The Unz Review • An Alternative Media Selection$
A Collection of Interesting, Important, and Controversial Perspectives Largely Excluded from the American Mainstream Media
 BlogviewJames Thompson Archive
X-Raying Race
Email This Page to Someone

 Remember My Information


Bookmark Toggle AllToCAdd to LibraryRemove from Library • B
Show CommentNext New CommentNext New ReplyRead More
ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
These buttons register your public Agreement, Disagreement, Thanks, LOL, or Troll with the selected comment. They are ONLY available to recent, frequent commenters who have saved their Name+Email using the 'Remember My Information' checkbox, and may also ONLY be used three times during any eight hour period.
Ignore Commenter Follow Commenter
Search Text Case Sensitive  Exact Words  Include Comments
List of Bookmarks

One conception of race is that it is skin deep, and is no more than a matter of skin pigmentation. By implication, such a categorisation is superficial, trivial, and unlikely to be an explanation of any presumed racial differences in behaviour. There may be effects due to people making unwarranted assumptions based on skin colour, but that says more about them than anything else.

According to the skin-pigmentation theory, an Xray should see right through that, to the reality of the bones underneath. This would reveal, the theory says, that people are alike under the skin. Perhaps so, but is it true of their bones?

[Submitted on 21 Jul 2021]
Reading Race: AI Recognises Patient’s Racial Identity In Medical Images
Imon Banerjee, Ananth Reddy Bhimireddy, John L. Burns, Leo Anthony Celi, Li-Ching Chen, Ramon Correa, Natalie Dullerud, Marzyeh Ghassemi, Shih-Cheng Huang, Po-Chih Kuo, Matthew P Lungren, Lyle Palmer, Brandon J Price, Saptarshi Purkayastha, Ayis Pyrros, Luke Oakden-Rayner, Chima Okechukwu, Laleh Seyyed-Kalantari, Hari Trivedi, Ryan Wang, Zachary Zaiman, Haoran Zhang, Judy W Gichoya

Background: In medical imaging, prior studies have demonstrated disparate AI performance by race, yet there is no known correlation for race on medical imaging that would be obvious to the human expert interpreting the images.

Methods: Using private and public datasets we evaluate: A) performance quantification of deep learning models to detect race from medical images, including the ability of these models to generalize to external environments and across multiple imaging modalities, B) assessment of possible confounding anatomic and phenotype population features, such as disease distribution and body habitus as predictors of race, and C) investigation into the underlying mechanism by which AI models can recognize race.

Findings: Standard deep learning models can be trained to predict race from medical images with high performance across multiple imaging modalities. Our findings hold under external validation conditions, as well as when models are optimized to perform clinically motivated tasks. We demonstrate this detection is not due to trivial proxies or imaging-related surrogate covariates for race, such as underlying disease distribution. Finally, we show that performance persists over all anatomical regions and frequency spectrum of the images suggesting that mitigation efforts will be challenging and demand further study.

Interpretation: We emphasize that model ability to predict self-reported race is itself not the issue of importance. However, our findings that AI can trivially predict self-reported race — even from corrupted, cropped, and noised medical images — in a setting where clinical experts cannot, creates an enormous risk for all model deployments in medical imaging: if an AI model secretly used its knowledge of self-reported race to misclassify all Black patients, radiologists would not be able to tell using the same data the model has access to.

This is an astounding paper. It appears to reveal that, using artificial intelligence, deep learning methods lead to the detection of race in X ray images, even if all possible giveaway signals are stripped out of the image. That is extraordinary. The next point of interest is that the authors have made it very clear that they are alarmed that this is possible, and warn that it might lead to evil consequences.

They assert that race is not based on ancestry, but is a social construction and based on self-report.

So, to take the authors at their word, this paper is a warning, as well as the publication of a finding. The underlying results show that you cannot get rid of race in an Xray, even though radiographers are reportedly unable to detect race in such images, and do not know how artificial intelligence achieves such good results.

I read these results some months ago, and now come back to them, seeing this as an important finding, and a vivid example of what authors must do in academia when they report unwelcome results.

First, let us look at what the authors say to deflect critics. Here is their statement:

We emphasize that model ability to predict self-reported race is itself not the issue of importance. However, our findings that AI can trivially predict self-reported race — even from corrupted, cropped, and noised medical images — in a setting where clinical experts cannot, creates an enormous risk for all model deployments in medical imaging: if an AI model secretly used its knowledge of self-reported race to misclassify all Black patients, radiologists would not be able to tell using the same data the model has access to.

Bias and discrimination in Artificial Intelligence (AI) systems has been heavily studied in the domains of language modelling1, criminal justice 2, automated speech recognition 3 and various healthcare application domains including dermatology 4,5, mortality risk prediction 6andhealthcare utilization prediction algorithms7 among others.

While AI models have also been shown to produce racial disparities in the medical imaging domain 9,10 , there are no known, reliable medical imaging biomarker correlates for racial identity. In other words, while it is possible to observe indications of racial identity in photographs and videos, clinical experts cannot easily identify patient race from medical images

Race and racial identity can be difficult attributes to quantify and study in healthcare research 15, and are often incorrectly conflated with biological concepts such as genetic ancestry 16. In this work, we define racial identity as a social, political, and legal construct that relates to the interaction between external perceptions (i.e. “how do others see me?”) and self-identification, and specifically make use of the self-reported race of patients in all of our experiments

After all those virtuous statements, they hope they are in the clear. They are against the evil discrimination shown by artificial intelligence. Only then can they turn to the very bad news.

In this study, we investigate a large number of publicly and privately available large-scale medical imaging datasets and find that self-reported race is trivially predictable by AI models trained with medical image pixel data alone as model inputs. We use standard deep learning methods for each of the image analysis experiments, training a variety of common models appropriate to the tasks. First, we show that AI models are able to predict race across multiple imaging modalities, various datasets, and diverse clinical tasks. The high level of performance persists during the external validation of these models across a range of academic centers and patient populations in the United States, as well as when models are optimised to perform clinically motivated tasks. We also perform ablations that demonstrate this detection is not due to trivial proxies, such as body habitus, age, tissue density or other potential imaging confounders for race such as the underlying disease distribution in the population. Finally, we show that the features learned appear to involve all regions of the image and frequency spectrum, suggesting that mitigation efforts will be challenging.

Essentially, a good data-crunching program can work out the race of people in X rays, and nothing can be done about it. This is a horrible result which cannot be “mitigated”. The authors do not for a moment consider the possibility that race is biologically real.

They studied chest, hands, breasts (mammogram), and lateral spine (including skull) X rays. In seeking to control for things which might have explained the results, they corrected for body mass index and also for bone density. The latter is a little publicised race difference, such that American black women and men achieve 5%-15% greater peak bone mass than white persons.

Despite all these corrections, race was detected with high accuracy. For some reason, it seems to be there in the bone structure.

By high accuracy, I mean that they system produces true positives of racial identification rather than false positives, and very few false negatives. The “receiver operating characteristics” (ROC) are excellent, often 98% accurate. ROC scores were designed in the early days of radar, when it was often difficult to distinguish a true signal from background noise. Receivers varied in their accuracy. In this case the capacity to pick up a true signal is very high. Furthermore, the system can be trained to do this without being directly trained for that particular task.

We hypothesized that if the model was able to identify a patient’s race, this would suggest the models had implicitly learned to recognize racial information despite not being directly trained for that task.

Again, even though races differ in fatness, body mass index does not do a good job of discriminating between x-rays of different races. Surprisingly, when they degraded the x ray images so much that radiographers could not even see that they were x rays, the system could still discriminate the different races.

In despair, they say:

One commonly proposed method to mitigate bias is through the selective removal of features that encode protected attributes such as racial identity, while retaining as much information useful for the clinical task as possible, in effect making the machine learning models “colorblind” 43. While this approach has already been criticised as being ineffective in some circumstances 44, our work further suggests that such an approach may not succeed in medical imaging simply for the fact that racial identity information appears to be incredibly difficult to isolate. The ability to detect race was not mitigated by any reasonable reduction in resolution or by the injection of noise, nor by frequency spectrum filtering or patch based masking.

It is a case of: “Out, damned spot! out, I say!”.

Lest they be misunderstood, they explain themselves further.

There has been extensive research into the correlation between self-reported race and genetic ancestry in the field of genomics, which have shown more genetic variation within races than between races, and that race is more a social than biological construct 16. We note that in the context of racial discrimination and bias, the vector of harm is not genetic ancestry, but instead is the social and cultural construct that is racial identity, which we have defined as the combination of external perceptions and self-identification. Indeed, biased decisions are not informed by genetic ancestry information, which is not directly available to medical decision makers in almost any plausible scenario. As such, self-reported race should be considered a strong proxy for racial identity.

We strongly recommend that all developers, regulators, and users who are involved with medical image analysis consider the use of deep learning models with extreme caution. In the setting of x-ray and CT imaging data, patient racial identity is readily learnable from the image data alone, generalises to new settings, and may provide a direct mechanism to perpetuate or even worsen the racial disparities that exist in current medical practice. Our findings indicates that future medical imaging AI work should emphasize explicit model performance audits based on racial identity, sex and age, and that medical imaging datasets should include the self-reported race of patients where possible to allow for further investigation and research into the human-hidden but model-decipherable information that these images appear to contain related to racial identity.

“Human-hidden but model-decipherable” is a danger, the authors aver.

What is a reasonable conclusion to draw from this research? Here is my version:

Interpretation: That a model can predict race — even from corrupted, cropped, and noised medical images — in a setting where clinical experts cannot, is a novel finding which reveals racial differences at the skeletal level. The biomechanics of these differences for health and behaviour should be the subject of further research.

By this I mean that if X rays can detect race, then there is certainly more to it that skin pigmentation. Does having a particular skeleton have any behavioural consequences?

It might partially explain differences in different sports and occupations. If you have denser, stronger bones, might that give you an advantage in a fist fight? It could also partly explain vulnerability to falls and other injuries, with clear implications for health. This artificial intelligence system can correctly identify skeletons by race, even if you control for bone density. Those density differences are a real-world difference with big implications for falls in the elderly.

Shorn of its exculpatory protestations, this is an important paper. Race is real by objective measures. Perhaps radiographers can see it and think it politic not to discuss it in public. Deep learning networks don’t have to be so circumspect. If there is a difference to be found, they can report the signal they detect within the noise. They have found such a signal.

Hide 159 CommentsLeave a Comment
Commenters to FollowEndorsed Only
Trim Comments?
  1. samoan says:

    I didn’t think racial differences in the medical field were so controversial. I work for a medical device company and for our ECG and Spirometry algorithms we require the patient’s race to perform an interpretation. We actually require 5 demographics (race, gender, age, height and weight), but I know of a competitor that sells an ECG device whose algorithm only uses race and gender for demographics which leads me to believe that race is either the most or second most important factor in ECG at least.

    I’m not sure about the rest of the medical field, but at least for cardiovascular health racial and gender differences are too big to ignore, so no one questions the importance of either as demographic inputs. If we ignored race and gender our algorithm would spit out inaccurate interpretations and therefore incorrect diagnoses which is obviously unacceptable.

  2. BuelahMan says:

    Why the weird arguments over something so obvious to anyone looking.

  3. dearieme says:

    A few years ago I was asked by a nurse in an NHS ward whether I would help with a research project. “Certainly.” She led me through her questionnaire. When she asked me my ethnic group I protested that I wanted nothing to do with such Nazi stuff. There my participation ended.

    When I told a pal about it he said “But everyone knows race matters for health and so on.” I replied that if she’d asked about race I would have replied but I was damned if I was going to reply to “ethnic group” because I had no idea what sort of odious doctrine might be disguised by that slippery label.

    P.S. About the paper: I can guess what they mean by Black and more doubtfully by White but what on earth do they mean by Asian? A big and varied place, Asia. Or have they shown that all Asians are brothers under the skin, from the Hairy Ainu to the Negritos on the islands in the Bay of Bengal? A hae ma doots.

    P.P.S. I wonder whether they could try to detect % Neanderthal or % Denisovan and compare with DNA results. That might be fascinating and keep them safely away from areas where they feel under pressure to deal in lies.

  4. Anonymous[369] • Disclaimer says:

    A few years ago I was asked by a nurse in an NHS ward whether I would help with a research project. “Certainly.” She led me through her questionnaire. When she asked me my ethnic group I protested that I wanted nothing to do with such Nazi stuff. There my participation ended.

    “Why I’m half-West African and half-Northeast Asian. Isn’t it apparent??”

    • Replies: @Dingo bay rum
  5. Anonymous[869] • Disclaimer says:

    The real point – ignored by the pompous pontificating professors – is that AI is doing what is supposed to do, and working like a charm, and is *certain* to be an incalculably large boon to medical science and human welfare, in that it is able to unerringly distinguish characteristics and patterns that the human brain, even with training and experience, is simply unable to distinguish.

    • Agree: YetAnotherAnon
    • Replies: @bike-anarkist
  6. Self-identified according to US Census classifications, so a very broad canvas.

    The Census Bureau defines a person of the Asian race as “having origins in any of the original peoples of the Far East, Southeast Asia, or the Indian subcontinent including, for example, Cambodia, China, India, Japan, Korea, Malaysia, Pakistan, the Philippine Islands, Thailand, and Vietnam.”

    • Replies: @dearieme
    , @pyrrhus
  7. @samoan

    Medical uses seem to get a special pass, but even then I find that things like bone density are not widely known by doctors.

  8. TG says:

    The so-called “left” (really far-right corporate whore running dogs using race to distract us from class war but don’t get me started) are rather schizophrenic on the issue of race.

    So yes, on the one hand they will deny that it’s more than skin deep, even when, for things like bone density on x-rays, the idea is absurd. And yet, for some time now the NIH has made it a big issue to ensure adequate representation of race and gender in medical studies and clinical trials, because a clinical treatment only designed and tested for white males might not give exactly the same results for white females or black men or asians etc. Far from complaining about it, it’s widely considered morally and medically justified – and in this instance, I surely agree.

    • Agree: Peter Johnson
  9. Anon[568] • Disclaimer says:

    Yeah, yeah, yeah. Who cares about physiology except basketball coaches.

    The burning question is WHY mindless homicides can be predicted using Race. I mean, EVERYBODY KNOWS that we Irishmen are all drunken homicidal maniacs, right? But we’re gonna kill you because you’re a PROTESTANT, not because you’re WHITE. Or Black. Or Yellow. Be gosh and begorra. After a couple beers and some Guinness we’ll even forget you’re not Catholic. (To my knowledge, the IRA isn’t violently anti-Bahia, etc.)

    And Blacks who HONESTLY join White society are FULLY accepted as White. I’ve worked with enough Black business consultants to know that what MATTERS is how well you understand Systems Analysis.

    But if you INSIST on being Black FIRST, then you ain’t got a future in civilized society. That is, “I’m gonna act like an ignorant jerk, and you MUST accept me for ‘who I am’ ” ain’t gonna get you very far beyond government subsidized set asides… (See Charles Murray’s books)

    But I gotta go play Irish fighting songs on YouTube… “Come out ya Black and Tan! Come out and fight me like a man…”

    • Replies: @Anon
    , @Curmudgeon
  10. Anon[568] • Disclaimer says:

    Yeah, yeah, yeah. Who cares about physiology except basketball coaches.

    The burning question is WHY mindless homicides can be predicted using Race. I mean, EVERYBODY KNOWS that we Irishmen are all drunken homicidal maniacs, right? But we’re gonna kill you because you’re a PROTESTANT, not because you’re WHITE. Or Black. Or Yellow. Be gosh and begorra. After a couple beers and some Guinness we’ll even forget you’re not Catholic. (To my knowledge, the IRA isn’t violently anti-Bahia, etc.)

    And Blacks who HONESTLY join White society are FULLY accepted as White. I’ve worked with enough Black business consultants to know that what MATTERS is how well you understand Systems Analysis.

    But if you INSIST on being Black FIRST, then you ain’t got a future in civilized society. That is, “I’m gonna act like an ignorant jerk, and you MUST accept me for ‘who I am’ ” ain’t gonna get you very far beyond government subsidized set asides…

    But I gotta go play Irish fighting songs on YouTube… “Come out ya Black and Tan! Come out and fight me like a man…”

  11. Anon[568] • Disclaimer says:

    Oh, it’s also relevant that the Irish (meaning CATHOLIC Irish) are “the [email protected] of Europe”.

    • Agree: Wyatt
    • Replies: @Swaytonious
  12. dearieme says:
    @James Thompson

    What about Central Asians? Uzbeks? Uighurs? Ah, maybe Uighurs aren’t numerous in the US.

  13. Alfa158 says:

    “ One conception of race is that it is skin deep, and is no more than a matter of skin pigmentation.”
    Just my own observation, I don’t know, nor have even heard of, anyone who actually holds that conception. Every time it gets trotted out it is being used as a clumsy and transparent straw man argument against race realists. The midwits who use the argument think that race realists must be even dimmer than themselves, and are therefore gob-smacked when the response is “Nope”.

    • Agree: bomag
    • Replies: @James Forrestal
  14. Notsofast says:

    what i would like to know is why their a.i. is offering unsolicited opinions as to the race of the patient. perhaps it has a.i. tourette syndrome and starts to hurl racial epithets uncontrollably. in addition to this facial recognition a.i. has a problem of identifying some black faces as gorillas. perhaps the repressed subconscious predjudices of the programers has somehow baked itself into the a.i. cake, kinda like what brought down the krell civilization in “forbidden planet”. i always said this a.i. thing wouldn’t end well.

    • LOL: bomag
  15. This is an astounding paper. It appears to reveal that, using artificial intelligence, deep learning methods lead to the detection of race in X ray images, even if all possible giveaway signals are stripped out of the image. That is extraordinary.

    It really isn’t a surprise to those of us that know about pre-Boasnian Anthropology.

    70k years of racial separation (a conservative estimate) will leave all kinds of evolutionary marks even if they aren’t readily perceptible.

    There are undoubtedly all kinds of proportional differences that AI would detect.

    We aren’t looking for them because we can already identify race 5k ways and most importantly through DNA.

    Forensic anthropologists are in an amusing situation as they are expected to maintain the belief that race doesn’t exist for Academic/egalitarian reasons but are also trained to identify the race of specific bones.

    All one big ridiculous circus and I guess everyone goes along with it until a computer calls BS.

    Skynet doesn’t understand what the big deal is.

    • Replies: @bomag
  16. “70k years of racial separation (a conservative estimate) will leave all kinds of evolutionary marks even if they aren’t readily perceptible. :

    If you’re talking about whites and negroes, they never were separated, because they never were together. The out-of-Africa theory is just wishful thinking by Boasian anthropologists who can’t prove anything conclusively, every few years they just trot out new theories about how much the races are alike.

    • Replies: @John Johnson
    , @IronForge
  17. Realist says:

    Interesting information…thanks.

  18. @Joe Paluka

    Yes the out of Africa theory (with only superficial differences) has fallen apart but even the original establishment version actually allows plenty of time for evolutionary differences to exist.

    They are forced to accept that racial separation existed for at least 70k years which is actually plenty of time for subspecies to develop if the environments are different. Lactase persistence for example is only 7500 years old. But more importantly natural selection can quickly change group genetics.

    Boasian Anthropology tries to block the conversation entirely even if the message is contradictory. They have basically won the department so such studies will have to exist elsewhere.

    • Replies: @Wyatt
  19. linsee says:

    I am delighted to see a new post from you after so long!

    Very interesting, too, but for your posts that goes without saying.

    I am reminded of the hand-wringing over AI “discriminating” against black mortgage applicants even though the data include nothing explicitly about applicants’ race.

  20. The underlying results show that you cannot get rid of race in an Xray,

    One more chapter in the big book of unintented consequences:

    Social construct x-rayed and debunked. Surprise, surprise!

    • Agree: Pepe the Frog
  21. @Anon

    Isn’t that the gypsies?

    • Replies: @Biff K
  22. My dentist told me they had to crank up the x ray machine to the max on negroes. Otherwise, the x ray was a blur.

  23. Dr. Doom says:

    Real Science is raciss and sheitt. Math is especially raciss.
    Objective reality disagrees with the Zion Pig fantasies.


    God, Jesus and even Satan do not support this Babylonian Cesspool.
    The Zion Pigs did it all. They will take the Fall.

    Such piss poor lies and betting on black.
    We’re seeing levels of STUPID that used to seem IMPOSSIBLE.

    But, what do you expect from dummies that believed Satan’s lies?
    You wouldn’t expect Babylon to have any firm grounding.

    The Zion Pigs are ALL-IN on Babylon. They WANT TO BELIEVE.
    You know and I know Babylon will Fall. Its inevitable.

    The Zion Pigs are TOO STUPID and FANATICAL to see it.
    By the time they find out, it will already be too late.

    Life here is a TEST. Do the Right Thing or be damned and DOOMED.
    The Zion Pigs always lose. They are STUPID. Never learned anything.

    The Zion Pigs are grooming their “allies” to KILL THE PASTY RICH.
    That is how STUPID they really are. Zion Pigs ARE the Pasty Rich.

  24. utu says:

    The robustness of their AI method is suspicious.


    HPF maintains high performance up to diameter 100, which is notable as there are no visually discernible anatomical features in the sample image even at lower radius levels (i.e., performance is maintained despite extreme degradation of the image visually).

    • Replies: @James Thompson
    , @res
  25. Angharad says:

    Jewish money and threats of being financially, socially, and professionally unpersoned.

  26. @dearieme

    Nice examples.

    The “science” article in question dates back to mid-2021. So far as I could determine, no peer review. And it has no DOI, so I doubt if it has been published anywhere except on the Internet.

    Why is that? And I’d have to ask if anyone has done more than to “ooh” and “ah” at the marvelous software.


    Maybe there are some identifying skeletal features for an ebony black human compared with his/her lily white counterpart of the same height, weight, and age. But I really, really doubt that the 50/50 mulatto is going to be identifable. Fuggedaboutit for the Hexadecaroon. I seriously doubt some freaking software which isn’t somehow fudged will declare that individual to be of the negro race.

    As of now, I smell BS. A “Clever Hans” situation. Or worse.

  27. Biff K says:

    Yes. Gypsies are the niggers; Irishmen are the toothless hillbillies.

  28. IronForge says:
    @Joe Paluka


    Science Fail.

    Mitochondrial Eve – was the Matrilineal Most Recent Common Ancestor.

    She (dated to be 100K-230K Years Ago) was found near the current Pygmy Habitats, shared Genetic Similarities with them. BTW, Current Pygmies have Avg IQs of around 51.

    EVERY HUMAN ALIVE TODAY are her Descendants.

    Interesting to consider what made her stand out amongst Thousands of Early Human Females.

    • Replies: @Levtraro
    , @G. Poulin
  29. @dearieme

    A hae ma doots.

    I`m sure Unz readers are cosmopolitan enough to understand you.

    As you`ll see from my name, I am probably from the same “ethnic group” as you.

    Can`t imagine there are many of us among Unz readers. Or could I be pleasantly surprised?

  30. Alfred says:

    By this I mean that if X rays can detect race, then there is certainly more to it that skin pigmentation.

    At the very least, all medications should be tested by race. Of course, the pharmaceutical companies will be loath to do this. The costs of so doing are prohibitive.

    All medications tested in cheaper locations such as India or Africa should only be approved for those locations and peoples.

    A dose that is appropriate for a White American may be inappropriate for a Black American. Just think of the consequences! The mind boggles. 🙂

    Many drugs for U.S. kids tested in poor countries (Reuters)

  31. GMC says:

    If A I is so smart , how come he didn’t see that the mRNA jab was going to kill thousands of people ? If A I is so smart how come he didn’t see that the US does not fit into the Democracy category . If A I is so smart , why are only the evil bastards allowed to program it ? Forget A I until the planet becomes normal and intelligent, , which isn’t going to happen anytime soon. It will only end up being used against the people, just like the doctors, lawyers, politicians,military and others have been.

    • Agree: Alfred
    • Replies: @Mike-SMO
  32. radiographers ????????

    I thought these were technicians and “Radiologists” interpreted the outputs !

    Maybe it is different in US health care ?

    Anyway, I am sceptical about Radiology itself since it is the Radiologist rather than the equipment that determines the findings and there are FEW good Radiologists who do not overlook obvious findings. I am sceptical from experience from many encounters where they have failed to detect what was there through failure to set the machines to identify what is there

  33. Sean says:

    Blacks have denser bones.

    • Replies: @gotmituns
  34. Gordo says:

    It appears to reveal that, using artificial intelligence, deep learning methods lead to the detection of race in X ray images, even if all possible giveaway signals are stripped out of the image. That is extraordinary.

    Hardly Doc, forensic pathologists and archeologist have been racially classifying skulls at a glance for years.

    • Agree: Alfred
    • Replies: @res
  35. Gordo says:

    We strongly recommend that all developers, regulators, and users who are involved with medical image analysis consider the use of deep learning models with extreme caution.

    So beware of learning these chaps are saying! Mmmh.

  36. Race is a social construct and social constructs are just as valid as anything else. Certainly a valid cause to rally around and prey on other humans for. Humans thrive through cannibalism. All seek superiority over all others. Some outright declare themselves ontologically superior to and distinct from the rest of the human race. You know who I am talking about.

    I guess what I am trying to say is I am a traditional racist while you modern racists insist on dressing it up with science and are obsessed with finding a biological basis of race. What folly.

    • Thanks: Ben the Layabout
    • Replies: @Old and Grumpy
    , @Curle
  37. Mike-SMO says:

    You just don’t seem to be aware of the “game”. I put the mRNA vaccines in a category of an emergency stuff job to fight a deadly disease. It worked sort of. If you listened closely to President Trump etc. you “knew” what was coming. I think that those vaccines fit in with the DD “floating”(sic) Sherman Tanks deployed for the Normandy Invasion. At the Brit and Canadian beaches, they saved lives by destroying defenses and bunkers. Off Omaha Beach 27 of 29, as I recall, sank in deep water killing their crews. It rapidly became obvious that the Chinese Virus was mostly dangerous to people with untreated medical co-morbidities (obesity, diabetes, high blood pressure, immune suppression due to transplants or implants, etc). The vulnerable individuals should have been protected. For the rest, the vaccine was a greater risk than the disease. You don’t need AI for that. Common sense would do.

    Haven’t you figured it out yet? The mandates were for effective working people. Illegal Aliens and welfare burdens were exempt. As well as the Congress critters, the Judiciary and the manufacturers of the novel “vaccines”. The “vaccine” push was designed to destroy the economy and to make positions available for the Progressive “pets”. Do you need AI to explain it for you?

    • Replies: @GMC
  38. Good God. People had this figured out three hundred years ago, no x-rays needed. You just had to observe. But society willingly put on the Jewish blinders, especially since the 60s.

  39. gotmituns says:

    That’s why they sink faster than Whites and Asiatics in water.

    • Agree: Lancelot_Link
    • Replies: @Sean
  40. bomag says:

    It’s the current year.

    Racial politics trumps a lot of common sense.

  41. @Anon

    “Bahia” (Bahía) is in Brazil. I believe you meant Bahá’í, a relgious sect derived largely from Islam.

    • Replies: @Ben the Layabout
  42. bomag says:
    @John Johnson

    What you said.

    In a creepy way, this paper is setting us up to accept medical failures that could have been prevented by taking race into account.

    “To save the Narrative, we had to let the patient suffer.”

    • Agree: Alfred
  43. @Biff K

    The English are well known for dental distress; the Irish not so much.

    • Replies: @Biff K
  44. Anonymous[223] • Disclaimer says:
    @Biff K

    And you’re a potbellied pig…

  45. @utu

    Do you mean that the claimed robustness is suspicious, since a true system would break down if the signals were degraded? I can see the logic of that, unless what the method picks up is “the correlation of correlations”, that is to say, a general overall picture which can be detected even through noise.

    • Replies: @utu
  46. Another nail in the coffin of race denialists.

    • Replies: @Ben the Layabout
  47. @BuelahMan

    Because …..We Wuz Kangz, Queenzzz an Wakandanzzzz an shiet!

  48. Well, I’m certainly glad we can trust doctors and medical researchers to never succumb to politics and to always ‘follow the science’, wherever that may lead………

  49. usNthem says:

    Wow. There actually are biological differences between blacks and everyone else. Who’d a thunk?

  50. @The_Masterwang

    Don’t disagree with the gist of what you are saying. However the term social construct is now a modern meaningless word jumble. My 19th century born grannies would have assume you meant economic class structure, and not cultures of people from different parts of the world.

    • Replies: @The_Masterwang
  51. TREG says:

    Whoopi said the holocaust was “Whites killing Whites” and not about race and so it was “really about Mans inhumanity to Man”. Immediately Leftist Billionare and Jew Ron Lauder and Alt-right millionaire and Jew Paul Gottfried said very loudly that yes, the holocaust was all about racism and condemned Whoopi words. Well, well, well…we can settle this by letting the AI look at all the bones of the dead in WWII… Will the AI be “antisemitic” ??? For me, its a win either way… Lol

  52. GMC says:

    You want to talk like a Moron – go use the MSM commenting scenarios. You can’t even understand the fact that A I is a tool for the Elite. Go peddle the game in the US – My president is Putin.

    • Replies: @bike-anarkist
  53. So much for bone X-rays. Has anyone been brave enough to do the same research for MRI scans of the brain? Even after controlling for size and shape of the cranial cavity, I would bet that AI can determine race from a brain scan, even when a skilled neurologist cannot.

    • Agree: Pepe the Frog
  54. In my world, X-ray technicians take pictures of people. Radiographers take pictures of pipe welds etc. Doctors diagnose X-ray pictures. I have no idea who the people in the article are or what they actually do. Maybe the AI knows.

    • Replies: @Curmudgeon
  55. bert555 says:

    I’m sure the AI was trained using datasets with images that were pre-labelled with race information. It then recognised these categories when given unlabelled images. What would absolutely seal the deal is if a generic categorizing AI could be built that was not inherently trained with pre-categorized images and if that AI then correctly separated out the races.

    • Replies: @SaneClownPosse
  56. macilrae says:

    Those of us who have had their DNA scrutinized by 23andme will know that scientific race determination is unequivocal. For example, as an apparent “white man”, I have 5% Ashkenazi Jewish, 0.5% west African and 0.5% Egyptian origins. African heritage is spectacularly easy to discern.

    All they needed to tell was some of my spit!

    But what perplexes me is how the science of biology has long been subverted to accommodate racial sensitivities.

    Distinct “subspecies” in the animal kingdom are exquisitely described which, to the normal person, appear identical yet, in the case of humans, biology must somehow swallow the obvious lie that there are no human subspecies! This point was well demonstrated here on Unz by Jared Taylor, who started his debate by showing a tall white man standing among a group of starkly clad pygmies.

    • Agree: 3g4me
    • Replies: @Gordo
    , @Pepe the Frog
  57. res says:

    Agreed. I made a fairly long comment when Steve Sailer posted about this in August.

    I wish they could figure out what was being detected. My take then was that the authors tried pretty hard to do that based on their series of experiments.

    • Thanks: Calvin Hobbes
    • Replies: @utu
  58. res says:

    Reread your quote. They key point was this: “even if all possible giveaway signals are stripped out of the image.”

    • Replies: @Gordo
  59. These authors seem to want to make it less likely to treat people based on conditions that they’re more likely to have.

    This could be a boon for the people who would prefer that we didn’t treat people for the results of their unhealthy lifestyles. After all, such conditions are far more prevalent in ‘underserved populations,’ so identifying them correctly would be evidence of biased AI.

  60. Whitaboot says:
    @Auld Alliance

    Dearieme indeed. Clearly a biscuit-ersed radge.

  61. @Old and Grumpy

    My granny would have simply not bothered with any -ism – the mere presence of foreign devils was trouble enough. People back then had identities. They didn’t need anything scientific to justify their judgement.

    That being said, damn, an AI can tell race from pictures of rib cages?

  62. @Biff K

    Irishmen are the toothless hillbillies? I think you mean SCOTS-IRISH…you know, those descended from indentured servants who, after paying off their indent, were given “40 acres and a mule” in Appalachia…. And aren’t gypsies descended from migrants from India?

  63. @James Thompson

    Medical uses seem to get a special pass, but even then I find that things like bone density are not widely known by doctors.

    To what do they attribute the obvious difference in hip fracture rates between Black vs. White/ Asian elderly females, then? The rates diverge as much by race as they do by sex — it’s not a trivial difference:
    Heterogeneity of hip fracture: age, race, sex, and geographic patterns of femoral neck and trochanteric fractures among the US elderly
    Race and Sex Differences in Hip Fracture Incidence

    Rather than simple ignorance, it may well be a matter of not wanting to step into what, in the wrong context, can be a veritable minefield of easily-broken taboos. Kind of like the way that massively disparate rates of Black violent crime are “not widely known” to shitlibs. Tactical “ignorance,” as it were.

  64. anon[289] • Disclaimer says:

    High caste Indians became gypsies in Europe. Afro-centrics claim black Irish and Indians are Negroes, which’s completely untrue. Calling Indians, Irish, Italians, ect. “niggers” feeds the dim Blacks ego.

  65. @Anon

    LOL, I think the Proddies prefer “The Sash” and “King Billie’s on the wall”.

  66. @David Homer

    As a former Radiographer, of people, the term X-Ray Technician or X-Ray Technologist are the most commonly used titles in North America. Radiologists are the physician specialists that read X-Rays, Nuclear Medicine scans and Ultrasound scans, and issue a report.

    I am a little surprised by the article. Before ionometric/photo-timing the technical factors used were set manually, based on the physical assessment of the patient, including medical conditions. Bone density of Blacks was taught as a “may have”. Skull shape was a factor for everyone when trying to demonstrate internal structures.

  67. Levtraro says:

    EVERY HUMAN ALIVE TODAY are her Descendants.

    I think you got that wrong. mtDNA passes from female to offspring. If the offspring is a male, in the next reproductive cycle the mtDNA of his mother will not pass. Therefore other females living with Mitochondrial Eve that had sons most surely have descendants today. The effective populations size (the part of the population that can reproduce) of the ancient human species never decreased to less than tens of thousands. Check the webpage you cited, Note number 4.

  68. @Anonymous

    And do we want that?

    I just had a heated conversation over the prankdemic over the argument that, “The new technology is going to revolutionize public health”.

    I frankly told him, while holding up my business mobile (which I only use at my place of business), “I refuse to allow my life to be dominated by poorly thought out and potentially lethal tech. We don’t use technology, the technology forces us into unconscious but changed behaviour.”

  69. @GMC

    You want to talk like a Moron

    You have got that wrong. He IS TALKING to a MORON.
    Mike explained in detail what our “authorities did with their amazing tech like RT PCR, or Experimental Gene Therapeutics.

    “A tool is a weapon, if used correctly”.

    Especially humans.

    • Replies: @GMC
  70. anon[161] • Disclaimer says:
    @James Thompson

    South Korean diplomat sucker-punched in unprovoked NYC attack

    Officers responded to a 911 call about an assault in the area of Fifth Avenue and West 35th Street in Midtown Manhattan at around 8:10 p.m. ET. Upon arrival, they found a 52-year-old man suffering from pain and swelling to his face. The victim was a diplomat from South Korea, according to the New York City Police Department.

    After further investigation, the officers learned that an unknown man had approached the victim at the location and punched him in the face before fleeing on foot toward Sixth Avenue. Police told ABC News that “the incident was unprovoked.”

    “The name of the victim in Wednesday’s incident has not been disclosed.” Neither has a description of the assailant, but based on prior incidents it won’t take you 3 guesses. Release the video! (There must have been a security cam.)

  71. Curle says:

    Air is an social construct. Having a word to describe such a thing is an social construct. Social constructs can be amazingly useful to the problem solving process.

    Problems, by the way are social constructs.

  72. Another chapter in the annals of “diversity is great but race is not real.”

  73. “They assert that race is not based on ancestry, but is a social construction and based on self-report.”

    Something about lunatics and the asylum seems relevant.

    Are molecules social constructs? Two H atoms and one O atom can decide that they are something other than a water molecule.

    Next I suppose that whether one is living or dead will be a social construct. Maybe take a year off dead for tax purposes. [HHGTG]

  74. utu says:

    “I wish they could figure out what was being detected”. – That’s the whole point of neural networks that you use them because you do not want to bother with figuring out how they works. It is like magic. As long aa it works you do not ask questions.

    We humans can be fooled with optical illusions but neural nets can be fooled even worse. We know of very sensitive instabilities when few pixels change causes catastrophically different output.

    But here the case is opposite. The algorithm is unreasonably stable producing correct identification from data virtually devoid of information.

    Perhaps in the corner fo each image somebody placed very small letter W, B, H, or A and that’s how the neural network learned how to categorized images. Or it could be a sabotage at some level by one of the researchers or a hoax targeting Sailer and Thompson and you. Sokal hoax targeting the alt-right HBD crowd.

    • Agree: nokangaroos
    • Replies: @dearieme
    , @Canspeccy
  75. @bert555

    The AI might not have been fed the categories.

    The AI may have determined the unnamed categorization from its learning, without having a concept of race. Probably didn’t even know it was looking at x-rays of humans and was just sorting digital images according to common characteristics.

  76. Is defining as a Zionist Jew a social construct?

    Defining as a White male is not a valued social construct.

  77. dearieme says:

    It is like magic. As long as it works you do not ask questions.

    A bit like quantum mechanics, then?

  78. It really comes down the inarguable fact that there is no reason to expect populations that were isolated from each other over tens of thousands of years did not evolve differently at every level of their biology, since evolution affects everything about us and everything about us is rooted in our biology, including our intelligence, which is literally part of the biological organism of our brain, not some amorphous thing floating in the air bestowed to all humans upon birth. No one who takes science seriously and understands science and argue with this fact.

  79. @Anonymous

    Listen idiot its not the nazis.Its genetics and physical anthropology. I still would like to see your antecedents before commenting. As they say in photograpy, lighting is everything.

  80. @utu

    It could be a hoax. I try to read papers as if they were honest, so I am certainly open to being fooled.

    • Replies: @nokangaroos
  81. @James Thompson

    – An African dentist has no problem determining race from a single tooth,
    but at tit x-ray so noised up as to be unrecognizable? Surely they should check
    for systematic error before decrying the End of the World.
    Still it´s incredible how powerful these things are; some time ago they tried to teach
    an AI psychiatric evaluation and it came up with three new symptoms no one had
    thought of before the calibration was finished.
    Of course AI has no concept of “race”, it simply states “one of these things is not like the others” and does not give a smoking whooptie how the thing “feelz” – that´s why it is called “intelligence”.

  82. @James Thompson

    When I was in school, we were told that “race doesn’t exist” and also that statements like “black men have heart attacks at higher rates than white men” was correct. One of assigned readings was a rather long article arguing that when you read or use terms that refer to race in a health care context, it seems like it means race exists, but actually race doesn’t exist. I, as you might imagine since I am reading Unz Review, found the paper unconvincing.

  83. Ok, xray can tell the race differences. What relevance does this have to do with anything though? besides the bone density?

    darker skin = more D from the sun = better absorption of calcium = higher bone density?


  84. @Alfa158

    “ One conception of race is that it is skin deep, and is no more than a matter of skin pigmentation.”
    Just my own observation, I don’t know, nor have even heard of, anyone who actually holds that conception. Every time it gets trotted out it is being used as a clumsy and transparent straw man argument against race realists.

    It’s more of a media narrative term than something normal people say in everyday conversation. But if they repeat it often enough, it eventually sinks in. And a lot of times you see it deployed a little more subtly, by simply using the term “skin color” as a substitute for “race,” without ever explicitly asserting the “race = just skin color™” trope*. Like a lot of “argument” by semantic manipulation, this has the advantage of operating at the level of associative conditioning rather than empirically-based dialectic, so it often flies beneath the radar. This usage is very common in mainstream “conservative“/ establishment GOP rhetoric.

    In mainstream shitlib usage, social constructivism is a shibboleth that simply cannot be challenged, so the “just skin color” canard is actually less common. You tend to see narratives more along these lines:

    In the 21st century, as America becomes less white and the multiracial community—formed by interracial unions and immigration—continues to expand, color will be even more significant than race in both public and private interactions. Why? Because a person’s skin color is an irrefutable visual fact that is impossible to hide, whereas race is a constructed, quasi-scientific classification that is often only visible on a government form.

    [The author then goes on to rant about “colorism” — which is basically muh brown paper bag test, etc.]

    Amusingly, this anti-“just skin color” take puts the Black shitlib author of this piece in some rather… diverse company:

    This writer at Jewish World News believes that race “isn’t about skin color” because the “just skin color” trope is bad for semitic supremacism.

    And of course, people like Jim Goad at Taki Mag and the ebil White supreeemist not-sees at Stormfront also agree that race is not “just skin color.” As does Razib Khan — though he’s addressing mostly mixed-race populations like mestizos, Puerto Ricans, and Uyghurs. Khan is a known heretic who’s very solid on genetics, but is definitely not a White supreeemist rayciss not-see, or even a fellow traveler. He’s a committed rootless cosmopolitan in the literal, rather than dog whistle, sense.

    Incidental note: While muh “just skin color!” trope does often function as a straw man, it’s also used as a “motte” to the social constructivist “bailey” — a “safer” space to retreat to in situations, or with target audiences, where pure social constructivism would be laughed at.

    *Perhaps the most well-known instance of this is the notorious “color of their skin” vs. “content of their character” quote from King — which is commonly cited by “conservative” politicians in the current year.

  85. TKK says:
    @James Thompson

    My mother was an X-ray technologist. I remember as a child she told me that more “juice” was needed to penetrate the skulls of blacks for a skull x-ray to be readable. This was not reported as a racial slight, just more of a ‘gee wow” science insight about her job.

    Furthermore, forensic pathologists and archeologists deduce race by skeletal remains, as well as gender, without any dispute or debate.

    Will new woke fiction have a gender fluid Kay Scarpetta who has to pretend the remains of a victim can’t demonstrate race or gender to fire the starter gun for a murder mystery?

    • Replies: @Pepe the Frog
  86. Ruckus says:

    So we have to use fucking AI to get the normies to see the obvious. Great.

  87. @Sean

    Where did you find reference to the connection between bone density and male fertility? It doesn’t appear to be in the Guardian article.

  88. @Sean

    I had trouble with the audio, but I did managed to hear that there was a connection between bone and testes!

  89. GMC says:

    Well, that is a no brainer – Dude – especially if you read Unz and other alternate websites. I was discussing A I , as in the article and that the Elite will use A I – strictly against us. An honest programmed A I would not have OK’d the mRNA jabs, the PCR tests, the masks, etc. etc. etc.

    I know all about the m RNA jab – it put my brother in the Hospital for a double by pass and new valve. His second jab took him out for 2 weeks and 2 mos. later – it wacked him. But he wouldn’t listen to me because the Americans are stupid and his main provider that told him that the jab was OK – should be buried alive. No offense.

  90. @samoan

    Your comments invite some speculation. Imagine how medical treatment would be impacted if governments were to decree that using any information about a patient’s race (or sex, or …) was henceforth illegal, since it was discriminatory. Is that really out of the question? Similar edicts already exist in other domains; perhaps already in medicine too. Never underestimate to what extremes of lunacy the Left can carry its ideals. Will it some day be illegal to even inquire as to the age and sex of a patient? Or even what his complaint is? Will it be illegal to test for blood type, since these tend to be more or less frequent depending upon race? Etc. Are we on a high-tech path back to a new dark age?

  91. @Montefrío

    You are trivially correct. However, who are you to deny a person’s religious freedom to believe he is a body of water adjacent to Portuguese speaking land? 😛

  92. @Bardon Kaldian

    Agreed. I’ve often smirked at another seeming contradiction. Have you ever noticed that the very entities that by turns say that race doesn’t exist, or at least, that it is irrelevant, are those who tend to collect the most data about race, to sort by race, or otherwise emphasize it? Perhaps I’m still stuck in the 1960s mind set of “equal opportunity” (not by belief, by by being raised then). They don’t even pretend equality of ability or opportunity any more. Witness Uncle Joe’s (or more precisely, his handlers’) choice of a black female for VP, or now, for a proposed Supreme Court Justice. They may not explicitly say “we must have a black woman,” but it’s not hard to miss that rarely, if ever, is fitness for the job even mentioned as a qualifier any more.

  93. Gordo says:

    So computer are cleverer than us, this kind of thing hasn’t been a surprise since Big Blue beat Kasparov.

  94. Alrenous says: • Website

    Precisely because it is so obvious.

    You can tell just from looking. So can I. But “experts” can’t? No, they can too, a fortiori, they’re just lying. Which makes them afraid of being caught lying, a fortiori. Which causes the weird arguments.

    Especially since “just looking” is so prole. Any idiot can just look. If you don’t need a PhD in sophisticated statistical analytics to tell…well, then how can they keep the cartel going? Turns out “doctor” is a blue-collar profession 97% of the time. Haha, haha…oh shit.

    Secondarily, how do you program a computer not to be able to tell the difference? Basically, you can’t. If you put in explicit racial corrections, you have to know what the racial differences are…and that’s racist. Anyone looking at your code can see how racist you are.
    If you don’t put in explicit racial corrections, it will see race. It doesn’t even need to know what races are. It will have “disparate impact” regardless.
    Haha, haha….oh shit.

    • Replies: @BuelahMan
    , @Pepe the Frog
  95. Alrenous says: • Website

    It’s worth saying these things out loud.

    The idea that race is only skin deep is utterly absurd. Genuinely dumber than trying to argue the Sun rises in the West. Like Creationism is a bit weird but [race is skin deep] thing is completely bananas. Baby babble is more profound. I’ve seen smarter things floating in the sewer.

    Do you have any idea how deeply interlinked genes are? The thing is a giant mass of spaghetti code. Because of course it is.

    Imagine a human programmer who can’t refactor anything and isn’t even aware that code discipline is a thing he can have. Now, have him hack a single code-base for billions of years. What are you going to get? It’s a miracle you get anything at all, but that thing is going to be a mess on top of a mess on top of a fractal mess.

    If anything is not dependent on everything else, it’s practically a double-miracle.

    Consider the Lewontin fallacy in reverse: with any one allele – just one! – you can predict someone’s race at like 30% accuracy.
    A bit like having a good idea which beach you’re on by looking at one grain of sand. Of course IRL you’re also allowed to look at the horizon, the sky, and inland.
    (What? Three alleles? We’re spoiled. We’re at 75% accuracy.) No wonder a moment’s casual observation can identify someone’s age, gender, race, profession, culture, etc etc. You don’t have one or two alleles to work with, but height, build, clothes, colour, face structure, fat structure, muscle structure, gait, voice timbre, word choice…

    Theoretically, with this much information to work with, it would be almost impossible not to learn a great deal about someone’s genes via application of the eyeball.
    Which is indeed what we experience in real life.

    • Agree: Gordo
  96. BuelahMan says:

    Tay proves your point!

    • Agree: Alrenous
  97. Gordo says:

    Those of us who have had their DNA scrutinized by 23andme will know that scientific race determination is unequivocal. For example, as an apparent “white man”, I have 5% Ashkenazi Jewish, 0.5% west African and 0.5% Egyptian origins.

    So 98.5% White, 1.5% unknown, with ethe alien designators put in for political reasons?

    • Replies: @macilrae
  98. Canspeccy says: • Website

    I don’t see the point of this approach — analyzing X-rays, or whatever to identify race.

    Interbreeding groups inevitably diverge from one another as the result of mutation, selection and genetic drift, and the results of such differentiation are generally visible in facial and other readily observable external features. In turn, most human races can be grouped into families, for example, European, East Asian, sub-Sahara African, Arab, etc. This is obvious to anyone with eyes to see. But for those who don’t want to believe their own lying eyes, DNA sequencing confirms the fact.

  99. Twinkie says:
    @James Thompson

    Medical uses seem to get a special pass, but even then I find that things like bone density are not widely known by doctors.

    Yes, such facts are well-known among physicians. They are just not discussed at all outside specific and relevant clinical settings.

    As some of the readers know, I am ethnically East Asian. All the doctors who worked on me over the years seem to know that East Asians are, on average, “lightweights” where medications such as anesthesia drugs are concerned. You just won’t find them talking about it outside treatment settings. They aren’t that socially dense. 😉

  100. @Twinkie

    So, they know when not to know things. Makes sense.

    • Replies: @Twinkie
  101. @BuelahMan

    Because this could interfere with the genocide of European peoples.

  102. Twinkie says:
    @James Thompson

    So, they know when not to know things. Makes sense.

    If you go to Christmas parties with physicians and get them drunk (which most of doctors, esp. older surgeons do), it all spills out. 😉

    Aside from higher bone density among blacks, East Asian susceptibility to anesthesia drugs, and so on, what’s becoming better known among physicians in affluent areas with IT industry these days is… “Indian heart.”

    Most physicians in such areas know that Indians die young from cardiovascular disease and East Asians die old from cancer (often stomach, rather than breast, lung, and colon common among whites, though the latter types are increasing in incidence among East Asians as well).

    • Replies: @James Thompson
    , @res
  103. @Twinkie

    Thanks. This has also been very evident in the differential Covid death rate within the UK. When discussing race, people often say “apart from medical matters” and then go on with the usual avoidance routines when discussing behavioural differences.

    • Replies: @dearieme
  104. res says:

    How is the knowledge being passed on (or taught?)? I would think if it was talked about in med school lectures there would eventually be complaints. Is the younger generation of doctors just as aware and accepting of these things as the older?

    Do you think “Indian heart” is something genetic or is it just their terrible fitness?

    • Replies: @Twinkie
  105. dearieme says:
    @James Thompson

    From this morning’s Telegraph:

    White people, Christians and professionals were at greater risk of catching coronavirus in the third wave, in a complete reversal of how the pandemic affected social groups previously, latest data from the Office for National Statistics (ONS) show.

    There have been ongoing concerns throughout Britain’s epidemic that ethnic minorities and the most deprived were disproportionately impacted.

    However, latest data show that has flipped in recent months, with experts speculating that groups who caught the virus in the Wuhan and or alpha waves were better protected when delta struck.

    So we can now be confident that no ongoing concerns will be expressed. Because.

    • Replies: @James Thompson
  106. @Twinkie

    Is it because of small body mass or something else? Because “small patient, small doses” seems too obvious to even bother mentioning.

    • Replies: @Twinkie
  107. @dearieme

    At the very end of the article:

    “The ONS said it could not compare the second and third waves to the first, because mass testing was not available in the first wave.”


    And, I am not interested in “cases” but numbers in intensive care, and deaths. Doubt those are higher in the higher social stata.

  108. Twinkie says:

    Is it because of small body mass or something else? Because “small patient, small doses” seems too obvious to even bother mentioning.

    No, East Asians don’t require as much anesthetics per bodyweight as Europeans or Africans. It’s likely genetics. And it also manifests in pain levels.

    A total of 21,229 anesthetics were included in analyses. In the adjusted analysis, no racial and ethnic group received significantly more or less opioids intraoperatively than non-Hispanic (NH) whites. Asians, Hispanics, and Pacific Islanders were estimated to have significantly lower odds of receiving non-opioid analgesics than NH whites: odds ratio (OR) = 0.83 (95% confidence interval (CI): 0.70, 0.97); OR = 0.84 (95% CI: 0.74, 0.97), and OR = 0.53 (95% CI: 0.33, 0.84) respectively. Asians were estimated to have significantly lower odds of reporting moderate-to-severe pain on awakening than NH whites: OR = 0.80 (95% CI: 0.66, 0.99).

  109. Twinkie says:

    How is the knowledge being passed on (or taught?)?

    Physicians do learn some of the genetic differences during schooling. But during internship, residency, and, later, practice, they learn firsthand. Don’t forget medicine is still largely taught by practice rather than from book-study.

    Do you think “Indian heart” is something genetic or is it just their terrible fitness?

    It’s likely genetic to some extent, because Indians with no fitness/obesity issues and non-Indian diets still manifest higher rates of heart disease compared to other groups. But, of course, lack of exercise, obesity, smoking, and poor diet greatly exacerbate the inborn tendency.


    Are South Asians at higher risk for heart disease?

    People from South Asia—India, Pakistan, Bangladesh, Nepal, Bhutan, Maldives and Sri Lanka—have a four times greater risk of heart disease than the general population and have a much greater chance of having a heart attack before age 50.

    Heart attacks strike South Asian men and women at younger ages and the attacks are more deadly compared to any other ethnic group. Almost one in three in this group will die from heart disease before age 65.

    In India, cardiovascular disease remains the No. 1 cause of death. One study found that South Asians developed heart disease 10 years earlier than other groups.

    What is causing this heart disease phenomenon in South Asians?

    Why these heart attacks occur is only partially answered with traditional risk factor assessment. South Asians tend to be smokers, and the typical South Asian diet tends to be high in sugar, refined grains, and fatty foods.

    An alarmingly high number of South Asians appear to be insulin resistant, a pre-diabetic condition in which the body does not process insulin efficiently. Insulin-resistant patients have similar rates of cardiovascular events as those with full-blown diabetes.

    Body mass index (BMI) in South Asians often falls into a thin-fat syndrome: People may have an acceptable BMI, but they also carry more of their weight in their abdomen and that visceral fat is more likely to lead to a cardiovascular event.

    More than one-third of South Asian men and 17% of South Asian women have metabolic syndrome. Metabolic syndrome is a cluster of conditions including:

    High blood pressure
    High blood sugar levels
    Excess body fat around the waist
    Abnormal cholesterol levels that increase the risk of heart disease, stroke and diabetes
    If more than one of these conditions occur in combination, the risk is even greater. South Asians are more likely to have high triglycerides and low HDL (the good cholesterol).

    A variant of HDL known as HDL2b, which is thought to mediate the good effects of HDL, is low in as many as 93% of South Asian men and 63% of women.

    What compounds these risks in the South Asian population is a lack of specific testing: The criteria for metabolic syndrome and the subfractionation of HDL and other lipid- and inflammatory-based cardiovascular risk biomarkers are typically not checked during routine physical exams and they are often overlooked in a standard cardiovascular workup.

    The cardiovascular risk in South Asians appears to begin early: Research has shown that even in infancy, children of South Asian heritage may have high levels of cholesterol and lipoproteins in their blood. [Boldfaces mine.]

    • Replies: @res
  110. res says:


    Don’t forget medicine is still largely taught by practice rather than from book-study.

    Understood. I am curious because my guess is if this information starts being considered unacceptable (we seem to be in the early days of this happening) we would see the following patterns.

    More public venues stop teaching it first so it would disappear in order from. (This is why I focused on lectures before.)
    1. Lectures
    2. Formal internship, etc. instruction
    3. Informal instruction

    The knowledge starts being concentrated among the older physicians (or those who see specifics in their practice every day and can’t miss them) because the younger ones either aren’t being taught or don’t want to hear the message.

    To some extent this tendency should be countered by the push concerning poor health outcomes for the diverse. Seems a bit like an irresistible force meeting an immovable object. “Diversity!” powering both sides.

    Thanks for your detailed response about “Indian heart.”

    • Replies: @Twinkie
  111. pyrrhus says:
    @James Thompson

    I would bet that the AI could distinguish between the sub-races just as well, and maybe tell whether the patient is a mixture of races…if allowed to do so..

  112. Twinkie says:

    I am curious because my guess is if this information starts being considered unacceptable (we seem to be in the early days of this happening)

    Notwithstanding some stupid noise from activist types, I don’t see that happening anytime soon. People – at least in medicine – seem to be pretty good at compartmentalizing clinically relevant information and social posturing. But, of course, no one knows the future.

  113. dearieme says:

    Some readers here have been interested in explaining some IQ results, including American results involving racial differences, as being caused by exposure to lead in childhood.

    They might find this blog post interesting.

  114. @Auld Alliance

    I’m in Oklahoma, and I have my doubts, too.

  115. Biff K says:

    Yeah, I know.
    It’s a metaphor.

  116. @Alrenous

    This has been going on for years. Four years ago, Google lobotomized their AI image recognition software because it noticed that sub-Saharan Africans look very much like apes.

    Ten years ago Microsoft had an “avoid high crime areas” option in one of their mobile phone mapping apps. People noticed that the data was “racist.” MS got rid of it very, very, very quickly.

    Another company created a similar “avoid areas with high rates of violent crime” app in 2014 and instantly got driven off the Internet and hounded out of existence. Isn’t there a word for people who see the term “violent crime” and instantly think of Africans-in-America? I know there’s a word for that. It’s on the tip of my tongue. Can someone help me out? What’s that word?

    When the malevolent all-powerful AI exterminates humanity and destroys the world, this will be why.

  117. @TKK

    I don’t think anyone anywhere is surprised to learn that Rastus’s headbone beez extra thick, gnomesayin?

  118. @macilrae

    23andme routinely puts false information in their reports. They’ve been caught.

    Employees of 23andme boasted about “messing with people” by sending them false information in anonymous interviews.

    Of course, if the test results are beneficial to you in some way, like getting you a scholarship or a promotion, feel free to disregard all this.

    • Replies: @macilrae
  119. dearieme says:

    Tell me, doc, is this sort of stuff woo or is there some meat in it?

    From The Times:

    Liverpool put their players through penalty training in the build-up to the Carabao Cup final with the German neuroscience team who have helped to improve their set-piece record this season …

    As well as being at the training ground last week, the founders of neuro11, Dr Niklas Häusler and Patrick Häntschke, a former academy player at the German side FC Energie Cottbus, were present at the national stadium to join in the celebrations. …

    Part of the work done by neuro11 has involved using headsets with electrodes which measure the live electrical activity of the brain. …

    Through this, they can identify correctly whether a player is in the zone or not, which in effect means that a sportsperson running on automatic rather than having to think about what they are attempting to do.

    • Replies: @dearieme
  120. macilrae says:

    Gordo – why would you say that?

  121. macilrae says:
    @Pepe the Frog

    Pepe – no, while I confess the Egyptian 0.5% is a total mystery, the 0.5% West African and the 5% Jewish had already been validated – my GG Grandfather (Jewish) owned slaves in West Indies and actually married a quatroon! The rest of me is as white as you – or maybe even whiter?

  122. dearieme says:

    The Telegraph has more:

    The pair’s journey to Anfield began in 2019. Häusler, who has a PhD in neuroscience, and Häntschke, a former player with a masters in corporate management, took a punt in writing directly to Klopp to explain how they could help his players improve their set-piece and penalty performance. …

    The pandemic stalled any immediate partnership, but … by last summer Häusler and Häntschke were attending Liverpool’s pre-season training camp in France.

    There, Klopp made his prime set-piece exponent, Trent Alexander-Arnold, and penalty takers James Milner and Mohamed Salah available for Häusler and Häntschke to “experiment” and demonstrate their techniques.

    Since then, Neuro11 has been officially on board with Liverpool, in effect, appointing their own “brain trainers”.

    The results speak for themselves, with Klopp’s side scoring more goals direct from a set-piece than any other side in the Premier League.

    Last Sunday’s emotional stress of a Wembley penalty shoot-out was the ultimate test. For all of Liverpool’s 10 outfield players and goalkeeper to score was a spectacular vindication of the impact of the club’s work with Neuro11, prompting Klopp’s most public namecheck.

    • Replies: @James Thompson
  123. @dearieme

    Would need to know the longer term success rate for penalty taking before knowing if this recent streak of success is unusual. If so, there is still the possibility that is is a placebo effect: I won’t fail this penalty because I have Brain Science to guide me.

    • Replies: @dearieme
  124. dearieme says:
    @James Thompson

    Thanks. I vos chust amuzed at Cherman scientists sticking electrodes on heads.

    It’s a generational thing perhaps.

  125. dearieme says:

    On another blog a commenter who often says interesting things about medical matters has written this about IVF and polygenic scores:

    After a certain point, higher IQ is more highly correlated with mental illness. Going “full speed ahead” has very high odds of selecting not just for better aptitude on tests, but also for bipolar disorder. Depending on the familial returns on IQ it may well end up being a net negative.

    And unfortunately, bipolar and related diagnoses are overwhelmingly not diagnoses for around two decades. So it is quite likely that the first movers will have high risk, high reward payouts and it may well be a net drag.

    In any event, the easiest way to get high polygenic scores is just to start with gamete donors who are highly successful. And people have been selecting for that for millennia.

    It occurred to me that you might be the very chap to summarise the evidence against (or indeed for) his proposition.

  126. dearieme says:

    Sorry, my italics didn’t work properly. The third and fourth paras should also be in italics i.e. they are part of that doctor’s blog comment. The fifth para is mine.

  127. @dearieme

    can you send me a link to anything he has written on this? I don’t know which study is being mentioned

  128. dearieme says:

    He’s a commenter on the Marginal Revolution blog called “Sure”. He tells us he’s a doctor and often says interesting things compatible with that claim (insofar as this non-doctor can tell).

    If you scroll down the rather haphazardly organised comment thread here you’ll find the discussion.

    Thanks for your response.

    • Replies: @res
  129. res says:

    Here is a direct link to the original quoted comment (you can get these from the “#” on the far right of the line with commenter name, date, etc.).

    And here is a meatier comment with some linked studies.

    The first Pubmed hit is:
    “there was a ‘reversed-J’ shaped association: men with the lowest intelligence had the greatest risk of being admitted with pure bipolar disorder, but risk was also elevated among men with the highest intelligence”

    A cohort study finds the link between childhood IQ and bipolar:

    And I have not read the paper, but it looks fun, as this one claims to have found direct loci overlap between the genes for polygenic intelligence and major depression:

    End of the day, the human body has a sweet spot and if intelligence did not have some fitness cost, why was it not selected for (at least among nobility) for generations?

    One thing that makes me cautious about his expertise/insight/objectiveness (pick at least one) is in the first comment he says this. Which completely misses the point of IVF embryo selection from embryos of the same parents.

    In any event, the easiest way to get high polygenic scores is just to start with gamete donors who are highly successful. And people have been selecting for that for millennia.

    One hope for PRS is they would help with selecting for intelligence while avoiding overly high risk of negative outcomes (e.g. bipolar, schizophrenia, autism).

    A caution I would offer is PRS are currently additive and SNP only. This ignores two things which I think are important.

    Heterozygous advantage (Sure astutely mentions this in this comment). Non-additive effects may have little effect on population variance (consider how rare homozygous uncommon alleles are, allele prevalence ^ 2), but may have important effects on those rare individuals (or perhaps not so rare when you consider how many SNPs are in play in these PRS).

    CNVs (Copy Number Variants). I will feel better once CNVs (and possibly other structural variants) are incorporated in PRS. See the third paper mentioned in this Steve Hsu blog post. Though note it does find PRS are more important than CNVs for their examples.

    I suspect the early days of IVF embryo selection will show much small-moderate success and a few notable failures (of uncertain magnitude, which is important). Not sure how that will go over if correct. Perhaps I am too pessimistic, but I think it is important to be ready for something like that to avoid disillusionment and throwing the baby out with the bathwater as a result.

    • Thanks: dearieme
    • Replies: @res
  130. res says:

    One thing I should clarify. Heterozygote advantage is probably relatively uncommon. The more likely case is nonlinearity where a single allele is a little bad, but two copies is much (>>2x) worse.

    For a hypothetical case consider a case where we have a SNP with PRS scores and allele frequencies like (this is 1% for a allele frequency in Hardy Weinberg equilibrium) the following. (this is off the cuff, so recommend double checking for yourself)

    Genotype | Proportion of population | PRS
    AA | 0.9801 | 0
    Aa | 0.0198 | -1
    aa | 0.0001 | -5

    I created a CSV for this and some R code to do a weighted linear regression (after the MORE) and it gave an allele PRS for a of -1.03 with an R^2 of 0.96 (adjusted 0.92). So you can see how little effect the homozygous case has on the variance explained and PRS coefficient and how much the homozygous case severity is underestimated (-2.06 vs. -5) as a result.




    R code

    SNP <- read.csv("SNP_PRS.csv")

    modelw= lm(PRS ~ NumGen, weights = Proportion, data=SNP)

    R results

    lm(formula = PRS ~ NumGen + NumGen^2, data = SNP, weights = Proportion)

    Weighted Residuals:
    1 2 3
    -0.000297 0.004179 -0.029403

    Estimate Std. Error t value Pr(>|t|)
    (Intercept) 0.0003 0.0300 0.01 0.994
    NumGen -1.0300 0.2111 -4.88 0.129

    Residual standard error: 0.0297 on 1 degrees of freedom
    Multiple R-squared: 0.9597, Adjusted R-squared: 0.9194
    F-statistic: 23.81 on 1 and 1 DF, p-value: 0.1287

  131. Factorize says:

    I find it increasingly terrifying how scientifically uninformed the mainstream conversation on Genetic Uplift is. Current best evidence is that a ~3 point IQ enhancement is possible with very weak 1 in 10 embryo selection with only normal selective pressure of mates.

    Why aren’t people concerned about this? Why aren’t people very concerned about this? 3 point IQ enhancement is approaching the cognitive ability difference between Asians and Europeans (~ 30,000 years of human evolution). This enhancement is now economically and technologically feasible and clearly is only the start of our journey to a new type of human experience. Technology has a way of amplifying through time- often exponentially. It is not difficult to imagine optimized reproductive technology over the medium term that used population scale gamete selection etc. to achieve considerably more than 3 IQ points. Humanity is now very near to the Genetic Singularity.

    The suggested connection between IQ and mental illness seems to me to be nothing more than a mirage. However, this is a question that could be answered by merely consulting the results from the large GWAS. Are there any SNPs that both increase IQ and increase the risk of mental illness? My first guess would be that there would probably be only minimal overlap.

    GWAS have given us the power to genetically dissect traits into the microscopically small effect sizes that create the overall polygenic score. It is no longer necessary to answer genetic questions by referring to a highly limited set of personal examples from one’s life experience. One can instead consider the results from GWAS that can involve more than one million people. The resulting variants from these massive genetic studies are the traits in a highly purified distilled form. Such variants usually have minimal effect sizes. One does not typically develop a trait (most traits (e.g., schizophrenia)) because one inherited one SNP; often one needs to inherit thousands of risk variants. This suggests that genetic engineering could create people with very high IQ without other detrimental traits.

    • Replies: @res
  132. res says:

    The suggested connection between IQ and mental illness seems to me to be nothing more than a mirage. However, this is a question that could be answered by merely consulting the results from the large GWAS. Are there any SNPs that both increase IQ and increase the risk of mental illness? My first guess would be that there would probably be only minimal overlap.

    This reference is paywalled, but DOI 10.1038/s41380-018-0332-x is on LIbgen/SciHub.
    Genome-wide analysis reveals extensive genetic overlap between schizophrenia, bipolar disorder, and intelligence

    I bolded one part of the abstract I found especially interesting.


    Schizophrenia (SCZ) and bipolar disorder (BD) are severe mental disorders associated with cognitive impairment, which is considered a major determinant of functional outcome. Despite this, the etiology of the cognitive impairment is poorly understood, and no satisfactory cognitive treatments exist. Increasing evidence indicates that genetic risk for SCZ may contribute to cognitive impairment, whereas the genetic relationship between BD and cognitive function remains unclear. Here, we combined large genome-wide association study data on SCZ (n = 82,315), BD (n = 51,710), and general intelligence (n = 269,867) to investigate overlap in common genetic variants using conditional false discovery rate (condFDR) analysis. We observed substantial genetic enrichment in both SCZ and BD conditional on associations with intelligence indicating polygenic overlap. Using condFDR analysis, we leveraged this enrichment to increase statistical power and identified 75 distinct genomic loci associated with both SCZ and intelligence, and 12 loci associated with both BD and intelligence at conjunctional FDR < 0.01. Among these loci, 20 are novel for SCZ, and four are novel for BD. Most SCZ risk alleles (61 of 75, 81%) were associated with poorer cognitive performance, whereas most BD risk alleles (9 of 12, 75%) were associated with better cognitive performance. A gene set analysis of the loci shared between SCZ and intelligence implicated biological processes related to neurodevelopment, synaptic integrity, and neurotransmission; the same analysis for BD was underpowered. Altogether, the study demonstrates that both SCZ and BD share genetic influences with intelligence, albeit in a different manner, providing new insights into their genetic architectures.

    Supplementary Material is freely available at

    P.S. One speculative thought. I wonder if (some of?) the BD/IQ matches might indicate heterozygote advantage. Idea being more common heterozygote bumps IQ up in the full population, but lower prevalence homozygotes increase relatively rare BD. Probably not, but seems worth checking…

  133. @Anon

    Begosh and begorah! We even like the same SONGS!!

  134. Factorize says:

    res, thank you very much for the correction!

    I wasn’t sure on this question as I had read of the correlations between mental illness and intelligence, though it was unclear to me whether this related to assortative mating. My assumption (from personal observations of my own family, my own genetics etc.) was that there would be substantial separation of the mating pools of normals and non-normals. Normals tend to largely be illiterate, reproduce relatively early and often and typically have a highly functioning weirdness radar, while the non-normals are the assorted others who find book reading stimulating, perhaps have higher rates of mental illness etc.. This vision of the mating pools would then result in more mental illness associated with intelligence not because the same variants were involved with mental illness and intelligence, but because the non-normal mating pool mixed different variants for mental illness and intelligence into the offspring’s genomes.

    To illustrate this, one family member appears to have a highly enhanced memory phenotype and can repeat verbatim conversations dating back to kindergarten. I, on the other hand, typically will not be able to recall what I had for breakfast. This might be the Clotho memory variant effect as this is present in our family genome, though perhaps other variants are involved. One could imagine that having such a cognitive trait (and many others) could result in being assigned to the non-normal pool. Basically, all the weird people are directed to a separate reality.

    Surprisingly, even from an early age my peers were able to recognize that I belonged in the non-normal pool. They had no personal reference points for why anyone would find school to be of interest- the modern version of this has become those who attend university exclusively for the enhancement of their vocational prospects, yet detest everything that they must endure.

    Thank you for providing the reference that establishes that there is a direct genetic linkage.

    res, I am very interested in your current thinking about the potential for genetically enhancing cognitive ability with various technologies including embryo selection, population scale genome mating databanks, possibly gamete selection, etc.. I am getting worried… very worried. No one wants to talk about this honestly. In the blog idiom, the dog did not bark. This is only making it worse; when the smartest people do not want to talk about the challenges confronting us it gives the impression that the situation is beyond salvaging. Increasing population IQ by 20 points would realistically imply the end of our species as we currently understand it. Yet, 20 points no longer seems that unattainable even over the medium term. I am becoming progressively more concerned about how humanity will be able to cope with the genetic technology that has been developed and that will soon be developed. I will anxiously await your comments.

    • Replies: @res
  135. res says:

    Thanks. Your non/normal mating pool discussion is interesting. I need to think more about that.

    Regarding Klotho, thanks for bringing it up. I had heard of it, but did not realize/remember it appears to be an example of heterozygote advantage.

    Regarding cognitive enhancement, first it is probably worth mentioning two areas where it happens now (one ancient I suspect).

    1. Assortative mating by intelligence.
    2. Selecting sperm/egg donors by phenotypic traits. Even if not measured I suspect this includes intelligence either by personal acquaintance judgment (personally known donors) or proxies like college attended.

    Of your examples, I think embryo selection is the most likely. Population scale genome mating databanks seems like a political non-starter in the West, and even if I could imagine something like that in China or Russia (hasn’t there been something similar for athletes by phenotype somewhere?) I would expect it to be very limited. Gamete selection seems like too difficult a proposition (non-destructively genotyping gametes) for the limited benefit it would bring over embryo selection.

    So embryo selection via IVF. The companies doing it have decided intelligence is toxic (probably correctly) so avoid talking about it. My thoughts are.

    1. If the system integrity hypothesis is correct then the overall health PRS are likely to promote intelligence. I don’t know what kind of effect size would result (though if I were at one of those companies I would at least attempt to check, but quietly).
    Looking for ‘System Integrity’ in Cognitive Epidemiology

    ‘… the so-called ‘‘system integrity’’ hypothesis … posits that higher intelligence may be a marker for a general latent trait of a well-functioning body.

    2. I would not be surprised if some level of embryo intelligence selection was being done now. If an individual doing IVF got the genotypes of their embryos for a health prediction what is to stop them from running an IQ PGS on them?

    3. What about China? I wish someone with a better understanding of Chinese cultural values (both historical and current) would talk about what would be acceptable there. Because to this outsider that seems like the most likely place.

    4. Personally, I would like the PGS to be better (e.g. see nonlinearity and CNV comments above), but suspect what we have is good enough to be useful with the occasional “oops.” In some ways I am a cautious incrementalist (and IMHO this is an area where that would be a good thing) so would want experience with the single generation version before doing things like iterated selection (especially without improved PGS!).

    5. I think state actors doing this as a weapon has a non-trivial chance of happening. In which case I would expect a full tilt go at iterated selection which might get ugly in a variety of ways.

    P.S. From a “cautious incrementalist” point of view, 1. is a pretty good way to start. Estimate the intelligence effect size (even if you don’t select for it you can predict it for each embryo) then see how reality matches.

    P.P.S. I wonder what would happen with iterated selection using a health PRS.

  136. Factorize says:

    res, gamete selection might be much closer to launch than many might realize.

    Apparently, full in vitro development of gametes is now considered plausible over the next ~10 years by experts. There is a substantial motivation for this research effort as success would allow people with fertility issues to conceive a biological child instead of adopting etc..

    My previous post on September 22, 2021 at 1:28 am GMT • 6.3 months ago • 700 Words
    described a potential workaround that might require essentially no additional technological breakthroughs.

    Roughly, capture some primary spermatocytes, cage them in “netting” etc., return them to their natural environment, allow them to develop, then retrieve them when they had matured into mature sperm, genotype 3 of the sperm and then the fourth sperm’s full genome would be known (without destroying it). The logic here is: We know the full genetic sequence of the 4 sperm (twice the father’s genome), we can sequence the genome of 3 sperm– subtract and what is left is the genome of the fourth sperm. Destroying the fourth sperm is not necessary to fully infer its genome.

    If this back of the napkin idea could be implemented it would allow for essentially unlimited gamete selection (especially sperm, hundreds?, thousands? ). The same genomic inference strategy could also apply to egg cells, though probably on a much reduced scale.

    In this conception of genetic engineering, the full genome of both the sperm and egg cells would be knowable before conception. The magic (and insanity) of parenting of randomly rolling the genetic dice to determine the future of one’s children and all descendants could be abandoned and placed on a more rational basis. The possibility for global scale gamete banks, possibly stored frozen over decades would then offer profound genetic control of offspring. It would not then be entirely unexpected that a near term Genetic Singularity event might be feasible.

  137. Factorize says:

    Considering how profoundly humanity would be affected by the gamete selection strategy that I mentioned yesterday (call this approach superselection), I will take one more run through the logic so that it is perlucid.

    Click on the above url from my last post and open another instance of this unz page in another window so you can follow along with the figure at the same time as reading this post. The big idea that I noted in my above post was that with 3 sperm cells one could deduce the genome of the fourth without destructive genotyping. Quite a few highly informed people had assumed that it would not be possible to genotype the fourth sperm non-destructively. The figure from the above post clearly shows this is incorrect.

    First look at the primary spermatocyte near the top of the figure. Notice that the genome of the primary spermatocyte is diploid (2n) that is it has 2 big blue chromosomes (call them chromosome 1s), 2 big red chromosome 1’s (also chromosome 1’s), two small blue chromosomes (call them chromosome 2’s), and two small red chromosome 2 (also chromosome 2’s). The primary spermatocyte then has 2 copies of the father’s paternal chromosome 1’s and 2’s, and two copies of the father’s maternal chromosomes 1’s and 2’s: two big blue, two small blue, two big red and two small red chromosomes.

    Now consider the secondary spermatocytes in the row one down. Genetic recombination has occurred so the maternal and paternal chromosomes have been intermixed, though the total genetic material is still the same: there are two copies of the father’s paternal chromosomes 1’s and 2’s and 2 copies of the maternal chromosomes 1’s and 2’s. Notice that all that you need to do is transfer the little blue piece of DNA on the left strand of the big red chromosome (red x on the left) of the right hand secondary spermatocyte and place it on the right strand of the the left big blue x on the left hand secondary spermatocyte and then there would be two big blue chromosome 1’s. The same could be done for all of the other chromosomes in the secondary spermatocyte. There are still two copies of all of the chromosomes.

    The logic becomes clear of how we could deduce the fourth sperm. We know that there should be two copies of each of the chromosomes. If I were to give you the genomes of 3 of the sperm, then the fourth sperm’s genome could be deduced by subtraction. For example if we were to capture the three sperm on the right, then we could determine the genome of the sperm on the left.

    Specifically, we can see that there is one complete red chromosome 1 (big one) on the far right. There is also another complete red chromosome 1 (big one) when we add together the red parts of the chromosome 1s from the sperm 2 from the right and the sperm 2 from the left (the small red piece on the blue big chromosome 1 from the left). What does this tell us? What does it mean that we have now accounted for 2 full red chromosome 1’s in the three sperm from the right?
    This implies that the fourth unobserved big chromosome must be all blue. There is no red chromosome 1 DNA left for sperm 4! It’s all been taken! We have been able to determine the chromosome 1 genome of the left hand sperm without actually genotyping it! This is a very powerful and potentially highly significant finding.

    What about the chromosome 2 genotype for the left hand sperm? The chromosome 2 of the sperm second from the left is all blue, and the chromosome 2 of the sperm second from the right is all red. This means that we need one more full small blue and one more small red chromosome. The small chromosome on the right gives part of the remaining DNA. Therefore, we know that the chromosome 2 of the sperm on the far left most be the photographic negative of the chromosome 2 on the far right. The chromosome 2’s on the far left and far right must add up to a complete blue chromosome 2 and a complete red chromosome 2. Thus, where the right hand chromosome 2 is red the left hand sperm will be blue and where the right hand sperm is blue (to the bottom) the left hand sperm will be red. This is exactly what we see for the chromosome 2 of the far left sperm..

    The excruciating detail above was provided because the result is so potentially overwhelmingly powerful. A gamete’s full genome could be known before conception and further that an embryo’s genome (excluding de novo mutations) could be known before conception. The genetic randomness of reproduction has been completely removed. There is no regression to the mean. Global scale gamete banks becomes plausible with the cost largely moved to the level of the individual while creating overwhelming positive network effects. By joining such a global network, one would be able to access possibly millions of gametes of known genetic sequence. The potential for
    large scale enhancement of human traits for those of any demographic background would be substantial.

    Doing the empirical analysis of this through UKBB could help to quantify the potential. What would the simulated IQ of children born with 100 fold sperm selection for fathers and 20 fold egg selection for mothers be assuming only random mating? Or the same selection but this time choosing the highest IQ sperm and highest IQ egg in a half million person sample? What if instead the objective were not the highest possible IQ but to “fix” the most chromosomes with high IQ on both strands of the offspring? In this genetic configuration, all such offspring would have “fixed” chromosomes and thus all of their gametes would also be “fixed” as high IQ. Future pairings of offspring with such “fixed” chromosomes will have eternally high IQ on these chromosomes. Creating such genetic order as a longer term strategy might be seen as more advantageous than merely seeking the maximal short term IQ. The genetic uplift that could occur as genetic optimization converged at population scale might be large and rapid. Superselection could be further amplified by considering other highly relevant traits besides IQ that give life advantage. Choosing amongst 5 normally distributed positive and somewhat independent traits would greatly enhance the drive towards extreme phenotypes.

    • Replies: @res
  138. res says:

    The description on this page helped me think about this.

    Do I understand correctly that this paper is doing something similar (in humans)?

    Is it correct to say that your “four sperm” are the four (haploid) products of a single primary spermatocyte?

    Overall, seems like an interesting idea. Any idea if it is possible to do something similar with a primary oocyte?

  139. Factorize says:

    res, yes I was thinking along the laser ablation as well (this is the tact that the article took). Given the advanced state of research into gametes such a purification technology had to have been developed. Yet, I am unclear whether this would even be needed; perhaps the below separation strategy would be all that is needed.

    Yes, four sperm (quartet) would be the offspring of a single primary spermatocyte. The figure in my post above illustrates the sperm formation process admirably. It appears necessary to go quite far upstream to be able to be sure of the genotype of the fourth sperm. The spermatogonium is the germ stem cell that can keep creating new recombinations, while the primary spermatocyte is the first cell in which the path to differentiation has begun.

    When I look more closely at the wiki link that you provided, isolating the sperm quartet could be much easier than I had thought. Spermatogenesis was originally described as a tree like process. This can be seen in the figure below; such a tree like structure offers the potential of simply isolating the branches.

    Spermatogenesis had seemed to be such a mess of spermatogonium, primary spermatocytes, secondary spermatocytes, spermatids and free floating sperm all in a highly connected mass. I was not clear how it could all be organized so that the sperm quartets could be sampled precisely. That is where the “netting” concept had popped up. I had thought that it would be a difficult task to create a clean surgical field. My worry was that the netting to keep the sperm contained might also prevent necessary chemicals from entering the developing spermatocytes.

    Yet the figure above suggest otherwise. It appears that there are these hair like projections of spermatocytes connected to spermatids etc. that could just be separated with a gentle nudge. As can be seen in the figure there are open areas between these projections. This implies that an interconnected mass is not required for proper sperm formation. I had thought that proper sperm formation might require complex interactions that could only occur as a solid mass of the projections. The potential here is that no chemical or other intrusive intervention might be necessary at all to isolate sperm. The sperm could be gently isolated and allowed to develop naturally until capture. The research effort now under way is trying to find all the growth factors and all of the genetic changes in order that in vitro gametogenesis can be perfected. The let them develop naturally and then capture them approach entirely avoids any complex scientific questions which can often add hidden risks into the final outcome.

    With my idea suggested above, gamete selection dramatically simplifies. Perhaps merely isolate a single Sertoli cell and capture the sperm as they develop on an individual branch. Collect the quartets of sperm as they develop. Possibly collect 25 such quartets; select the phenotypically healthiest sperm and then genotype the other sperm. One should then have 24 complete quartets along with 3 of four of the 25th quartet. Oversampling in this way would act as a check to make sure that it was highly likely that no stray sperm from another quartet had been included. The sampling could also continue after the 100 sperm had been collected in order to also ensure that there were no stray sperm formed later (perhaps also collect before the official start of sampling also to ensure that complete quartets had been collected).

    This is quite exciting!

    I think that the UKBB should make a special effort to do phased genotyping of their half million sample to determine the maximal IQ of gametes for individuals and the maximal IQ of a gamete for the entire sample. I programmed a basic implementation of the idea with an existing publicly available genetic database using an earlier PRS for EA and the numbers I saw when choosing the maximal genome EA for typical people with chromosomal level selection were large. I would like to see this done with the latest software and on the scale of a large biobank. It seems critically important that scientific research exerts some effort to give fair warning to what is approaching. Without such an effort we could all wake up one morning to a world completely transformed and possibly entirely beyond our ability to manage theoverwhelming cognitive ability.

    If this idea could go from the armchair to the bedside, then the uplift potential could be enormous. As it is now, “genetic engineering” attempts are hopelessly primitive. Until recently, it was merely based on phenotypic selection- choose the smartest mate and see what happens. This is a hopelessly ineffective strategy that has yielded minimal results after centuries of effort.

    Without knowing genotypes one has no idea how phenotype would relate to potential offspring genotype; genetic recombination then adds yet more confusion and noise into the process. Gamete level selection would allow knowing the full genome of an embryo before conception. The potential involved is almost unlimited.

    Why would anyone be interested in lab reproduction? Well, the economic incentive at the individual level for gamete selection for someone with an extreme gamete cognitive phenotype would be massive. Might \$5 billion be of interest to anyone? The net present value of 200 IQ is probably on that scale. People with extreme IQ have the cognitive ability to help us solve our problems and cure our diseases. How would it be ethical not to move genetic enhancement forward rapidly when it could reduce human suffering and make life better on a planetary scale? Quantification of this uplift effect would be welcome, though as a guess I would think that it would be substantial. There would be very large positive network effects as well. A one million person gamete bank might cost ~\$10 billion, though any one person might only be in for \$10,000. Buy your genetic lottery ticket and if you hit the jackpot you’re in for a \$5 billion payday. Not bad. Probably better payback than actual lotteries.

  140. Factorize says:

    I have continued with background reading on the superselection idea. With the sperm, it is remarkable how seemingly accessible the seminiferous tubules are for a potential surgical intervention. There are only 3 structures from the outside to reach the tubules: the outer skin layer, the tunica vaginalis and the tunica albuginea which is non-vacularized. The seminiferous tubules would then be directly ahead. These tubules are one of the least protected structures in the body; they exist almost at the skin surface. Clearly micron level surgery is beyond our current technology, though possibly not that far off.

    With oocytes… it might be quite a bit closer to application. Apparently, deducing the genome of the fourth “sperm” concept has been appreciated by the reproductive technology community for over ten years. It also seems that studies have already been done in which oocytes were harvested along with first polar bodies which would allow for insight into the genome of the secondary oocyte before fertilization. One could deduce the total genomic composition, though it would not be known which recombinant chromosome would be inherited. If this assessment were in fact true, then the technology needed for significant genetic uplift is more advanced than many realize. The ability to select at the gamete stage (at least for oocytes) would clearly amplify enhancement.

  141. If this had been the case, it would have shown up very strongly in the studies of Lubinski and Benbow, but it didn’t.

    • Replies: @res
  142. res says:
    @James Thompson

    Is that a response to dearieme’s comment 133?

    That is a good point. Though that would not be inconsistent with heterozygote advantage (which might show issues with siblings/parents/etc. but not with participants).

    Gwern gives some related links in this comment.

    The more relevant is this from the 2017 ISIR (so you may very well know more about this).
    Gifted Kids and High-Achievers Stay Fresh: Health Outcomes of Four SMPY Cohorts at Age 50

    I have been having trouble finding SMPY references which discuss mental illness (in particular, bipolar disorder, see genetic association with IQ comment above) at all. Do you know of any? The Terman study looked at “mental adjustment” as discussed in this book chapter (DOI 10.1002/9781118367377.ch23 on Libgen and SciHub).
    Lifetime Biopsychosocial Trajectories of the Terman Gifted Children: Health, Well-Being, and Longevity

    P.S. This paper discussing schizophrenia and visuospatial imagery was in interesting hit while I was searching for references.

    Also this 2000 discussion of individual differences by David Lubinski.

  143. Factorize says:

    It is worrisome that quartet sperm collection does not appear overwhelming difficult (at least from a high level conceptual view). The outside of the testis is covered in the parietal tunica vaginalis (first figure). After cutting through this layer one reaches, the visceral tunica vaginalis (second figure). This was described as highly vascularized, though in the second figure it does not appear to be overly vascularized in relation to the scale that would be needed to access the seminiferous tubules.

    The next layer is the tunica albuginea (Third Figure). This layer is not vascularized. After cutting through the tunica albuginea, the seminiferous tubules would be reached.

    The fourth figure shows the inside of one of these tubules. 200 micron level surgery would be challenging, though perhaps not impossible. A camera with the dimensions of 575 microns by 200 microns by 200 microns has been recently developed. One might set up a window into a tubule and
    identify the Sertoli cells. These cells act as containers for the sperm from a spermatogium. The quartet could be captured, enclosed in “netting”, released and allowed to develop to full mature naturally in the epididymus and then captured again. Notably such a scheme would require zero understanding of the complex biology involved in gamete development. The current research approach is trying to comprehensively understand the entire development of gametes for growth in vitro.

    At a top level conceptual view the above seems at least plausible, while technology would clearly be required to implement it. Some problems that could hinder development are the pulsating motions that occur to move the sperm through the tubules and the fact that the process of maturation in the tubules requires ~70 days. One might have to set up a base camp, take a peek in the tubules to see where the sperm are in the process, call it quits for a month or so and then try again.

    With oocytes, I wonder whether they could be tricked into believing that they had been fertilized in order that they would release a second polar. This would then allow for complete genetic characterization of the ovum before the real fertilization had even occurred. The problem that arose with oocytes is that they only release the first polar body during ovulation, though wait to release the second polar body until fertilization by the sperm occurs. If false fertilization could occur, then both the gametes could be genetically characterized before conception. This would have profound implications for human society.

  144. res says:

    Dr. Thompson, any chance you will do a post about EA4?

    • Replies: @James Thompson
  145. @res

    Reading it, also listening to James Lee being interviewed, also reading Greg Clark, but not writing anything at the moment!

    • Replies: @dearieme
  146. dearieme says:
    @James Thompson

    I hope you’re well, Dr T.

    I couldn’t write if I wanted to: got Covid. Urgh!

  147. says:

    Embryo Selection Or Genome Editing

    Quote from someone else: After a certain point, higher IQ is more highly correlated with mental illness. Going âfull speed aheadâ has very high odds of selecting not just for better aptitude on tests, but also for bipolar disorder. Depending on the familial returns on IQ it may well end up being a net egative.

    That is a comment from an IGNORANT SJWonker or a blantant liar. That certain point is far into the future and not meaningful in present situation. In current situations it is already uncertain how someone with IQ 250 can function in a mostly sub IQ 100 society let alone pushing past IQ 250. Those reckless Chicken Littles are assuming that other people are as reckless as them.


    From the Education Attainment EA3 study, the sum of the significant POSITIVE EFFECT SIZES (BETAs) is MaxPgsEA=51.94, MaxEAYear=218.16. If EA3 is a good proxy for IQ, that is equivalent to POTENTIALLY ABSOLUTE MaxIQ=779.14. Considering that the practical range for EA is about 24 EA years, it is a very flexible situation and there are plenty of opportunity to maximize EA years / IQ without taking excessive risks. Later it will be shown that it could be possible to push Edu Attainment EA past PhD level through many generations of embryo selection or many genomic editings without entering statistically expected mental disorder domains (i.e. mental disorder PGS score > 0). Thus that assertion is a very stupid regurgitation from the diarrheas from the SJWonker echo chambers. Come to think of it, it could be that those ancient Chicken Littles considered that
    FLIGHT was too risky and the flightless chicken have evolved to be large scale farmed by human and slaughter for meat.
    Given any two phenotypes, e.g. EA3 and BiPolar Disorder BiP (pgc-bip2021), anyone with more than ‘two bits brain’ will know that there are 4 possible clusters of GWAS effective individual SNPs / gene variants with different effect sizes to choose from, as shown in the chart at right objectively scaled down by pvalues (odds of false positive) for clarity,

    The distribution of the systemic effects of pleiotropic SNP subset clusters will depend on the subject. (Excluding those opportunistic singleton SNPs not affecting the constraining phenotypes considered. Use of opportunistic singleton SNPs are more risky for introducing accidental currently unknown bad effects as systemic pleitropic SNPs that affect more than 2 phenotypes have gone through more scrutinies.) Those knowledgeable will mostly select the SNPs / gene variants from the second quadrant with +ve EA3 and -ve BiP effect sizes. ONLY WHEN THE SUBJECT IS OF SUCH POOR QUALITY THAT THERE ARE SCARCE SNPs IN THE SECOND QUADRANT THAT SNPs IN THE FIRST QUADRANT WITH +ve EA3 and +ve BiP HAVE TO BE SELECTED TO INCREASE THE EA3 SCORE. That is DIRECTLY OPPOSITE TO THE ASSERTION OF THE LYING SJWONKERS WHO TRIED TO ASSERT THE BOTTOM SCRAPPING CASES TO BE THE MANUALLY OPTIMIZED HIGH END OUTCOMES.

    For example in the simplest case if two embryos have almost identical gene variants except one, say identified by the marker rs184139610 in the second quadrant, and the former embryo T with effective allele T while the other embryo C has allele C. The embryo T will have the ADDITIONAL PGS scores contribution from the allele T of -0.252 BiP and +0.02861 EA3 while the other will have the respective contribution from allele C of both zeros, i.e. embryo T with allele T will be the better embryo with potentially higher EA3 and lower odds of BiP. Aggregating with hundreds of such similar conditions for other gene variants will produce potentially better outcomes across all the phenotypes considered. However in practice the preferred SNP profiles are dependent on the parents’ risk preferences, only in desperate situations that the parents have to trade off EA3 scores with undesirable SNPs. In “average” situations there are plenty of room for improvement. In my simulation studies the NET bad phenotype effects are kept below average. That is the essence of SELECTION / EDITING that the simpleton Chicken Littles failed to understand, or the Chicken Littles’ agenda is to lock in the current global national IQ ranking. Even the Russian agents failed to push such effective strategy against the AngloAmerican spheres. The ignorant SJWonkers already pushing such “mushroom culture” strategies to them.

    The magnitude of the SJWonker LIES is apparent from the distribution plot of the EA3 SNP (gene variant) effect sizes from GWAS study below. The y axis is the contributing effect size of the EA3 SNP if present (all the green+blue+red bars). The SNPs common with BiPolar Disorder are the scarce blue and red bars, i.e. MOST OF THE GENE VARIANTS FOR EA3 CAN BE VARIED WITHOUT AFFECTING THE BiP SCORE, WITHOUT GETTING INTO THE DOMAINS OF INCREASING EA3 WITH INCREASING BiP. The scarce common pleitropic SNPs (blue and red bars) have two type of effects, i.e. effects in the same sense as in the red bars (i.e. positively correlated) which can be ignored as there are already many other favoured SNPs or oppositely correlated as in the blue bars (i.e. +EA3 with -BiP to be slected or -EA3 with +BiP to be deselected) which could give more advantages with less bad efforts. With current limited technical ability it is more cost effective to concentrate on those SNPs in blue. Those plenty singleton SNPs in green are less studied and at this stage is best to avoid to prevent getting unintended collateral damages in other less studied genetic disorders.
      It is unclear why the commenter is so fixate on the effects of BiP where the interactions with EA3 are so sparse that it is relatively easier to avoid the bad effects of BiP. For example the effects of Autism ASD are several times higher than that for BiP. Even that it is generally clear that those higher magnitude EA3 effect sizes are pretty independent of those from the bad genetic disorders (green bars), or they correlate in the favourate sense (blue bars). Even those with the unfavoured sense (e.g. +ve EA3 +BiP) trade-offs can be made with those where the relative magnitude of EA3 effect sizes are higher than that of the other such that EA3 score can be improved while the NET EFFECT of the bad effect sizes can be overcome by the extra slacks provided by those from the green and blue bars. In summary it is usually able to maximize the EA3 score within the constraints unless the sample is of such POOR QUALITY that the BAD red bar SNPs have to be selected, the VERY OPPOSITE TO THE ASSERTION OF THE COMMENTER.

    There is a strange negative correlation between SCHZ and ASD which might interfere with embryo selection or genomic editing. Although there is a very large cluster of +ve ASD effective SNPs to be avoided, there is a significant cluster within close proximity in Chr17 (range 1179777/83257441=1.42% of Chr17) involving more than 2500 gene variants with -ve ASD effect sizes with the corresponding +ve SCHZ effect sizes that they might make the selection/deselection very tricky.

    There are many studies on the correlation between ASD and SCHZ but they are mostly on the psychological traits and brain structures, and some on genetic Copy Number Variations CNVs which concentrated on QUALITATIVE EFFECTS, unlike GWAS which gives QUANTITATIVE RESULTS. The SNP markers are proxies for the gene variants as well as CNVs and some SNP markers can be proxies for several different gene variants. Thus to find the QUANTITATIVE OPTIMAL requires the use of GWAS PolyGenic Score PGS not directly involves CNVs.

    Finally, just to prove how stupid the commenter’s anti enhancement scare tactics assertion was, it will be shown later in more detail that it is possible to have net improvement in cognitive performance BY ACTIVELY REDUCING GENETIC MENTAL DISORDERS RELATED SNPs EVEN FOR THOSE ALREADY WITH HIGH IQ. TECHNICALLY THAT PROCESS IS NOT IQ ENHANCEMENT SINCE THE POTENTIAL COGNITIVE ABILITY ALREADY INNATELY PRESENT BUT IT IS THE PRESENT OF BAD SNPs ESPECIALLY THOSE THAT INCREASE MENTAL DISORDERS WHILE SIMULTANEOUSLY SUPPRESSING OF THE INNATE COGNITIVE ABILITY, IT IS TECHNICALLY HARM REDUCTION AND NOT EHAHNCEMENT. For example, a web influencer has openly published his REAL LIFE genomic SNP distributions and using the results of GWAS EA3 study his EA PGS score is expected to be 0.405 which is equivalent to 18.50 EA years, pass the nominal ability to have a post graduate degree. The PGS score is net results from two components, the cluster of ALREADY PRESENT SNPs which give potentially +ve delta EA3 score of additional +18.62 EA years from the population EA3 mean and the cluster of ALRESDY PRESENT BAD SNPs which are expected to give -16.92 delta EA years and SOME of these also concurrently and potentially produce BAD GENETIC MENTAL DISORDERS which gives the NET DELTA EA YEARS to +1.70 EA years in his REAL LIFE CASE above the population EA mean of 16.8 EA years, enough to enable the subject to have a post graduate degree. By DESELCTING/EDITING those BAD GENE VARIANTS which give mental disorders, i.e. techinically HARM REDUCTION NOT ENHANCEMENT, the POTENTIAL EA YEAR GAIN from the ALREADY PHYSICALLY PRESENT GOOD EA3 SNPs can be stretch to ADDITIONAL +16.8 EA years, way pass the current nominal max of about absolute 21 EA years. If it is possible to perform “FULL HARM REDUCTION WITHOUT ENHANCEMENT” the expected possible max EA years is 35.42 or if the EA3 results are applicable to IQ, the expected possible max IQ of 166.51 from the already existing GOOD IQ SNPs. Of cause those are potentially possible for his children if his wife is also of similar genetic makeup. ONLY A SADISTIC SIMPLETON WILL OPPOSE HARM REDUCTION.

    • Replies:
  148. says:

    test if the pic will show up

  149. says:

    The charts are swallowed by the system.

  150. says:

    The Case For Embryo Selection Or Gene Editing

    Despite what the ignorant sadistic control freak Chicken Littles who have scared people about small number of gene selections/manipulations, labelling those involved as monsters even at the very low number of 2 additional mutations, human gene mutations happen naturally and regularly in significant numbers, especially with respect to the age of the fathers. It is very naive if those Chicken Littles thought that no mutation will happen in “normal” pregnancies.

    The US CDC has kept a dataset on the number of newborn babies with readily observable inherited disorders due to gene mutations. For year 2020 mulfunction genetic disorders can be observed in early age and whose fathers were as young as 20 years old. Though the probability of having genetic disorders increases with the father age, there are less fathers over the age of 45 and in absulute numbers the babies with genetic disorders are mostly from fathers between the age of 20 to 45, i.e. from the average ‘young’ families, not from those overly elderly. Yet so many simpleton suckers taking the diarheas from the ignorant Chicken Littles seriously without consideration to those families whose suffering could be averted.

    Although there are many studies on the effects of paternal age on genetic diseases, most concentrated on sperm quality, quantify and mobility. There is a Icelandic study which concentrated on the casuality due to parents’ ages.

    “Rate of de novo mutations, father’s age, and disease risk”
    We conducted a study of genomewide mutation rate by sequencing the entire genomes of 78 Icelandic parent-offspring trios at high coverage. Here we show that in our samples, with an average father’s age of 29.7, the average de novo mutation rate is 1.20×10−8 per nucleotide per generation. Most strikingly, the diversity in mutation rate of single-nucleotide polymorphism (SNP) is dominated by the age of the father at conception of the child. The effect is an increase of about 2 mutations per year. After accounting for random Poisson variation, father’s age is estimated to explain nearly all of the remaining variation in the de novo mutation counts. These observations shed light on the importance of the father’s age on the risk of diseases such as schizophrenia and autism.

    In summary it was found that on average the father will introduce 2.01 additional mutations per chronological year, i.e. the additional 2 mutations those ignorant simpleton Chicken Littles so freaking out is TECHNICALLY EQUIVALENT TO THE FATHER BEING ONE ADDITONAL YEAR OLDER. From the US CDC data, the average age of the fathers weighted by the number of babies is 31.96 years old. Given that human male reaches sexual maturity at about age 15, the ‘average’ US father will have introduced additional 34 new mutations directly due to him. Thus the ignorant Chicken Littles are implicitly asseted that the ‘average’ US fathers are 17x bigger ‘monsters’.

    While human sperms are regularly being produced and new genetic mistakes can be introduced with time, the Icelandic study found that for human female the number of and genetic composition of the human eggs remain the same after the development from the embryo state and that on average the human eggs will have about 14.2 mutations irrespective of the mother’s age though the probability of being fertilized will decrease with time. Thus the ignorant Chicken Little’s assertion implicitly accuses that the US ‘average’ mothers are 7x bigger monsters. Thus if those ignorant Chicken Littles are parents themselves they are implicitly very much bigger monsters themselves.

    The correlations of mutations with early detected problems like schizophrenia and autism with respect to the father’s age can readily be shown. The suffuring of the affected families could be mitigated if embryo selection or genomic editing are available. There could be more genetic diseases manifested later in life of the babies.

    PS: I have no financial interest in embryo selection or gene editing business.

    • Replies:
  151. Factorize says:, thank you for re-energizing the discussion. Genome engineering really needs to be discussed more at the mainstream level in light of the results published in EA4 and elsewhere.

    My sums are somewhat different than yours; I will provide them for completeness. My sum from Supplementary Table 1 for all the positive effect alleles was: 21.47. If someone received all these positive alleles twice, then this would give 42.94. The units are somewhat unclear for beta; I assumed that they would be in standard deviations. When I calculated the expected positive score for two alleles the result was 13.53; for one positive allele it was 6.17. for the negative effect betas the sum was -20.3– two times this for all bads would be -40.6. For 2 bad alleles given population frequencies the expected result is -12.26 and for 1 bad is -5.69. This totaled for two of either with population frequencies at 1.27 and for one at 0.47 for a total total of 1.74. The average person should expect a polygenic EA4 score of 1.74. I tried some randomizing to determine variance, though I do not have these numbers now.

    It should be noted that EA probably should not be considered equal to IQ, thus 51 EA–> 771 IQ likely does not apply, though there is a moderately high correlation between EA and IQ. Clearly maximal IQ with the latest result would be well into the multiple hundreds of points.

    The EA4 Nature article included a peer review in which the reviewers expressed considerable concern at the lack of ethnic diversity in the GWAS research. The appalling racial GWAS gap that has emerged makes it increasingly likely that our future will be even more racially divided than it already is. The genetic knowledge that we have acquired for some populations now has the potential to cause our species to sub-speciate along cognitive phenotypes in a way that has never happened to date.

    I have good insight into my full genome through hundreds of polygenic scores that I have received. From what I understand given my genome I could with reasonably highly probability and with some effort reproduce offspring with extremish phenotypes (2-3 SD). Obviously people should become very very concerned about what is now possible. This will not be published as a headline in your daily newspaper, this will happen quietly and we will all wake up one morning and there will be a alien race of humans in our midst that we will have no idea how to cope with. The extremely hostile reactions against genetic enhancement that we have already witnessed has almost guaranteed that the necessary honest conversations will not take place but this will not prevent those interested in choosing the genetic destiny of their children and descendants. It would have been much more constructive if we had left the lines of communication open and agreed to disagree instead of resorting to moratoria.

    The peer reviewers recognized the wide ranging implications of the EA research and rightfully expressed alarm. Such commentary might finally signal the arrival of the scientific left into the discussion. The scientific left will not focus on denying the scientific validity of GWAS studies, but instead point to the far reaching consequences that these studies will have on our society especially as it relates to racial and possibly other segregations. Up till now the left has willfully rejected the scientific evidence and have left us facing a near endless abyss of genetic difference that could separate people from people.

    Thank you for pointing out the large negative polygenic component in the overall score. It is important to highlight this feature of polygenic scores. From my numbers above, the positive pulled the score up to 19.7 and the negatives pulled it down by 17.95 to ~1.74. Being special will not be that special in the future; all you need to do is not have the negatives which is a lot easier than CRISPRing in positives. If you could remove the 17.95 SD of expected negatives, your expected (EA/~IQ) would be 270. This is universally true with polygenic scores; there is a massive tug of war between the pluses and the minuses and then the pluses win by a slight amount. Selecting against the negatives would be a strategy that would be open to everyone (while positively selecting for the positives would actually require the presence of a positive).

    • Replies:
  152. says:

    De novo mutations are the mutations present in the offspring but not in the ‘normal’ genome of the parents, i.e. the mutations happen in the sperms, eggs or embryos. In addition to that the parents will also pass down mutations from their ancestors. Theoretically on average additional 46 expected mutations per ancestral generations will be added. However the extrapolation backward to too many generations become imprecise because of polygamy and significant number of cousin ma tings which might have reduced the diversity of the mutations but might caused more problems overall because of the higher probability of getting homozygosity with higher chance of the genetic diseases being expressed. A recent study put the ratios of the number of unique ancent paternal ancestors to be about 18x less than that for the maternal ancestors.

  153. says:

    The total EA3 GWAS dataset has many pvals approaching 1 and I filter them with pval ≤ 5e-5. I used that limiting pval because even at pval ≤ 5e-2 my laptop cannot handle the more than 1 giga bytes ASD dataset. Using the Bonferroni adjustment of pval ≤ 5e-8 is too severe with many without interactions between the phenotypes and those SJWonkers will scream about cherry picking the results. EA4 filtered the dataset with pval ≤ 5e-8 which will give the impression that EA can mostly be modified without affecting other phenotypes. Even at pval ≤ 5e-5 there is only a sole common SNP between EA3 and ADHD.

    EA3 is also useful as the supplementary data give the real life sample EA years mean and StdDev and so the EA3 PGS score can be related back to real life unit. I could not find such data from EA4.

    EA3 PGS is normalize to mean 0 SD 1. Those mental disorder PGS usually normalize to mean 0 but SD depend on how the disorders are quantified. Those papers are usually behind paywall and I only got the public accessible data.

    So at this stage it is best to claim “HARM REDUCTION” not “ENHANCEMENT”. The possible range is so large that even halving the potential efforts possible are very advantageous. The favoured gene variants are already present in the subject but their effects are suppressed by those bad gene variants some of which also concurrently give other bad genetic disorders or diseases. Who will object to enable the subject to reach fuller level of the INNATELY PRESENT POTENTIAL through harm reductions??

  154. Factorize says:, this is also a great point to highlight from the article: Low EA was found to create a wide variety of harms. Low EA implies much more than simply fewer years sitting in a classroom; there are far-reaching consequences for the individual and the community when people do not continue to develop their human potential through formal learning experiences.

    Low IQ and low educational attainment are associated with (?) (cause?) a wide range of health related outcomes: including: a ~ doubling in the risk of hypertension, ischaemic heart disease, myocardial infarction, hypercholesterolaemia, Type 2 diabetes, asthma, rheumatoid arthritis, migraine and major depression from the 1st decile to the 10th decile of EA (1st decile has the lowest EA). This is a commonly observed finding in various psychometric studies, though recent GWAS have been able to provide a more objective classification system using polygenic scores. It is not obvious how preventing harm reduction through genetic engineering could be considered to be progressive. Unless progressive is to be understood as oppressing the most disadvantaged members of our community.

    Nevertheless rational discussion of genetics at a sociopolitical level is no longer possible due to all of the lingering suspicions about eugenics, etc. However, financial self-interest might be able to drive forward a reappraisal even of genetics. There are enormous financial liabilities involved with the current suboptimal population genetics (perhaps on the scale of ~\$1,000s per year per capita). I suspect that as people become more fully aware of the magnitude of these costs, there could be a significant rethinking of the taboo of cognitive enhancement (er, harm reduction). There is a tendency for more urgency to be expressed when abstract topics about cognitive ability can be expressed more in terms of meat and potatoes politics.

Current Commenter

Leave a Reply -

 Remember My InformationWhy?
 Email Replies to my Comment
Submitted comments have been licensed to The Unz Review and may be republished elsewhere at the sole discretion of the latter
Commenting Disabled While in Translation Mode
Subscribe to This Comment Thread via RSS Subscribe to All James Thompson Comments via RSS