Reframing the evolutionary benefit of sex

either evolution is surprisingly forward-looking

Interestingly, this relates to our discussion about episodic vs non-episodic learning algorithms. In this case it seems clear that evolution is not episodic and assuming very large population sizes ought to maximize long-run inclusive fitness. So the puzzle here is that if it takes 300 generations for sex to break even, then if a mutation caused some member of a sexual species to start reproducing asexually, the sexual population would crash to 0 before it could recover.

My idea for solving this (which I just thought of now so take it with a grain of salt) is, because a sexual species can maintain a genome against a much higher mutation rate than an asexual species can (see past discussion [LW(p) · GW(p)]), an asexual species needs to have a much lower mutation rate (i.e., much more machinery to prevent/repair mutations) to survive. When an asexual population arises from a sexual species, it doesn't have the extra protective machinery and therefore quickly succumbs to accumulation of harmful mutations, perhaps before the sexual population can go extinct.

Or if it does drive the sexual population extinct first before itself going extinct, if most species are sexual then a phenotypically nearby species can just come occupy the now vacated niche.

Mutations of germline cells come with huge fitness penalties. Taking 0.01% of your genes introduces an extremely small amount of variance.  And unilaterally replacing 50% of my genes with yours is equivalent to a 50% drop in fitness

99.9% of those genes will be identical, hence why I used the 0.01% number, if you want to induce more mutations you can (see bacteria) and if you want to introduce more mutations in a controlled way (i.e. not break anything important) you also can, and humans actually already do this, as do most multi-celular eukaryotes (see B cell selection in response to antigens, for example)

The bacterial method itself also doesn't seem to work in humans (because you need to have the genes during development, and less importantly you need to spread them throughout your body). So it seems to me like sex adds very significant benefits over these alternatives.

I'm not sure if it works in humans, but humans did not invent sex. Simple multi-ceulular eukaryotes did. In those species HGT works just fine and is indeed a contributor to evolutionary variance that got them to where they are. I had a great paper on this but can't find it, see for example: https://onlinelibrary.wiley.com/doi/pdf/10.1002/bies.201300007

It's not a big thing in practice, because sex works better, so it makes sense to protect against HGT in all but a few niche cases (transfer from bacteria-like organels such as mitochondria)

Is your view here coming from some quantitative estimate or further reasoning you didn't include in the comment, or is it reflecting the consensus view in some field? In either case, it would be great to see a pointer. If this is just your guess based on the reasoning in the comment, that's fine and I'm happy to leave the argument here (or with your rebuttal).

It reflects a consensus view in the field in so far as nobody has tried to justify sex in terms of variance in genetics since the 70s (see the very wikipedia article cited) and even that tried to justify it was being variance under very specific conditions that allow it to be less dangerous and/or with a higher chance of varying in a "right" direction.

Variance is for sure an advantage. But it's not the advantage. Since there are 1001 ways you can introduce variance.

So both under common sense and under all accepted models I know of the intuition presented here is wrong.

And you don't need much to figure that out, again, just try to answer the question:

If variance is the main reason we have sexed reproduction than why not do it bacteria-style, or more realistically why not do it B-cell-selection style but for germinal cells?

That also means the variance could be better controlled by environment, you could literally say "I want exactly x% variance".

Additional questions to ponder might be:

• Why always 2 sexes (outside of weird fungi)
• Why is the sex stable (outside of a few niche animals)

Those will also point you in the right direction of actually figuring out why we have sexed reproduction.

Again, this is not to say variance is not a benefit. Much like "locomotion over flat ground" is a benefit of crab legs. But "variance" is something boring that can come about 1000 different ways. The reason crab legs are interesting and special and worth evolving is showcased when looking at how crabs move on steep terrian. Similarly, the reason sexed reproduction is interesting and special and worth evolving only becomes obvious when thinking of things that you can only accomplish using it (or at least that are much more costly to accomplish otherwise, and that no other lifeforms do in other ways)

***

If the above doesn't make sense please tell me and I'll try to rephrase

If you want a good pop-science book that goes over how biologists think of the role of sexed reproduction you could check out "The Red Queen: Sex and the Evolution of Human Nature". I read it like 4 years ago, when I knew nothing on the subject, and while in hindsight it has some bad points (and I don't agree with the central hypothesis) it does a good job of showcasing the different views in the field in a way that you can get with limited knowledge of the molecular level processes involved.

As in, given that the information here is seemingly false, I'm unsure why it appeared under the "Recommended" section of the website.

Please ask them to auto-crosspost ai-alignment.com too, unless you're not doing that intentionally? (I have some comments on your latest post there and would rather discuss here than on Medium.)