If we get to the point where we can distinguish genes that cause autism to be able to remove them, we'll also be able to distinguish 10% Nobelist case.

But even if that was not the case, with gene editing some parents who did not have a chance to make that gamble, will be able to make it intentionally.

The point of my thought experiment is to imagine that these are the same genes. Remember that about half of population variation is "unshared environment" meaning noise in the translation of DNA into an adult: genes will never allow perfect prediction.

I guess I believe that very few parents would make this gamble, because the risks are very personal and the benefits are mostly to others. But I could be wrong. One real live example is that there's been very little push-back against eliminating Tay-Sachs via testing.

In your thought experiment, even if very few parents made the gamble, it would be enough to keep the gene around until we figured out what in the environment caused autism, after which the many more would want to have it.

For the Tay-Sachs example what would be the reason to push-back against eliminating it? And who should have pushed back? people who had that gene, or people who didn't?

IMO the solution is to simply have more people, if we have 1000 billions on earth (which is possible with seasteading and terraforming deserts), 10 billion on moon 100 billion on mars we'll have enough place to try out all kinds of solutions, for people to decide how much to edit or not to edit their own genes, and many new random mutations arising simply because there are more new people than were in all of the history combined.

Other thought experiments are certainly possible, but mine is about the stated gamble, in which you cannot predict. That's what's meant by "unshared environment", it's poorly named, the environment you talk of is "shared environment". As I said, I don't know that this is true in the real world, but I don't think it wildly implausible. (Maybe tortured mad artists are a better example than nerds.)

Tay-Sachs is the first example I know about, of some genetic strain of humans being deliberately bred out of existence. Phrased like that I think the idea would alarm many people. But it has not. (I should remind myself more of the details of this story.)

Thought experiments aside, note that it's not obvious we need a single new mutation to make much smarter people. We just need the shuffling of the deck to place more of the existing + variants into one body. This will obviously be the goal of embryo selection, and I think would also be the goal of explicit editing. I'd wager that many more of the Nobel-sized brains alive at the end of this century (and the next) will owe their existence to one of these kinds of engineering, than to population growth.