That is incredibly awesome, if it actually turns out to be helpful. I find it hard to believe it is more helpful than a controlled (and much smaller) study of human problem solving algorithms, but I suppose sometimes quantity really does beat quality for analysis.
Well, that's what I meant. A controlled study of the ways humans approach the problem of folding proteins, where they can survey the people and find why certain methods are used. But, I guess they can watch the progress of the humans who score the best and see what sorts of things they do to the protein in question, which is probably more useful in this case.
The point is to find the most likely way these proteins are folded, which is beyond our current problem-solving algorithms. The process of finding those unpredictable folds, while interesting (I hope they're collaborating with a psychologist who can get useful data from that information), is entirely beside the point. Which is good, because I strongly doubt a bunch of molecular biologists are going to want to come up with a study of human problem-solving that over a hundred years of looking into the matter hasn't tried yet (because if they were they'd be psychologists, not molecular biologists).
Ah, I think I'm completely misunderstanding the purpose of the test. I thought they were trying to improve their AI by augmenting their algorithms with human problem solving techniques as applied to this specific problem. But you're right, that's more CS/Psych than molecular biology despite the application. I guess they just need proteins folded various ways.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
brynndragon) wrote2009-01-02 01:36 am
no subject
no subject
no subject
no subject
no subject
no subject
no subject
Yet.
no subject
no subject
no subject
no subject