from contemptslot:
The Lie Detector That Could Decide if You’re Getting On The Plane
Haven’t we already decided that lie detection isn’t reliable? Isn’t this a massive step backward?
Let’s make the huge assumption that this will work outside the lab, and run the numbers. The claim is of 90-100% accuracy, so we’ll split the difference at 95%. Since it’s a binary determination (either you’re lying or not), to be “inaccurate” would either be a false-positive (innocent person detained) or false-negative (liar gets on the plane). We don’t know if there’s a bias one way or the other, so we’ll split them down the middle.
In 2006, the TSA performed over 708 million screenings. Rounding to 700 million and using the failure rates described above, we’d have 35 million false results. That would be 17.5 million liars that got on board, and 17.5 million truthsayers detained in error.
But wait. Potential terrorists don’t make up fifty percent of the traveling public.
The government’s terror watch list is reported to have 755,000 names on it. Let’s pretend it’s been compiled by the most efficient government program ever, and every name on it is correct. On top of that, let’s assume there are 245,000 “unknown unknowns”, giving us an even million terrorists and that every single one will try to fly. With a five percent failure rate, that’s 50,000 that’ll get by the “brain machine”. That also means 34,950,000 innocent travelers detained.
Even if we reduce the number of terrorist flight attempts to 10,000 and bump the accuracy up to 99%, we’re still talking about 100 terrorists getting on board. So nearly one hundred thousand people detained every day, and still a terrorist gets through every three days or so.
Let’s just invest the money in real human intelligence activities, and spare ourselves the pointless exercise in security theatre.
