Radical Educational Meritocracy: An Ethical and Empirical Defense

A review of behavioral genetics, education policy, and our priorities.

“The natural distribution is neither just nor unjust; nor is it unjust that persons are born into society at some particular position. These are simply natural facts. What is just and unjust is the way that institutions deal with these facts.”

John Rawls, A Theory of Justice (1971)

The behavioral geneticist Paige Harden gave an interesting talk, “A World with Equity in Educational Success.” Speaking to an audience of educators and educational policymakers, she argues that though genetic differences have historically been seen as an enemy of educational equity, our rapidly growing genetic knowledge can be used in service of educational equity and social justice.

Harden rightly points out that the study of heredity has historically been used to odious ends: forced sterilization, genocide, etc. She claims that these historical atrocities have been motivated by three incorrect and dangerous ideas:

  1. Genetic essentialism: people have an essence, a true self, that is defined by their genes. This idea is dangerous because it links our moral worth to our DNA. 

  2. Genetic determinism: genes determine your life outcomes. This idea is dangerous because it saps our will to create a better world. 

  3. Genetic and scientific racism: some races/ethnicities/populations are better than others due to innate genetic differences. 

Harden notes that the audience might bristle at this talk of genetics. However, this needn’t be the case: she believes there’s real potential for using genetics to establish educational equity. The upshot of her talk is that we can recast these three hereditarian ideas into the following more justifiable claims:

  1. Genetic luck: though we are all dealt a different genetic hand, no one hand is more praiseworthy than another since it is all a matter of luck. No one has earned their DNA sequence.

  2. Genetics as a tool for identifying potential for growth: genetics can be used to identify untapped potential and assess if schools are adequately dealing with this potential. 

  3. Prioritizing diversity in the genomics research world: though 79% of participants in genetic research are of European ancestry, genetics can be used to empower historically oppressed populations by involving members of these groups, both as study participants and researchers. 

According to her, in the face of the DNA revolution, we should “take the science of genetic differences and peel it away from these [three dangerous] ideas that have plagued any discussion of inherited differences for the last century and actually use the tools of modern genomics as a tool for equity.”  

I largely agree with Harden on the empirical research on genetics and educational attainment. However, I’m far more pessimistic and don’t believe genetics can or should be used to advance an equity agenda. I’ll argue that despite our best efforts, we will be unable to reconcile the deluge of genetic research coming out with goals of equity. That said, ultimately I don’t believe there’s a lot of daylight between Harden and me on this issue, other than for a few key ethical assumptions.

Note: This essay presumes familiarity with behavioral genetics. For a brief primer, check out this review of Robert Plomin’s book, Blueprint. To make a long story short, in the past 5-10 years, due to a steep decline in the cost of genotyping (sequencing a set of genetic markers) and studies done on this genetic data, we are quickly developing genetic predictors (called polygenic scores, PGS, or PRS) of traits like educational attainment (EA). These predictors currently explain around 13% of the variance in educational attainment, which is as predictive as knowing one’s parental income level.

Steelmanning Harden on Equity

The OECD definition of educational equity that Harden cites has two components: inclusion—“ensuring that all students reach at least a basic minimum level of skills”—and fairness—“personal or socio-economic circumstances, such as gender, ethnic origin or family background are not obstacles to educational success” (p. 15). On its face, both aspects of educational equity seem laudable. We want an informed citizenry that is able to do basic math, reading and science; and we don’t want there to be personal or socio-economic impediments to educational success. 

Harden only briefly sketches how genetics (and her three principles) will aid her goal of educational equity. So, I’ll attempt to flesh out a charitable vision of (i.e., steelman) what she might have in mind: 

Everyone is dealt a genetic hand which they had no control over (genetic luck). With respect to education, our hands differ in intelligence, conscientiousness, and other aspects of personality. Though twin studies pin the heritability of educational attainment in Western countries at somewhere between 40% (for total years of education) and 60% (for GCSE exam scores), and GCTA pins the SNP heritability (disregarding rare variants, which aren’t included in typical SNP-chips and probably explain the discrepancy between SNP-based heritability estimates and twin study-based heritability estimates) anywhere between 20% and 40%, there’s a lot of wiggle room left, even if much of the remaining variability is due to non-shared environment. Because of this, we may be able to address achievement gaps with better social programs that target these disparities in shared environment. Genetics will aid us in doing so, by identifying the points in development when intervention will be most useful.

That said, the SNP heritability figures might be overestimates, since within-family heritability seems to be markedly lower than between-family heritability. This suggests the role of shared environment (including shared environmental effects that are passively correlated with additive genetic effects, called “genetic nurture”) might be more important than previously believed. That is, the polygenic predictors from molecular genetic studies might be capturing, for example, gene-environment correlations, because those with smarter parents not only inherit their parents’ genes but the home environment their parents create. 

Our current, rudimentary genetic predictors of educational attainment are useful at the population level, not the individual level. On average, individuals with higher polygenic scores for educational attainment tend to be more economically mobile regardless of childhood socioeconomic status, for example. As stated in a paper Harden was co-lead author on, “Integrating genetic information in educational research also has the potential to introduce a novel source of data for researchers and policymakers interested in estimating schooling effects on the distribution of student outcomes.” For example, we might be able to use school-wide PGS heritability figures to ensure as level a playing field as possible; this will maximize the chance that all students reach the basic minimum level of skills (inclusion) and that family background won’t be an impediment to educational success (fairness). 

Equality of Outcome and Opportunity

I agree with most of what I stated in my recapitulation of Harden’s views on behavioral genetic research. I hope she’d agree, too. However, our intuitions differ on how this research can and should be applied to the goal of equity. Here I’ll outline the doubts I have.

The inclusion clause of the OECD definition of equity stipulates that all students reach at least a basic minimum level of skills; in other words, there must be a minimum level of equality of outcome. The fly in the ointment is that many individuals are simply not capable of meeting these basic minimum levels of skill, no matter how many resources we pour into them.

For example, consider the work of the National Assessment of Educational Progress (NAEP), a US Department of Education project that since 1971 has been putting out The Nation’s Report Card, “the largest ongoing assessment of what U.S. students know and can do.” This report card finds that in 2017, among public school students, only 39% and 35% percent of 4th graders were at or above the NAEP Proficient level for mathematics and reading, respectively. The outlook looks equally bad if not worse for 8th and 12th graders. And the standards aren’t especially high, as sample questions for each proficiency level (basic, proficient, and advanced) show.  

But suppose we relax our criteria and look at students who are at or above the lowest of NAEP standards in math and reading, the Basic level. 19%, 30%, and 40% of public school students in Grades 4, 8, and 12 fail to meet these math standards; for reading, the percentages are 32%, 25%, and 29%. 

So, a basic minimum of educational attainment is untenable as a benchmark for justice, or ‘equity’, unless we decrease our academic standards to accommodate a significant proportion of the student population.

The fairness clause of the OECD definition of equity, on the other hand, seems much more reasonable: the removal of personal and family-background obstacles. We might call it “equality of opportunity.” This seems like a desirable goal. But is it possible? Unfortunately, no. The idea of equality of opportunity is not as quixotic as the idea of equality of outcome but it is untenable nonetheless. This is for a couple of reasons.

The first reason is that the genetic lottery places an upper limit on the extent to which we can achieve equality of opportunity. To quote the behavioral geneticist Robert Plomin:

“[G]enetic inequality…leads to an inherent inequality of opportunity. That is, children dealt a lucky genetic hand have a better chance of doing well at school and getting a better job and making more money. This inequality in outcome is not going to be tackled indirectly through the educational system. As mentioned, if all children were taught exactly the same, their genetic differences would still lead to differences in their achievement, which would lead to differences in occupational outcomes.”

Blueprint, Robert Plomin

The second reason true equality of opportunity is impossible is that we don’t have unlimited educational resources, so we can’t realistically give any individual all the help and time they need. At the elementary and secondary level, total US public education spending amounted to $700 billion (nearly $14,000 per student) in 2015-16. Bryan Caplan and others have argued that much of this spending is bloat. Yet even if you don’t buy that argument, it seems that throwing more money at the problem won’t solve it, considering how much money is already being spent.

US Educational Reform: Attempts to Achieve Equality of Opportunity

One might retort that we don’t know that equality of opportunity is impossible since we’ve never really attempted it. However, this is largely wrong. Efforts to help the worst students improve their academic performance—i.e., our best approximations of equality of opportunity—have been numerous, yet all these efforts have failed. 

In the US, they began with the Elementary and Secondary Education Act (ESEA), signed into law by President Lyndon B. Johnson in 1965, as part of the War on Poverty. Title 1 of this act disbursed money to school districts with a high proportion of disadvantaged students, in an attempt to close achievement gaps. Unfortunately, the ESEA didn’t achieve its intended effect (if it had, we likely wouldn’t be talking about equality of opportunity right now). 

The NAEP only began collecting data on average mathematics test scores disaggregated by parent’s highest level of education in 1978, so it’s slightly difficult to determine the effect of the ESEA on the achievement gap. Fortunately, the NAEP bins the test score data into three age groups—age nine, thirteen, and seventeen—so we can look at members of the oldest bin, who would have entered elementary school in 1967, two years after the ESEA came into effect. In 1978, the difference between the average math test scores of seventeen-year-olds whose parents graduated college vs. those whose parents didn’t finish high school was 37 points, equivalent to more than a whole standard deviation difference. By 1996, the mean difference was essentially the same (p. 74-75). The data on reading scores goes all the way back to 1971, where the picture is similar (p. 125-126). 

The other notable wide-scale attempt to close achievement gaps was No Child Left Behind (NCLB) Act, a reauthorization of the ESEA enacted in 2001 under President Bush. NCLB threatened and imposed sanctions against low-performing schools that didn’t demonstrate year-on-year improvements in test scores. Unfortunately, NCLB failed. Its impact was nil at worst, and small at best (though only for math scores). Even the US Department of Education admits that the excessive emphasis placed on standardized tests under NCLB had perverse effects. 

This isn’t to say no progress has been made: from 1990 to 2017, the 4th-grade average NAEP math scores increased from the Basic level to almost the Proficient level. However, the average reading score has barely improved during the same period. But lest we attribute these gains to NCLB, the majority of mathematics gains were made from 1990-2003, before NCLB came into full effect. Worst of all, it seems that the highest performing students are the ones improving the most, both in reading and math, whereas the lowest-performing students’ floundering only worsens. That is, the gap between the best- and worst-performing students continues to widen.

Deconstructing the Debate

Though disheartening, none of these findings on the failures of educational interventions are terribly surprising given what we know about the behavioral genetics of educational attainment and intelligence.

That said, even if many of the premises of the OECD definition of equity are demonstrably false, equality of opportunity still seems like a worthy goal. For even if educational resources are scarce and true equality of opportunity is impossible, we should try to minimize the unfairness we have control over by allocating educational resources most effectively. And even if the ESEA and NCLB failed to improve the average test scores of the disadvantaged, this doesn’t mean these policies didn’t aid precocious children on the margins.

In an ideal world, we’d have equality of opportunity modulo genetic inequality: the bright kid would get the resources he or she needs, irrespective of his or her parents’ income. The crux of the disagreement between Harden and me is how this should be done and what we’re willing to do it at the expense of. In other words, what sorts of tradeoffs do we want to make, and at what point have we gone too far trying to create an equality of outcome?

Genetics research and the history of educational interventions considerably constrain the space of reasonable answers to these questions. That said, there’s still room for difference of opinion on the empirical and ethical assumptions underlying our answers to these questions. Here I’ll lay out the substantive dimensions along which these assumptions can differ.

Empirical Dimensions To Disagree On 

Human Capital Development

How much of education is about developing real human capital and how much is merely signaling? How large are the positive externalities from education?

Utility and Tractability of Broad, Scalable Educational Intervention

Is variation in educational attainment due mostly to genetic variation, shared environmental variation, or non-shared environmental variation? Given our answer to this question, how useful/tractable do we believe educational interventions will be? 

If we are able to improve educational outcomes, are these improvements psychometrically valid? That is, are we merely teaching to the test or are we actually improving intelligence? (Is this even an important question to be asking?)

At what point do we reach diminishing returns to education? What constitutes utility, the dependent variable, in this analysis?

Does there exist some unknown educational intervention that reliably improves educational outcomes and/or intelligence? Will throwing money at the problem make things better or are the prospects truly as grim as they seem?

Utility and Tractability of Educational Acceleration for Bright Students

How well does educational acceleration work? Is education particularly useful for kids on the far right of the ability distribution? 

Will Fixing Education Cure Underlying Inequality?

Do you believe in “educationism”—the idea that social inequality is due to differential access to education—or do you believe the problem runs deeper?

How Pressing is the Problem of Automation? 

Are “robots really taking our jobs” and is “software eating the world”? Is this trend going to accelerate? Will this trend disproportionately affect those with little education and prospects of retraining for a new job?

Ethical Dimensions to Disagree On

Discounting Rate of Future Lives

Should we prioritize economic growth and improvements in quality of life for all over short-term inequality; i.e., how heavily do we discount the value of future human lives? (This is a question of population ethics.)

Ethical Values

Is fairness/justice more important than average well-being?

The Value of Education

Is education merely an instrumental good or does it have inherent value (e.g., moral edification, intellectual fulfillment, etc.)?

Will education be able to prepare students for the jobs of tomorrow? Do we even know what those jobs will be?

Educational Austerity

How much should we cut the educational budget? Where do we situate education in our list of priorities?

Three Visions for K-12 Public Education

Here I situate three visions—two commonly held and one novel—along the dimensions I’ve outlined. 

The Egalitarian Vision

Let’s get unrealistic views off the table. Many people, who are unaware of behavioral genetics research, believe some element of what we can call “the egalitarian vision.” This vision states that, on the empirical front, intelligence is completely malleable and the achievement gap will be closed if we use enough educational resources, and, on the ethical front, that we should attempt to redress educational inequality at all costs. Charles Murray called this ideology “educational romanticism…The lie is that every child can be anything he or she wants to be” (Real Education, p. 11). Anyone who proffers these sorts of views is the ideological suicide bomber being invited to dinner: they don’t deserve a place at the table because they won’t engage with the empirical evidence, stifling fruitful conversation. So, we needn’t give this scheme any more thought.

The Rawlsian Vision and Noble Lies

What I’ll call the “Rawlsian vision” attempts to reconcile egalitarian ethics with realism about behavioral genetics. One might also call this luck egalitarianism. According to luck egalitarianism, we should create institutions that strive to minimize opportunities and power that come from undeserved advantages, such as genetic endowment or other happenstance of birth (like being born in a first-world country). Per Rawls, this type of fairness constitutes a just world. 

Where Rawls’ theory of justice becomes interesting is in its second principle, the principle concerning social and economic inequalities. The first part of this principle—that people of equal ability should have the same educational and economic opportunities—takes into account varying genetic endowments and makes good sense. However, the second part of this principle, the difference principle, which states that social and economic inequalities must be of greatest benefit to the most disadvantaged members of society, smuggles in certain normative assumptions: namely, that we ought to minimize inequality at the expense of other values, such as maximizing total utility. 

I’d ascribe the Rawlsian vision to Paige Harden. Robert Plomin seems to subscribe to this sort of view, too, though he seems to believe genetics play a stronger role than Harden does, asserting that education won’t address inherent inequality of opportunity:

“My value system suggests that we need to replace meritocracy with a just society…economic inequality could be dealt with directly through a redistributive tax system that reduces the gap between rich and poor.”

Blueprint, Robert Plomin

In this excerpt of Blueprint, Plomin doesn’t flesh out what sort of policy prescriptions follow from genetics “because policies depend on values.” His one interesting idea on this front is that “instead of genetics being antithetical to equal opportunity, heritability of outcomes can be seen as an index of equality of opportunity.” I agree wholeheartedly with this idea and will get to its implementation later.

However, in Plomin’s earlier book on genetics and education, which he co-authored with Kathryn Asbury, G is for Genes, he offers some rather anodyne policy suggestions such as:

Offer free, high-quality preschool education to disadvantaged children from age 2, free, high-quality preschool education to all children from age 3 to 4, and extra support to children in low-SES families from birth.

G is for Genes, Kathryn Asbury and Robert Plomin (p. 170)

Unfortunately, preschool interventions like Headstart seem to have no lasting impact on cognitive abilities (as measured by test scores) and only a small positive impact on non-cognitive abilities like social skills, self-control and educational attainment, though meta-analysis suggests publication bias in the literature.

“However,” Plomin says, “a focus on developing a growth mindset, IQ, social, and thinking skills, and self confidence, would seem to be a good idea” (170). 

These ideas don’t hold water, either. Though the effects of growth mindset are mixed in studies of children and adolescents, they tend to be small. In a large sample of university students, growth mindset wasn’t significantly related to scholastic aptitude. In a randomized control trial of a goal-setting intervention for first-year undergraduates, the treatment had no effect on GPA, course credits, or second-year persistence. And the most recent replication attempt of growth mindset training flopped spectacularly.

Clearly, Plomin understands behavior genetics; as he has been working on the Twins’ Early Development Study for 25 years. Yet, for whatever reason, he chooses to tell noble lies (or, in the case of Blueprint, chooses to stay silent about the policy implications of his research). 

Radical Meritocracy: A Pessimistic, Utilitarian, Care- and Fairness-Insensitive Take on Education

The responsibility for the creation of new scientific knowledge—and for most of its application—rests on that small body of men and women who understand the fundamental laws of nature and are skilled in the techniques of scientific research. We shall have rapid or slow advance on any scientific frontier depending on the number of highly qualified and trained scientists exploring it. 

Vannevar Bush, Director of the Office of Scientific Research and Development, July 1945

Here I’ll make a case for the empirical and ethical premises I believe are justified and what the best possible education system would look like given these premises. In his 2008 book Real Education, Charles Murray outlines “Four Simple Truths” which are wholly consistent with my argument (p. 12-13):

  1. Ability varies.

  2. Half of the children are below average.

  3. Too many people are going to college.

  4. America’s future depends on how we educate the academically gifted.

I hope to both build on Murray’s claims and root them in deeper ethical and empirical foundations, paying particular attention to the fourth claim.

I’ll call this system radical educational meritocracy (which I’ll henceforth refer to as “radical meritocracy”). The term meritocracy was coined by Michael Young in his 1958 science-fiction satire, The Rise of the Meritocracy. Many consider the book’s vision to be that of a dystopia, demonstrating the folly of an administrative state’s attempt to use intelligence testing to optimize collective performance. However, if radical meritocracy were executed properly, something which is becoming more feasible, I believe it would maximize collective utility over the long-run.

Motivations Underlying Radical Meritocracy

Radical meritocracy is motivated by a sense of urgency about the following two trends:

  1. The increasing threat of automation and attendant “technological unemployment,” a term Keynes coined nearly 90 years ago. Automation will likely obsolesce a substantial portion of our workforce in the coming decades, deepening the economic and social fissures between the highly educated and everyone else, causing massive social disaffection.
  2. Growing existential risks to humanity, which are magnified every year by our increasing technological capabilities.

Here I’ll quickly address why we should feel a sense of urgency about these two issues:

Automation

Though the U.S. unemployment rate is at a record low, 3.6%, the labor force participation rate has been steadily declining since the mid to late 90’s. From 1979-2007, employment rates fell most for males who dropped out of or only graduated from high school. And though we’re currently in the midst of a jobs boom, one in which weekly nominal earnings have been increasing for the lowest earners, many of the jobs being created are low-skill or temporary jobs (which few people want to do). The prospects of job reskilling programs are dismal—so it seems the truckers and journalists won’t “learn to code” any time soon—and for the majority of workers, automation has become more a force of labor displacement than of labor augmentation. Whatever the cause of these trends, the low-skilled, poorly educated worker is taking the brunt of it, and a substantial proportion of white-collar workers might soon be as well. Average will soon be over. That said, it bodes well for those who perform cognitive, non-routine work and skilled manual trades.

Existential Risk

For a variety of reasons—the existence of the Fermi Paradox, our close brushes with nuclear war in the second half of the 20th century, the ease with which bad actors can now develop dangerous bioweapons, etc.—we should be worried about existential risk.

The relationship between humanity’s technological capabilities and our risk of extinction is probably close to an inverted U-shape. We are on the left-hand part of the curve and quickly nearing the peak, where existential risk is highest; thus, getting over the peak and reducing existential risk should be a priority, provided we care about the possible lives of our future descendants. Leopold Aschenbrenner explains this well in his paper “Existential Risk and Growth”:

In a model of endogenous and directed technical change, with moderate parameters, existential risk follows a Kuznets-style inverted U-shape. This suggests we could be living in a unique “time of perils,” having developed technologies advanced enough to threaten our permanent destruction, but not having grown wealthy enough yet to be willing to spend much on safety. Accelerating growth during this “time of perils” initially increases risk, but improves the chances of humanity’s survival in the long run. Conversely, even short-term stagnation could substantially curtail the future of humanity.

Leopold Aschenbrenner

The way we get over the peak of the existential risk curve is by developing better institutions and technologies. Doing this requires nurturing innovative, highly intelligent individuals who will build such technologies and institutions. Our and our descendants’ future depends on them.

If we don’t build these technologies and institutions that allow us to crest the hill and reduce the ambient level of existential risk, given a long enough time horizon it seems the only alternative outcomes are stagnation (which is unlikely given our expanding technological capacities) or annihilation (i.e., we are hit with an existential risk which either wipes out humanity or reverts us back to an earlier technological age). Neither of these outcomes is desirable, especially the latter, so we should highly value getting over the hump of the existential risk curve.

Empirical and Ethical Premises

On the empirical front, my other premises are: education is a zero-sum, positional good mostly about signaling and selection, not the development of real human capital; that said, educational interventions (like talent identification programs) are particularly useful for children on the far right end of the ability distribution, as these children go on to achieve disproportionate creative and professional success compared to their peers; for the majority of students, after teaching basic skills, we quickly reach diminishing marginal returns to education; variation in educational attainment is mostly, though certainly not completely, caused by genetic variation, not variation in access to educational resources; given the behavioral genetic research and the repeated failure of educational interventions to demonstrate long-term, reliable improvements in educational outcomes, it’s most parsimonious to think we’ll never discover a scalable intervention that fully closes achievement gaps between the best and worst students; finally, educationism is wrong—the relationship between education and social inequality is largely mediated by genetic variation in traits that predispose one towards economic mobility, not family background privilege.

On the ethical front, my premises are: we assume an indirect version of consequentialism, which means that we assume we should create institutions that tend to maximize social welfare over the long-run; we ought to only minimally discount future lives and therefore should radically expand the temporal scope of our thinking; this means that we should prioritize economic growth over short-term inequality, because this increases technological development, getting us over the peak of the technological-development/existential risk curve; this strategy maximizes the likelihood our species reaches the 22nd century and that our descendants exist at all; finally, education is an instrumental good, one which most people are unable to make use of, not an intrinsic good.

Fully fleshing out all of these premises is beyond the scope of this essay. Nonetheless, given these premises, contra Plomin and Harden, I believe we should embrace radical meritocracy even at the expense of short-term inequality and injustice.

What is Radical Meritocracy?

Radical meritocracy takes on board the Rawlsian idea that we should make choices about social systems (such as what constitutes educational equality of opportunity) as if we were behind a veil of ignorance, not knowing where we’d end up. But behind the veil, radical meritocracy permits greater variance in life outcomes if it means that total utility is higher, which flies in the face of the difference principle of Rawls’ second principle of justice. Radical meritocracy also takes the long-term view, considering the utility of those currently alive and of our yet-to-be-born descendants (that is, it is not a person-affecting view), meaning that behind the veil of ignorance one doesn’t know if they’ll be born in the 21st century or will be one of our (possibly existent) descendants living in the 31st century, colonizing the galaxy. This requires rejecting the luck egalitarians’ preoccupation with short-term inequality in favor of focusing on long-term gains in quality of life for all.

Radical Meritocracy in Practice

Under our current K-12 system, a child’s access to educational resources is largely a matter of chance, and not correlated with their innate ability. Poor, smart kids tend to end up in the low-quality public schools near their house. Rich kids more often get sent to private schools, even if they’re a bit dull. Yes, highly competitive specialized schools like Bronx High School of Science and Stuyvesant High School (both of which are more selective than Harvard, accepting 3% of applicants each) allow the poor but bright kids to get ahead. But these schools are currently few and far between. Under a system of radical meritocracy, there would be many more such competitive specialized schools. However, this is just the tip of the iceberg.

Highly competitive public schools like Bronx Science and Stuyvesant work because they ruthlessly select for intelligence via standardized tests. The key question is if we can identify intelligence (and other elements of academic potential like creativity, curiosity, etc.) before high school. This is a signal detection problem and one of the two focuses of radical meritocracy. We’d like to nurture those rare students who have the potential to make a massive positive impact, but first, we must find them. 

One way would be to use age 13 test scores on the SAT, as is done in talent search programs like the Study of Mathematically Precocious Youth (SMPY) and the Duke Talent Identification Program (TIP). I’ll expand on this later. 

Polygenic scores present another way of identifying these rare students. Though it might seem like something out of the movie Gattaca, the heritability of intelligence is low during childhood and slowly increases till peaking in late adolescence, so it’s conceivable that we may be able to use polygenic scores (PGS) early in life when PGS are more predictive of future educational attainment than noisy test scores.

There’s a massive caveat, though: these PGS aren’t currently useful at the individual level, as they have little to no incremental validity over knowledge of a student’s prior educational attainment and little precision at the individual level. Additionally, some of the signal the PGS picks up on is in fact noise—indirect parental effects on the phenotype of interest—meaning that the PGS is partly confounded by environmental effects. However, given the SNP heritability values derived from GCTA (which estimates the theoretical maximum variance our PGS could capture) and the rate of increase in the variance captured by PGSs, someday we might be able to use them to identify positive outliers. If this were possible, such PGS could be used to identify children with academic potential and place them in gifted programs, for example.

The second focus of radical meritocracy, after detecting talent, is matching it with the resources it needs. In other words, we need to “merge an individual’s potential (abilities) and passion (preferences) with educational experiences tailored to each student’s unique promise (readiness to learn).” The key question in this domain is if schools and talent identification programs have a large, causal impact on student achievement. If they do not—that is, if their achievement is merely due to selection effects—then placing so much emphasis on acceleration programs is misguided.

The findings on the value added by elite schools are mixed. Some papers suggest that the higher average achievement in elite schools is due to selection effects, not the school itself. Other papers argue that school quality significantly impacts student achievement. The findings on various forms of tracking are mixed, too; for example, one study on full-grade acceleration (i.e., grade-skipping) only found a small to moderate impact on earnings, as compared to ability-matched, non-accelerated students.

But the findings on talent identification and acceleration programs like SMPY and TIP are much clearer. Such programs identify extremely precocious children and provide them with the academic stimulus and support that’s appropriate for their abilities and interests. These students go on to achieve above and beyond their peers. Even among these gifted children, increases in intelligence produces non-linear returns in creative and professional achievement (i.e., intelligence keeps on giving). Most interestingly, accelerative programs like SMPY and TIP have demonstrated small to large causal effects on academic achievement, including among minority students. For example, as compared to non-accelerated matched control students, SMPY participants who skipped a grade earned twice as many STEM PhDs, produced 55% more stem publications on average and earned over 50% more doctorates. Other talent identification acceleration interventions show similar results.

We shouldn’t be surprised by the SMPY and TIP findings on IQ and creative achievement. IQ is strongly linked to the probability of being an inventor, even after controlling for parental income. That said, parental income seems to have an effect on the probability of becoming an inventor, even after partialling out the effects of IQ, suggesting there exist “lost Einsteins” that we need to identify. This is the point of radical meritocracy: to identify and nurture these precocious yet underserved children, as doing so will maximize aggregate welfare. 

One may still have doubts about testing, though. For example, consider Campbell’s Law: “The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor.” Campbell’s Law manifested itself during the Bush administration in increased time spent on standardized testing in public schools that faced sanctions due to No Child Left Behind. When a school district’s livelihood depends on improving test scores, they’re going to spend a lot of time prepping for them. 

One might also fear that increasing the role of test scores in education might lead to a hyper-competitive educational system, like that in South Korea, one in which students study for dozens of hours per week. However, I believe these fears are misplaced. The reason why talent identification programs have been so successful is that students don’t study for the SAT as twelve- and thirteen-year-olds. Yet even if they did, meta-analysis and other studies suggest the gains on test scores to be had from studying, at least among those ages seventeen to eighteen, are quite small. Because the test is hard to game, it’s a good indicator, one resistant to Campbell’s law. 

Navigating the Tradeoff Landscape

From a cost-benefit perspective, one important detail is how our signal detection devices (test scores, PGS, etc.) should be calibrated; for example, how many true geniuses are we willing to miss (false negatives) in order to maximize the number of educational resources given to bright kids? Obviously, children don’t fit into the binary of bright or dull—academic potential falls along a continuum—but it’s a useful toy model for thinking of how to allocate scarce resources.

Suppose we have 1000 units of educational resources and 1000 students to educate. If we could reliably identify outliers on the far right tail on the ability distribution—whether via standardized testing or PGS—would we be justified in giving them more than 1 unit of resources, thus taking away from other students? 

If the goal of our radical meritocracy is to maximize innovation in order to improve quality of life for all, and if you buy my premise that the marginal utility of education is substantially higher for the bright child than for the dull child—because there is a threshold effect for innovation; because children who do poorly in school tend to not like school anyway; and because improving test scores among poorly performing students has proven incredibly difficult—then it makes sense to spend these limited resources on the brighter children rather than the laggards.

I’m certainly not claiming we ought to cut all education spending on students in the X-th percentile of intelligence and below. I’m merely stating that a more optimal allocation of education spending is possible, and we’ll get the most bang for our buck (in terms of technological innovation and long-term improvement in our collective quality of life) by focusing our efforts on the extremely smart kids.

You might be asking yourself: what do we do with the overwhelming majority of students who aren’t selected into a talent identification program or a highly selective, specialized high school? I’ll leave this one to Plomin and Asbury:

Increase the number and range of options available for work- and college-based vocational training; make apprenticeships more affordable for and attractive to employers, and educate pupils so they have mastered basic skills, found their true interests, and are more attractive to employers.

G is for Genes, Kathryn Asbury and Robert Plomin (p. 174)

In Germany, for example, nearly ⅓ of students pursue vocational training. Beyond vocational training, though, I think a radical reduction in public education spending and a renewed focus on reading, writing, and arithmetic (plus increased focus on other important topics, such as how to be in a relationship, basic Western history, etc.) would be the best course of action.

Dimensionality Reduction of Intuitions Around Radical Meritocracy

My radical meritocracy scheme is lacking in specifics and leaves many questions unanswered. There are many topics to touch on: school choice (vouchers vs. public), how K-12 education relates to higher education (and the pernicious signal inflation that is occurring), how to change pedagogy, what to do about all the bright “learning-disabled” children that fall through the cracks of our educational system, what to do with the students not on the right tail of the ability distribution, etc.

You might react to my vision of radical meritocracy with disgust. You might find it callous. This is understandable. It’s a highly counterintuitive suggestion, predicated on ethical assumptions that we did not evolve to intuit easily (viz. utilitarianism—as opposed to thinking in terms of fairness—and extreme counterfactual and long-term thinking—as opposed to thinking parochially and falling prey to status quo bias).

And you needn’t agree with my scheme; there exists a range of reasonable views between the Rawlsian take and radical meritocracy, depending on one’s opinions about the aforementioned empirical and ethical assumptions.

However, I bet that beliefs about each of these assumptions aren’t orthogonal. In fact, I imagine there’s quite a strong coupling among these dimensions, and knowing one’s opinion on one will allow us to reliably predict their opinions on others. This is because people’s beliefs on any issue typically aren’t motivated by pure reasoning; rather, most people arrive at their beliefs via intuitions and then justify them with post hoc reasoning.

If these dimensions are highly coupled, then we should be able to reduce them to fewer dimensions. Jonathan Haidt’s Moral Foundations Theory provides a potential means of doing this dimensionality reduction. I believe the primary ethical dimension— the first principal component or factor of these intuitions, so to speak—correlates strongly with the care vs. harm and fairness vs. cheating moral foundations. Individuals high in this care and fairness factor tend to favor an ethical framework that emphasizes fairness and equality; believe education is not just an instrumental good, but a good in and of itself; and are optimistic about the malleability of intelligence and our ability to improve educational outcomes. These folks, whom others have called “equalitarians,” likely skew politically left.

Conversely, the “utilitarian realists” (which include the radical meritocrats) are low in this care and fairness factor; believe we should not heavily discount future lives; are willing to unequally distribute educational resources based on ability, if it means that it will improve economic growth; and are skeptical of educational interventions working.

Whatever your ethical and empirical beliefs, it seems any education system rooted in the best information from behavioral genetics involves trading one form of inequality—that of family background—for another—genetic inequality, or inequality in the ability to develop skills the job market will find valuable. Under radical meritocracy, we do so to maximize collective utility and the probability that our species survives through the 22nd century (and does so with ever-increasing quality of life). 

Radical meritocracy isn’t fair, in the traditional sense. But it’s fair (and, in my opinion, desirable) because it generates positive externalities by matching smart kids with resources, which increases the potential for innovation and economic growth, benefiting humanity as a whole in the long-term. Perhaps counterintuitively, radical meritocracy is very Rawlsian in this respect:

All social values—liberty and opportunity, income and wealth, and the bases of self-respect—are to be distributed equally unless an unequal distribution of any, or all, of these values is to everyone’s advantage.

Justice as Fairness, John Rawls (p. 9)

But why not maintain the educational status quo? Philosopher Evan Williams has suggested that we may currently be engaged in various moral catastrophes that we’re completely unaware of; education may be one such catastrophe. A potential solution, per Williams, is “refocusing our educational system on identifying the most talented children and nurturing them to their full potential, not on trying to shepherd every last mediocre student into a four-year college” (9). If we truly care about the long-term future of humanity and believe that technological innovation is the means of achieving such a future, then we ought to take this idea seriously. Forestalling scientific progress (and the development of better human institutions) could result in the counterfactual deaths of trillions of lives, squandering huge amounts of utility.

The Educational Inequality Treadmill

Discussing the progressive conversation about racism in the United States, Coleman Hughes writes: 

The disparity fallacy and the denial of cultural factors conspire to create a dynamic that I call the Racism Treadmill: as long as cultural differences continue to cause disparities between racial groups, and as long as progressives imagine that systemic racism lies behind every disparity, then no amount of progress in reducing systemic racism, however large or concrete, will ever look like progress to progressives.

We can apply this same principle to disparities in educational outcomes among the rich and poor, which we might call the Educational Inequality Treadmill. 

As long as [genetic differences] continue to cause disparities [in educational attainment] between [the rich and poor], and as long as [egalitarians] imagine that [lack of equality of educational opportunity] lies behind every disparity, then no amount of progress in [improving equality of educational opportunity], however large or concrete, will ever look like progress to [egalitarians].

I am certainly not claiming that the sort of equality of educational opportunity I’ve argued for currently exists—there’s clearly room for improvement—or that all disparities in educational attainment are due to genetics. However, I am urging those who sympathize with the egalitarian or Rawlsian take on education to consider the point at which they will cease their efforts towards equality. What does the end game look like given what we know from behavioral genetics about the natural distribution and mutability of intelligence and other education-related abilities? At what cost are they willing to pursue their goals of equality?

Though Harden and I largely agree on the empirical research, we disagree on what to do in light of it. But if there’s anything we can agree on, it’s that genes do not determine one’s moral worth, though they do constitute a form of privilege, a privilege that should engender compassion for those who didn’t receive the same genetic luck

I’ll leave open the question of whether the genetically lucky have a responsibility to give back to those less fortunate—either via redistributive taxation or some other means—because as I’ve stated, we should quibble less about short term inequalities and focus more on the long-term future of our species. This isn’t meant to diminish those among us being left behind in this rapidly changing world. It’s merely to say that improving the quality of life for all human beings will, in the long run, do more to improve their plight than will palliative noble lies or finger-pointing.