Today's post, Natural Selection's Speed Limit and Complexity Bound was originally published on 04 November 2007. A summary (taken from the LW wiki):


Tried to argue mathematically that there could be at most 25MB of meaningful information (or thereabouts) in the human genome, but computer simulations failed to bear out the mathematical argument. It does seem probably that evolution has some kind of speed limit and complexity bound - eminent evolutionary biologists seem to believe it, and in fact the Genome Project discovered only 25,000 genes in the human genome - but this particular math may not be the correct argument.

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 Evolutions Are Stupid (But Work Anyways), 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.

New Comment
8 comments, sorted by Click to highlight new comments since: Today at 9:26 PM

I'm rather skeptical of the whole "one mutation, one death" thing. That maybe could seem plausible, if mutation is viewed as a black box that randomly makes changes to the macroscopic properties of an organism, but from a biochemical view it seems rather bizarre.

First, each protein is coded for by a set of 3 base pairs in DNA, a codon. Since there are 4 bases, there are 64 possible codons. There are 20 (common) amino acids, so there are, on average, 3.25 possible codons per amino acid. Since the codons for a particular acid frequently differ in only one base pair, sometimes a single mutation will not change a protein at all. There will literally be no changes whatsoever. And in fact, biochemists have discovered that humans bearing the same genes, coding for the same proteins, frequently have different DNA sequences. A mutation of this sort seems unlikely to cause even one death.

Secondly, it is entirely possible to change the primary sequence of a protein (the sequence of amino acids) without changing the three dimensional structure at all, or in minimal ways. When biochemists are studying a new protein, a frequent test is to change single amino acid residues, and observe the impact on protein function. Most of the time, it really doesn't do much. There are always a few amino acid residues which, when changed, completely prevent the protein from functioning, but there are usually only a handful of that type of residues in an entire protein (which contain hundreds of amino acids). And then, if a mutation did something like swap out an asparagnine for a glutamine would do very little, since their chemical properties are very similar. (as a side note, I do not endorse, nor did i even read, the pages where those links go. I just did an image search). Many mutations like this won't really do anything.

Mutations become even less harmful if the target is some sort of folded RNA. You would have to have dozens of mutations within a single short strand of DNA before you could even notice a change in the secondary structure of a folded RNA molecule.

Could someone point me to something on evolutionary biology that discusses the one mutation, one death principle in more detail? I'd like to know if I'm missing something here.

"One mutation, one death" is, of course, silly if taken literally - but the idea might have some utility if the purpose is to establish an upper bound on the quantity of information that selective deaths can provide.

The general principle of harmful mutations being eliminated from the population is good. But in the simulations that appear to have been run, it looks like the "mutation rate" and thus the death rate, was the rate at which a copying error makes a mistake in one DNA base pair. The actual rate you would want to use would be massively lower.

I looked at the wiki, as recommended in the original article and it said several times that in the models being used, mating is assumed to be random. What happens when you alter the model so it has mate-choice, as many species do in the wild?

Also, the wiki article also said something about assuming perfect selection. Does modelling selection as imperfect gain you anything?

What sorts of species don't have mate choice? What is the most complex/evolved one (I have no clue what the proper terminology is)?

Species that just dump their sex cells into the water/air to fertilise randomly, like sessile sea-creatures, plants (not knowing the bio-terminology, I don't know if bees etc. transferring pollen could count as some sort of choice or not)

As far as 'most complex/evolved' - even if I could figure out how to decide how evolved or complex something was... I don't have the knowledge.

Then we should check the model against those cases it claims to model perfectly, since there are so many.

From what I can gather, Eliezer made an error, and figured out what the error was. Why can't he redo the calculations?

Exactly how likely is a mutation to kill you? If it's a low probability, it seems like having two mutations would make you about twice as likely to die, so this wouldn't have any real effect.