Open Thread – 12/20/2020

19 thoughts on “Open Thread – 12/20/2020”

1. I’m planning on writing about intelligence, the environment in “STEM” higher-ed, and the aggregation of ethnic America, in the future.

   And

   Since some people are now paying for the privilege to read me you may not some differences in tone, style, and word-choice.

   Would I be wrong to think that up until now you have been kinda holding back? 🙂

2. not consciously. i haven’t changed my views, but a lot of stuff is boring to write about over and over

3. 2002 https://www.amacad.org/publication/unequal-nature-geneticists-perspective-human-differences

4. Yeah, the White middle class is threatened by the success of Asians so they are getting rid of the testing. Honestly I don’t even blame them. Its pure self-interest which I can respect. They will benefit from the new admission process all the whole convincing Blacks and Hispanics that its for their benefit.

5. The White Tiger
   https://www.youtube.com/watch?v=oM-Nw9XzqVM&feature=emb_title

   A Netflix movie about asian people of color!

6. More early crop exchange news – https://phys.org/news/2020-12-food-south-asia-revealed-east.html – tumeric, bananas, sesame oil, even soy, through Near East in 2000-1000 BCE period. “Exotic spices, fruits and oils from Asia had thus reached the Mediterranean several centuries, in some cases even millennia, earlier than had been previously thought,”

   Complement of earlier reports like this https://www.sciencedaily.com/releases/2019/02/190206091431.htm – “Prehistoric food globalization spanned three millennia”

   https://www.sciencedaily.com/releases/2020/02/200214134705.htm – “5200-year-old grains in the eastern Altai Mountains redate trans-Eurasian crop exchange”

   https://phys.org/news/2020-05-global-cooling-event-years-spurred.html – “Global cooling event 4,200 years ago spurred rice’s evolution, spread across Asia”

   https://www.sciencedaily.com/releases/2019/06/190610100650.htm – “Dramatic change in ancient nomad diets coincides with expansion of networks across Eurasia”

7. Razib, in your substack post “Applying IQ to IQ” I think you may have made a typo:
   “But all good things come to an end. Around 200,000 years ago the growth of our brains leveled off, probably due to biological constraints. The large head of the human fetus runs up against the limits of the mechanics of childbirth. Human infants are already born notably early in relation to their development. Natural selection tends to run into obstacles if it keeps being pushed in a single direction. Chickens grown huge for their meat eventually become infertile. And H. sapiens and its brain found its barrier 200,000 years ago. In fact, the largest brained humans seem to have lived during the Ice Age, not today. Since the transition to farming our brains have been shrinking”

   I think it should be 20,000 and not 200,000.

8. dathang, last i checked the rate of change really shifted around 200K, though there was some growth in all lineages up to LGM

9. 15 residents dead in 1 week where i work. if FDA had immediately approved that vaccine they might be alive…9 months of “it’s not that bad” from conservative politicians and then they ALL cut in line to take the vaccine manufactured *in my hometown* even though my grandma doesn’t have it yet (she also lives here.) Rage inducing, to say the least…

10. I guess so speaking in terms of ratios. The average from 200K years ago would have approached 1400cc, and it would have grown about 200cc or more in the next ~100,000 years (various archaics from Europe to China and AMH in Africa). The ratio change slowed down over time.

    But in more absolute terms, Erectus from a million to three quarters of a million years ago were about 900cc to 1000 cc and archaic-moderns like Jebel Irhoud from 200,000 to 300,000 years ago were 1300 to 1400cc. 300 to 500cc increase in 500K to 400K years vs 300 to 200 cc increase in 200K to 300K years. The absolute change doesn’t look all that different.

11. Couple of things on using siblings to understand trait genetics and trait genetic change over time, one from 2020 but which I haven’t seen posted anywhere before, one a couple years old:

    – August 2020: https://www.nature.com/articles/s41598-020-69927-7 – “Sibling validation of polygenic risk scores and complex trait prediction” (one of co-authors is the well known Steve Hsu!)

    This kind of follows on from the issues raised in the papers by Berg and Sohail in 2019. Those are where UK Biobank analysis, which allowed for very large ancestry homogenous (relatively) and sibling cohorts, led to finding that many variants identified in GIANT which a pan-European panel just seemed to discover population structure across Europe that was linked with height (possibly for environmental reasons) and did not replicate as height linked in homogenous ancestry set, using White British full siblings within WB (who are absolutely rock solid in shared ancestry), as a training/testing population.

    In this paper they set aside subsets of siblings from Biobank and then use scores trained on main set to look at the % where the sibling with >trait is correctly predicted by >PRS and the correlation within sibling sets, and then also compare this to correct prediction among general population. That tests to what degree a predictor functions in the absence of ancestry differences and to which it may be capturing differences which just reflect residual population structure. Do this both for disease risk predictors, biomedical traits and general quantative phenotype traits.

    Discussion comments: We compared validation results obtained using non-sibling subjects to those obtained among siblings, and found that most of the predictive power persists in between-sibling designs..

    One exception is the Educational Attainment (EA) predictor, which exhibits a very strong reduction in power when applied to sibs. This is not entirely unexpected as effects like the violation of the equal-environment hypothesis may be found for EA, and EA can depend on complicated correlations between environment and genes. Interestingly, the corresponding reduction for the Fluid Intelligence predictor is much less than for EA.

    (Above: Educational attainment (EA) shows an especially large between-sibling attenuation in performance relative to the other predictors. This has been noticed in other studies. The results suggest that at least some of the observed power in polygenic prediction of EA among non-sibling individuals comes from effects such as subtle population stratification (perhaps correlated to environmental conditions or family socio-economic status), genetic nurture, or other environmental-genetic correlations. Interestingly, the decrease in power seems to be not as large for the phenotype Fluid Intelligence (measured in UKB using a brief 12 item cognitive test).).

    We emphasize that predictors trained on even larger datasets will likely have significantly stronger performance than the ones analyzed here … As shown in earlier work, we expect the predictors to improve substantially as more data become available for training. This is conditioned on genotyping that captures a sufficient part of the predictive regions of the genome. It seems clear that with the possible exception of the height phenotype (for which we start to see diminishing returns; most of the common SNP heritability is captured already in the predictor), training is limited by sample size (specifically for risk predictors: number of genotyped cases) and not by algorithm performance or computational resources.

    For educational attainment as distinct from IQ, the scores do not really replicate well in siblings (possibly because confounded by capturing differences in environment between siblings, or because predictors are capturing population structure remaining in UK Biobank). (Possibly worth reading https://www.nature.com/articles/s41562-019-0757-5 – “Genetic correlates of social stratification in Great Britain” with this in mind ;-)? )

    Height getting to the point where most the common variants are exhausted and the remaining score is probably increasingly going to be predicted by family specific or otherwise fairly localized and recent rare and ultra rare variants of individually comparatively large effect (which can’t really be identified by these GWAS methods?).

    On very large, relatively homogenous cohorts they describe, China may have an advantage in putting them together (although a slight disadvantage in sibling testing – would have to use cousin and parent-child testing?). Possibly some fraction of those those will generalize to other populations. Same variants do not necessarily need to have the same frequencies in between populations, as long as the effect size is the same (e.g. https://twitter.com/chen_cmh/status/1175085877220929537 shows some generalization of height scores calculated in outgroup (Japan) to Sardinia). Since most common variants are shared between populations, so long as the effect sizes are consistent there could be portability.

    – 2018 – https://www.pnas.org/content/115/26/6674 – “Flynn effect and its reversal are both environmentally caused”. Uses large Norwegian registry to show that cumulative tracking of the effects between full siblings over time recreates Norwegian declines in full-scale IQ seen in general population. Therefore declines in IQ not due to changing composition of parents. Links up on theme to this new January 2021 paper – https://www.sciencedirect.com/science/article/abs/pii/S0160289620300908 – claims “No negative Flynn effect in France” – “Tested the claim that intelligence decreases in France (negative Flynn effect). Performance only decreases on tests involving declarative knowledge, not reasoning. This is attributable to measurement bias for older items, due to cultural changes.”

12. @Razib: Couldn’t start your Insight episodes both on
    https://insitome.libsyn.com/
    and
    https://www.stitcher.com/show/the-insight
    with different browsers. Any recent problems anyone? Did work fine before.

13. https://phys.org/news/2020-12-ancient-dna-retells-story-caribbean.html – ancient dna in caribbean. Previously a preprint at https://www.biorxiv.org/content/10.1101/2020.06.01.126730v1

    Another long comment- The more exciting new bit seems to be using ancient dna as a window into census population size, using the new methods developed by Ringbauer and collaborators to recover RoH (Runs of Homozygosity) in ancient samples.

    A technique developed by study co-author Harald Ringbauer, a postdoctoral fellow in the Reich Lab, used shared segments of DNA to estimate past population size, a method that could also be applied to future studies of ancient people. Ringbauer’s technique showed about 10,000 to 50,000 people were living on two of the Caribbean’s largest islands, Hispaniola and Puerto Rico, shortly before European arrival. This falls far short of the million inhabitants Columbus described to his patrons, likely to impress them, Keegan said.”

    Ancient dna can resolve some of the discussions between “high counters” and “low counters” as to the size and decline of different Native American populations upon encounter with Europeans.

    A comparison from recent literature: https://www.sciencedirect.com/science/article/pii/S0277379118307261 – “The primary source for most modern estimates of the contact population (of Hispanola) is a contemporary eyewitness report by the Dominican friar Bartolomé de las Casas who reported a Hispaniola population of 4 million at the time of contact (Denevan, 1992a). However, studies based on this same report range from 60,000 that neglects contemporaneous reports of post-contact deaths (Verlinden, 1973, in Henige, 1978), to acknowledging some degree of depopulation with initial populations estimated at between 100,000 (Amiama, 1959; Rosenblat, 1976) and 8 million (Cook and Borah, 1971, Table S1) … Most estimates are between 300,000 and 500,000 people in the Caribbean before European contact (Córdova, 1968; Dobyns, 1966; Morison, 1948; Moya Pons, 1979; Williams, 1970).”

    As new paper’s abstract states: “(W)e estimate to be a minimum of 500–1,500 and a maximum of 1,530–8,150 individuals on the combined islands of Puerto Rico and Hispaniola in the dozens of generations before the individuals who we analysed lived. Census sizes are unlikely to be more than tenfold larger than effective population sizes, so previous pan-Caribbean estimates of hundreds of thousands of people are too large“

    (Another previous adna and modern dna paper – https://advances.sciencemag.org/content/4/11/eaau4921 – estimated effective population size reductions for highland Peruvian vs lowland Native Americans as follows: 27% reduction in highland Andes, vs 94-96% in lowland Mexico and lowland Peru. This is not census population size, though it seems like if you look at the model parameters in the supplement and then assume the same working constraint that census unlikely to be >10x effective population size, these groups would also likely be in the thousands before contact?).

    Some more comments by Reich at the above link (navigating the isthmus between the science and contemporary social mores). And here’s the NYT if anyone wanted the mainstream media message take – https://www.nytimes.com/2020/12/23/science/dna-caribbean-islands.html (including “Colonization delivered a huge shock to the Caribbean world, drastically changing its genetic profile. But the Ceramic Age people still managed to pass on their genes to future generations. And now, with a population of about 44 million people, the Caribbean may contain more Taino DNA than it did in 1491.”).

    These results are interesting in part because they flesh out some comparison with other peoples hit by the diseases from large Eurasian agricultural populations as they expanded, but who diverged from Eurasian populations at different points – such as the Polynesians – or who were always in contact with them – such as the populations of Siberia. One idea is that the Native American populations more suffered relative to these populations by having branched off earlier, before diseases that evolved in the agricultural era were encountered, and so suffered from a lack of selection for HLA diversity etc. Getting more information on census sizes in groups of individuals where documentation may be unreliable and biased or absent lets us test those ideas. (Also to compare to other patterns that mostly look like population replacements, like in Britain with “Bell Beaker” people!). It also lets us understand how much the establishment of economic and political complexity in ancient societies was simply a function of population size (vs ideas like Turchin’s that a specific need to tap and organize populations for warfare, only somewhat correlated with population size, was more of a driver).

14. I was aware of 2000 election truthers, people holding to this very day that Gore would have won if Bush hadn’t tricked the Supreme Court to stop the recounting in Florida. I was not aware of 2004 election truthers: it’s like a mirror image of Trump’s claims today.

15. Commentary on recent report from US Department of Labor on annual household income by different ethnic groups:

    https://quillette.com/2020/12/22/a-peculiar-kind-of-racist-patriarchy/

    One argument I constantly see being made about these types of findings (basically Asian/immigrant POC communities having greater wealth than white Americans) is that Asian immigrants are self-selected as the wealthier/upper classes of their respective native countries. I know this is sort of true for Indians, as I think I’ve read that Brahmins constitute something like 25% of the Indian-American population while they only make up 5% or so of the population of India.

    But I do remember someone once posting a study here (I can’t remember if it was Razib or maybe Matt?) about actual East Asian immigrants (Chinese, Koreans, Japanese) and I believe that contrary to the received wisdom, most immigrants from those countries to the US in the 20th century did not come from particularly high SES backgrounds at all. Does anyone know what I’m talking about and have a link handy?

16. @Mick, for my part, can’t say as I have a really good rock solid reference for how people who were descendents of Chinese/East Asian migrants from the early 20th century before the selective post-1960s waves did in the US in education and SES. It would be hard as well to track how those people’s descendents would do today because of course they’ve intermarried into the general population and to post-1960s migrants and so it’s not really separable and maybe not a meaningful question.

    Following as is as I understand it for present day migrants from East Asia: Firstly, yeah, of course they are definitely selected for education accomplishment well over average for source countries.

    One quick reference for that here: https://jsmp.dk/posts/2019-09-26-braindrain/

    It’s just how the US immigration system works, and that’s true as far as I know for all the “Big Wave” of Asian migration post-1960. These are not systems that just select people who are the mean accomplishment for their nations of origin (the only current migrants in the US who are average or close are Mexican, and that’s basically illegal immigration selecting for people from below average in the education distribution in Mexico to fill blue-collar jobs).

    (Interestingly, within the category of East Asia, it looks like Chinese migrants tend to improve on parental education while that’s not the case for Japanese. Ref – https://imgur.com/a/4vdDiWk . That seems to me to point to their families being more “pro-education”/smarter for a given level of education accomplishment in the incoming stream. Which makes sense in that China is a less developed country, so any given level of education would be harder to achieve and anyone who had it would have a harder time getting there. And thus Chinese are functionally more selected than Japanese.)

    But, it’s also true that today not all East Asian migrants are high education (even though most are, and much more so than in proportion to source population), and there is some strong evidence that those who are from low education backgrounds still perform pretty well – see the Success Frame stuff – https://link.springer.com/article/10.1007/s12552-014-9112-7.

    It does seem like even among the low education background segment today, they do pretty well, and that’s probably some mix of East Asian people being pretty smart, plus some kind of migrant selection within the low education category (they’re actually more hyperselected than their education seems), plus the pro-education achievement cultural influence from the disproportionately high education migrants.

    So for current migrants it’s neither the Woke thing of “They’re all from privileged backgrounds of course! Poverty explains everything! And they’re still underprivileged relative to their educational accomplishment because of racism.”. That’s more wrong that right it seems. But nor the thing of “Of course, it’s just unselected Chinese doing what comes naturally”. Somewhere between and a mix of other effects.

    (There’s some suggestive evidence to me that the selective effect accelerates over time too – Asians move away from Whites in the SAT. That seems like the increasing effect of selection over time and the feedback of success frame and assorting into more competitive schools and so on.).

    Does that kind of help answer any of it? Appreciate you’re more interested in the 20th century, pre-60s unselected migrants, but that’s a difficult one to answer (hence I’ve answered a bunch of other stuff that may be related/unrelated to what you’re really asking about, as I roughly understand it to be).

17. @Matt

    Thanks, I thought I read something on here once that Japanese or Chinese immigrants weren’t from particularly higher SES backgrounds but maybe I mis-remembered because the data you pulled seems to refute that. And I suppose trying to find out more about the SES of pre-1960s Asian immigrants probably wouldn’t be that informative in the larger story of Asian immigration since the vast majority of Asians came after the 60s anyway, and the ones in the US before that would have been almost exclusively concentrated in California or Hawaii. Maybe Hawaii is actually what I was originally thinking of, because I know the Japanese and Filipinos who immigrated there were there to work on the plantations, that cohort definitely would have been more blue collar than the post-60s type I would imagine.

    I guess I’m not even really sure what the Woke talking point even is in regards to the SES background of Asians – Asians might do well economically, but that’s only because they come from high SES backgrounds themselves, so America is still white supremacist because only the most successful Asians are allowed to come and thrive in the US? If the US was such a white supremacist hell-hole why would it even allow any Asians or non-Europeans to immigrate to it at all?

18. @Mick, yeah, I think you’re right about Hawaii. If you’re looking for a multi-generation, unselected group, that is the best bet. I think there possibly some stuff there – it’s been a long time since I looked.

    Honestly I’m doing some mind reading on what the Woke stance would be on this topic (perhaps inaccurately), because mostly Woke folk just seem evasive on Asian SES in the US rather than having a clear tendency to argue a certain way. Like, I’m less sure there is an well-defined argument that they make.

    When it comes to Asian-Americans, Woke seem more comfortable on: “cultural appropriation” (down to the sort of “Wypipo should not be allowed to make ramen” sort of thing), orientalism, emasculation of Asian men, historical injustices against Asian communities (railroads, internment, exclusion act), present-day cultural chauvinism in arts, supposed underrepresentation of Asians who do not fit “model minority” (or is it just more that successful Asian groups represent themselves a lot?), historical European imperialism in Asia and some other things. Things where there is perhaps some level of truth and fact to it, that most people reasonable and intelligent people would agree on, but just really hyper-amped up and simplified in service of the ideology. The present-day SES topic is just not so easy to put in harness to serve the ideology, which has radicalized to the point that anything less than race being the overwhelming factor in SES seems almost hard for them to consider.

19. A new paper clarifies that the medieval river civilizations in the Turan collapsed due to drought before the Mongols invaded them. http://dispatchesfromturtleisland.blogspot.com/2020/12/drought-destroyed-river-civilizations.html

Comments are closed.