Last night I saw The Road, a truly bleak post-apocalyptic movie.  [Warning: Minor spoilers.]  As I watched, I realized that I was witnessing a mighty counter-example to my views on the propriety of statistical discrimination.  In the movie, about 80% of the people seem to be murderous cannibals.  This is common knowledge.  As a result, everyone is tempted to shoot first and ask questions later.  After all, even if two perfectly innocent human beings bump into each other, each can rationally assume the worst about the other.

Notice the tipping point.  Once p(a stranger is a murderous cannibal) gets high enough, morally confident statistical discrimination spirals out of control.  Even if the stranger down the road isn’t a cannibal, he has a strong motive to preemptively murder you – which gives you a strong motive to preemptively murder him.

Given a grim starting point, statistical discrimination can quickly lead to mass human extinction even if decent folk adhere to a “beyond a reasonable doubt” standard to convict.  For example, if your threshold is 99% certainty, everyone in a world of 99% murderous cannibals has a license to shoot on sight.  And if, like me, you accept the preponderance standard for conviction, the situation’s even worse.

My best reply, I’m afraid: Just because you probably have a right to do X, doesn’t mean that it’s right to exercise that right.  Non-cannibals in The Road usually have a right to shoot first, but it’s morally wrong to do so when they can gather information at moderate cost, or simply avoid conflict by hiding or running. 

Got a better story?