profiling is effective in this type of case is generally rooted in the naive security approach of treating malicious adversaries as random accidents. Most people are better at defending against tornadoes than hijackers. Tornadoes are random and dumb, they just pop up here and there and don't adapt to your defenses so defending against them is a 1-step process. Malicious adversaries, however, make the game multi-step. We reveal our defenses, they probe them and adapt to them. We then have to probe and adapt to those as well. And so on. You have to constantly out-think the attacker who is trying to out-think you. Having had many conversations on the topic of airport security with random people here and there, it would seem that most people's threat model assumes that terrorists, inside and outside airports, make their plans barely a day in advance with little to no knowledge of the defense system, stand helplessly by making no active attempts to thwart the defenses, have little to no creativity or originality, and are kind of dumb. In all honesty, that's the attacker that most people seem to have in mind. Those kind of people do exist, but they've always existed an law enforcement is long since accustomed to dealing with it. They aren't the threat. It's the smart, probing, adaptable, well-funded attackers who we always have to think about. All that said, to be fair, some degree of profiling can be advantageous under two conditions: when the target has actually a known non-uniform distribution that can be exploited (not the case for airport security), and only if the profiling is kept secret. When the attacker evades a profiling system, it's because they a) know a profiling system exists, and b) knows how the profiling system works. (Sometimes knowing (a) is itself a strong hint at (b).) If both are kept secret, the attacker may not be able to adapt. But the TSA couldn't keep it a secret, they have way too many employees that could leak information (intentionally or accidentally) and all their screening decisions are public, so an attacker would quickly learn the pattern anyway.
- serverascode 15y agoWe should be stopping possible terrorist attacks before they get to the airport, not after.