3 ms·

How do you measure IQ in this context? You can make some guesses based on the known population, but there’s no way you can capture the full range of possibiliti

by nis0s 11mo ago

How do you measure IQ in this context? You can make some guesses based on the known population, but there’s no way you can capture the full range of possibilities. I worry that this type of tailoring might lead to an average of averages, and make human populations more susceptible to pandemics, or other issues due to reduced genetic diversity.

moktonar 11mo ago
We already choose each other based on genetic diversity. Smell for example is a big factor, did you notice that people you’re not attracted to usually smell worse than those you’re attracted to? And I guess there are many other non-rational ways the body-mind chooses for us. As always we think that science can do better than evolution. We’ll see, but if I had to bet..
polski-g 11mo ago
Previously PGT only got us to 9% accuracy in detecting predicted intelligence. The latest literature suggests they're now at 40-50%. The "missing heritability" has been solved.
- nis0s 11mo ago
  Your 40-50% variance range isn’t supported by recent literature, as far as I can tell. Secondly, per my understanding you can only get a sense of embryonic PGS via PGT, and that doesn’t necessarily relate to intelligence. https://pmc.ncbi.nlm.nih.gov/articles/PMC8280022/ https://pmc.ncbi.nlm.nih.gov/articles/PMC8280022/
  - polski-g 11mo ago
    You shouldn't like papers from 2021 when people are talking about "recent". This stuff is moving very fast. https://osf.io/preprints/psyarxiv/jfhtu_v1 https://osf.io/preprints/psyarxiv/jfhtu_v1
    - nis0s 11mo ago
      I can’t put much stock in something that’s not peer-reviewed, nor something that’s not confirmed as reproducible. Herasight, the company, has a deep conflict of interest when publishing a paper like this, don’t you think? It’s like the tobacco industry publishing about the effects of smoking. Two other issues: do the cognitive tests control for socioeconomic status? Otherwise you’re just measuring the effect of nurture, not nature. Second, the UKB fluid intelligence test isn’t a rigorous or reliable measure to fully capture latent GCA.
      - polski-g 11mo ago
        We already know from twin studies that g is 80% heritable, so the effect of nurture on intelligence is slim to none.
        nis0s 11mo ago
        Again a couple of thoughts occur: 1) you’re greatly discounting the effects of epigenetics on development and trait expressiveness, so even if you tailor for everything as you wish, you’re still only going to guarantee outcomes in a certain range, and 2) heritability is population-level, not individual level. So a figure like “80%” means that that much variation in intelligence in a given *population* is associated with genetic variation. This does NOT mean that 80% of a person’s intelligence comes from their genes. 3) you’re also completely neglecting the effects of genetic nuture, besides environmental effects.
tptacek 11mo ago
You don't. The claims are based on PGS correlations from things like biobank data. Broken record: this is a brilliant product. It promises a marginal improvement in a volatile stat measured meaningfully only when the grown child applies for college. It's purchased exclusively by parents that were already going to fold space like a Guild Navigator to get their kids into selective universities. Heads: you take the credit for them getting in somewhere they were getting in anyways. Tails: there is no tails. It's like a financial derivative product where the underlying is upper-class parental status anxiety. There should be a ticker symbol for it.

by nis0s 11mo ago

moktonar 11mo ago

We already choose each other based on genetic diversity. Smell for example is a big factor, did you notice that people you’re not attracted to usually smell worse than those you’re attracted to? And I guess there are many other non-rational ways the body-mind chooses for us. As always we think that science can do better than evolution. We’ll see, but if I had to bet..

polski-g 11mo ago

Previously PGT only got us to 9% accuracy in detecting predicted intelligence. The latest literature suggests they're now at 40-50%. The "missing heritability" has been solved.

nis0s 11mo ago
Your 40-50% variance range isn’t supported by recent literature, as far as I can tell. Secondly, per my understanding you can only get a sense of embryonic PGS via PGT, and that doesn’t necessarily relate to intelligence. https://pmc.ncbi.nlm.nih.gov/articles/PMC8280022/ https://pmc.ncbi.nlm.nih.gov/articles/PMC8280022/
- polski-g 11mo ago
  You shouldn't like papers from 2021 when people are talking about "recent". This stuff is moving very fast. https://osf.io/preprints/psyarxiv/jfhtu_v1 https://osf.io/preprints/psyarxiv/jfhtu_v1
  - nis0s 11mo ago
    I can’t put much stock in something that’s not peer-reviewed, nor something that’s not confirmed as reproducible. Herasight, the company, has a deep conflict of interest when publishing a paper like this, don’t you think? It’s like the tobacco industry publishing about the effects of smoking. Two other issues: do the cognitive tests control for socioeconomic status? Otherwise you’re just measuring the effect of nurture, not nature. Second, the UKB fluid intelligence test isn’t a rigorous or reliable measure to fully capture latent GCA.
    - polski-g 11mo ago
      We already know from twin studies that g is 80% heritable, so the effect of nurture on intelligence is slim to none.
      - nis0s 11mo ago
        Again a couple of thoughts occur: 1) you’re greatly discounting the effects of epigenetics on development and trait expressiveness, so even if you tailor for everything as you wish, you’re still only going to guarantee outcomes in a certain range, and 2) heritability is population-level, not individual level. So a figure like “80%” means that that much variation in intelligence in a given *population* is associated with genetic variation. This does NOT mean that 80% of a person’s intelligence comes from their genes. 3) you’re also completely neglecting the effects of genetic nuture, besides environmental effects.

tptacek 11mo ago

You don't. The claims are based on PGS correlations from things like biobank data. Broken record: this is a brilliant product. It promises a marginal improvement in a volatile stat measured meaningfully only when the grown child applies for college. It's purchased exclusively by parents that were already going to fold space like a Guild Navigator to get their kids into selective universities. Heads: you take the credit for them getting in somewhere they were getting in anyways. Tails: there is no tails. It's like a financial derivative product where the underlying is upper-class parental status anxiety. There should be a ticker symbol for it.