I try to read many different kinds of books each year. Outside of my own fields, I try to consume high quality popular science, but mainly read academic books within my own areas. However, sometimes it makes sense to read popular science from my own area. I’ve been reading a number of Hans Eysenck books, for instance, which are somewhere between academic and popular science books. Since dysgenics and eugenics are perennial topics, it made sense for me to read the only popular science book on the topic published in recent years, namely At Our Wits’ End: Why We’re Becoming Less Intelligent and What it Means for the Future by Ed Dutton and Michael Woodley. The book was also written by two fellows that are my coauthors, so there’s that.
The book own description is to the point:
We are becoming less intelligent. This is the shocking yet fascinating message of At Our Wits’ End. The authors take us on a journey through the growing body of evidence that we are significantly less intelligent now than we were a hundred years ago. The research proving this is, at once, profoundly thought-provoking, highly controversial, and it’s currently only read by academics. But the authors are passionate that it cannot remain ensconced in the ivory tower any longer. With At Our Wits’ End, they present the first ever popular scientific book on this crucially important issue. They prove that intelligence ― which is strongly genetic ― was increasing up until the breakthrough of the Industrial Revolution, because we were subject to the rigors of Darwinian Selection, meaning that lots of surviving children was the preserve of the cleverest. But since then, they show, intelligence has gone into rapid decline, because large families are increasingly the preserve of the least intelligent. The book explores how this change has occurred and, crucially, what its consequences will be for the future. Can we find a way of reversing the decline of our IQ? Or will we witness the collapse of civilization and the rise of a new Dark Age?
Why? Why is it that we used to be able to fly from the USA to London in less than 4 hours but now we can’t? Why is it that we used to be able to put people on the Moon but now, it seems, we can’t? The answer is surprisingly simple. We are no longer intelligent enough to be able to do these things. We have become too stupid to keep Concorde in flight; let alone go back to the Moon.
It fell from the sky on 25th July 2000. Air France flight 4590 took off from Paris’s Charles de Gaulle Airport and promptly smashed into a hotel, killing all 100 passengers and 9 crew, as well as 4 people on the ground. The crash was essentially due to incompetence. A titanium ‘wear strip’ had been attached to the back of the thrust reverser of a Continental Airlines DC 10 Airliner as part of an operation to repair it. Not only was the wear strip badly produced, but it had not been made by the Airliner’s original equipment manufacturer. Accordingly, when the plane took off—just minutes before Concorde did on the same runway—the wear strip fell off. As Concorde accelerated over this sharp bit of metal at high speed, it punctured Concorde’s tyre, causing rubber to spin off the wheel and break open the fuel tank. This led to a fuel leak, which in turn led to a fire, resulting in one of Concorde’s engines being shut down. At the speed it was going, it was committed to take off, but Concorde couldn’t gain enough velocity or altitude to remain in the air. So Concorde crashed; all because an earlier aeroplane hadn’t been maintained properly. There had been problems with Concorde before—such as part of the rudder breaking off on a 1989 flight—but never a crash. The system had always worked. The pilots, in the heat of the moment, had always realised how to save the plane; the ground crew never made any major mistakes. Public confidence was shaken and, by 2003, Concorde was permanently grounded.[1] We were back to how it used to be. Flights between the UK and the USA were once again interminable.
So the error was with another aircraft. It’s not clear how this would show that we are too dumb to resume Concorde flights. Rather, Concorde flights are not resumed because of lack of demand. Aircraft in general have been getting slower over time, but a lot cheaper. The demand is simply that consumers prefer to fly more comfortably and cheaper rather than faster. Furthermore, the internet has decreased the need for personal meetings. Busy business people don’t fly over the Atlantic in a rush to have some meeting, they just have a video call on their phones. Easy.
With regards to space technology. I also have no doubt that if we wanted to, we could resume moon landings. Many countries have claimed they will do so in the near future. I don’t know if they actually will. Governments routinely claim all sorts of things. We still have terrorism, poverty, and children left behind despite governments declaring these to be ended in one way or another. The reason that we haven’t in general been going back to the moon is that it is not as cool anymore as it has been done, and it’s also absurdly expensive. I found a price of 280 billion USD in modern dollars. As far as I can tell, this is about the same as the entire state budget of California for a year. Something like 400,000 people worked on this project. It took out a large part of the state’s budget and Nixon shut it down after the popularity waned.
The authors are on better footing with some other claims. For instance, the decline in the rate of important innovations (‘macro-innovations’):
These rates are based on compilations made by other people of important innovations, and tend to show this 1850-1900s peak the authors are proposing. The reason for this peak is that dysgenics began in about 1850 in Britain, so the quality of people would be declining after that, and since it takes humans about 20-40 years to grow up and make their impact, this would imply a start in the decline in innovations around 1870-1890.
One annoying feature is the prominence given to that WORDSUM study:
It’s based on this never replicated Woodley et al study from 2015, By their words ye shall know them: Evidence of genetic selection against general intelligence and concurrent environmental enrichment in vocabulary usage since the mid 19th century:
It has been theorized that declines in general intelligence (g) due to genetic selection stemming from the inverse association between completed fertility and IQ and the Flynn effect co-occur, with the effects of the latter being concentrated on less heritable non-g sources of intelligence variance. Evidence for this comes from the observation that 19th century populations were more intellectually productive, and also exhibited faster simple reaction times than modern ones, suggesting greater information-processing ability and therefore higher g. This co-occurrence model is tested via examination of historical changes in the utilization frequencies of words from the highly g-loaded WORDSUM test across 5.9 million texts spanning the period 1850–2005. Consistent with predictions, words with higher difficulties (δ parameters from Item Response Theory) and stronger negative correlations between pass rates and completed fertility declined in use over time whereas less difficult and less strongly selected words, increased in use over time, consistent with a Flynn effect stemming in part from the vocabulary enriching effects of increases in population literacy. These findings persisted when explicitly controlled for word age, changing literacy rates and temporal autocorrelation. These trends constitute compelling evidence for the co-occurrence model.
So really there are 10 words in a vocabulary test, they authors split them into 2 groups, hard and easy. For the hard words group, they look up their frequencies on Google’s n-gram, and their weighted average (factor score) is what you see on the plot above. This of course seems very fishy, so why hasn’t this been replicated before being used as important evidence of dysgenics in numerous papers and even books? I don’t know. I have access to at least a few other vocabulary datasets that provide the same item information (like this one, see my study), that is, word difficulty and g-loading (discrimination in item response theory terms). As a matter of fact, the use of vocabulary is a bit tricky because a single item has not one, but two words. In the WORDSUM test, there is a target word and 5 options, and one has to pick the synonym. There is a 20-item version here you can try. From what I can tell, these are the 10 items used in the study (posted on this blog in 2010):
Another 2010 blogpost gives this alternative set of words:
- A. SPACE 1. school 2. noon 3. captain 4. room 5. board 6. don’t know
- B. BROADEN 1. efface 2. make level 3. elapse 4. embroider 5. widen 6. don’t know
- C. EMANATE 1. populate 2. free 3. prominent 4. rival 5. come 6. don’t know
- D. EDIBLE 1. auspicious 2. eligible 3. fit to eat 4. sagacious 5. able to speak 6. don’t know
- E. ANIMOSITY 1. hatred 2. animation 3. disobedience 4. diversity 5. friendship 6. don’t know
- F. PACT 1. puissance 2. remonstrance 3. agreement 4. skillet 5. pressure 6. don’t know
- G. CLOISTERED 1. miniature 2. bunched 3. arched 4. malady 5. secluded 6. don’t know
- H. CAPRICE 1. value 2. a star 3. grimace 4. whim 5. inducement 6. don’t know
- I. ACCUSTOM 1. disappoint 2. customary 3. encounter 4. get used to 5. business 6. don’t know
- J. ALLUSION 1. reference 2. dream 3. eulogy 4. illusion 5. aria 6. don’t know
Looking up the frequencies over time, they report this figure:
FIGURE 1. Temporal trends of the 10 WORDSUM words based on the fitted mixed models. Solid lines represent more difficult and dashed lines less difficult words. Black lines represent words that are more associated and gray lines less associated with rFPR (strength of association was based on a median split). WORDSUM codes are displayed to the right.
So J should be sedulous or allusion. So let’s check:
Based on this, I would guess the word is allusion. Still, it is obvious that this analysis has to be redone with more words from other tests, but without changing the methods. Contact me if you would like to join this project (emilowk@proton.me).
I also think their implied formal model fails. As far as I can tell, the way to derive the predictions that easy words increase in use while harder words decrease, one has to posit a declining vocabulary size over time. A shrinking vocabulary size combined with an assumed constant rate of words used per day will result in the rarer words being used less often as a smaller proportion of speakers know them, while common words increase in frequency. However, extant data on vocabulary size show very small secular changes, not the predicted decline. One would then have to posit some kind of vocabulary-specific co-occurrence model to account for this…
These are in my opinion the worst parts of the book. In general I buy the overall claims:
- Intelligence has been declining, which we know from studies that examine fertility and intelligence as well as its various social status proxies.
- In 2018, this wasn’t much done, but now we know that polygenic scores bear out this prediction of declining intelligence, a risky prediction that was confirmed, so it counts as a lot.
- There is some decent cognitive data showing declines, reaction times and color acuity.
- There are some fairly convincing third party metrics like innovation rates that also decline (shown above)
- Flynn effect has stopped and now been replaced by a decline even in overall IQ scores (anti-Flynn effect, two metas already)
The most interesting part of the book is the speculation about the historical changes in other civilizations. With the realization that our civilization is in decline and this relates to the genetic basis of intelligence, we might ask ourselves: can the rise of fall of empires more generally be explained in terms of ups and downs of genetic scores? I think it is likely! This brings us into the territory of doomsayers like Spengler but with a modern genomic angle. But this kind of thinking goes back a long time:
However, it was the Greek philosopher Polybius (200–118 BC) who was the first to advocate, albeit implicitly, a cyclical philosophy of the rise and fall of civilisations wherein there was no metaphysical dimension. Polybius was from the city state of Megalopolis in Arcadia. The son of a senior politician, Polybius rose to be highly influential in politics himself. In analysing the rises and falls of dominant societies in the history of Rome and Greece, Polybius noticed that the same pattern could be observed again and again. Societies rise when they are religious, have a deep reverence for the past and for older generations, are prepared to engage in noble acts of self-sacrifice, and follow clear moral rules. These qualities ensure that they have a sense of superiority, a sense of their own destiny, that they are a cohesive community, and that they can be motivated to defend their society, even unto death. When they lose these qualities—which they inevitably do—then they fall. People become too rich and when this happens they lose their ‘fear of the gods’ and with it their selflessness and community spirit, their sense of eternal destiny, their reverence for older generations, and the strict moral rules which bind them together. By the time Polybius was writing, he was of the view that Roman society was itself in decline, as we will see in Chapter Eleven. Polybius also noticed that this process coincided with the same demographic decline that we have witnessed in the second half of the 20th century in the West: people, and especially the most intelligent people, simply stop having children.[15]
Moving into the medieval period, we meet the Islamic scholar Ibn Khaldun (1332–1406). Ibn Khaldun was born into an aristocratic Andalusian family that had emigrated to Tunisia after the fall of Seville to the Reconquista in 1248. He worked as an adviser or prime minister to various political leaders and established himself as a great philosopher. In 1400, he was caught up in the siege of Damascus. The leader of the siege, Timur, was so keen to meet the famous philosopher that Ibn Khaldun was lowered in a basket over the city wall and spent seven weeks in Timur’s camp, lecturing him on the theory of history.[16] Ibn Khaldun argued that central to civilisation was the concept of ‘Asabiyyah’, which translates as something like social cohesion or social solidarity. Asabiyyah will increase and reach a peak as civilisation advances but, ultimately, it will go into decline and, with it, the civilisation will go into decline and be displaced by another one in which Asabiyyah is stronger. It can be seen just how similar Ibn Khaldun’s theory is to that of Polybius, though Ibn Khaldun directly spells it out.[17] For Ibn Khaldun, conditions of something like group selection were strong among people who lived in the deserts. This meant they could only survive if they were high in Asabiyyah and manifestations of it such as religiousness and martial values. This high Asabiyyah allowed them to flourish and create cities. However, here the selection for Asabiyyah was lower because conditions were more luxurious. As such, after a number of generations Asabiyyah declined to an extent that they would be invaded by desert tribes that were higher in Asabiyyah and the cycle would begin all over again.[18]
The authors conclude their book with a discussion of eugenic solutions to the problem of dysgenics. As followers of Dutton’s channel will know, they don’t expect embryo selection to save us:
Advances in understanding the genetic basis of traits like g, educational attainment, and various diseases have led some to propose a liberal eugenics which is based on voluntaristic (rather than coercive) approaches to improving the inborn characteristics of one’s descendants.[6] Certain bioethicists, such as the Australian bioethicist Julian Savulescu, who is based at Oxford University, have even argued that individuals are morally obliged to use genetic enhancements on their descendants, so as to bring about greater human flourishing. A key problem with this liberal eugenics is that it is unlikely to ever ‘catch on’, owing to what bioethicists call the ‘yuck factor’—this is a basic and visceral rejection of meddling in human nature that colours much of the debate about the desirability of eugenic intervention among Western populations in particular. Gerhard Meisenberg (who we met earlier) conducted a study into the attitudes of 1,464 medical students on whether or not and also under what conditions reproductive genetic intervention should be acceptable.[7] He found that ‘the strongest and most consistent influence [on attitudes towards the desirability of reproductive genetic intervention] was an apparently moralistic stance against active and aggressive interference with natural processes in general.’ In other words the sample had negative attitudes towards reproductive genetic intervention, especially if the objective was human enhancement. This suggests that the majority of individuals would likely fail to take advantage of ‘genetic enhancements’, even if they were cheaply and legally available to prospective parents, these simply being too ‘yucky’ to contemplate. The fact that leading a horse to water doesn’t necessarily entail it drinking is less of a problem, however, than the uses to which the increasingly distant and unaccountable globalist elites—the ones that Spengler predicted would come to dominate the political life of civilisations in winter time, and indeed did—may put such technologies. Recall that the historical period associated with rising levels of g was also associated with group selection—essentially g could only rise to the extent that it benefitted the group via provision of geniuses, whose innovations could create new opportunities for group expansion. An elite that is anti-group selected (i.e. purely self-interested) is likely to enhance in their offspring those traits that were most important to its success—traits such as psychopathy.[8] Thus such a ‘liberalised’ eugenics is more likely than not to make things worse for civilisation in its winter years.
They don’t really have any technical objections. Their hostility is based on their strong belief in multi-level/group selection, which they think is the only way to select for intelligence and related traits. It seems at odds with obvious current technology. IVF was once considered yucky, but now it’s normal. I think IVF-based genetic enhancement/betterment will be the new norm in the future. How far future? We can project the future based on the growth of IVF. Sometime ago I made this plot, which shows the general direction:
The data are delayed for several years unfortunately. But in 2014, Denmark was already at 6% of birth involving some kind of help. In 2020, Danish media reported that the IVF rate hit 10%. Can this value get to, say, 50%? C-sections used to be rare, but now C-sections are already more than 50% in at least one European country in 2017:
So with these trends in mind, I think we can confidently forecast that IVF% of all births will grow and become 50% sometime this century, probably the first part. It’s a good Metaculus question, which I have made here (link will start working when it goes public).
Can artificial selection select for intelligence? What will parents choose? We already know based on sperm donors that prospective mothers pick men with more education and better physiques, so clearly, this is about the same as natural selection used to do before 1850. The trouble is more with any other non-intelligence traits that used to be under selection but whose selection we may now be changing in a suboptimal direction. For instance, extroversion is strongly related to fertility in modern societies, but this was perhaps not always the case. It may not be good for society to become increasingly extroverted. High levels of extroversion are generally associated with positive outcomes, such as higher salaries, but are also associated with drug abuse, infidelity and some other issues. For a defense of this kind of selection for the future of mankind, read Jonathan Anomaly‘s book Creating Future People: The Ethics of Genetic Enhancement (2020). This was published 2 years after the present book, so the authors cannot be faulted for not replying to it. Overall though when it comes to forecasting the future, I think the authors are needlessly bleak. Yes, civilizational cycles in the past were probably real, and we do stare at the abyss. However, we are the first civilization to have attained the direct means to change the course of selection. What will we do with this knowledge? That is the question.