Report comment

A much larger share of talented people went into STEM fields in the Soviet Union, while managers were marginalized (or had to be engineers).

http://www.airpower.au.af.mil/airchronicles/aureview/1979/jul-aug/heuer.html

Report comment

Report comment

It is common for selective breeding to move a trait 30 standard deviations. It would be easy to breed humans that are 9 feet. It is unlikely that one could easily move human height 30 standard deviations, but it is probably easy to move weight 30 standard deviations. That means that in the population before breeding there exists variation enough to move 30 standard deviations. If the trait is mainly controlled by mutational load, then spell-checking the genome should move it the whole way all on its own.

IQ 300 is meaningless. What it really means is that one could find in the typical individual 4 sets of variants, each of which, when corrected would yield IQ 150. It is not clear what would happen if all 4 sets were corrected. It might have bad effects, since the brain is designed under the assumption of mutational load. This concern is more plausible in the case of height than IQ.

Steve Hsu gives various models and reality checks. There is no direct measurement of the mutation load hypothesis. But there has been fairly direct measurement that no common variant contributes more than 1/100 part of the variance, probably not even 1/1000. So there must be lots of relevant genes and it is hard to find a model compatible with that that does not predict that everyone has many bad variants.

Report comment

Report comment

Report comment

Naively I would guess Scott’s talking about the research areas that Steve Hsu tends to write about?

My basic model from reading abstracts and watching videos of research talks (over a year ago, so maybe other people’s models have clarified since then in response to better data) is that brains are constructed by lots of genes and when a few genes go awry they become less functional but still basically work. There are tens of thousands of ways for brains to break down like this, and each way reduces IQ a little bit.

Basically all humans have many such slightly broken genes (in a unique pattern), losing fractions of a point here and there, but people who are “smarter than average” have slightly fewer.

Once we’ve figured out all the ways that brain genes can be “misspelled”, it should be theoretically (if not pragmatically) possible to produce human genomes with no broken brain genes… “locally modal genomes”… like running spelling correct on a document. Rough calculations suggest that a brain produced by a “fully spellchecked genome” would have an IQ of ~300.

Aside from this potentially being economically costly, and politically infeasible, there might be a “real” barrier to this in that real genomes are often full of compromises, often in response to diseases. Europeans have mutations to fight tuberculosis (a double dose of which gives cystic fibrosis) and Africans have anti-malarial mutations (a double dose of which gives sickle cell anemia) and so on. Losing a few IQ points to avoid a terrible disease can be worth it, and a 300 IQ spell checked genome might end up susceptible to everything, and thus be a pragmatically losing proposition?

Evolution is pretty clever. Doing better than it will take serious work 😛

Report comment

Report comment

Report comment

It would be charitable to assume 3. Alas, I lean towards 2. You can’t prove deliberate dishonesty of course. But, in recent years ‘high impact’ scientific publications, like Science magazine, and Nature magazine, seem to have become nothing more than propaganda arms of the liberal ‘consensus’. About the only subjects you can trust them on anymore is the hard sciences — stuff like molecular biology, and chemistry.

Report comment