3 ms·
Let's say you have 200,000 suspected criminals, some % of which are criminals. How many of them are you willing to put in prison? If it's looking that most like
by iopq 10y ago
Let's say you have 200,000 suspected criminals, some % of which are criminals. How many of them are you willing to put in prison? If it's looking that most likely around 30% are guilty, do you put 60,000 in prison and say "oh well, some false negatives, some false positives, whatever"
or do you just put 30,000 in prison to prevent most innocent persons from going to jail? Then you are letting half of the guilty people go free
It's a question of how many innocent people you are willing to jail to catch more real criminals
- cortic 10y agoI see what you are getting at, but it would still be; 60,000 = "oh well, some false negatives, some false negatives AND false positives, whatever" 30,000 = "oh well, some false negatives" Although i get the feeling we are moving into semantics, its just always bothered me the way this was worded. Anyway thanks for that perspective, made me think.
- angry_octet 10y agoOr to give it an engineering flavor, the ROC: https://en.wikipedia.org/wiki/Receiver_operating_characteristic https://en.wikipedia.org/wiki/Receiver_operating_characteris... It seems we accept a higher FAR (False Alarm Rate, a diagnostic which gives a false positive) for certain communities (poor people, minorities) than others. The consequence matters too: even people who support the retributive aspect of the death penalty may reject it in practice because of the excessive false conviction rate.