3 ms·
You are not thinking sufficiently Bayesian. While you could say that there is an elevated chance that the terrorist is of Arabic descent (and frankly, the odds
by jlgreco 14y ago
You are not thinking sufficiently Bayesian.
While you could say that there is an elevated chance that the terrorist is of Arabic descent (and frankly, the odds there are not that strong. Maybe better than 50%, but I wouldn't put my money even on that...) that is only a small part of the picture. You must also consider the possibility that any given person of Arabic descent is a terrorist (an absurdly small probability). When you combine these two factors then you can decide how probable it is that he is the terrorist, given his descent.
If you actually work out the numbers here, you will undoubtedly discover that the probability he is responsible falls well below any rational standard for harassing the man.
Edit: To get a grip on what sort of standard you would usually demand: consider a murder in a small isolated town. With no clues at all, lets say the murderer could be anyone. In a town with 3 people, each person could be said to have a 50/50 chance of being the murderer. Is that enough for search warrants for both (not under our legal system, but just ethically/morally)? Sure, I'd accept that. If there were a hundred people in that town? No way. So, does "terrorist, given arabic descent" go over 1%? I HIGHLY doubt it.
- vadman 14y agoI don't know much about Bayesian statistics, but wouldn't it be "terrorist, given Arabic descent AND participation in the targeted event AND proximity to the explosion AND high probability of a follow up explosion (judging from past events) AND ..."?
- jlgreco 14y ago> wouldn't it be "terrorist, given Arabic descent AND participation in the targeted event AND proximity to the explosion AND high probability of a follow up explosion (judging from past events) AND ..." The only thing that man has over the likely thousands of others that in some way participated in the event is his skin color. Everyone else there had those same properties. So if we are considering the population of people in the general area at the time, then no. They used only his descent. They, and djanogo[edited], used P(arabic|terrorist) to justify a search. In order to justify a search you need a reasonable P(terrorist|arabic). Assign some reasonable numbers to each factor off the top of your head, work it out on a napkin, and come back to me with whatever figures work out to be over whatever a reasonable threshold might be.
- danenania 14y agoGiven that the coordinated way the two bombs exploded indicates that they probably didn't malfunction, it seems like proximity to the explosion should be a negative indicator, no?
- jlgreco 14y agoIt's probably a tossup. How much time can you usually put on a kitchen timer, an hour? Cut off the few minutes the timer would have had to be set early (since the timer presumably was not started when the bombs were placed, since that would be pretty conspicuous). That puts the guy at most an hours distance away, which granted, could be many miles. If the bombs were really placed in backpacks that were left sitting on the ground, I would further suspect that the bomber placed them much later an hour before. Wouldn't want someone to steal your bomb... Furthermore, maybe the sick fuck wanted to watch so he stuck around at a short distance instead of booking it. Who knows.