Thank you.

focusing on higher education in the middling twenties and then after fifty spending most time on grandchildren whose parents followed the same patter and focus their late 20s and 30s on other goals.

Proposed algorithm:

1. Reproduce, early childcare
2. Higher education (informal or formal)
3. Other goals
4. Raise multiple pre-teen/early-teen kids

Step one ends at about the early 20s, just as cognitive performance peaks. Still leaving time to make use of it for other achievements in the middling 20s. The obvious exception is really hard math, but fortunately most people in a viable population don't need to deal with really hard math.

Step three builds up resources and acheives other goals you estimate you won't be able to acheive after step four.

Step four raise your grandchildren or if you don't have those someone else's pre-teen/early-teen kids.

Current algorithm (for the demographic many LWers find themselves belonging to):

1. Higher education (formal)
2. Other goals
3. Reproduce, maybe sort of
4. Raise one or two teen kids maybe

I would argue that the quality of childcare isn't significantly diminished (if it is at all). There is indeed some opportunity cost here since is sub optimally employed cognitive horsepower. But the neat trick is that the new algorithm increases available cognitive horsepower every generation, while perhaps doubling or even tripling the number of highly trained rationalists. Also raising small children may leave you physically exhausted and be a time sink, but if you are provided resources by your parents or a community, that still leaves plenty of time to familiarize yourself with what you'll be doing in step two or even get the higher education done with. If parental funds are unavailable or insufficient there is work you can do to help cover the shortfall.

People can still do horizontal meme and mindware transmition with minimal losses in efficacy during "other goal" period or after they are done with phase four. Wasting peak reproductive time for that, is a really big waste of resources unless you are that rare exceptional person that started building the bedrock of the community.

LWers would not be the only ones to try this tactic, though. There's a whole movement based on producing more evangelical Christians.

Well, yes. I did use Amish as the example of a group already using this tactic. But rationality is about winning, so clearly if rationalists put their minds to it they would be better at it. ;)

Another tactic, better at changing ratios, would be to adopt existing children.

Hey it worked for Shakers. Until government said no more mass adoptions for you.

What is P that government agencies might try to interfere with organizations that helped rationalist families adopts lots of kids? What is the P that some group might lobby and warn parents against giving away their children to those creepy/Godless people?

This could propagate rationalists' memes, though not genes, and it's unpredictable how much those two things impact whether your children come out like you want (e.g. rationalist, or evangelical).

Obviously enough evangelicals stay evangelicals for their group to keep growing. Amish reject many modern comforts, something evangelicals don't do, yet their losses aren't as great as people seem to think they would be.

But basically to the first approximation kids are like their parents in temperament and ability. And in the environment that LWers are likley to raise their kids its basically settled that nurture plays only a negligible role when explaining the differences between biological and adopted kids in the same family. Now lets consider which set of people have the temperaments and abilities better on average suited to our needs? Set of biological parents of children legally up for adoption or set of LWers?

And it assumes you want more rationalists compared to others, as opposed to just more rationalists, period.

More rationalists period sounds pretty awesome to me. Human brains are pretty much dirt cheap supercomputers that can actually do incredible things when hooked up together.

Another tactic, better at changing ratios

I'm responding a bit lightly and teasingly, so I should clear up that I'm not claiming this is most probably the best strategy, I haven't done much more than Fermi estimates, but so far its clear that its hard to deny it is a workable strategy.

And it keeps getting better and better looking the further away one puts his most likley singularity date, or even if one is sceptical of it ever occurring. But most LWers wouldn't think of it straight away as a potential strategy. I think the reason is that this strategy is currently employed by people LWers don't like, not only that it is a strategy that is low status in modern society. Perhaps we may be on average biased against it? No?

Again, my main point was this:

People underestimate just how much "dumb" replication can do to carry memes forward.