Comments on: I think his name was Chips Ahoy

By: Professor Coldheart

Professor Coldheart — Fri, 09 May 2008 02:13:24 +0000

Should I say “more people know about how unreliable eyewitness testimony is than how hard it is to quantify the meaning of a DNA match?”

By: Phanatic

Phanatic — Thu, 08 May 2008 23:31:46 +0000

I think people already know instinctively about the nature of false positives in eyeball IDs.

I don’t think they do. Eyewitness testimony in trials is hugely effective, even though it’s the most unreliable sort of evidence short of astrology.

By: Professor Coldheart

Professor Coldheart — Thu, 08 May 2008 03:52:25 +0000

I think people already know instinctively about the nature of false positives in eyeball IDs. DNA science is treated as more exact (which it is, provided you understand the probabilities involved).

And Bayesian priors are not the only grounds on which I object to a national DNA database, but you probably could have guessed that.

By: Byrne

Byrne — Thu, 08 May 2008 00:57:43 +0000

A DNA database is much easier to quantify. How do we Bayesify the question of how likely it was that Witness A saw the suspect running from the scene, Witness B saw him discard a handgun two blocks away, and Witness C heard him make a death threat the day before. People can already get convicted on that basis, even though the possibility of false positives and false negatives is there.

By: Professor Coldheart

Professor Coldheart — Wed, 07 May 2008 23:26:56 +0000

“Everybody” knows? You’ve seen some of the arguments judges buy, right?

By: Phanatic

Phanatic — Wed, 07 May 2008 23:07:02 +0000

In one sense, it’s better to have a massive DNA database, since it makes the false matches obvious. Test the suspect’s DNA and get 345 hits, and everybody knows that’s leaving a lot of reasonable doubt open. Test it and get 1 hit, and he’s convicted.