Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

J_Thomas comments on An Alien God - Less Wrong

80 Post author: Eliezer_Yudkowsky 02 November 2007 06:57AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (137)

Sort By: Old

You are viewing a single comment's thread.

Comment author: J_Thomas 14 November 2007 12:15:47PM 0 points [-]

Douglas, your ideas are reasonable but unproven.

It certainly makes sense that new proteins with new functions should arise by recombination among old proteins with old functions. Start with functional groups that do things -- hold a calcium ion, hold a magnesium ion, fit to a lactam group, etc -- and fit them together in just the geometry that gets a result, and then fiddle with the details to change that geometry slightly. Sure, that makes sense.

And to get brand new protein structures you need to evolve them special -- to get selected starting with a protein that has a different structure you need to go through intermediate stages that are likely to have no function at all. But if the protein is under selection that won't happen, and if there's no selection it's sill quite unlikely.

So it's plausible there should be some mechanism to do all that.

However, the trouble here is that this mechanism would have to be mostly inactive. Do we create novel new proteins every generation? Every hundred generations? Not as far as I know. What maintains this mechanism that produces a good result in one individual per thousand generations (maybe), and produces no result or bad results the rest of the time? It would have to be a side effect of something else, something that does get used a lot. And that may be possible, but would you expect it to be the same mechanism in prokaryotes and eucaryotes? We haven't observed it yet in that case. So two different mechanisms, like they have vastly different ways to have sex.... And we still haven't particularly found those mechanisms.

I don't regard your tuberculosis story as useful. OK, lots of things are preadapted toward making certain discoveries. But not other discoveries.

Sure, it isn't random. But the assumption of randomness was a mathematical convenience. Mutation shouldn't be random, There's been 3 billion years of selection to encourage mutation at the times and places it does the most good -- to the extent that organisms can predict that.

There's a lot about the way populations respond to natural selection that isn't known yet, there are exciting discoveries waiting for us. You and I both have hints about what those discoveries will look like. But the discoveries haven't been made yet and we're only guessing about them at this point. I can guess pretty well about some genetic mechanisms that would increase the rate of evolution, but I can't guarantee that nature hasn't found even better mechanisms that outcompete the ones I imagine.