How one man’s witch hunt made racist pseudoscience mainstream… again
The Nathan Cofnas Story
A Manufactured Authority
Over the past month, an American postdoctoral researcher at Ghent University in Belgium has become one of the most quoted men in British public life. Nathan Cofnas built his current platform on a single achievement: publicising plagiarism allegations against Professor Jason Arday, the youngest Black professor in Cambridge’s history, in the weeks before Arday resigned and was found dead at his London home. Since then, Cofnas has done the rounds of newspapers, podcasts and talk shows, treated by almost everyone who interviews him as a credentialed expert delivering inconvenient truths about race and intelligence.
He is nothing of the sort. Cofnas is a philosopher, not a geneticist and not a psychometrician, and he is not, by any serious academic reckoning, a figure held in good standing within his own field. His signature academic contribution, a 2020 paper in the journal Philosophical Psychology titled “Research on group differences in intelligence: A defense of free inquiry,” was met on publication by a formal letter from nine academics who did not consider it worth a reply, only a demand that the journal explain why it had printed it at all. One of the journal’s own editors resigned over the affair. None of that history has made it into the coverage currently giving Cofnas a platform to argue, in essence, that Black people are less intelligent than white people and that science is being suppressed from proving it.
It is worth setting the record straight, on the paper and on the man.
The Paper: Free Inquiry as a Cover Story
Cofnas’s central claim is that a moral squeamishness among scientists and philosophers is preventing the world from confronting a genetic explanation for group differences in IQ. The framing does a great deal of work for him. It allows him to present himself as a martyr for open enquiry rather than as someone whose sources simply do not hold up.
The tell is in his own text. Cofnas repeatedly hedges his claims with words like “likely” and “partly,” yet builds an entire argument for urgency on the back of those hedges. He also concedes, in the same breath as his central thesis, that there is more genetic variation within any given racial group than there is between racial groups. That single admission dismantles the entire project. If the variation inside a population dwarfs the variation between populations, chasing a racial average is a statistical exercise in futility, not a scientific discovery waiting to happen.
Cofnas also narrows his comparison almost exclusively to Black and white populations. He does not ask why, by his own logic, university lecture halls are not disproportionately Chinese or Japanese, both of which outperform white cohorts on the same tests he treats as gospel. The selectivity is the argument. This is not a man following data wherever it leads; it is a man choosing the one comparison that flatters a predetermined conclusion.
What Britain’s Own Exam Data Shows
The most direct refutation of Cofnas’s thesis is sitting in the Department for Education’s own records, freely available and drawn from entire national cohorts rather than the handful of subjects that make up most of the “race science” literature.
Analysis of GCSE key stage 4 results broken down by pupils’ home language and ethnic background, covering hundreds of thousands of pupils in a single cohort, shows white British pupils achieving five or more good GCSEs including English and maths at a rate of roughly 57%. That places them close to the middle of the national table, not the top. Above them sit several Black African ethnic groups: pupils of Igbo heritage at around 73%, Yoruba at around 71%, a full fourteen to sixteen points ahead of white British pupils, with Luganda, Amharic and Twi-speaking pupils also clearing the white British figure. Meanwhile, several white ethnic groups sit at the very bottom of the same table: Czech, Slovak and Romani pupils scored in the low teens, a full forty points behind the Serbian and Bosnian pupils sitting near the top of the white cohort, despite the two populations being geographically and genetically about as close as any two European groups can be.
This is not a cherry-picked sample. It is whole-cohort, government-collected data, covering pupils across every kind of school and every kind of household. It shows precisely what genuine population-level testing always shows once socioeconomic circumstances are held roughly constant: performance scatters across every ethnic group, with no consistent racial hierarchy of the kind Cofnas requires for his thesis to mean anything. If race explained intelligence the way he claims, no plausible model would put Igbo and Yoruba pupils ahead of white British pupils while putting Czech and Slovak pupils, who are white by any definition Cofnas would recognise, at the bottom of the entire country.
The Guesswork Behind “Race and IQ”
Cofnas leans heavily on two sources: Richard Herrnstein and Charles Murray’s The Bell Curve, and Richard Lynn and Tatu Vanhanen’s Race and IQ, later revised with Vanhanen. Both have been picked apart by working scientists for decades, and the reasons why are worth spelling out plainly, because they are rarely explained to the audiences currently being fed this material as settled science.
Lynn’s national IQ dataset, the backbone of most modern “race realist” claims, assigns a single average IQ score to entire countries. Independent reviews of his sourcing found that over half of his national estimates were not derived from any testing at all. Where no data existed for a country, Lynn simply interpolated a figure from the scores of its geographic neighbours. Papua New Guinea is a useful case study, because its history in Lynn’s data is unusually well documented. Two small studies from the 1970s and 1980s, one of 152 teenagers using a general reasoning test and one involving children given a visual-motor coordination task called the Bender-Gestalt, were never designed to measure IQ or to represent the national population. Lynn converted their results into an “IQ-equivalent” score through his own devised formula and applied the result, drawn from fewer than 400 people, to a nation of roughly ten million spread across more than a thousand distinct ethnic groups and eight hundred languages. The resulting figure, hovering around 65, was then treated in the literature as a fixed national trait, despite field researchers who had actually spent time with Papua New Guinean communities describing them as unusually resourceful and quick-witted.
The same pattern recurs across the dataset: refugee children tested in a language foreign to them standing in for entire nations, sample sizes in the tens or low hundreds standing in for populations in the tens of millions, and studies never designed to measure intelligence repurposed after the fact to produce an intelligence score. Cofnas also applies his standards inconsistently. He discounts more recent studies narrowing the Black-white IQ gap in the United States on the grounds that they included children, while relying without complaint on Lynn and Vanhanen estimates gathered largely from children in Somalia, Papua New Guinea and elsewhere. A methodology is either sound or it is not; it cannot be sound only when it produces the answer you were looking for.
A simple exercise illustrates why none of this ever had a chance of being rigorous. Take a single UK secondary school of a thousand pupils and try to produce a meaningful average IQ from a budget that allows testing of thirty of them, a 3% sample, far larger than most of the national samples underpinning Lynn’s book. Test the six most able pupils in each year group and the average is inflated. Test six at random and there is no way to know whether the sample happens to have caught an unusually strong or weak cohort. Every method of selection introduces bias, and there is no way around it short of testing the entire population. If a 3% sample cannot yield a trustworthy average for a single school, a sample of a few hundred cannot yield a trustworthy average for a nation of tens of millions. The entire “national IQ” literature fails this basic test before a single genetic argument is even considered.
Genetics Says Nothing of the Kind
Cofnas leans on the geneticist David Reich, quoting him to argue that the genetic separation between human populations is old enough and deep enough to plausibly account for cognitive differences. What Cofnas omits is that Reich has also stated, in the same broader body of work, that population geneticists went looking for distinct, separable human races and did not find them. What they found instead was what Reich has described as a “braided trellis”: a continuous history of migration, contact and interbreeding between populations, with no clean, long-isolated branches of the kind racial hierarchies require.
Cofnas’s genetic argument also depends on a sleight of hand between two very different kinds of DNA evidence. Multi-locus autosomal analysis, the method that lets researchers estimate which broad population a person’s ancestors are drawn from, works by comparing patterns of relatedness across large modern reference samples. It can tell you who a person is related to. It cannot tell you what an “ethnic group” is in any fixed biological sense, and it certainly cannot tell you anything about a non-physical, polygenic trait like intelligence, for which no reliable predictive genetic markers currently exist at the population level. Cofnas writes as though ancestry-inference technology and cognitive-trait genetics are the same field at the same stage of maturity. They are not.
The wider scientific consensus, formalised by the American Association of Physical Anthropology in a statement revised in 2019, holds that human populations have never been biologically discrete, isolated or fixed, and that patterns of genetic variation do not map cleanly onto socially recognised racial categories. This is not a fringe or a “woke” position within the field; it reflects what has been standard understanding in population genetics and physical anthropology since the completion of the Human Genome Project. Modern racial categories are, in large part, an artefact of a few centuries of imperial classification rather than an evolutionary reality that predates it, which is precisely why the genetic evidence keeps failing to find the clean divisions the theory requires.
Twins and the Heritability Trap
Cofnas’s paper leans heavily on twin and adoption studies to argue that intelligence is substantially heritable, and on that point he is not wrong in isolation. Identical twins raised together do tend to resemble each other in measured IQ more than fraternal twins do. But heritability within a population tells you nothing about the source of a difference between populations, a distinction the psychologist Arthur Jensen himself was careful to draw and that Cofnas’s paper glosses over. Two groups can each have highly heritable traits while differing from each other for reasons that have nothing to do with genetics, in the same way that two fields of genetically identical wheat will grow to very different heights if one is planted in rich soil and the other in poor.
Anecdote is not data, but it illustrates the point cleanly enough: identical twins, sharing one hundred percent of their genes and near-identical upbringings, regularly diverge sharply in academic performance, one placed in a top set and the other in a bottom set at the same school. If genetics alone determined outcomes to the degree Cofnas implies, this should be close to impossible. It is, in fact, common. The environmental variables that produce that divergence, confidence, motivation, the particular teacher a child gets, the way a setback is internalised, are exactly the kind of factors Cofnas’s paper waves away as an “X factor” that decades of research have failed to isolate, when the far simpler explanation is that no single X factor exists because dozens of them are operating at once, unevenly, on every individual.
Scientists Respond: “More Provocative Than Scientific”
The most damning verdict on Cofnas’s paper came not from journalists but from nine academics working in philosophy, biology and anthropology: Rasmus Rosenberg Larsen, Helen De Cruz, Jonathan Kaplan, Agustín Fuentes, Jonathan Marks, Massimo Pigliucci, Mark Alfano, David Livingstone Smith and Lauren Schroeder. Their open letter to the editors of Philosophical Psychology was addressed to the journal, not to Cofnas, and said so explicitly: they had no interest in inviting him into further dialogue. The paper, in their assessment, simply did not meet the standard expected of a peer-reviewed publication.
Their letter methodically dismantled the paper’s foundations. They pointed out that the absence of any engagement with the extensive literature on human genetic variation, including the 2019 AAPA statement, suggested Cofnas had either not read it or had chosen to ignore it. They identified what amounts to a straw man at the heart of his framing: the suggestion that the field avoids this research purely out of moral squeamishness, when in reality serious researchers have been actively working on the genetics of intelligence and the race and IQ question since at least Jensen’s 1969 paper in the Harvard Educational Review, and journals in the field publish on it regularly. If anything, they noted, the sheer number of citations in Cofnas’s own paper undercuts his claim that the topic is being suppressed.
They also went after the journal’s editors directly, quoting the editorial note defending publication on the grounds that Cofnas’s positions “deserve to be disputed rather than disparaged” provided they are backed by logical argumentation and empirical evidence. The letter’s response was blunt: none of it is backed by logical argumentation and empirical evidence, which is precisely the problem. Their closing argument was a utilitarian one that mirrors Cofnas’s own reasoning back at him. If resources are scarce, as he argues, then researchers are right to spend them on lines of enquiry that might actually produce something useful, rather than revisiting a hypothesis about racial hierarchy that has already been tested to exhaustion and found wanting, and whose closest historical precedent for being taken seriously by a state was the eugenics movement that preceded the Second World War. One of the journal’s editors, Cees van Leeuwen, resigned in the aftermath.
A Career Built on Controversy
Cofnas’s current media profile rests on an image of a lone truth-teller punished for his convictions. His actual employment history tells a more familiar story of an academic whose views have repeatedly proven incompatible with the institutions that have hired him.
He first attracted wide attention for a public dispute with Kevin MacDonald, a psychologist known for an antisemitic conspiracy theory holding that Jewish people act as a coordinated ethnic interest group. Cofnas, who is Jewish, argued against MacDonald not on the moral failings of the thesis so much as its empirical basis, countering that Jewish overrepresentation in elite fields is better explained by higher average measured IQ within that population. It was, in other words, the same hereditarian toolkit he would later apply to Black and white populations, deployed on a different group entirely, which rather undercuts the idea that his interest in group differences is a disinterested pursuit of scientific truth wherever it leads.
In February 2024, a blog post by Cofnas arguing that a genuine meritocracy would sharply reduce Black representation at elite universities, and that Black people would “disappear from almost all high-profile positions outside of sports and entertainment,” triggered a formal dispute with Emmanuel College, Cambridge, where he held a research associateship alongside a Leverhulme fellowship in the university’s philosophy faculty. The college terminated his affiliation that April, judging the post a rejection of the institution’s stated commitment to diversity and equity and a source of reputational harm. Cofnas, backed by the Free Speech Union and roughly £85,000 in crowdfunded donations, sued the college for belief discrimination. In January 2026, the court found that his views were a legally protected philosophical belief but ruled that the college’s decision to dismiss him was proportionate given the harm the post had caused; he lost the case and was appealing at the time of writing.
Months later, in March 2026, Ghent University in Belgium appointed him as a postdoctoral researcher, ostensibly to work on a project about the future of liberal societies. More than three hundred staff and students signed letters objecting to the appointment, and a petition against him gathered over two thousand signatures. Following Arday’s death and a wave of public interviews in which Cofnas discussed the affair, Ghent suspended him in August 2026 pending a disciplinary investigation, on the grounds that his public statements on race realism had strayed well outside the scope of the research project he had actually been hired to conduct. He has said he expects to be dismissed.
Taken together, that is two academic institutions, independently concluding within eighteen months of each other that his conduct had become incompatible with continued employment. It is the record of a man whose ideas keep generating the same reaction wherever he takes them.
The Arday Affair and the Cost of a Campaign
It was Cofnas who first raised plagiarism allegations against Professor Arday, publishing his claims to his own newsletter in July 2026. What followed was a scale of media attention that bore no relation to the scale of the allegations. Broadsheet titles that pride themselves on measured, non-tabloid coverage ran dozens of articles apiece in the weeks that followed, a volume of coverage that plagiarism allegations against white academics, even proven ones, have rarely if ever attracted. Arday resigned from his post on 5 August 2026, describing the scrutiny as more than any person should reasonably be expected to endure, and was found dead at his home in south London a week later. Police stated they had reason to suspect he had died of unnatural causes; a formal cause of death has not been confirmed at the time of writing, though the working assumption in most reporting is suicide. Arday’s family described him as having been subjected to a sustained campaign of abuse that became too much for him to bear.
Cofnas has publicly rejected any responsibility for the outcome, framing his own role as having simply exposed academic misconduct that others were too afraid to raise. That defence deserves to be tested on its own terms rather than dismissed outright, and reasonable people, including some who share none of Cofnas’s politics, have made the case that whistleblowing on genuine misconduct should not itself carry consequences. But that argument does not require anyone to accept the substance of Cofnas’s separate, and separately false, claims about race and intelligence, and it does not erase the plain disparity in how the press has treated this case compared with others involving white academics accused of the same conduct. In this instance, a formal university committee ultimately cleared the late professor of the plagiarism charge, yet major outlets continued to report and comment on the story as if guilt were established, rarely foregrounding or even mentioning the exoneration. Those are two different questions, and the second one is not remotely close.
Where This Leaves Us
None of what Cofnas has been given a platform to argue withstands contact with the evidence: not Britain’s own exam data, not the genetics he selectively cites, not the twin studies he leans on, and not the scholarly literature his own paper failed to engage with. Nine academics said as much in the plainest terms they could find. Two universities, on two countries, have separately concluded his conduct was incompatible with their institutions. None of that has stopped newspapers and interviewers from treating him as a credentialed authority on a subject his own field has already closed the book on.
If you have been affected by any of the issues raised in this article, support is available in the UK from Samaritans on 116 123, free and confidential, any time.
Leave a Reply