Friday, June 20, 2008

What is it with this demand for certainty?

I recently suggested that we should never be 100% sure of anything. It's pretty obvious but on average we suffer pretty serious biases in favouring certainty. Drugs must go through ridiculously touch testing while people who could be saved, die and we often are more disturbed by a 100% to 90% shift in probability than a 80% to 20% one. What? There are loads of examples and they're part of the big literature on biases.

I was struck by this while watching Minority Report again. It's just taken for granted that if one error is made, by a cop, precog whatever, the whole system will collapse. It's never spelled out, but doesn't this imply that no error has ever been made in the regular criminal justice system? If not, why hasn't it been shut down?

If reducing the error rate in crime detection and prevention by zillions of orders of magnitude isn't enough to recommend this new system to the author, then I don't think I'd have much luck in persuading him that he's an idiot.

No comments: