[SEQ RERUN] Evolutions Are Stupid (But Work Anyways)
Today’s post, Evolutions Are Stupid (But Work Anyway) was originally published on 03 November 2007. A summary (taken from the LW wiki):
Evolution, while not simple, is sufficiently simpler than organic brains that we can describe mathematically how slow and stupid it is.
Discuss the post here (rather than in the comments to the original post).
This post is part of the Rerunning the Sequences series, where we’ll be going through Eliezer Yudkowsky’s old posts in order so that people who are interested can (re-)read and discuss them. The previous post was The Wonder of Evolution, and you can use the sequence_reruns tag or rss feed to follow the rest of the series.
Sequence reruns are a community-driven effort. You can participate by re-reading the sequence post, discussing it here, posting the next day’s sequence reruns post, or summarizing forthcoming articles on the wiki. Go here for more details, or to have meta discussions about the Rerunning the Sequences series.
I don’t think the chance of fixation formula given (2s, where (1+s) is the fitness) is correct. It seems to fail some basic sanity tests.
First of all, it’s possible to imagine a mutation that makes you have 10x the number of children of a regular person. This gives us s=9. Is the probability of fixation then… 18?
Second, it seems like this would have to depend on other factors, such as the base probability an individual has children in the first place. In a dangerous environment where (as a base rate) 99% of all individuals die before they can reproduce at all, the individual has 0.01(1+s) chance of surviving, which is an upper bound for the probability of fixation. In an environment where everyone is guaranteed to have at least one child (but some have more), the mutation should (if dominant) stick around no matter what.
Good spot. The P(fixation) = 2s formula is actually an approximation that holds only when s is small but positive and the population size is very large. A more precise formula (here’s a publicly-accessible paper deriving it) is P(fixation) = (1 - exp(-2s)) / (1 - exp(-4N**s)), where N is the population size. When s is small one can approximate the numerator by 2s and when s is positive and N is infinite the denominator becomes 1 — so when all of these conditions hold the formula reduces to 2s. In your example s is large and the P(fixation) = 2s approximation is no longer a good one.
Really? Can you give an example? Because I don’t see how that could be done.
Edit: Oh, right. Nevermind.
I think he means that in the sense of a theoretical possibility that shows that the math must be flawed. Not something likely to actually crop up.