I’ve been thinking this for a while: here on LW evolution is often portrayed as a weak algorithm that procedes to explore the genetic landscape fumbling around with random mutations. And this is certainly true for natural selection. We have though also sexual selection: ‘suddenly’ you can choose which genome gets to reproduce thanks to a brain! In principle, brains are Turing complete, and this means that evolution could be much smarter. Of course, even that program is determined by your genetics, and sometimes things can go awry. Still, and more cogent for humans, there is the possibility of ‘smarting up’ evolution.
Random mutations aren’t natural selection. Natural selection is the process that prevents harmful mutations from spreading and that encourages beneficial mutations to spread.
Sexual selection is part of natural selection in the way the term “natural selection” is usually used by biologists.
In principle, brains are Turing complete
No, they aren’t. Neural nets don’t work with 1′s and 0′s the way computers do.
Actually, because a human can simulate a Turing machine’s execution using pencil and paper, humans are Turing complete. (I realize the original statement was that brains are Turing complete, but since each human brain usually come equipped with an attached human, it seems reasonable to discuss whether humans, rather than human brains, are Turing complete.)
I don’t think I know a human who would have a zero error rate doing 1,000,000,000 Turing operations
Nor do I. However, this is irrelevant. In determining whether a system is Turing complete, physical limitations are usually ignored. From Wikipedia:
To show that something is Turing complete, it is enough to show that it can be used to simulate some Turing complete system. For example, an imperative language is Turing complete if it has conditional branching (e.g., “if” and “goto” statements, or a “branch if zero” instruction. See OISC) and the ability to change an arbitrary amount of memory locations (e.g., the ability to maintain an arbitrary number of variables). Since this is almost always the case, most (if not all) imperative languages are Turing complete if the limitations of finite memory are ignored.
If we did not ignore physical limitations, no actual computing system would be Turing complete.
It’s a model. Models have it’s use. It makes sense to model a computer as an ideal Turing machine. It doesn’t make much sense to model a human that way.
Nobody suggested modeling humans as Turing machines. The question was whether humans are Turing complete and you implied that they are not because they make errors. By the same standard, no physical device is Turing complete.
In principle, brains are Turing complete, and this means that evolution could be much smarter.
If by “evolution being smarter” you mean things like “brains becoming intelligent, developing science, and doing genetic engineering”, then yes. But it’s the brains who are smart, not the evolution per se.
The evolution would still fumble around the genetic landscape, except that with brains the local landscape becomes much more complicated, something like a fractal mountain instead of the smooth hills in most of the plain. On a different terrain, the weak algorithm may produce more interesting results. That does not make the algorithm more intelligent.
Very little of eugenics’s bad reputation dates from the Nazis. Some of it dates from before and some from long after. In particular, the winners continued their pre-war programs for decades after the war. Eugenics became unfashionable around 1960, a bit late to blame on the Nazis. And I think the emphasis on Nazi associations is even later (maybe 1970 or 1980), after eugenics was clearly losing.
People have had this idea before. It’s called “eugenics”.
Not quite: eugenics is a set of techniques that isn’t certainly limited to sexual selection. Plus only humans could practice eugenics. Instead, anything with a brain and sex can practice sexual selection.
BTW, ferns have feromons, they might be able to choose their partners. At least, we don’t know yet if they do. Would you call it sexual selection, if it leads to preferential inbreeding/outbreeding/between-species breeding?
I’ve been thinking this for a while: here on LW evolution is often portrayed as a weak algorithm that procedes to explore the genetic landscape fumbling around with random mutations. And this is certainly true for natural selection.
We have though also sexual selection: ‘suddenly’ you can choose which genome gets to reproduce thanks to a brain! In principle, brains are Turing complete, and this means that evolution could be much smarter. Of course, even that program is determined by your genetics, and sometimes things can go awry. Still, and more cogent for humans, there is the possibility of ‘smarting up’ evolution.
Random mutations aren’t natural selection. Natural selection is the process that prevents harmful mutations from spreading and that encourages beneficial mutations to spread.
Sexual selection is part of natural selection in the way the term “natural selection” is usually used by biologists.
No, they aren’t. Neural nets don’t work with 1′s and 0′s the way computers do.
Actually, because a human can simulate a Turing machine’s execution using pencil and paper, humans are Turing complete. (I realize the original statement was that brains are Turing complete, but since each human brain usually come equipped with an attached human, it seems reasonable to discuss whether humans, rather than human brains, are Turing complete.)
I don’t think I know a human who would have a zero error rate doing 1,000,000,000 Turing operations.
Nor do I. However, this is irrelevant. In determining whether a system is Turing complete, physical limitations are usually ignored. From Wikipedia:
If we did not ignore physical limitations, no actual computing system would be Turing complete.
Making errors means not behaving as a Turing machine. It’s separate from limitations of memory.
Any physical Turing machine will make errors.
To the extend that it does it’s no ideal Turing machine.
Ideal Turing machines, being, y’know, ideal, do not exist in reality.
It’s a model. Models have it’s use. It makes sense to model a computer as an ideal Turing machine. It doesn’t make much sense to model a human that way.
Nobody suggested modeling humans as Turing machines. The question was whether humans are Turing complete and you implied that they are not because they make errors. By the same standard, no physical device is Turing complete.
That’s largely irrelevant, since even the first computers didn’t work with 0′s and 1′s the way modern computers do.
If by “evolution being smarter” you mean things like “brains becoming intelligent, developing science, and doing genetic engineering”, then yes. But it’s the brains who are smart, not the evolution per se.
The evolution would still fumble around the genetic landscape, except that with brains the local landscape becomes much more complicated, something like a fractal mountain instead of the smooth hills in most of the plain. On a different terrain, the weak algorithm may produce more interesting results. That does not make the algorithm more intelligent.
No, I meant that by sexual selection you can have an algorithm exploring the genetic landscape, you’re not limited to random mutations.
People have had this idea before. It’s called “eugenics”.
It has a bad reputation from its implementation by the Nazis, who might have corrupted it a bit for their other political goals.
But I think even a pure implementation of eugenics is not as good as the other options we have for improving the lives of future humans.
Very little of eugenics’s bad reputation dates from the Nazis. Some of it dates from before and some from long after. In particular, the winners continued their pre-war programs for decades after the war. Eugenics became unfashionable around 1960, a bit late to blame on the Nazis. And I think the emphasis on Nazi associations is even later (maybe 1970 or 1980), after eugenics was clearly losing.
Not quite: eugenics is a set of techniques that isn’t certainly limited to sexual selection. Plus only humans could practice eugenics. Instead, anything with a brain and sex can practice sexual selection.
BTW, ferns have feromons, they might be able to choose their partners. At least, we don’t know yet if they do. Would you call it sexual selection, if it leads to preferential inbreeding/outbreeding/between-species breeding?