Contra Paul Christiano on Sex
I have come to think that many people don’t understand evolution. Not in the “believe that a sentient deity created the universe after a template” sense. A lot of people seem to “get” the idea of evolution in broad strokes but then fail at explaining its various oddities or making inferences based on it.
I count myself and basically everyone I know among these people because evolution is actually surprisingly hard. The reasons why its hard to get are exactly why understanding it better helps in a multitude of other domains.
Evolution is a simple principle, but when faced with the oddities of reality, it gets exponentially complex. If you were to think of evolution with causality backwards, it’s like a programmer trying to write a highly optimized piece of software that, in theory, could fit into 100 lines of lisp. But, when faced with the realities of hardware and customers, it turns into a sprawling codebase with thousands of files, hundreds of dependencies and a mixture between multiple languages.
It starts “simple”, literally, in that it can start with just RNA, a model under which “survival of the fittest” works in its most simplistic form, sometimes in a very literal lock-and-key sense of “fit”. But then it quickly (d)evolves into things like apes and cetaceans, which make absolutely no sense when viewed in this framework.
This mirrors the real-world process of engineering, and it provides a very good foil for any sort of “positive feedback” theories ala intelligence explosion.
I am in no position to explain evolution, but maybe I can enlighten some people a bit by showcasing how people get it wrong. In this case, I want to harp on something Paul Christiano (coincidentally an advocate of intelligence explosion type theories) wrote about why sex is useful. Paul Christiano is really smart. So if he can make these kinds of errors, I will assume they are fairly common, and thus this correction might be rather useful, rather than just attacking a strawman.
He writes about the role of sex, to me, his explanation sounds similar to how most biologists in the 20th century would have (wrongly) thought about it:
From the perspective of an organism trying to propagate its genes, sex is like a trade: I’ll put half of your DNA in my offspring if you put half of my DNA in yours. I still pass one copy of my genes onto the next generation per unit of investment in children, so it’s a fair deal. And it doesn’t impact the average fitness of my kids very much, since on average my partner’s genes will be about as good as mine. (ETA: but see the discussion below, in which case the costs might be much bigger.)
But the trade has transaction costs, so I’m only going to do it if I get some benefit. In this post, I’ll tell a particularly simple story about the benefit of sex. I think this is basically equivalent to the standard story, but I find it much clearer. It also makes it more obvious that we don’t require group selection, and that the benefit is very large.
Why doesn’t sex change the average fitness of my kids? The possibility of a “lucky” kid who gets the better genes from both of us is offset by the possibility of an unlucky kid who gets the worse genes from both of us. If the effects of genes are linear, the average fitness will be exactly the same as the parents. In practice I expect it to be slightly lower because of convexity and linkage disequilibrium.
But sex increases the average fitness of my grandchildren, because my fittest children will be responsible for a disproportionate fraction of my grandkids. More precisely, if my if an organism with fitness dX has (1+dX) kids per generation, then the total fitness of my grandkids is E[(1 + dX)^2] = 1 + 2 E[dX] + E[dX^2]. So increasing variance by 1 unit is as good as increasing average fitness by 0.5 units.
Reproductive decisions are naturally a tradeoff between average fitness and variance. Sex slightly lowers the average but increases the variance. If you try to get the same amount of variance with random mutations, you’ll have to totally tank your kid’s expected fitness, because your current genome is well-optimized, and you’ll also pass on fewer genes to the next generation (since some of yours got destroyed). In fact, it’s hard to think of any way to get similar benefits without exchanging genes.
Variance becomes linearly more important over time. For the fitness of my grandkids, 1 unit of variance is worth 0.5 units of fitness; for great-grandkids, they are equally valuable; for great-great-great-grandkids variance is twice as valuable. I don’t think we have to look too many generations ahead before the variance bonus from sex outweighs the costs. For example, if genetic fitness differs by 5% between my offspring, and sex reduces fitness by 1%, then sex breaks even within 6 generations.
This view is “correct” in outlining variance as one of the benefits of sexed reproduction. But utterly fails in thinking that variance is THE benefit that makes sexed reproduction worth it.
It’s like saying “locomotion over flat ground” is the benefit of crab legs. But “locomotion over flat ground” is something boring that can come about 1000 different ways, and crab legs are a silly complex and expensive way to do it. That’s not to say crab legs haven’t evolved, in part, for that. But focusing on that misses the point; The reason crab legs are interesting and special and worth evolving is showcased when looking at how crabs move on steep terrain.
There are much easier ways to get variance in your genes than sex, with much lower costs. Sex is expensive, in some species, such as peacocks and apes, the vast majority of resources over an organism’s lifespan are spent on finding mates. Even in primitive organisms, sex introduces a whole lot of issues. (more on this later)
You can get variance by simple mutations of your DNA, or in more complex organisms, by a mutation in the DNA of germinal cells (e.g. ovums). This is something that does happen, with huge variation in frequency among species. You can also get variance via gene transfer, which is the main mechanism for bacteria and does happen in some simple eukaryotes. But the more “complex” an organism gets, more and more of the variance is sex-related (a few odd species of reptiles aside).
Mutation in germinal cells is awesome in many ways, for one you can control which genes or bases of a gene are more likely to get mutated. For example, assuming you have a really awesome set of genes around this or that enzyme, you can protect it from mutations, and assuming you have a gene that could benefit a lot from variation, you can make mutation more likely.
This is hard in simple organisms, but easy in complex ones. As an example, the adaptive immune system does just this. It engages in “selection events” when presented with enough of an antigen, where a specific set of genes become much more likely to mutate. It’s not impossible to assume such a mechanism could happen in germline cells. Think for example of mitochondrial mutations being selected based on the amount of available glucose. In practice, these environment-based changes seem to mainly happen in-utero and don’t affect DNA directly, but in theory, if we wanted more variance, we might imagine it as a mechanism that could be evolved.
Based on sheer numbers, environments, long-lasting configurations and length of existence, bacteria are much more “fit” than eukaryotes. And this is kinda what they do, they react to the environment and “drop” bits of DNA or uptake DNA dropped by neighbouring bacteria.
So why pick sex instead of one of these low-cost mechanisms?
The answer is… I don’t know, nobody really does, that’s why evolution is complex.
One theory is that sex is a hidden trait preserving mechanism.
The chromosomal setup that allows for sex means we can have hidden traits that only show up in a small % of offspring (e.g. simple story is you need 2x SNPs which are present only on one of the chromosomes for each parent => 25% occurrence). This allows for e.g. sickle cell anaemia to be a thing that can be selected for when malaria incidence is high in the population and selected against when incidence is low. So you have “fitter” hunters in no-malaria times and fitter malaria-resisters in loads-of-malaria times.
You don’t get rid of either genotype entirely, because it can be carried around and only manifested in a small % of kids (which are going to be selected out early on) until it’s needed, and then the small % with the adaptive mutation are heavily selected for.
This “hidden trait” explanation is what the chromosomal setup we have provides over bacteria, which have a “looser” structure for their DNA in order to facilitate more mutation and gene exchanges. It’s not “better” per se, but it might facilitate more “complex” organisms evolving.
The question that remains is something like:
- Why not carry around 4x different ones in your germinal cells all the time but only “activate” two while allowing all 4 to mutate?
The answers might be something like:
might be harder to evolve, since you want to keep the embryonal stage as simple as possible, that’s when most mistakes in growth happen
that requires carrying around extra DNA
asymmetric selection has benefits
actually < these weird species of fungi>
But again, the real answer is that we can’t really know.
Remember, this hidden trait theory is just that, a theory. There are other theories about why sex works, and none of them yields a definitive answer. You’d likely have to simulate evolution “from scratch” in a bunch of Earth-like environments to get an answer with any degree of certainty, given how complex sexually reproducing organisms are.
Maybe it was sheer chance that introduced sex, and it’s actually a really bad selection mechanism, but it stuck because sexed organisms and intermediaries happened to also carry around a bunch of useful mutations, which were really hard to evolve (i.e. low probability of), and the sex-related genes were too hard to get rid of. This is similar to the way many mammals carry around a bunch of deactivated or even outright harmful viral DNA… it’s not that it’s good, it’s just that “getting rid” of it is really hard, and those mammals have other advantages that allow them to carry this genetic burden.
But the takeaway here should be that sex doesn’t exist for the sake of variance, it’s one of the benefits of sex, but it’s a benefit you could also get in 1001 other ways that are seemingly cheaper, and most lifeforms do.
I wrote something like this as a reply to the original article, and Paul responded, though I don’t think his response addresses my main point. The reply:
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 (!). 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.
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).
To address the last point, I honestly have no idea if my view here reflects a consensus among most (or among most good) biologists. I’d be hard to tell because my view boils down to:
- The problem is too complicated to get a definitive answer, and any explanation currently on the market can be rebutted with a <but then why not use this simpler mechanism that already exists?> style argument.
And saying “I don’t know” doesn’t really generate any paper. I’d further claim that consensus is a really bad measure because the vast majority of researchers will default to a “popular” opinion without giving it much thought. But I do believe that nobody has really tried to justify sex in terms of variance since the 70s, people are now focused on other aspects precisely because variance doesn’t explain its advantages over much simpler and less expensive mechanisms.
The more interesting part of this reply is:
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 (!).
I’m honestly not sure what the argument being made here is. As in, it seems to me that you can control the amount of variance in both mutations and sex. In mutation, you can do so via mutation-promoting proteins and DNA repair proteins, as well as by the way DNA is structured, which can lead to more or fewer mutations from e.g. radiation in various regions. in sex, you can do so via mate selection.
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 think this overlooks the point that sex exists in much simpler organisms. It even exists in colonial animals. These are simple enough to be comparable with bacterial colonies. Indeed, there are many simple multi-cellular eukaryotes in which horizontal gene transfer works just fine as a mechanism for introducing variance in a better-than-random way (for more details, see, for example, this paper)
How would asexual variance induction look in something as complex as humans? Besides random germline mutations, I don’t know, there aren’t things are complex as humans that rely on anything but sex to induce variance (again, some potential *s here, but mainly edge cases or organisms where “sex” looks different enough that it might be argued it doesn’t quite fit the common definition).
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)
Additional questions to ponder might be:
- Why always 2 sexes (outside of weird fungi)
- Why is the sex stable (outside of a few weird animals, and probably fungi… damn fungi, always messing up nice models of multicellular life)
I’m not saying they can be answered, but trying to do so might shed some more light on the potential advantages of sex, or at least on the fact that the current configuration might just be a stroke of randomness and not something particularly good.
I’m paraphrasing my arguments here a little bit, so that might point Paul’s reply in a bad light. I’d have finished the conversation with him, but he seem to have gone silent on it, so I thought I’d publish it as an article, since, as state above, I think it’s a useful example.
I should say that I am by no means 100% convinced that I am correct here, and obviously welcome a reply from Paul or anyone else that believes that variance is the cause-du-jour for sex, or that there is any theory that compellingly explains sex which hasn’t been thoroughly falsified.
Somewhere in the background, I am thinking of a more complex model that explains this phenomenon, but I doubt my explanation will be better than the current theories. I am simply trying to disprove things that seem wrong, I don’t claim to have a right answer.
When you get an allele from sex, there are two sources of variance. One is genes your (adult) partner has that are different from yours. The other is additional de novo mutations in your partner's gametes.
The former has already undergone strong selection, because it was part of one (and usually many) generations' worth of successfully reproducing organisms. This is much better than getting variance from random mutations, which are more often bad than good, and can be outright fatal.
Selecting through many generations of gametes, like (human) sperm do, isn't good enough; it doesn't filter out bad mutations in genes that aren't expressed in sperm cells.
Lateral gene transfer might be as good as sex, but I don't see how higher mutation rates can compete. I believe that empirically, mutations that weaken one of the anti-mutation DNA preservation mechanisms in gametes are usually deleterious and are not selected.
Neither of which are guaranteed to yield viable offspring, the latter won't carry all or maybe any of the benefits when mixed with your genes. Indeed, chances are most won't.
On the other hand just getting random mutation on a constant set of genes seems like it has a much higher chance of still yielding a viable combination.
How many generations of reproduction do you get in-absentia of recombination with other lineages between they are no longer compatible sexually? The answer varies based on the mutation rate, but it boils down to "surprisingly few".
A lineage getting a new mutation often results in speciation.
Also, see, most mammals mate with individuals on average 1 to 10 generations removed from them with few exceptions in rather "primitive" animals.
Why? The vast majority of potential children in most mammals die because of structural issues that are "detected" very early on (i.e. at the stage when they are still a clump of cells). One reason why old animals become infertile even if germinal cells are still present.
Could this mechanism not do any better?
More importantly, I think you're missing the point when I say "in some species the vast majority of resources go towards mate selection". Get rid of that and allow every individual to reproduce and you'd get the ability to have many more offsprings to "test" stuff in.
Some bacteria and archaea do little to not LGT, many viruses don't either. They have been here for potentially billions of years and will likely outlive.
The same can be said for many plants that reproduce mainly via cloning.
If you want to take a homo-centric POV and assume we are the end-all-be-all of biological life, fine, but even then you ought to keep in mind that sex might not be a requirement for that. Other strategies exist, and they don't involve sexual reproduction, the fact they did not evolve us may be chance
At any rate, I think your view might be mostly right, but it's not the view I was arguing against. But I think you're still missing the point if you have any reasonable certainty this explains sex in a satisfactory (see my reply for why). But you can be mostly right and still miss most things that are interesting, critical, give predictive power and best describe a phenomenon.
I think Dan is correct.
Sex is necessary to avoid Müller's ratchet if you can't have a gazillion offspring. Müller's ratchet is not avoided by introducing more variance (at least that's a very weird way to look at it). It is avoided by getting "undamaged" genes for your "damaged" ones and "fixing" your chromosome by recombination. (Of course this only happens randomly, but then there's selection on top.)
The strategy of trying many mutations in many offspring just doesn't work for very complex organisms. And higher mutation rates just speed up Müller's ratchet.
Without sex and recombination you'd need an insane amount of selection to counteract mutation.
having pairs of chromosomes and crossover are sufficient to resolve it
Do you mean cloning instead of sexual reproduction but with two chromosomes and crossover? That wouldn't be enough to avoid mutational meltdown.
How so? single-chromosome mutations can account for all variations one gets from the opposite sex, bad configurations can be selected against inside the germinal cells themselves or when the new organism is just a clump of a few thousand cells, which is how most "really bad" configurations get selected against in sexual organisms too.
Many genes and downstream effects are only expressed (and can be selected on) after birthing/hatching, or only in adult organisms. This can include whole organs, e.g. mammal fetuses don't use their lungs in the womb. A fetus could be deaf, blind, weak, slow, stupid - none of this would stop it from being carried to term. An individual could be terrible at hunting, socializing, mating, raising grandchildren - none of that would stop it from being born and raised to adulthood.
There's no biological way to really test the effect of a gene ahead of time. So it's very valuable to get genes that have already been selected for beneficial effects outside of early development.
That's in addition to p.b.'s point about losing information.
Let's say there is a section in a chromosome with 10 genes. In one chromosome 8 of these have damaging mutations. In the other chromosome these 8 are good copies but the other two are damaged. Now crossover of that section could fix the first chromosome by replacing 8 bad copies with 8 good copies and only 2 good copies with 2 bad copies. But going forward the resulting organism only has bad copies of these two genes.
In sexual reproduction there would be a large pool of correct copies out there and at some point these would be swapped back into this line. With cloning the information is lost for all descendants until random mutation recreates it.
Positive mutations would have to achieve for each germline what in sexual reproduction they have to achieve for just a few members of the entire species.
I think I get your point here, though I think this assumes a lot about how much cross-over mechanisms can actually "detect" genetic damage.
If this damage can mostly be detected only once the organism is mature enough to be selected for/against by "environment" then I think that kind of goes back into the "red queen" style theory that I'm a fan of (i.e. "hidden traits" that occasionally manifest in the population instead of dying out)
If this damage can mostly be detected at cross-over time or when the organism is still very young or in the germ cells themselves... then I'd expect this is also the kind of damage that won't be present in germ cells to being with, or not in many because there's already intra and inter cellular mechanisms to correct for this by inducing apoptosis in the damaged cell.
But maybe I'm missing something and I don't understand the finer details of cross over well enough.
Narrow Roads of Gene Land Volume 2: Evolution of Sex by W. D. Hamilton claims to cover this topic, and I just got a copy. I plan on reading and writing a book review, because I suspect that Hamilton has some good theories that LW would be interested in, given this post and recent interest in the evolution of sex.
Book review is now up here for those interested.
Maybe interesting related read: The sparrow with four sexes (Nature)
My competing article: Why We Worship Thee, O Great Sex! It doesn't cover as much ground as this, but I'm trying a new technique I'm calling "humor", let me know how it worked out. Unless I shouldn't quit my day job, in which case keep quiet. I like quitting goddammit
I think it's super important to stress that (1) evolution doesn't really have goals or "pick" things, and if you think of it as goal-directed or choice-making, you're missing something important; and (2) an extensive and routine gene recombination method was developed only once in 4 billion years. This indicates it is something that is highly unlikely to develop (and a candidate "great filter"). However, once it does develop, it improves grandchildren and so there is an exponential compounding of the benefits from that over time.
To say "one theory is that sex is a hidden trait preserving mechanism" is to suggest that evolution wanted to preserve hidden traits and therefore built a mechanism to do it, which inappropriately reverses causation. Instead, the theory (hypothesis?) would be better described as "sex was first created by chance*, but once created, acted as a hidden trait preserving mechanism, which helped it sexual organisms survive, thrive and fill the globe in the eons afterward." (* in phases, surely not all at once)
I imagine that for some species "the vast majority of resources over an organism’s lifespan are spent on finding mates", but is that a normal case? (I'm married, so I spend ~0 on that.) In any case, when it comes to creatures with elaborate mating features, like peacocks, I suppose it's because the environment has enough slack to allow it. That is, if there aren't a lot of predators around, the evolutionary process won't optimize for what we tend to think it should.
But I don't really know. I bought a textbook on evolution and it's gathering dust. Textbooks are boring, or at least that's what I evolved to think.
Epistemic background: I studied bioinformtics a decade ago. More recently I played around with an evolutionary simulation to attempt to get it to recreate hunter gatherer lifespan patterns and was surprised about how little prior art there is. But I'm not read up on all current discourse on evolution.
Sex allows two populations of a species to combine their genes in a way that results in the species getting roughly the best versions of the genes from both populations. Asexual bacteria do something similar by passing plasmides around but that strategy doesn't work for multicellular organisms and likely also not for one-celled organism that most of the time aren't in proximity to other members of their species.
If you have millions of individuals of a species, a species that has a way to combine the best genes from multiple organisms is going to develop new benefitial genes much faster then a species that doesn't have a mechanism for gene exchange.
The current estimates I can find for the amount of mice in the world is even bigger at 20 billion which results in even bigger gains from being able to combine genes across the population.
I expect that there are mostly two sexes because you don't get additional gains by introducing more then two. I'm not quite sure what you mean with sex being stable.
How would horizontal transfer even work when you go from a prokariot with millions of base pairs of genetic material to a eukariot with billions of base pairs and all the complexity of RNA splicing?
Sex being stable i.e. lack of gender switching or hermaphrodites in most animals (with exceptions ala snails).
Also, the recombinant benefits seems like they could also come by via increased varrisnce from e.g. mutation and 'survival of the fittest' style mechanism.
I feel like there's an obvious hypothesis about sex: it's for gaining the information encoded by genes being prevalent in conspecifics. Since your conspecifics have roughly the same niche as you, their genes are selected for being useful in your niche. To check this hypothesis we might want to see if it can explain the origin of sex in terms of local gradients for asexual bacteria with horizontal transfer: are genes that have a +1% chance per generation of being thrown overboard, and a +1% * [surrounding rate of evolution] per generation rate of organismal fitness increase (or something like that), selected for or against?
I think we can at least answer, why have 2 sexes rather than, 3, 4, or whatever.
Assuming the benefit of sex is to mix up genes with others' (seems reasonable, as that's what it does!),
In one generation 2 sexes mixes in 50% others' genes to 3 sexes mixing in 66%; not a huge difference. In 4 generations, it's 94% to 99%.
So the benefit of the extra sex isn't huge, but the cost of getting the third may be (just as the cost of finding one mate can be high, esp. if you're somebody's prey and need to both be noticed and not be noticed at the same time).
If the hidden trait theory is correct, shouldn't mutations that are only useful in a particular context be disproportionately recessive? That seems like a hypothesis that could be tested.
In Biology classes, I was taught that it's because the more "different" the sexes were, the more viable their offspring, so if a species started with 10 sexes, the 2 extreme ones outcompeted the others. I don't remember the reason given for the value of "difference", maybe some chemical reactions going faster. I don't know if it's still the consensus (or ever was).