Today is the publication date of Hive Mind, a book by economist Garett Jones on the intimate relationship between average national IQs and national success, first and foremost in the field of economics.
I do intend to read and review it ASAP, but first some preliminary comments.
This is a topic I have been writing about since I started blogging in 2008 (and indeed well before I came across Steve Sailer or even HBD) and as it so happens, I have long been intending to write a similar sort of book myself – tentatively titled Apollo’s Ascent – but one that focuses more on the historical aspect of the relationship between psychometrics and development:
My basic thesis is that the rate of technological progress, as well as its geographical pattern, is highly dependent on the absolute numbers of literate high IQ people.
To make use of the intense interest that will inevitably flare up around these topics in the next few days – not to mention that rather more self-interested reason of confirming originality on the off chance that any of Garett Jones’ ideas happen to substantively overlap with mine – I have decided to informally lay out the theoretical basis for Apollo’s Ascent right now.
Assume that the intellectual output of an average IQ (=100, S.D.=15) young adult Briton in the year 2000 – as good an encapsulation of the “Greenwich mean” of intelligence as any – is equivalent to one nous (1 ν).
This can be used to calculate the aggregate mindpower (M) in a country.
Since sufficiently differing degrees of intelligence can translate into qualitative differences – for instance, no amount of 55 IQ people will be able to solve a calculus problem – we also need to be able to denote mindpower that is above some threshold of intelligence. So in this post, the aggregate mindpower of a country that is above 130 will be written as M(+2.0), i.e. that aggregate mindpower that is two standard deviations above the Greenwich mean.
2. Intelligence and Industrial Economies
There is a wealth of evidence implying an exponential relationship between average IQ and income and wealth in the United States.
There is likewise a wealth of evidence – from Lynn, Rindermann, La Griffe du Lion, your humble servant, etc. – that shows an exponential relationship between levels of average national IQ and GDP per capita (PPP adjusted). When you throw out countries with a legacy of Communism and the ruinous central planning they practiced (China, the Ex-USSR and Eastern Europe, etc), and countries benefitting disproportionately from a resource windfall (Saudi Arabia, the UAE, etc), there is an amazing R2=0.84 correlation between performance in the PISA international standardized student tests and GDP (PPP) per capita. (In sociology, anything about R2=0.3 is a good result).
The reasons for this might be the case are quite intuitive. At the most basic level, intelligent people can get things done better and more quickly. In sufficiently dull societies, certain things can’t get done at all. To loosely borrow an example from Gregory Clark’s A Farewell to Alms, assume a relatively simple widget that requires ten manufacturing steps that have to be done just right to make it commercially viable. Say an 85 IQ laborer has a failure rate of 5% for any one step, while a 100 IQ laborer has a failure rate of 1%. This does not sound like that big or cardinal of a difference. But repeated ten times, some 40% of the duller worker’s production ends up being a dud, compared to only 10% of the brighter worker’s. Consequently, one is competitive on the global markets, whereas the other is not (if labor costs are equal; hence, of course, they are not).
Now imagine said widget is an automobile, with hundreds of thousands of components. Or an aircraft carrier, or a spaceship. Or a complex surgery operation.
More technical way of looking at this: Consider the GDP equation, Y = A * K^α * L^(1-α), in which K is capital, L is labour, α is a constant that usually equals about 0.3, and A is total factor productivity. It follows that the only way to grow per capita output in the longterm is to raise productivity. Productivity in turn is a function of technology and how effectively it is utilized and that in turn depends critically on things like human capital. Without an adequate IQ base, you cannot accumulate much in the way of human capital.
There are at least two further ways in which brighter societies improve their relative fortunes over and above what might merely be implied by their mere productivity advantage at any technological level.
First, capital gets drawn to more productive countries, until the point at which its marginal productivity equalizes with that of less productive countries, with their MUCH LOWER levels of capital intensity. First World economies like Germany, Japan, and the US are extremely capital intensive. It is probably not an accident that Japan, Korea, and Taiwan – some of the very brightest countries on international IQ comparisons – also have by far the world’s highest concentrations of industrial robots per worker (and China is fast catching up). Since economic output is a function not only of pure productivity but also of capital (though subject to diminishing returns) this provides a big further boost to rich countries above the levels implied by their raw productivity. And as the age of automation approaches, these trends will only intensify.
Second, countries with higher IQs also tend to be better governed, and to effectively provide social amenities such as adequate nutrition and education to their populations. Not only does it further raise their national IQs, but it also means that it is easier to make longterm investments there and to use their existing human capital to its full potential.
All this implies that different levels of intelligence have varying economic values on the global market. At this stage I am not so much interested in establishing it with exactitude as illustrating the general pattern, which goes something like this:
- Average IQ = 70 – Per capita GDP of ~$4,000 in the more optimally governed countries of this class, such as Ghana (note however that many countries in this class are not yet fully done with their Malthusian transitions, which will depress their per capita output somewhat – see below).
- Average IQ = 85 – Per capita GDP of ~$16,000 in the more optimally governed countries of this class, such as Brazil.
- Average IQ = 100 Per capita GDP of ~45,000 in the more optimally governed countries of this class, or approximately the level of core EU/US/Japan.
- Average IQ = 107 – Per capita GDP of potentially $80,000, as in Singapore (and it doesn’t seem to have even finished growing rapidly yet). Similar figures for elite/financial EU cities (e.g. Frankfurt, Milan) and US cities (e.g. San Francisco, Seattle, Boston).
- Average IQ = 115 – Largely a theoretical construct, but that might be the sort of average IQ you’d get in, say, Inner London – the center of the global investment banking industry. The GDP per capita there is a cool $152,000.
Countries with bigger than normal “smart fractions” (the US, India, Israel) tend to have a bigger GDP per capita than what could be assumed from just from their average national IQ. This stands to reason because a group of people equally split between 85 IQers and 115 IQers will have higher cognitive potential than a room composed of an equivalent number of 100 IQers. Countries with high average IQs but smaller than normal S.D.’s, such as Finland, have a slightly smaller GDP per capita that what you might expect just from average national IQs.
These numbers add up, so a reasonable relationship equilibrium GDP (assuming no big shocks, good policies, etc) and the structure and size of national IQ would be:
Equilibrium GDP of a country ≈ exponent (IQ) * the IQ distribution (usually a bell curve shaped Gaussian) * population size * the technological level
Which can be simplified to:
Y ≈ c*M*T
… where M is aggregate mindpower (see above), T is the technology level, and c is a constant denoting the general regulatory/business climate (close to 1 in many well run capitalist states, <0.5 under central planning, etc).
To what extent if any would this model apply to pre-industrial economies?
3. Intelligence and Malthusian Economies
Very little. The problem with Malthusian economies is that, as per the old man himself, population increases geometrically while crop yields increase linearly; before long, the increasing population eats up all the surpluses and reaches a sordid equilibrium in which births equal deaths (since there were a lot of births, that means a lot of deaths).
Under such conditions, even though technology might grow slowly from century to century, it is generally expressed not in increasing per capita consumption, but in rising population densities. And over centennial timescales, the effects of this (meager) technological growth can be easily swamped by changes in social structure, biome productivity, and climatic fluctuations (e.g. 17th C France = pre Black Death France in terms of population, because it was Little Ice Age vs. Medieval Warm Period), or unexpected improvements in agricultural productivity e.g. from the importation of new crops (e.g. the coming of sweet potatoes to China which enabled it to double its population over the previous record even though it was in outright social regress for a substantial fraction of this time).
All this makes tallying the rate of technological advance based on population density highly problematic. Therefore it has to be measured primarily in terms of eminent figures, inventions, and great works.
The social scientist Charles Murray in Human Accomplishment has suggested a plausible and objective way of doing it, based on tallying the eminence of historical figures in culture and the sciences as measured by their prevalence in big reference works. Societies that are at any one time intensively pushing the technological frontiers outwards are likely to be generating plenty of “Great People,” to borrow a term from the Civilization strategy games.
To what extent does the model used for economic success apply to technology?
4. Intelligence and Technology Before 1800
A narrow intellectual elite is responsible for 99%+ of new scientific discoveries. This implies that unlike the case with an economy at large, where peasants and truck drivers make real contributions, you need to have a certain (high) threshold level of IQ to materially contribute to technological and scientific progress today.
The Anne Roe study of very eminent scientists in 1952 – almost Nobel worthy, but not quite – found that they averaged a verbal IQ of 166, a spatial IQ of 137, and a math IQ of 154. Adjusted modestly down – because the Flynn Effect has only had a very modest impact on non-rule dependent domains like verbal IQ – and you get an average verbal IQ of maybe 160 (in Greenwich terms). These were the sorts of elite people pushing progress in science 50 years ago.
To really understand 1950s era math and physics, I guesstimate that you would need an IQ of ~130+, i.e. your typical STEM grad student or Ivy League undergrad. This suggests that there is a 2 S.D. difference between the typical intellectual level needed to master something as opposed to making fundamental new discoveries in it.
Moreover, progress becomes steadily harder over time; disciplines splinter (see the disappearance of polymath “Renaissance men”), and eventually, discoveries become increasingly unattainable to sole individuals (see the steady growth in numbers of paper coauthors and shared Nobel Prizes in the 20th century). In other words, these IQ discovery thresholds are themselves a function of the technological level. To make progress up the tech tree, you need to first climb up there.
An extreme example today would be the work 0f Japanese mathematician Shinichi Mochizuki. At least Grigory Perelman’s proof of the Poincare Conjecture was eventually confirmed by other mathematicians after a lag of several years. But Mochizuki is so far ahead of everyone else in his particular field of Inter-universal Teichmüller theory that nobody any longer quite knows whether he is a universal genius or a lunatic.
In math, I would guesstimate roughly the following set of thresholds:
|Intuit Pythagoras Theorem (Ancient Egypt)||90||120|
|Prove Pythagoras Theorem (Early Ancient Greece)||100||130|
|Renaissance Math (~1550)||110||140|
|Differential Calculus (~1650+)||120||150|
|Mid-20th Century Math (1950s)||130||160|
|Prove Poincare Conjecture (2003)||140||170|
|Inter-universal Teichmüller theory (?)||150||180|
This all suggests that countries which attain new records in aggregate elite mindpower relative to their predecessors can very quickly generate vast reams of new scientific discoveries and technological achievements.
Moreover, this elite mindpower has to be literate. Because a human brain can only store so much information, societies without literacy are unable to move forwards much beyond Neolithic levels, their IQ levels regardless.
As such, a tentative equation for estimating a historical society’s capacity to generate scientific and technological growth would look something like this:
Technological growth ≈ c * M(>threshold IQ for new discovery) * literacy rate
ΔT ≈ c * M(>discovery-threshold) * l
in which only that part of the aggregate mindpower that is above the threshold is considered; c is a constant that illustrates a society’s propensity for generating technological growth in the first place and can encompass social and cultural factors, such as no big wars, no totalitarian regimes, creativity, etc. as well as technological increases that can have a (generally marginal) effect on scientific productivity, like reading glasses in Renaissance Italy (well covered by David Landes), and the Internet in recent decades; and the literacy rate l is an estimate of the percentage of the cognitive elites that are literate (it can be expected to generally be a function of the overall literacy rate and to always be much higher).
Is it possible to estimate historical M and literacy with any degree of rigor?
I think so. In regards to literacy, this is an extensive area of research, with some good estimates for Ancient Greece and the Roman Empire (see Ancient Literacy by William Harris) and much better estimates for Europe after 1500 based on techniques like age heaping and book production records.
One critical consideration is that not all writing systems are equally suited for the spread of functional literacy. For instance, China was historically one of the most schooled societies, but its literacy tended to be domain specific, the classic example being “fish literacy” – a fishmonger’s son who knew the characters for different fish, but had no hope of adeptly employing his very limited literacy for making scientific advances, or even reading “self-help” pamphlets on how to be more effective in his profession (such as were becoming prevalent in England as early as the 17th century). The Chinese writing system, whether it arose from QWERTY reasons or even genetic reasons – and which became prevalent throughout East Asia – surely hampered the creative potential of East Asians.
Estimating average national IQs historically – from which M can be derived in conjunction with historical population sizes, of which we now generally have fairly good ideas about – is far more tricky and speculative, but not totally hopeless, because nowadays we know the main factors behind national differences in IQ.
Some of the most important ones include:
- Cold Winters Theory – Northern peoples developed higher IQs (see Lynn, Rushton).
- Agriculture – Societies that developed agriculture got a huge boost to their IQs (as well as higher S.D.s).
- Inbreeding – Can be estimated from rates of consanguineous marriage, runs of homozygosity, and predominant family types (nuclear? communitarian?), which in turn can be established from cultural and literary evidence.
- Eugenics – In advanced agricultural societies, where social relations come to be dominated by markets. See Greg Clark on England, and Ron Unz on China.
- Nutrition – Obviously plays a HUGE role in the Flynn Effect. Can be proxied by body measurements, and fortunately there is a whole field of study devoted to precisely this: Auxology. Burials, conscription records, etc. all provide a wealth of evidence.
- Parasite Load – Most severe in low-lying, swampy areas like West Africa and the Ganges Delta.
All this means that educated guesses at the historic IQs of various societies are now perfectly feasible, if subject to a high degree of uncertainty. In fact, I have already done many such estimates while planning out Apollo’s Ascent. I will not release these figures at this time because they are highly preliminary, and lacking space to further elucidate my methods, I do not want discussions in the comments to latch on to some one figure or another and make a big deal out of it. Let us save this for later.
But in broad terms – and very happily for my thesis – these relations DO tend to hold historically.
Classical Greece was almost certainly the first society to attain something resembling craftsman level literacy rates (~10%). Ancient Greeks were also unusually tall (indicating good nutrition, for a preindustrial society), lived in stem/authoritarian family systems, and actively bred out during their period of greatness. They produced the greatest scientific and cultural explosion up to that date anywhere in the world, but evidently didn’t have quite the demographic weight – there were no more than 10 million Greeks scattered across the Mediterranean at peak – to sustain it.
In 15th century Europe, literacy once again begun soaring in Italy, to beyond Roman levels, and – surely helped by the good nutrition levels following the Black Death – helped usher in the Renaissance. In the 17th century, the center of gravity shifted towards Anglo-Germanic Europe in the wake of the Reformation with its obsession with literacy, and would stay there ever after.
As regards other civilizations…
The Islamic Golden Age was eventually cut short more by the increasing inbreeding than by the severe but ultimately temporary shock from the Mongol invasions. India was too depressed by the caste system and by parasitic load to ever be a first rate intellectual power, although the caste system also ensured a stream of occasional geniuses, especially in the more abstract areas like math and philosophy. China and Japan might have had an innate IQ advantage over Europeans – albeit one that was quite modest in the most critical area, verbal IQ – but they were too severely hampered by labour-heavy agricultural systems and a very ineffective writing system.
In contrast, The Europeans, fed on meat and mead, had some of the best nutrition and lowest parasitic load indicators amongst any advanced civilization, and even as rising population pressure began to impinge on those advantages by the 17th-18th centuries, they had already burst far ahead in literacy, and intellectual predominance was now theirs to lose.
5. Intelligence and Technology under Industrialism
After 1800, the world globalized intellectually. This was totally unprecedented. There had certainly been preludes to it, e.g. in the Jesuit missions to Qing China. But these were very much exceptional cases. Even in the 18th century, for instance, European and Japanese mathematicians worked on (and solved) many of the same problems independently.
But in the following two centuries, this picture of independent intellectual traditions – shining most brightly in Europe by at least an order of magnitude, to be sure, but still diverse on the global level – was to be homogenized. European science became the only science that mattered, as laggard civilizations throughout the rest of the world were to soon discover to their sorrow in the form of percussion rifles and ironclad warships. And by “Europe,” that mostly meant the “Hajnal” core of the continent: France, Germany, the UK, Scandinavia, and Northern Italy.
And what had previously been but a big gap became an awning chasm.
(1) In the 19th century, the populations of European countries grew, and the advanced ones attained universal literacy or as good as made no difference. Aggregate mindpower (M) exploded, and kept well ahead of the advancing threshold IQ needed to make new discoveries.
(2) From 1890-1970, there was a second revolution, in nutrition and epidemiology – average heights increased by 10cm+, and the prevalence of debilitating infectitious diseases was reduced to almost zero – that raised IQ by as much as a standard deviation across the industrialized world. The chasm widened further.
(3) During this period, the straggling civilizations – far from making any novel contributions of their own – devoted most of their meager intellectual resources to merely coming to grips with Western developments.
This was as true – and consequential – in culture and social sciences as it was in science and technology; the Russian philosopher Nikolay Trubetzkoy described this traumatic process very eloquently in The Struggle Between Europe and Mankind. What was true even for “semi-peripheral” Russia was doubly true for China.
In science and technology, once the rest of the world had come to terms with Western dominance and the new era of the nation-state, the focus was on catchup, not innovation.This is because for developing countries, it is much more useful in terms of marginal returns to invest their cognitive energies into copying, stealing, and/or adapting existing technology to catch up to the West than to develop unique technology of their own. Arguments about, say, China’s supposed lack of ability to innovate are completely besides the point. At this stage of its development, even now, copying is much easier than creating!
This means that at this stage of global history, a country’s contribution to technological growth isn’t only a matter of the size of its smart fractions above the technological discovery IQ threshold. (This remains unchanged: E.g., note that a country like Germany remains MUCH more innovative per capita than, say, Greece, even though their aveage national IQs differ by a mere 5 points or so. Why? Because since we’re looking only at the far right tails of the bell curve, even minor differences in averages translate to big differences in innovation-generating smart fractions).
It also relates closely to its level of development. Countries that are far away from the technological frontier today are better served by using their research dollars and cognitive elites to catch up as opposed to inventing new stuff. This is confirmed by real life evidence: A very big percentage of world spending on fundamental research since WW2 has been carried out in the US. It was low in the USSR, and negligible in countries like Japan until recently. Or in China today.
Bearing this in mind, the technological growth equation today (and since 1800, more or less) – now due to its global character better described as innovation potential – would be better approximated by something like this:
Innovation potential ≈ c * M(>threshold IQ for new discovery) * literacy rate * (GDP/GDP[potential])^x
I ≈ c * M(>discovery-threshold) * l * (Y/Y[P])^x
in which the first three terms are as before (though literacy = 100% virtually everywhere now), and potential GDP is the GDP this country would obtain were its technological endowment to be increased to the maximum level possible as dictated by its cognitive profile. The “x” is a further constant that is bigger than 1 to reflect the idea that catchup only ceases to be the most useful strategy once a country has come very close to convergence or has completely converged.
Japan has won a third of all its Nobel Prizes before 2000; another third in the 2000s; and the last third in the 2010s. Its scientific achievements, in other words, are finally beginning to catch up with its famously high IQ levels. Why did it take so long?
Somebody like JayMan would say its because the Japanese are clannish or something like that. Other psychometrists like Kenya Kura would notice that perhaps they are far less creative than Westerners (this I think has a measure of truth to it). But the main “purely IQ” reasons are pretty much good enough by themselves:
- The Nobel Prize is typically recognized with a ~25-30 year lag nowadays.
- It is taking ever longer amounts of time to work up to a Nobel Prize because ever greater amounts of information and methods have to be mastered before original creative work can begin. (This is one consequence of the rising threshold discovery IQ frontier).
- Critically, Japan in the 1950s was still something of a Third World country, with the attended insults upon average IQ. It is entirely possible that elderly Japanese are duller than their American counterparts, and perhaps even many Europeans of that age, meaning smaller smart fractions from the Nobel Prize winning age groups.
Japan only became an unambiguously developed country in the 1970s.
And it just so happens that precisely 40 years after this did it begin to see a big and still accelerating increase in the numbers of Nobel Prizes accruing to it!
Extending this to South Korea and Taiwan, both of which lagged around 20 years behind Japan, we can only expect to see an explosion in Nobel Prizes for them from the 2020s, regardless of how wildly their teenagers currently top out the PISA rankings.
Extending this to China, which lags around 20 years behind South Korea, and we can expect to see it start gobbling up Nobel Prizes by 2040, or maybe 2050, considering the ongoing widening of the time gap between discovery and recognition. However, due to its massive population – ten times as large as Japan’s – once China does emerge as a major scientific leader, it will do so in a very big way that will rival or even displace the US from its current position of absolute primacy.
As of 2014, China already publishes almost as many scientific papers per year as does the US, and has an outright lead in major STEM fields such as Math, Physics, Chemistry, and Computer Science. (Though to be sure, their quality is much lower, and a significant fraction of them are outright “catching up” or “adaption” style papers with no new findings).
If we assume that x=1, and that c is equal for both China and the US, then it implies that both countries currently have broadly equal innovation potential. But of course c is not quite equal between them – it is lower for China, because its system is obviously less conductive to scientific research than the American – and x is higher than 1, so in practice China’s innovation potential is still considerably lower than that of the US (maybe a quarter or a third). Nonetheless, as China continues to convege, c is going to trend towards the US level, and the GDP gap is going to narrow; plus it may also be able to eke out some further increases in its national average IQ from the current ~103 (as proxied by PISA in 2009) to South Korea’s level of ~107 as it becomes a truly First World country.
And by mid-century it will likely translate into a strong challenge to American scientific preeminence.
6. Future Consequences
The entry of China onto the world intellectual stage (if the model above is more or less correct) will be portentuous, but ultimately it will in its effects on aggregate mindpower be nowhere near the magnitude in global terms of the expansion in the numbers of literate, mostly European high IQ people from 1450 to 1900, nor the vast rise in First World IQ levels from 1890-1970 due to the Flynn Effect.
Moreover, even this may be counteracted by the dysgenic effects already making themselves felt in the US and Western Europe due to Idiocracy-resembling breeding patterns and 80 IQ Third World immigration.
A lot of the techno-optimistic rhetoric you encounter around transhumanist circles is founded on the idea that observed exponential trends in technology – most concisely encapsulated by Moore’s Law – are somehow self-sustaining, though the precise reasons why never seem to be clearly explained. But non-IT technological growth peaked in the 1950s-70s, and has declined since; and as a matter of fact, Moore’s Law has also ground to a halt in the past 2 years. Will we be rescued by a new paradigm? Maybe. But new paradigms take mindpower to generate, and the rate of increase in global mindpower has almost certainly peaked. This is not a good omen.
Speaking of the technological singularity, it is entirely possible that the mindpower discovery threshold for constructing a superintelligence is in fact far higher than we currently have or are likely to ever have short of a global eugenics program (and so Nick Bostrom can sleep in peace).
On the other hand, there are two technologies that combined may decisively tip the balance: CRISPR-Cas9, and the discovery of the genes for general intelligence. Their maturation and potential mating may become feasible as early as 2025.
While there are very good reasons – e.g., on the basis of animal breeding experiments – for doubting Steve Hsu’s claims that genetically corrected designer babies will have IQs beyond that of any living human today, increases on the order of 4-5 S.D.’s are entirely possible. If even a small fraction of a major country like China adopts it – say, 10% of the population – then that will in two decades start to produce an explosion in aggregate global elite mindpower that will soon come to rival or even eclipse the Renaissance or the Enlightenment in the size and scope of their effects on the world.
The global balance of power would be shifted beyond recognition, and truly transformational – indeed, transhuman – possibilities will genuinely open up.