Reader questions from 13th Feb 2025

And a new comment section for the next round of questions

Back in February I posted the first of a series of paid subs ask me questions threads. Readers has some questions, so I'll attempt to answer them. This is in effort to boost the paid subscriber against, as even bloggers must eat at some point.

Mutational load and leftism

Reader Luc:

What's your opinion on bronskis hypothesis of mutational load being the true origin of "wokeness"?

I don't think it's likely. However, it is also not well studied, so it is too early to say for sure. I've previously written a compilation of evidence that paternal age affects the chance that offspring get mental illness based on large studies of sibling comparisons. Sibling comparisons are necessary because otherwise causes of older parental age will have genetic correlations with offspring traits, which will then appear to be caused by mutational load, proxied by paternal age. Using full siblings, the expected genetic values for the offspring are all identical (they are not of course, actually identical, however, there is no self-selection bias). The problem with the paternal (and maternal) age studies in the prior post is that they often show nonlinear patterns, strange reversals, and other findings that don't make sense from a genetic model of these arising from mutation load. For instance, there was a strong effect of paternal age on chance of ASD (autism), but no effect on the genetically related trait of ADHD. That is odd, since psychiatric problems are in general hypothesized to be related to bad genetics in general, which is what you (slightly) get from having older parents, and fathers in particular. Men produce semen their entire lives, but the cells that produce these accumulate random, small, randomly placed mutations, and on average these will disrupt genes.

But even if we accept that mutational load from aging parents causes some increase in psychiatric issues, we also have to assume this is for genetic reasons. As mentioned above, many of the findings don't make sense from a genetic perspective due to lack of consistent effects across diagnoses and odd nonlinear patterns. But even if we ignore these, we also have to accept that leftism is markedly increased in the mentally unfit. Again, we have data on leftism and various measures of mental and physical health. Leftism does seem to relate to poor health in general, but the effects are not strong. It is not so clear that increases in rates of random mutations can explain such large shifts in political ideology that has happened in the West the last 100 years or so. Rather, it looks like the path to leftism started much before the aging of parents began. The French revolution and the ensuring regime of terror is plausibly seen as a kind of proto-communism and egalitarian political push. Napoleon followed up, and while he did manly military things (usually right-wing coded), he also abolished laws favoring nobility in conquered areas, instituted more meritocracy and so on, which can also be interpreted as leftist politics (other interpretations are possible). Of course, it is somewhat foolhardy to try to classify all historical political movements and changes into a simple left-right dimension based on current Western politics, especially considering how fast these can change:

• Being pro-war is left-wing coded now, as long as it concerns Ukraine.

• Being pro-Russia is right-wing coded now, but when Russia was communist, it was left-wing coded.

• Being pro-NATO is left-wing coded now, but was right-wing coded the last many decades.

• Being anti-elite is right-wing coded now, but was left-wing coded for many decades.

• Being pro-free speech is right-wing coded now, but was left-wing coded when McCarthy was attacking the communists in Hollywood and the spies in the American government.

Overall, it seems that what policy is left or right-wind coded just changes depending on who currently would benefit from that policy, left or right-wing regimes, friends, or 'protected groups'. Russia is right-wing right now, so anything that hurts Russia is now a left-wing opinion; free speech is right-wing because left-wing people are in power and censor right-wing speech, and so on.

My other beef with mutational load theory is that the central concept is vague and not defined in statistical terms. Yes, we can reason that new random mutations will result in higher mutational load, but we don't generally measure this directly and there are no agreed upon genetic definitions or construct. Advocates of mutational load need to face this challenge and come up with genetic measures they think are included in their concept. There are many ideas, such as taking databases of variants known or suspected to be related to severe disorders and calculating a composite score of such metrics, and showing that these are 1) higher in leftists, and 2) increased over time, especially with a large spike since parent ages started increasing a lot in recent cohorts. I recognize that deep whole genome sequencing (WGS) data is needed for this research, which is generally not available. The best thing currently would be to study public genomes of high quality, and check whether these show an increase in recent cohorts compared to, say, genomes from the 1800s (if any such are public), or at least the medieval ones (100s are public). Studying the correlation with leftism requires also having given some kind of questionnaire to the subjects. Almost no datasets collected by researchers ask about political ideology, since most of these studies concern medical issues or diet etc. One would have to carefully read over the documentation of e.g. HRS or All of us and see if there is anything useful. Possibly one can impute political ideology based on other items, but then this would rest upon assumptions about the accuracy of this method. The best way forward would be to somehow convince academics to get biological samples from existing studies like the NLSY studies (which have politics questions already), and then get the WGS data, and make it available for study. This would be years into the future. I don't know of any suitable datasets right now for examining this question unfortunately. However, here's a graphical overview of the model, which I haven't seen published much before, but which is useful for guiding our thinking:

This is my impression of what this overall model is, not something I've seen explicitly proposed in a single post (Closest is the book by Sarraf et al 2019).

From my perspective then, mutational load as a cause of leftism is speculative and much more research is needed to reach firm conclusions, but the data needed to do this don't seem to be available. Seb Jensen also has a major post on this theory. I don't spend much time on this because I lack the means to study it. I also think we should be suspicious of models that model political opponents as genetically deficient in some way (self-serving bias is the strongest bias in scientists I think).

Policy implications of high heritability

Peter asks:

Frankly, your fairly convincing evidence that there are significant differences in average intelligence between races and countries (a racial proxy) is terrifying. Accepting this is all correct, what does this imply, what can be done, and how is a cohesive society and world to be maintained? I hate to say it, but is a highly selective DEI policy seeking out the most intelligent members of a disadvantaged population warranted or does this simply devolve into more reverse racism? Are there no environmental factors that can be manipulated to improve outcomes? Outside of eugenics are there other interventions to increase intelligence starting in pregnancy and continuing through maturation? I know you are fighting the first battle to simply get to a true understanding of population differences and have this scientific work not suppressed, but is it perhaps more productive to think about these larger implications and possible solutions at the same time? Big questions, beyond the scope of your research, but certainly necessary to consider.

Empirical findings don't strictly speaking imply any policy, as optimal policy choice always depends on priorities and values. Since readers don't necessarily agree with my values, I generally refrain from discussing policy implications of findings (if you nonetheless want to know my political opinions in general, click here). Let democracy sort it of (I know I am naive, blue eyed Scandinavian!). The optimal policy will also depend on the country and the values of the inhabitants.

With regards to improvements that can be made for intelligence that don't involve genetics, I am afraid I don't know of any. All the usual proposed causes are very minor when studied rigorously or haven't been studied proper much. Take fluoride in the drinking water, a popular theory. There are cross-sectional studies from poor countries comparing regions of e.g. China high and low in fluoride in the water, but when Swedish researchers looked at their register data, they basically found nothing for intelligence, but did find the usual protective effects on dental health. Their design is much stronger because they can analyze people who moved around between areas with high and low levels of fluoride in the drinking water. Or take lead poisoning, it seems overrated based on high quality research. Maybe we spent decades removing lead from paint and fuel for not much gain. We know for certain that the usual cognitive training methods do not work, so there's no hope that way either. I don't know what to try next.

Social desirability and halo effects

Hi Emil. I value your posts. I would be interested in hearing about methods of accounting for social desirability or halo effects in personality testing. I think it was Peacock who suggested looking at adjectives that had the same (or similar) meaning but different degrees of social desirability. It seems that one could take the degree of discrepancy as an index of social desirability bias. Thanks, Norm.

Social desirability is when survey takers (and others answering questions) tend to give responses that they think the listener or reader want to hear. Using self-rated personality data, one can often find a latent factor corresponding to this idea and thus statistically control for it, or include it in models to see what it relates to. Ulrich Schimmack has some a lot of research on this (he is otherwise obnoxious but he is a good scientist). So that is one method, use more complex statistical methods to try to adjust results for it. Other researchers think we should be using ipsative scales for personality self-report. They look like this:

Scalar:

Are you a hard worker? [Agree, slightly agree, neutral, slightly disagree, disagree]

Are you friendly? [Agree, slightly agree, neutral, slightly disagree, disagree]

Ipsative:

Which of the following is most accurate of you?

a) I am a hard worker
b) I am friendly

So instead of asking people "Are you a social butterfly?" rate yourself 1-5, they ask people to choose 1 of a number of descriptions or adjective, all of which may be positively worded. Thus, a tendency to select positively worded descriptions for oneself cannot cause particular personality dimensions to increase. Researchers who advocate this approach say that such scales are better for personality testing for hiring purposes. Yeah, maybe. They come with other statistical issues, which you can read about many places.

Psychiatric reliability, 23andme, ethnicity of readers, and assassins

Mel asks various questions:

1. Can you make a post exploring intra-psychologist diagnostic reliability?

2. Have your 23andMe results been updated since you last posted them?

3. In a hypothetical pie chart representing the race and ethnicity of everyone you ever cited on your blog, what would the biggest slice represent?

EDIT:

4. Can you do a post on assassins? This post is trying to use statistics to show that assassins tend to be short, but I don't believe that's the correct conclusion to draw at all:

Diagnostic reliability has been studied many times, both for psychiatry and medicine in general. My general understanding of psychiatry is that they have improved their game over the years. Older research shows not impressive amounts of reliability, as measured by specific diagnosis by different doctors:

Second generation, from World War II to the publication of DSM-III in 1980. During this generation, psychiatric nomenclature and diagnosis expanded gradually with the publications of ICD 6, 7, 8, and 9 and DSM-I and -II. As opposed to the first generation, studies of this generation relied on direct interviews with subjects utilizing the psychiatric nomenclature developed by the WHO and APA.¹⁴ In addition, a review of the general trends in psychiatric research between 1953 and 1983 by Reich, et al., found increasingly sophisticated research design.¹⁹ As a result, several investigators studied the reliability of psychiatric diagnoses with disappointing results. Sandifer, et al., conducted a study that involved 91 cases that were diagnosed by 10 experienced psychiatrists. The overall likelihood of a second opinion agreeing with the first was 57 percent.²⁰ Another study by Schmidt, et al., involved 426 state hospital patients who were diagnosed independently by a pair of psychiatrists. Agreement with respect to specific diagnostic subtypes occurred only in one half of the cases.²¹ Beck, et al., examined 153 outpatients and found 54-percent agreement for specific diagnoses.²² Kreitman, et al., found 63-percent agreement on 90 outpatients examined independently at approximately three-day intervals.²³ A review of six studies between 1956 and 1972 by Spitzer and Fleiss using Kappa statistic showed that the reliability of psychiatric diagnoses is still a major problem.⁷ Another observation was that psychiatrists had definite preferences in the selecting diagnostic categories.^24,25

However, this is a difficult bar to meet. As more and more diagnoses are added to manuals, it is statistically less likely they would agree by random chance, thus, reliability would have to decrease over time. When many similar diagnoses are added, it is not surprising if doctors can't agree on the exact variant. Take for instance obsessive-compulsive disorder (OCD):

Obsessive-Compulsive and Related Disorders
Obsessive-Compulsive Disorder
Body Dysmorphic Disorder
Hoarding Disorder
Trichotillomania (Hair-Pulling Disorder)
Excoriation (Skin-Picking) Disorder
Substance/Medication-Induced Obsessive-Compulsive and Related Disorder
Obsessive-Compulsive and Related Disorder Due to Another Medical Condition
Other Specified Obsessive-Compulsive and Related Disorder
Unspecified Obsessive-Compulsive and Related Disorder

There are now 9 different variants, and similar numbers for the other classes. Thus, depending on how one computes the agreement, the value may be high if one considers getting the right class, or low if one considers only getting the right subclass, exact diagnosis. Furthermore, because approximately all psychiatric diagnosis have overlap in symptoms and seem to be genetically correlated, at a given time, a patient may seem more OCD-like, and 2 weeks later when examined by a different physician, may appear more anxiety-like. Is this an error or true variation that was accurately captured? Who knows. It is probably better to focus on more general index scores than specific diagnoses (ironically, given all the work in expanding the list to include just about everything, except whatever leftists don't like).

My current 23andme results:

It says last updated 2021, so I don't know if it has updated, but it doesn't look like it. I can't find any support for the French-German component in my family tree, but there are some individuals with relatively unknown ancestry (everybody seems Danish except for some Jews far back), so maybe that is an error or maybe it is true.

The ethnicity of my blog readers was examined in my last blog survey:

Readers were asked to estimate their ancestry components, and these are the means. It looks fairly typical of an elite European group, with a few outsiders. Nordics are over-represented, unsurprising given that I am Danish.

Assassins? I don't know anything about assassins. I would scrape all the Wikipedia pages for assassins and compare them to other people from the same time and place and see if they are different in height. For the numbers you linked to, recall that people in the past were much shorter, so you shouldn't compare their heights to modern heights, or to their victims necessarily. One could generalize this effort to any other categories of humans on Wikipedia. But how much do we care about the height gaps of occupations like that? Maybe not enough to do this project.

Anti-obesity drugs and their other effects

Leo asks:

Question 1: does this point to a biochemical cause for something like the "p" or "general psychopathology" factor? I get it - craving isn't the same thing as psychopathology, but also there seems to be a connection that from the lay point of the view is kind of "obvious". I would say offhand that the ability to control one's cravings, and/or a lack of morbid craving, or even better yet, not having cravings at all (only "likes", "desires", "preferences") seems kind of clearly like it'd be a big factor in having a good life, including mental/psychic well-being.

Question 2: Emil, I didn't read, nor know of you, back in the day when you wrote more about philosophy. This is more a point to ponder than a question. First - it's kind of well-known that it's part of human nature to never be satisfied. Buddhism tells us that this constant, persistent, unrelenting dissatisfaction is something to be gotten rid of; yet obviously the great achievers of the world were powered by just this "dissatisfaction with the way things are today".

The matter of speculation: if these drugs work effectively to reduce cravings - which they clearly do - then are we actually changing human nature? For the first time ever? And if that larger question isn't quite the right one to ask, then "are we changing (major parts of) human nature?

There are many ongoing trials for GLP-1 agonists (Ozempic and friends). Some of these also examined other things aside from weight, and it seems they reduce cravings in general, suggesting shared biological pathways between addictive-behaviors and obesity. It seems some people are just better at dealing with cravings or feel them less strongly. One can tie this into Buddhist philosophy, with the goal of completely deactivating desires and achieving nirvana (and doing nothing in life). Some researchers scraped Reddit to see what people reported, and yes, they reported drinking less (but this is not an RCT, so may just reflect that people who write comments on Reddit report that anything works). There is more of this kind of evidence, but I couldn't find an RCT so far.

I don't think making changes to cravings a modifying human nature. It is a temporary environmental effect that will soon go away if the treatment is stopped. Modifying human nature is possible through genetic selection and editing, of course.

Who to support?

PT asks:

Who or what else should I support besides yourself? That is, how to be an "effective altruist" in this sphere?

Well, there's a lot of good Substacks around these days, most of which have subscriptions enabled. What you want to support is up to you. Of course, I think you should consider just increasing your support for my blog first! If you want to spread out your money, here's some suggestions:

• Seb Jensen

• Aporia magazine

• Joseph Bronski

• Peter Frost

• Meng Hu

• Uncorrelated

These are all relatively small Substacks, and you could of course also give money to Steve Sailer and so on.

The best way to ensure this kind of research gets done is to convince a rich friend to make a major donation.

Future of genetic enhancement

Realist asks:

Emil, this is a great idea, thanks.

I strongly believe in the potential of genetic research, especially the advancement of positive human traits. What are your thoughts on current and future research on genetic enhancement? What countries, if any, will lead the way? Who are some other researchers in this field?

Keeping with this line of thought, as I remember some time back, I had replied to a commenter that I thought China was researching human genetic enhancement, and you replied that from your Chinese contacts, that was not the case. A few days ago, I came across an old 2013 article by Geoffrey Miller, who believes that the Chinese are deeply into human genomic research. But he gives no sources. Please comment.

https://www.edge.org/response-detail/23838

I am pro-genetic enhancement. If we can prevent the birth bringing a person into the world who would get diabetes and instead bring into the world another very similar person (a sibling) who wouldn't, it seems difficult to come up with a reason why this is not a good thing. Depending on your ethics, it may be morally mandatory to prevent such preventable harms. I don't see any difference between avoiding bad states and upgrading humans for more intelligence etc. Many of these traits are genetically correlated anyway, so minimizing health problems will lead to increased intelligence and so on anyway, we might as well optimize directly for everything of value (according to those parents, I am against coercion).

I think USA will lead the way. It has lenient regulations, whereas other European countries mostly outlaw any optimization. This could change if Trump's conservative judges come up with some religious bullshit or overturn existing laws that prevent regulations (like Roe vs. Wade). On top of this, USA also has a very good environment for business and a can-do attitude. Europe-Europeans are very fearful, and don't even have much GMO-food, so who is going to allow GMO-humans?

Richard Lynn predicted (in his 2001 book on eugenics) that China would lead the way based on general trends and probably lower reluctance to do this. So far, they did have the first case of genetically edited babies for enhancement, but their reply was to imprison him (because he violated regulations). He is out again now, and promoting this idea. In general, Asians are also conservative, so it seems unlikely they would push ahead with such experimental, futuristic ideas. I guess it comes down to whatever Chinese leadership decides. If they decided tomorrow, they need to genetically enhance humans, I am sure it would happen immediately. So I expect slow but steady gains in USA (and whoever travels there to do it), and unlikely, but potentially large gains in China.

That's it!

I've set this post also to be subscriber only comments. You can post your new questions in the comments for the next episode.