What if we lost our smartest 5%?

If a deadly virus infected the smartest 5% of society, what would we witness?

Summary:

The following article explores the concept of a "Gifted Virus" that selectively targets the top 5% of the population in terms of intelligence, likening the impact to the devastation seen in historical events like the Cambodian genocide. It delves into the nature and measurement of intelligence, its societal value, and the potential catastrophic effects on society's infrastructure, technological advancement, and overall well-being if such individuals were lost. The piece also highlights the importance of the division of labor in modern societies and the irreplaceable role of highly intelligent individuals in maintaining and advancing civilization.

---

Written by Matthew Archer.

What is honoured in a country will be cultivated there.

—Plato, The Republic

How long could a nation survive without its most intelligent citizens? If a deadly virus infected the smartest 5% of society, what would we witness? How many scientists, surgeons, senior military officers, senior civil servants, pilots, judges, or engineers would be left standing? Infected, then, a few days later, gone. Perhaps the smarter people are, the more deadly the virus is. While the concrete poured over mass graves was still wet, the remaining elite would need to grapple with a near-total loss of top scientists. Naturally, this would mean no vaccines were forthcoming. Even with vaccines, the logistical nightmare of rolling them out would be made all but impossible without senior planners. As chaos ensued, the remaining gifted, unsure if they’d been targeted by a bioweapon, would barricade themselves away. Dead or in hiding, the end result would be much the same: a country stripped of its most important citizens teetering on the edge of extinction. 

While the Gifted Virus may sound like a plot to a dystopian novel, historical parallels do exist. During the Cambodian genocide, Pol Pot’s Khmer Rouge regime exterminated anybody suspected of being an intellectual, including those who wore glasses or spoke multiple languages. However, given that nearly a quarter of the Cambodian population was murdered (between 1.5-2 million people), it’s difficult to specify the long-term effects of losing the intellectual elite alone.¹ Yet, as we shall see, there certainly are ways of quantifying the value gifted people bring to modern societies. First, however, we need to understand a more fundamental set of questions: What is intelligence? Can it be measured? What predictive power do such measurements have? And how valuable are other traits compared to intelligence?

What is intelligence?

Most of us recognised the clever kids at school. What made them stand out? Well, they were capable of learning new information, recalling it, and applying it more easily and accurately than anybody else. If your school grouped people by ability, you probably noticed a few other things. First, the smart kids at age eight were often the smart eighteen-year-olds. Second, those who were in the top set for one subject were also in the top sets for others. There were plenty of exceptions (there always are), but that seemed to be the trend. If you’re reading this there’s a good chance you were in at least one top set. I can be confident about that prediction thanks to more than a century of scientific work on intelligence. We’ll come back to what intelligence research says about your classroom observations, but let’s start with a simple definition. It comes from Professor Linda Gottfredson and has been quoted by intelligence researchers ever since:

Intelligence is a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings—“catching on,” “making sense” of things, or “figuring out” what to do.

Note that this definition is fairly neutral, meaning it is perfectly possible to call people intelligent even if they believe very stupid things. For example, the bankers who crashed the global economy in 2008 were clearly intelligent by Gottfredson’s definition. You might take issue with this, but the neutrality is important. Without it, we are subject to the classic problem of conflation — the act of combining two separate concepts into one. In this case, the smuggled concept is wisdom. 

For most intelligence researchers, intelligence is about logical or analytical reasoning ability. Wisdom, on the other hand, requires emotional awareness and an appropriate application of knowledge to questions that probably don’t have one right answer. Because of this, it’s unclear whether wisdom can even be scientifically studied — perhaps it’s more like one of those you know it when you see it concepts. Of course, there’s a saying in social science that we should measure what we value, not value what we can measure. And while it would be useful if we could measure wisdom, there’s no point making perfection the enemy of the good. And measuring intelligence is a considerable good! Also, if we know who the really intelligent people are, perhaps we can teach them wisdom, an idea that traces back to Plato’s philosopher kings. 

That said, people are perfectly entitled to define words however they wish (though they might have some trouble convincing others to use more esoteric definitions). And it is admittedly slightly odd to describe someone as intelligent if their whole life involved an unending series of bad — even evil — choices. For example, were the Nazis intelligent? Undoubtedly many of them were, and it should cause us no consternation to say so, because intelligence, again, is about making the correct logical decisions. It is an instrumental, goal-directed cognitive process to get us from A to B. And the Nazis — like any leaders — faced many instrumental problems. Much has been written, for example, about their need to transport and kill Jews more efficiently. That problem required considerable logistical and engineering intelligence to solve. Indeed, at the Nuremberg trials, twenty-one leading Nazis were given IQ tests. The average score was 128. That’s the top three per cent of the German population. Three men had scores in the top one per cent: Hermann Göring (Hitler’s second in command) scored 138, Arthur Seyss-Inquart (the chancellor of occupied Holland, where he spearheaded the deportation and murder of tens of thousands of Jews) scored 141, and Hjalmar Schacht (Hitler’s banker and later an important part of the resistance) scored 143. 

Other examples of conflation include statements like, “Well, many of the smartest people are destroying the planet via climate change, so they can’t be that clever!”

Even if we grant that this is the case, it would be a fallacy to assume the individuals involved are stupid. The ‘stupidity’ here is what philosophers call an emergent property, something that exists in a system but not its constituent parts. For example, individual neurons do not possess consciousness; it is an emergent property of their network. The fallacy of composition is when one infers that just because something is true of the whole, it must be true of some part of the whole (or vice-versa). One of my favourite examples comes from Bertrand Russell who argued that just because everything within the universe has a cause, it does not automatically mean the universe itself has a cause — you cannot logically get from one to the other, you have to argue for it on separate grounds.

Can we measure intelligence?

Based on Gottfredson’s definition, the answer is an unequivocal yes. IQ tests are the most predictive instruments in all of psychology. Granted, they’re nowhere near as accurate as a ruler or thermometer, but a very low or high score usually means one will lead a certain type of life.

That’s why The Diagnostic and Statistical Manual of Mental Disorders (DSM) uses an IQ cut-off around 70 as a guideline for intellectual disability (formerly known as mental retardation).² 

So what is an IQ test? Forget about the type of questions you might encounter, a standard undergraduate psychology textbook will tell you that IQ tests are measurements of responses that are imperfectly correlated with consequential behaviours that are imperfectly correlated with intelligence.

Let’s walk through that with examples:

IQ tests are:

Measurements of responses

Example: How many logic puzzles you can solve in ten minutes. Your result will then be imperfectly correlated with:

Consequential behaviours

Example: If you can solve more logic puzzles than 99% of people, you are more likely, on average, to have a well-paid job. This, in turn, is imperfectly correlated with:

Intelligence

And we’re back to Gottfredson’s definition.³

A professionally administered IQ test will consist of many different subtests: choosing the correct symbol in a pattern recognition sequence, memorising a list of numbers and repeating them back in reverse order, mental arithmetic, verbal comprehension, a vocabulary test, verbal fluency (how many words beginning with K can you think of in under a minute?), reaction time, and rotating shapes in your mind’s eye. We then compare your performance to others. All IQ tests are arbitrarily ‘normed’ to make the average score 100. In other words, the test designers ensure that most people score close to the average.

Two objections are usually raised at this point. The first: do you have to get the same score each time for it to be valid? This is what’s called test-retest reliability. Thankfully, subtests are only included if they correlate highly from session to session. That is, if you take two memory tests one week apart, how similar are the scores?⁴ The correlation will never be perfect because factors other than intelligence play a role in performance, such as motivation, hunger, health, and distraction. The second objection: what if the subtests are measuring different abilities? If that’s true, then adding them up to create an artificial construct called “intelligence” is highly misleading. However, recall our empirical observation from earlier: more often than not, top-set maths students tended to be top-set English students. This should give you a clue as to one of the first major findings in intelligence research.

In 1904, the English statistician Charles Spearman discovered that performance in different school subjects correlated with each other to a considerable degree. Researchers had created tests that focused primarily on verbal memory, arithmetic, and pattern recognition. Yet Spearman showed that someone good at one test was, on average, very likely to be good at another. This phenomenon became known as the g factor, the letter referring to general intelligence or general mental ability. Analogies are often made to the concept of fitness. If your squat is in the 90th percentile, you are very likely to be able to deadlift considerably more than the average person! You are also likely to be a faster runner than the average person, even though sprinting is quite different to weightlifting. Just like intelligence, these measurements (how high you can jump, for example) imperfectly correlate with consequential outcomes (how much you weigh), which, in turn, imperfectly correlate with a concept we call fitness. In other words, there is a g factor for fitness, just like intelligence.

In 1993, at the age of seventy-seven, a leading American psychometrician named John Carroll published an 800-page book titled Human Cognitive Abilities. Reflecting on the accomplishment years later, Carroll would call it ‘the most bracing event of my life’. In the book, he looked at 460 sets of intelligence-testing data. He, too, found that all of the tests were positively correlated. The contemporary intelligence researcher Stuart Ritchie calls Carroll’s book ‘incontrovertible evidence for the g factor’. And, as Ritchie says in his own book, this is all rather peculiar when you think about it:

It didn’t have to be this way. It might have turned out that people who are excellent problem solvers are only so good because their skills in other areas have suffered: perhaps they have poorer memories because they have a ‘reasoning’ brain rather than a ‘remembering’ one. It might be that people who react quickly also tend to have poorer vocabularies. These scenarios are possible in theory and would result in negative correlations between the scores on some cognitive tests. But they’re not true in practice. It’s worth stepping back for a moment and realising how impressive this is—before we’d done the research, would we have predicted that people who are faster at moving their finger when a light blinks will be more likely to know the definition of defenestrate? Not necessarily. The positive manifold is remarkable, and it demands an explanation.

The general factor of intelligence is now recognised by almost every researcher out there as a real thing, not merely a statistical construct, but something representing a physical fact about how the brain is organised. Recall the first words of Gottfredson’s definition: intelligence is a very general mental capability. But let’s get back to the issue of prediction.

If you want to know how well someone will do in life and you can only measure one thing, an IQ test administered by a professional over several hours is the gold standard. By “well”, we mean almost anything that correlates with a common sense understanding of the good life, including happiness and your choice of marital partner, but the highest correlations are for educational and occupational success. Below is an extract from an undergraduate psychology textbook:

[…] job performance correlates more highly with intelligence than with factors such as performance during a job interview or education. The conclusion to be drawn from almost a century of research on the topic is that ‘for hiring employees without previous experience in the job, the most valid measure of future performance and learning is general mental ability’ (Schmidt and Hunter 1998, p.252). In the US, where these tests are such good predictors of job performance, the cost of not using them would equal total corporate profits, or 20% of the federal budget (Hunter and Hunter, 1984).

This isn’t to say that there’s nothing to the idea that IQ tests are something of a self-fulfilling prophecy: modern societies obviously do somewhat create the correlations between IQ and success. Nobody doubts that. But here’s the rub: it’s probably impossible to imagine a desirable society in which the abilities that IQ tests measure were not absolutely essential to the success of said society. The late Harvard psychologist Richard Hernnstein put it best on the TV show Firing Line:

Does society have to value those things that show up in IQ tests, however crudely? Can’t there be a society that values other things? […] Whenever these studies are done elsewhere, the data come out about the same and it makes no difference whether the political system is socialist so-called, or capitalist, or some mixed society. That is to say these data turn up in the Soviet Union, and Sweden, and the United States without very much difference. I think that it’s no accident that society values these traits. I think there are very real reasons — the work that needs to be done in society calls on the abilities that are measured (to some extent) by IQ tests. People will insist, for example, that their airplane pilots meet some standard of performance […] because airplane pilots play a very consequential role in our lives. That’s a characteristic, I think, of a complex society. But if you argue that society need not value these things, then you should go further and say well what kind of society might it be in which these things that are correlated with IQ tests are no longer valid?   

Finally, we should spend a little time discussing the word correlated. How happy you are and whom you marry is correlated to a lesser extent with IQ than occupational and educational success. Clearly, other traits and variables have a much greater influence over the former outcomes compared to the latter. Even with something like occupational prestige, the correlation is modest. One large longitudinal study called Project Talent followed Americans born between 1933-1936. Childhood IQ explained 15% of the variance in occupational prestige at age 28.⁵ Adult income is an even lower correlation in most longitudinal studies. This is not to criticise IQ. Rather, it serves to show just how hard it is to do good social science. 

Indeed, let’s compare IQ to the second-best predictor of occupational and educational success. According to most studies, that would be the personality trait known as conscientiousness. In 2014, eleven leading psychologists and geneticists published a fascinating study, the results of which are essential reading for anybody interested in education. The team obtained the GCSE scores for 13,306 twins in the UK. By comparing identical and non-identical twins, they determined the respective roles played by intelligence and non-cognitive traits (such as personality) in determining educational success. The pie chart below simplifies their findings:

Fig 1. The role of IQ vs non-cognitive variables in explaining GCSE results.

A single childhood IQ test explains an impressive 34% of the variance. And that’s just the average. Other studies have looked at IQ’s effect on different subjects. In a five-year longitudinal study of 70,000 English children, an IQ test taken at age eleven explained 58.6% of the variance in GCSE Mathematics grades, 48% in English, and even 18.1% in Art and Design (GCSEs are taken at age 16). Meanwhile, all the non-cognitive scores combined — meaning all the personality traits (openness, conscientiousness, extraversion, agreeableness, and neuroticism), plus scores for health and school and home environment, as well as parent- and child-reported behavioural problems — explain just 28% of the variance in GCSE scores. This means that the predictive power of any single personality trait is negligible. 

That surprises people. Not least because many have heard about the power of personality traits like perseverance. Indeed, psychologist Angela Duckworth popularised the importance of conscientiousness in her six-minute TED Talk on Grit, a catchy colloquialism for the type of determined person who identifies their goals and has the staying power to relentlessly pursue them in the face of adversity. However, psychologists have long critiqued the studies on Grit for using non-representative samples, such as army officers in training. These individuals inevitably share similar levels of intelligence, meaning one would expect to find that individual differences in perseverance make a large difference to success. When psychologists use representative samples, they find that intelligence contributes 48–90 times more than grit to educational success and 13 times more to job-market success.

Before you read the next section, I’m going to break the fourth wall and talk directly to you, yes you! These five-thousand-word articles take a lot of time to write. I wrote this one to be an engaging and accessible way for normal folk to understand the value of intelligence. And Aporia puts this — and indeed all of our articles — out there for free! If you think this is a worthy cause, I hope you’ll do the right thing and become a paid subscriber!

So what about giftedness? 

Now we have made the case for intelligence and IQ, it’s time to tackle the gifted and their role in society. Like intelligence, giftedness is also a contested concept. Some scholars prefer definitions emphasising potential, while others focus on outcomes. The former prioritises IQ scores, the latter actual achievement (often relative to a child’s peer group). Let’s focus on the most important trait differentiating the gifted: intellectual aptitude.

First, there is no perfect cut-off point for defining giftedness. Our Gifted Virus thought experiment uses the top 5%, other scholars consider the top 2% more useful. As far back as 1922, Carl Seashore, a Swedish psychologist, argued that the top 5% of college freshmen in the United States could assimilate five times as much information as the bottom 5%. Regardless of the cut-off, there will always be people who fall outside of it whose intellectual achievements would warrant the gifted label. Likewise, there’s probably not much of an observable difference between someone at IQ 125 versus 130, though a teacher would easily notice a 10-point gap between students.

Gifted researchers separate people into different levels of ability. The table below gives an example of a gifted classification system:

The categories above overlap with certain professions. Indeed, as one would expect, the predictive power of IQ for occupational success increases with the job’s complexity. Note that the top 5% is one in twenty people or an IQ of around 125. That’s the upper end of the mildly gifted category. These are the smartest kids in a normal classroom in a normal school, that is to say, a non-selective school in an average-income area. We can think of the mildly gifted as the ‘normal’ clever people most of us meet in our daily lives — some accountants, managers, teachers, etc. The moderately gifted will include many doctors, lawyers, and business executives. Unless you work in these fields, you’re unlikely to meet these people that often. The top 0.1% and above are people at the forefront of engineering, programming, law, journalism, and academia. Those above IQ 160 are mainly elite scholars in the hard sciences. A few of these people will leak out into industry or head up large public bodies, such as investment firms or central banks, but they are generally only stimulated by the most complex of problems.

Thinking about the work the gifted do is the best place to start when considering the effect of a Gifted Virus. Psychiatrist Bruce Charlton conceives of a pyramid of technology to illustrate this. At the top is the breakthrough, where some genius or (more likely today) a team of geniuses develops a once-in-a-generation tool, a paradigm shift, such as the Saturn V rocket that carried humans to the moon and launched America’s first space station, Skylab. The people capable of producing truly novel knowledge and technology with this type of impact are a fraction of a per cent. Below them are the people who refine the theory or technology, making incremental improvements in efficiency. Then we have those who replace the technology, which is harder than fixing it. Below the fixers are the operators. The final level is the sub-operators who cannot even use the technology. Culling the top 5% would likely mean losing nearly all geniuses and most of the people who refine their inventions. And we wouldn’t just lose the ability to innovate but to maintain and fix. Think of the major infrastructure that would be severely disrupted or face extinction without replacers and fixers. These are the people capable of replacing something as sophisticated as the hidden guts of the internet, the ones who fully comprehend the intricacies of server rooms and underwater cables; they are the people who can fix and replace nuclear power plants; they keep planes in the air and satellites in space. In short, they avert catastrophe and enable prosperity.

What does this mean in terms of hard numbers? At the national level, the German psychologist Heiner Rindermann has written much about Smart Fraction Theory, the idea that a country’s prosperity depends on the number of clever people it has. Rindermann and his British colleague and fellow Aporia contributor James Thompson used data from over ninety countries to show that the top 5% have the largest impact on national wealth. Other analyses suggest that the top 5% are between 30-40% as important as the mean citizen when it comes to national well-being (average health, educational level, creative achievements, scientific output, and so on). National IQ appears to be the best predictor of economic growth, with around 70% of the variation in the wealth of nations being due to IQ alone. The economists Garett Jones and Joel Schneider have used hundreds of thousands of models to conclude that each national IQ point increased GDP per capita by approximately 6%. 

What about the division of labour, though? Doesn’t a successful society require everybody, from operators to geniuses? Indeed, economists have shown that the most successful economies do have a high level of basic skills and that this high foundation is essential for implementing the innovations of the gifted, just as Charlton’s pyramid suggests. During the COVID-19 pandemic, there was much-needed praise for the roles ordinary people played in keeping society going. Despite being at or near the bottom of Charlton’s pyramid, nurses, delivery drivers, mechanics, factory workers, police officers, teachers, vicars, and cleaners deserve recognition for their essential roles in the profoundly complex reproduction of daily life. The magic of the division of labour was brilliantly captured by the economist Leonard Reid in his famous 1958 essay, ‘I, Pencil’, in which he described how no single person on the face of the Earth could produce a modern pencil.

Examples of the invisible ancestors of the humble pencil are the waitress who serves the logger his midday meal and the thousands of people involved in producing the food he eats before using — in comparison to the pencil — an unfathomably complex tool, the chainsaw, to chop down the tree that will be hauled onto a vehicle that drives on roads designed, built, and maintained by innumerable people. That’s before the lumber has been brought into a mill to be processed and before we enquire about the origins of the graphite mined in Sri Lanka, the rubber for the eraser, or the compound metal band holding it in place.

The pandemic showed us how a few cuts to this irreducibly complex web can cripple our ‘just-in-time’ supply chains. Moreover, despite the occasional Good Will Hunting type of recluse genius, most mathematicians do not want to sweep the floors at MIT. For that matter, many of the people who do sweep the floors probably don’t want to either. Yet they do. Mathematicians thus owe much to their ancestors and contemporaries who enable them to think and teach.

Nevertheless, this does not alter the harsh reality of the Gifted Virus thought experiment. Namely, if we care about maintaining advanced societies, most of us are replaceable. The gifted are not. While it’s true that modern societies require a diverse set of skills to function, losing the smartest 5% would be a catastrophe. Would it cripple society? Thankfully there’s no hard data. But losing these people — whether all at once or incrementally — would be a fair conservative guess.

In Sum

Absent cataclysmic natural disasters, our fate is ultimately always entwined with that of the gifted. How we treat and cultivate our most brilliant minds determines the majority of a nation’s successes and failures. The great intelligence researcher Arthur Jensen understood this well:

The quality of a society’s culture is highly determined by the very small fraction of its population that is most exceptionally endowed. The growth of civilisation, the development of written language and of mathematics, the great religious and philosophic insights, scientific discoveries, practical inventions, industrial developments, advancements in legal and political systems, and the world’s masterpieces of literature, architecture, music and painting, it seems safe to say, are attributable to a rare small proportion of the human population throughout history who undoubtedly possessed, in addition to other important qualities of talent, energy, and imagination, a high level of the essential mental ability measured by tests of intelligence.

Matthew Archer is the Editor-in-Chief of Aporia.

---

If you enjoyed this, you’ll probably enjoy another of my articles:

How about a Genius Basic Income?

1. Some have tried. James Flynn estimated that the Cambodian IQ was barely lowered by the massacre. However, Cambodia also has the lowest IQ in the region. The impact on the smartest citizens was indeed probably larger than the mean.

2. For brevity, this article won’t address national IQs. However, I thought a footnote might be useful for readers who have heard that African countries have IQs in the intellectually disabled range and are sceptical. The short answer is that most Africans who score poorly on IQ tests do not have any kind of mental disability, because the IQ score itself tells you nothing about the origins of the individual’s ability. See this short piece from Aporia’s Noah Carl for more.

3. Note: this model is simplified. In reality, there is often bidirectional causation: the consequential behaviours can make you better at the tests. We will develop this thought later.

4. For statistically-minded readers, the correlation tends to be around r = 0.8. 

5. Simplifying a little, the variance is the distance between the mean of a set of data to any point in the data. Variation refers to the amount of difference between a normal expected output to the observed output. We can measure that variation using variance. 

---

Enjoyed this article? Consider supporting Aporia with a paid subscription: