3 ms·
I think the problem with racial profiling is a Bayesian one. Even if P(terrorist|brown looking person) is slightly higher than P(terrorist|white looking person)
by 467568985476 10y ago
I think the problem with racial profiling is a Bayesian one. Even if P(terrorist|brown looking person) is slightly higher than P(terrorist|white looking person) (which I'm not even convinced is true), P(terrorist) is so insanely low that the result is singling out and victimizing (by the shame of being pulled aside, the intimidation of being aggressively searched, the violation of being treated like a criminal because your ethnic group looks similar to another one, etc.) mostly innocent people based on the arbitrary judgement of what a terrorist should look like. It would cause enormous disruption and social harm, and not really move the needle on successfully thwarted plots.