Further reflections on Covid-19

The make or break period for Western countries is pretty much over. Most of Europe managed to control the virus. Except Sweden, which is paying for its special strategy. This doesn’t mean these countries are out of the woods. Covid-19 will keep simmering and mismanagement or mishaps might lead to the occasional bigger outbreak. The US is pretty much done for. If they don’t science their way out of the hole they’ve dug for themselves, fatalities will reach several hundred thousand by the end of the year. With both sides of the political spectrum now not giving a damn, shut-downs and social distancing will not get the virus under control. And with riots ongoing and the election coming up this will not change.

In the third world Covid-19 is taking off. Unfortunately the low average age in these countries doesn’t seem to mitigate the death toll much if at all. Healthcare and the general health status of the population seems to matter too much. For the really lowest tier countries we won’t even know how many died. I expect estimates to be all over the place.

I still think the probability that Covid-19 escaped from WIV is substantial, but I don’t know whether it’s closer to 20% or 80%. I assume we will never know, because the matter is complicated enough that the general public will not be able to assess arguments made by evolutionary virologists and of course in the absence of a slam-dunk proof politics will rule the day.

The forecast website Metaculus has the question „Will it turn out that Ovid-19 originated inside a research lab in Hubei“, with a median prediction of 16% down from 20% a few days ago. Of course this is not the same as to ask whether it actually originated in a Chinese lab, so the Metaculus-derived probability for that should be significantly higher. Consider this a sanity check.

However, there is the possibility that this state of affairs will not be good enough for China and at the end of this year or maybe at the start of 2021 they will publish the ultimate proof that Covid-19 is of natural origin. Maybe they’ll find a predecessor of the virus with the furin-site but without the spike-protein, maybe even among samples taken from the Wuhan wet market. If that happens, I’ll update my probability estimate to >95% for a lab accident.

The mirrored contradictions in leftist and right wing ideology

The left is anti-war, anti-violence, anti-racism, anti-sexism, anti-nationalism. The anti-racism pillar has led to leftist generally preferring minorities over white people. Especially underprivileged minorities, such as Africans, Hispanics, Muslims, who all have to deal with a considerable amount of prejudice in the general population. Unfortunately for the left, this package of beliefs leads to the central contradiction, that these preferred demographics are generally much more sexist, racist, nationalist and violent than Whites.

The cope is a form of magical thinking: The badness of Whites makes non-Whites do bad things. Systemic racism, which can’t be quantified. Stereotype threat, which ascribes the power to Whites to change other peoples abilities just by having an opinion. Colonialism, which is still the reason for the existence of most of the railroad infrastructure in Africa, has retarded Africas development. Etc.

As I’ve discussed in a previous blogpost this is mirrored by the central contradiction of far-right ideologies: Despite their raison-d’etre being the awesomeness of Whites, the far-right is shifted along all ideological dimensions in the direction of the hated minorities. The complain about „pathological altruism“ and all the other left-wing traits that are exactly what differentiates Whites from almost all other groups.

They try to square the circle with antisemitic conspiracy theories. Whites are awesome, but Jews brainwashed them into population suicide. This again mirrors the left-wing cope: All the unwanted qualities of the preferred group are due to the evil meddling of another group.

The difference is of course, that the left has won the culture war (at least in the west) and can pour their particular blend of bullshit into the minds of all people via education, media and politics. There is also the long-term possibility that once minorities have reached critical mass, the left-wing delusion will seamless transform into the right-wing delusion with Whites being booted out of the coalition and getting handed the role of the Jews.

The New York Times riots

I try to keep this blog non-political, so I am just going to offer a few facts and observations that are orthogonal to the media coverage.

Fifteen unarmed black men were killed by the police in 2019 in the US. (Compared to 25 unarmed white men.) [1]
Unarmed is a somewhat poor proxy for harmless or innocent. If George Floyd is representative even these unarmed victims are ex-cons for armed robbery, high on fentanyl and trying to pay with counterfeit money. Which doesn’t mean they should be killed, but it does mean that their fate does not allow generalization to the non-criminal population.

Given that mostly armed criminals are shot by the police, you can’t take over representation compared to the population as indication of police bias. Instead you have to compare the number fatal shootings with the percentage of violent criminals of each racial group. Given that African Americans commit roughly half of all violent crimes in the US, despite being only 13% of the population [2], it is pretty astounding that they only account for 25% of the victims of police shootings.

The number of 15 killed unarmed black men and even the total number of 236 black men being shot by the police in 2019, is dwarfed by the 20.000 black victims of the Corona virus. The protests will almost certainly kill more innocent African Americans via increased virus transmission than the police would kill in the next couple of decades.

Of course one also has to consider the aim of the protests to reduce racism in the US. If there is one way to reduce racism in the US, it’s to provide a ton of videos of black men vandalizing and looting stores with the occasional assault on white people thrown in. Now that’s for sure.

The trend in police shootings of black men has been 258, 234, 224, 229, 236 from 2015 to 2019, so nothing really changed. If we observe some kind of trend coming to a head and leading to widespread violence, looting and burning, it must be some other trend …

By Zach Goldberg
By Zach Goldberg

[1] Washington Post police shooting database

[2] Wikipedia – race and crime

Testosterone and IQ

Almost twenty years ago I encountered the two-pronged IQ dilemma for the first time, when I realized that almost every second famous chess player was of Jewish descent. Around the same time I began to wonder why there are zero African mathematical or technological geniuses. This was genuinely troubling to me and in the first case I quickly accepted the reason brought forth by a book of chess history, that the systematic study of the Talmud uniquely prepared Jewish kids for playing high-level chess. I now know that this is at most superficially plausible as a closer look reveals that players like Gary Kasparov, Bobby Fischer or Levon Aronian likely never even set a foot in a synagogue.

My cope for the absence of African genius in technical areas was more ingenious and anticipated this blog to a certain degree. I had read about the Eysenck-study which is still a favorite of newspapers to trot out whenever the need arises. The Eysenck study was published in several steps so the numbers are different depending on which part one cites, but the way I remember the newspaper article, the kids of African-American GIs in Germany had an IQ of 99 for the boys and of 101 for the girls. I had also read that testosterone levels correlate negatively with IQ (for men). Despite almost two decades going by I still remember the claim, that this relationship was so robust that it was easily replicable with shoe size and math scores in a random class. Apparently there was an ideal testosterone level for mental achievements, which was in the low male range.

So I took these two disparate facts and combined them into the theory that higher testosterone levels in Africans raised the IQ of African women by bringing them closer to the low male level, but lowered the IQ of African men. The Eysenck-study was evidence in that direction, because the kids were only ten years old and only half African, so it was plausible that this bifurcation would show in full force after puberty. This fit perfectly with the evidence and the leftwing worldview I was raised in: The average IQ of all groups where the same, but patriarchy in Africa suppressed the expression of the superior mental abilities of African women.

Very happy with that theory I avoided checking any facts that might disabuse me off it. Of course the reality is that African women do not tend to do better in mathematics or IQ tests than African men. The testosterone-IQ-connection does not seem to have gained much traction either. However, it still strikes me as very plausible. Rarely are mathematicians or chess professionals particularly well muscled and rarely does it make sense to listen to what a physical athlete has to say. I was reminded of this again, when I saw the AlphaGo documentary recently released on youtube and listened to the unusually high voice of Lee Sedol.

That same high voice I had already noticed among top level chess players, for example, of the top of my head, Viswanathan Anand, Peter Leko, Veselin Topalov, Ivan Cheparinov and Anatoly Karpov. Now, there is such a thing as confirmation bias, so this list containing no less than three world champions does not constitute much evidence. But it would be an interesting project to quantitatively assess the pitch of the voice of top chess players compared to some baseline.

Covid19 developments

The emerging picture seems to be that most nations have not taken measures strict enough to nip the epidemic in the bud. New Zealand and Iceland being notable exceptions. In most countries the number of new cases declines, but not fast enough to actually eliminate the virus within a realistic timeframe. Re-openings will likely happen before the remaining cases can be controlled with tracing and testing, leading to a resurgence of the epidemic.

By now, judging from the results of serological studies in NYC where likely between 10% and 20% have been infected, it is even conceivable that large parts of the US will have „achieved“ herd immunity by the end of the year. Paid for with hundred of thousands of deaths.

On the other hand there are still a lot of things that could be done. Germany now has the capacity to test 800.000 people per week, but uses only 400.000 tests. Pooled randomized testing in every municipality could give a very fine-grained picture of what is going on. At risk populations could be tested monthly. Masks are now starting to be mandatory. Measuring body temperature before being allowed to enter shops or offices is a possibility. As we learn how and where the virus spreads targeted measures might become much more effective than general lockdowns.

And of course treatment will continue to improve. Apparently Covid19 trashes the lungs ability to absorb oxygen without similarly reducing the ability to remove CO2 from the bloodstream. Because high CO2 is responsible for the feeling of needing air, patients tend to visit the hospital too late for optimal outcome. This can be counteracted with automated oxygen measurement, which might reduce mortality in a significant way.

The other big development seems to be that the theory that the virus originated in the Wuhan Institute of Virology starts to become mainstream. As I understand it [1], the virus is very similar to a known bat virus, except for two positions in the genome. These two positions determine which cells can be entered. One position is very similar to the same section in a known pangolin virus. The other position has a nucleotide sequence that is known to make the virus more dangerous, which does not occur in either the pangolin or the bat virus.

Now the kicker is that both the pangolin and the bat virus where found by the WIV and the same institute has conducted gain of function experiments changing exactly these positions in viruses to study the increased virulence in humans. This makes it extremely plausible that there was yet another gain of function experiment conducted at WIV, which unfortunately couldn’t be contained.

The alternative is that these very specific and purposeful combinations of virus sequence happened by chance, because somehow a pangolin virus, a bat virus and some third virus exchanged just the right sections of their genome to create a virus that is especially dangerous to humans. Then this new virus started an epidemic in Wuhan, not known to be home to bats or pangolins, with or without wet market (the bat virus was isolated roughly 1000 miles away from Wuhan).

For once I side with intelligent design.

[1] Lab made cov2 genealogy

Covid19 – make or break

Some countries will wrack their economies and others won’t.

Some countries will stop the epidemic in its track and others will be overwhelmed.

It seems pretty unclear how these two bifurcations will correlate. It is entirely possible that a quick collapse with 100.000s of deaths during maybe two horrible months will leave the economy largely unscathed, while an extended lockdown might have much more severe economic consequences.

But of course quickly controlling the epidemic might lead to minor economic consequences while it is also possible to combine all the economic damage of an extended lockdown with not managing to keep the virus from spreading through the population.

There probably won’t be much middle ground in the epidemic bifurcation.

On a walk recently I was thinking about AGI and about what it would mean to slow the epidemic to such a degree that the healthcare system isn’t overwhelmed, without really stopping it. No, these two topics are not related.

In Germany there are 30.000 ICU units, I was calculating with 20 percent severe cases, of which 10 percent require a ICU unit. This is probably too high. Hey, I was taking a walk, had to work from memory here.

Anyway, with 10 percent requiring ICU to stay underwhelming the epidemic wouldn’t be allowed to hit more than 300.000 people at once. With a mean hospitalization duration of 3 weeks we end up with 80.000.000 / 300.000 * 3 weeks = 15 years. Which means that that scenario is never going to happen. Either the epidemic is stopped or it overwhelms the healthcare system.
In the same evening I came across an article making the same point. It used 2.5% ICU rate and was written by an AGI-researcher.

Of course my point still holds even with 2.5%, though that strikes me as too low, being even below some estimates for the completed mortality rate.

In the economic bifurcation things might be more gradual. If only because the plunging world economy will take everybody down a notch. Still some countries will enact a full scale lockdown, possibly over many weeks or even months. And others won’t.

The best case scenario seems to be that ramping up testing (PCR, X-ray, antigen + possibly test pooling), tracking and mandatory masks, gloves and disinfection will allow containment of the virus without extended lockdowns.

In any case, the next month is probably make or break for many countries. Unchecked doubling every three days will lead to 1000-fold increase, taking those countries who have a couple of hundred cases or thousand cases right now into disaster zone. Slowed doubling every 6 days would lead to a 32-fold increase, overloading the health system of most of those same countries.

Covid19 in Germany

There is a lot of data freely available, unfortunately it seems to reflect testing more than the actual spread of the virus and so far it is hard to know whether the Chinese death rates are representative of the overall lethality. But a comparison might be enlightening.

In the winter of 2017/18 there was a severe flu epidemic in Germany. Partly because the less expensive vaccine didn’t work against it and the more expensive one wasn’t paid by health care system. Within 3,5 months an estimated 10 million cases lead to 25.000 deaths. A lethality of 0.25%, somewhat high for the flu but not unusually so. 60.000 people were hospitalized.

The lethality of Covid19 is somewhere between 3-12 times higher than that flu. So a similar epidemic might lead to 75.000 to 300.000 deaths in Germany. This is quite possible, because Covid19 seems to be at least as contagious as the flu.

Additionally, the flu is generally somewhat similar to strains that have been common before, so part of the population is somewhat immune even without vaccination. Covid19 is completely new and therefore might achieve much higher penetration, with 50-60% of the population being infected at some point being a distinct possibility. In that case the worst case scenario is up to a million deaths.

In fact the toll might be even higher than that, because the German population is significantly older than the Chinese population. So the percentage of particularly susceptible people is quite a bit higher than in China.

Of course Covid19 is taken much more seriously than the flu and I assume that this worst case scenario would be avoided by draconian measures probably implemented several weeks or even months too late for optimal effect, that will completely crater the economy.

Standard deviations in population traits

Standard deviations are a somewhat neglected topic when it comes to the statistical analysis of group differences. And when it comes up, it usually only for the explanation of some tail effects: A larger standard deviation leads to more outliers and beats a higher mean if you go far out onto the tail of the distribution. Brilliant example: La Griyffe du Lion’s analysis of crime rates and serial killers [1].

But standard deviation are interesting beyond these tail effects. For example environmental hardship or strong environmental influences on a trait should generally increase the standard deviation. If half of all kids in a village get a disease that costs a few IQ points this will increase standard deviation in IQ, compared to a country where this disease has been eradicated. You can often see this effect in scholastic achievement studies where the standard deviation of second generation immigrants can be notably lower than that of first generation immigrants. In second generation immigrants language ability, health and malnutrition varies a lot less than among first generation immigrants and so does every trait downstream of these.

This makes it striking that the standard deviation of IQ in African Americans and of Africans generally, is usually lower than the SD of white Americans or European populations, despite the undoubtably worse environmental conditions. African American standard deviation in IQ for example varies between 11 and 14 points compared to a white SD of 15. Given that a worse environment should increase the SD, this lower SD most likely is due to genetic reasons. In this post I want to discuss possible influences on these differences in standard deviations.

Strikingly all tests in this study show lower SDs for the African students
In this study most subtests show lower SD for the South Africans with a four point differences in SD in the Performance and Full Scale IQ

One possible influence on the standard deviation would be admixture. If a population is a relatively recent mix of two populations with a different mean, the new population would have a higher SD. Basically the variation in admixture percentage would add to the trait SD. This can be observed in Hispanics, see my blogpost [2]. Of course African Americans are an admixed population with roughly 20% white admixture, while white Americans aren’t, so purely African African Americans should have an even lower standard dev than the current population.

In non-admixed populations the standard deviation of a trait ultimately directly depends on assortative mating for that trait. It is intuitive that random mating minimizes differences because people high on a trait and people low on a trait mix genes often. Strong assortative mating sees a widening of the bell curve up to a steady state influenced by the heritability of the trait.

So one interpretation of this observation would be that environments that select for a trait are environments in which this trait is valued, which means that assortative mating is strong. In that case we would expect to see populations with a high mean to also have a high standard dev and vice versa, which is kind of what we see in IQ. But as the blogpost linked above shows, the standard deviation of violent crime is higher in Whites although the mean is lower. This seems to constitute a counter example, until we realize that the trait under selection here might be peaceful behavior.

But there are also possible explanations that don’t invoke selection pressure. For example a population that has local mating, but a global cline in the trait in question, will have globally a higher standard deviation. Such a cline is often observed in IQ where the Northern parts of many countries are higher in IQ than the Southern parts, though occasionally it is the other way round. Nigeria is probably an extreme example for such an IQ cline, see [3]. So Nigerians as a whole population might have quite a high standard deviation. However, the resulting distribution in Nigeria would not be gaussian, but multimodal, because the different ethnic groups are very much endogamous. So Whites might have higher standard deviations simply because they have historically formed larger endogamous groups or rather endogamous groups that stretch over more terrain. This explanation would predict tails that are slimmer than expected, because the distribution is not fully gaussian. This scenario is somewhat comparable to the admixture case mentioned above.

A third and maybe most convincing scenario combines aspects of the other two ideas: Maybe standard deviations depend on the historical sophistication of societies. More advanced societies lead to stronger social stratification and this in turn leads to stronger assortative mating even without changing the preferences of the people involved. Assortative mating would partly be a byproduct of assortative socializing in socially stratified societies.

[1] Why most serial killers are white men.

[2] Hereditarianism III: Discussion

[3] An answer to Chanda Chisala

Demographic Change in France – Prenoms Rare

In my blogposts on demographic change in France I have discussed the growing percentage of kids of African origin in France. If estimated via the number of kids that are tested on sickle cell anemia, this percentage has surpassed 40% and has more than doubled since the year 2000. I counted the change in how many newborns are given typical Muslim names and could validate at least the growth rate of the sickle cell data, roughly a doubling between 2000 and 2015. It’s now 1.5 years since I analyzed the data and I decided to revisit the newest given name database, which is updated every year by the French bureau of statistics INSEE.

Re-checking the percentage of muslim names, I made the surprising discovery that since 2016 this percentage has stopped growing. For comparison, the number of names covered by my short list of muslim names is 13003, 26926, 25873 for the year 2000, 2015 and 2018. Given that the number of births has been dropping slightly, this translates into the percentage staying steady over 2015, 2016, 2017 and 2018.

If the muslim population from the Maghreb were the only sickle cell tested kids, this would translate to a steady 40% of kids with African origin. However, this is not the case. Subsaharan immigrants to France usually aren’t Muslims, so even if the percentage of Muslim kids has stopped growing, the original „sickle cell“ percentage might still be growing. (The sickle cell statistic has long been discontinued, for obvious reasons.)

Looking through the data I came across the prenoms rare, the rare first names. Under this moniker the kids are counted that have been given a relatively rare name. Contrary to the Muslim names, the percentage of prenoms rare has been steadily rising even in the last few years. I looked back a bit and made the discovery that this percentage has been doubling every twenty years or so since world war 2.

Given that the Muslim names had stopped increasing recently and given that this exponential growth seems to have started very early, surely before mass immigration, I was ready to interpret it as a trend of the French society towards greater and greater individualism or something. Then I decided to check whether the percentage of Muslim names in the different departments correlates with the percentage of prenoms rare:

It turns out that there is a significant correlation starting in the year 1949, that steadily increases as more and more departments have a count of Muslim names above zero until it reaches almost 0.80. In the early 2000s it suddenly starts dropping and has vanished by 2014. So what’s going on?

One thing that might contribute to killing off the correlation is that names of a 20% minority just aren’t as rare as names of a <10% minority. So Muslims might have slowly outgrown the prenoms rare marker. But the prenoms rare keep rising exponentially, so if Muslim names are getting less rare there has to be another driver of that growth. If we look at the intercept of the linear fit between the Muslim names and the prenoms rare percentage, that is the extrapolated prenoms rare for a department with zero Muslims, we get the following pattern over time:

There is something like 1% of rare names in the original French population and this probably doesn’t change too much. But starting in the 1970s there is an increase in the rare names independent of Muslims names that shows roughly a doubling every ten years or so. It seems probable that this is driven by non-Muslim immigration. However, when I look for the names that correlate strongest with prenoms rare over the different departments I get anything but rare names:

(0.47688124948411037, 1.8063429505759013e-76, ‘GABRIEL’),
(0.470505323840092, 3.2246147618498544e-74, ‘LIAM’),
(0.4416925115802408, 1.2859076223422156e-64, ‘MAËL’),
(0.4326625169376073, 8.57024632603287e-62, ‘MIA’),
(0.4206020094129121, 3.743600297108489e-58, ‘TIMÉO’),
(0.4190831094973139, 1.0503598293466153e-57, ‘ADAM’),
(0.4149723540036518, 1.6689711293740196e-56, ‘NOÉ’),
(0.4079050152026234, 1.7728242577990175e-54, ‘TIAGO’),
(0.4039208782138888, 2.342116862082597e-53, ‘KAÏS’),
(0.3975213210364356, 1.3754944258800354e-51, ‘EDEN’),
(0.39080880395597334, 8.963524732093063e-50, ‘ISSA’),
(0.39039340121367977, 1.1570834678410338e-49, ‘MILA’),
(0.3895565952224957, 1.9331033907637143e-49, ‘ISMAËL’) …

This seems to mostly pick up on urbanisation, which is in line with prenoms rare being driven by immigration. However, the hope to find actual names that represent the prenoms rare population comes to nothing. So again we are left without a sensible measure of the non-Muslim African population growth in France. But with the massive increase in prenoms rare it seems unlikely the entire „sickle cell“ percentage has petered off like the Muslim names.

Why I do not believe in a big dysgenic effect in the West

There are now several convincing papers that show a dysgenic effect in Western countries, but when I say big dysgenic effect what I mean are the estimates given by Woodley of Menie. He claims that in great Britain over the last hundred years average IQ has dropped by more than 10 points, that is 1 point per decade.

There are several reasons why I find these estimates unrealistic. One reason is that for such a big dysgenic effect presumably limited to western countries, the IQ gaps we see today are remarkably similar to Galton’s estimates 160 years ago.

Another reason to be skeptical is that we actually live in the golden age of mathematics. It seems unrealistic that after a drop of more than 10 points we would still have the geniuses to solve century old problems like the Poincare conjecture or Fermat’s last theorem.

I would also assume that some of the normal IQ tests used in the Wechsler test, would show a negative Flynn effect over the last decades. In actual fact the Flynn effect of the different subtests ranges from 0.07 to 1.59 standard deviations for the second five decades of the last century in the US. If the effect of environmental improvements can range from 0 to „a lot“, it seems a priori unlikely that the subtests with the weakest Flynn effect more or less exactly cancels a large dysgenic effect.

Flynn effect in the US between 1947 and 2001, don’t ask me why vocabulary is marked.

Woodley’s shtick is to find different traits that correlate with IQ and show that some sample many decades ago scored better than the average person does today. Unfortunately, this amounts to cherry picking and the long time between the studies makes sampling problems impossible to rule out.

Myopia for example correlates with IQ and has become much more prevalent in the last century. The correlation is even due to an overlap of genetic factors. Does the increasing prevalence of myopia prove a eugenic effect? Hardly. The true dysgenic effect is probably 2 to 5 times weaker as estimated via polygenic scores for the population of Iceland.