Airline Screening List Mathematics

You may also like...

3 Responses

  1. ac says:

    A healthy false positive rate is expected even with a very high quality list, given the number of travelers, the frequency of actual threats, and the nonuniqueness of names. The odds are stacked against the scheme. Biometrics have some prospect of reducing the number of false positives.

  2. Adam S says:

    I think your math is off.

    30,000 people have gone through the process of adding their names to the cleared list. We can presume some fraction of them (let’s say 100) are named David Nelson. All of these 100 people are actually a single match against an unknown name in the list. It might be David Nelson, or it might David Ellison, or something else. Without knowing the names of the removed, we can’t guess how many of the names on the lists are being matched.

    Also, we don’t know from the 30,000 number how many matches occur. (TSA has repeatedly claimed to keep no statistics on this, which I find odd, because it means they can’t evaluate if Secure Flight improves anything.) That 30,000 could be (a) everyone who matched, or it could be (b) everyone who matched who also flew in the relevant period, or (c) everyone in ‘b’ who didn’t decide that the new process was so disgusting they’d never fly again, or (d) some subset of ‘b’ who asked the right questions after 2003 and was given the form and believed that the government would do something useful with it.

    I’d go with option ‘d,’ and suggest that its a class whose size is difficult to measure, but is likely far smaller than ‘a’ or ‘b.’

  3. Paul Gowder says:

    AC has a very good Bayesian point… which doesn’t make the list justified, mind.