The Arday affair left one question unanswered by everything written about it. When a man opens somebody’s footnotes and finds that the citation does not support the sentence, is he tracking truth or tracking an enemy?
His defenders said scholarship. His opponents said fraud.
Judging by the evidence, the case for plagiarism by Arday is overwhelming. The meaning of the plagiarism combined with Arday’s Cambridge status is contested.
There is a control available.
Richard Lynn (1930–2023) built the national IQ datasets that underlie a large secondary literature. With Tatu Vanhanen he published IQ and the Wealth of Nations in 2002 and IQ and Global Inequality in 2006 in 2006, and he later worked with David Becker on updated estimates. The datasets assign an average score to each country and correlate those scores with economic performance. Lynn described himself as a scientific racist. Ulster University stripped his emeritus status in 2018 after student protests.
The current status of that work is contested. The European Human Behavior and Evolution Association issued a formal statement in 2020 discouraging use of the datasets and calling them unscientific. A campaign led by Rebecca Sear has been pressing journals to retract articles built on them, and is tracking well over a hundred such studies. Proceedings of the Royal Society B retracted a 2010 paper, “Parasite prevalence and the worldwide distribution of cognitive ability,” in March 2024, saying the underlying between-country IQ data had been criticized for inaccuracies and biases.
This is an essay about who is willing to check whose work.
Lynn’s conclusions are congenial to the people who defended him against the retraction campaign. If criticism of scholarship follows coalition lines, they should have defended the work along with the man. Many did not.
Nathan Cofnas is the most consequential of them, because he is the person who later checked Kevin MacDonald and then audited Jason Arday (1985–2026).
In 2024, responding to the campaign against Lynn, Cofnas argued in a comment thread under Steve Sailer’s post that the papers should not be retracted, that retraction belongs to fraud, plagiarism, or something equivalent to mathematical error, that the charge of biased practice against Lynn was true and beside the point at issue, and that the targeting was political. Then he turned on his own side. He warned that many hereditarians would feel an impulse to defend Lynn’s work, called that a mistake, and said it was what the campaigners wanted. He described the national IQ work as profoundly flawed and told his readers not to make excuses. In a 2024 essay, Cofnas said Lynn and Rushton produced flawed scholarship and often exercised poor judgment about their associations. He repeated the criticism in 2025, calling Lynn sloppy while maintaining that some of the findings survive.
Was Lynn politically targeted? In his view yes. Should the papers therefore be pulled? No. Was the work bad? Yes. Should his allies defend it because they like the conclusions? No.
That is the distinction which collapsed in the first week of the Arday affair, running in the opposite direction. There, the identity of the accuser was offered as an answer to the accusation. Here, the identity of the accusers is offered as a reason to reject the retraction while conceding the substance of the methodological complaint.
Jason Malloy is another example.
When Malloy launched Human Varieties in 2013 he described himself as a hereditarian who had spent years in that world and had defended James Watson. Then he opened Lynn’s references country by country. The Equatorial Guinea figure was the spectacular one. Lynn and Vanhanen had assigned the country a score drawn from a study of 48 subjects. The subjects had been tested in Spain. The estimate was eventually withdrawn. Working through “HVGIQ: Dominican Republic,” Malloy said the number of errors was disturbing him, found a cited UNESCO source that did not contain what it was said to contain, and questioned the inconsistent handling of achievement data and the omission of an unusually high Cuban result. He wanted a better database and set about building one.
That is the same method Cofnas later applied to Arday’s thesis, performed by a man on Lynn’s own side, thirteen years earlier.
Emil Kirkegaard was close to Lynn. Lynn helped place his first paper and invited him to a conference, and his 2023 eulogy treats Lynn as a mentor. His 2019 review of The Intelligence of Nations, “Solid numbers, missed opportunities,” raised measurement invariance, the question of whether the same test measures the same construct across very different populations, and criticized the causal analysis as making little use of the methods that would be needed to establish national-level causation. The eulogy repeats the point. He still thinks the program has been productive.
Heiner Rindermann accepts that cross-national cognitive measures track economic outcomes, and has said that Lynn and Vanhanen’s data had serious problems: unrepresentative samples, tests from different periods requiring Flynn-effect adjustment, missing countries filled by estimate, heterogeneous instruments, the Equatorial Guinea error. He then showed that international school-assessment data produced broadly similar relationships. He is a party to the retraction fight. Discover Education pulled a 2024 article of his, saying post-publication review found fundamental flaws in using student assessment data as a measure of cognitive ability. He disagrees with the retraction.
David Becker became Lynn’s collaborator and rebuilt the dataset with more standardized procedures, writing that earlier estimates contained ambiguities, errors and deviations, correcting further mistakes after publication when outside reviewers found them, and acknowledging the absence of conventional peer review.
Russell Warne’s 2022 assessment argues a middle position: enough validity to justify continued work, alongside substantial weaknesses, with the imputed scores as the worst of them. Where a country had no usable data, a figure was sometimes estimated from its neighbors, and those imputations correspond poorly to later country-specific measurements.
Steve Sailer conceded errors, questionable inclusion decisions and some personal bias, while arguing the broad findings replicate. Crémieux argues that newer datasets validate the estimates. Noah Carl reviews the dispute and states in the open that some hereditarians count among Lynn’s critics, naming Cofnas.
Behind all of it sits Jelte Wicherts and colleagues, who reviewed the sub-Saharan African literature systematically and argued, in their rejoinder to Lynn and Meisenberg, that Lynn’s inclusion and exclusion decisions were not systematic, that judgments about whether a sample was representative appeared to track the result the sample produced, and that the African estimate was therefore biased downward. Lynn and his collaborators disputed the reading. Carl, who accepts the general research program, has said that this particular criticism is valid.
So the world that produced Cofnas contains a man who audited the founding dataset and found the numbers did not match the sources, a protégé who published the causal weaknesses, a collaborator who rebuilt the file, and a spectrum of defenders running from qualified to unqualified.
What does that add to the Cofnas story?
The claim that Cofnas audits scholarship only when the audit injures political opponents is hard to hold once you have the 2024 and January 2026 material in front of you. He told his own coalition that its canonical figure was unreliable, while that coalition was under attack, at a moment when solidarity was the obvious move. Whatever else he is, he is not a man who has never paid a price inside his own tent.
Cofnas’s stated rule holds sloppiness outside the retraction threshold and plagiarism inside it. Lynn’s documented problems are selection, imputation, source handling, measurement and causal inference. The central allegation against Arday was copied text. On the face of the rule, the two positions cohere. But consistency of a stated rule and equality of its enforcement are different claims. Nobody has shown that Cofnas would spend three weeks running a right-coded scholar’s thesis through detection software, publish the passages, and press that man’s employer. In the Lynn case he declined to defend the methods while opposing any sanction. In the Arday case he initiated the process that ended in sanction. Both can follow the rule while the effort runs one way.
Find a case where a figure inside or adjacent to hereditarian circles faced a documented allegation of plagiarism or fabrication, and ask what Cofnas and the others did. If no such case exists, that absence is the finding.
Kenan Malik wrote in the Observer that Cofnas’s racial views settle nothing about whether the evidence was true. The Lynn material shows the same operation performed by the least likely people, on their own side, years earlier and out of public view. It can be done. People with strong political stakes in a conclusion have looked at the workmanship behind it and said the work was flawed. That is a higher standard than most of the commentary on Arday reached in three weeks.
The Lynn file also hands the left an argument. The strongest version of the case against the Arday coverage is that the intensity of scrutiny, the willingness to sustain it for weeks, and the choice of target were selected by politics even where the individual findings held. That argument survives even if every claim against Arday were sustained.
There is one more asymmetry. Malloy audited Lynn’s references and found errors and nothing happened to him. Cofnas told his own coalition its founder was unreliable and nothing happened to him. Cofnas audited Arday and now faces dismissal at Ghent. The difference lies in where the subjects sat. Lynn was a marginal figure whose institution had already withdrawn his title. Arday held a chair at Cambridge, a publishing contract, and the protection of a defamation firm that had already stopped one publication from running the story.
When you check the footnotes of a man the institutions have abandoned and nothing follows. Check the footnotes of a man the institutions are invested in and the consequences arrive, first from his lawyers, then from the police, then from your own employer. Whatever Ghent eventually says about its reasons, that is the pattern a young academic will read off the record.
In January 2000, Slate ran a week-long dialogue between the journalist Judith Shulevitz, the Lingua Franca editor Alex Star, and the evolutionary psychologist John Tooby (1952–2024). The subject was how to deal with fringe academics. The particular academic was Kevin MacDonald (b. 1944), whose trilogy on Judaism argued that Jews pursue a group evolutionary strategy at the expense of gentiles and that antisemitism is a rational response to it. MacDonald had just volunteered to testify for David Irving (b. 1938) in the libel case Irving brought against Deborah Lipstadt.
Shulevitz’s position was that a field which welcomes all ideas cannot then ignore the bad ones. Tooby answered with a taxonomy. You can denounce an idea as immoral, show it to be untrue, or ignore it. The three are close to mutually exclusive, and each carries a cost. He then stated the principle that governs everything that followed, in this affair and in the two that came after it. Inside a truth-seeking culture, his opinion that MacDonald was wrong counted for nothing unless he showed readers the reasons. Demonstrating that a claim is false, he wrote, is the most effective possible answer to it, better than calling it immoral. He added the corollary that ought to be printed above every campus controversy of the past decade: denunciation subverts refutation, because it implies you could not win a fair fight, and therefore that something in the silenced position must be true.
Steven Pinker (b. 1954) supplied the counterweight in a closing letter. Science generates a thousand bad ideas for every good one, and doing battle with all of them is beyond mortals. Engaging one is a tacit concession that it cleared a threshold of attention-worthiness. MacDonald’s had not, for reasons Pinker listed: the argument that Jews advance hypotheses because they are Jewish is ad hominem and self-sealing, and MacDonald had already announced in advance that Pinker would reject his ideas for being Jewish; the axioms about group selection and the genetic cohesiveness of ethnic groups run against large bodies of theory and data; the theses collectively produce an invidious portrait couched in disparaging language; and the argument lacks a control group and a comparison with alternative explanations such as Thomas Sowell’s account of middleman minorities.
Then Pinker wrote the sentence that has hung over the case for twenty-six years. He had not plowed through MacDonald’s trilogy, and he acknowledged the risk that this made him unfair to the arguments and insufficiently resounding in refuting them.
Tooby chose differently. He said he had dropped everything to help Lipstadt’s defense team, and he promised a careful refutation of MacDonald’s books, to appear under the Critical Eye section of his center’s website as soon as it could be drafted. He estimated a month or two.
It never appeared. Tooby died in 2024.
In 2001 David Lieberman published “Scholarship as an Exercise in Rhetorical Strategy: A Case Study in Kevin MacDonald’s Research Techniques,” which compared MacDonald’s representations of Jaff Schatz’s study of Jewish communists in Poland against the underlying text and charged selective quotation and omission. George Michael’s 2006 survey in the Journal of Church and State describes Lieberman’s work, reports that Daniel Kriegman produced a critique of roughly fifty pages and that Ken Jacobs prepared a manuscript of some three hundred pages cataloguing alleged errors, and records that Jacobs’s manuscript was never published and that Tooby’s promised refutation had still not appeared.
Before Cofnas’s review in 2018, the audits of Kevin MacDonald existed. They were scattered, partly unpublished, and mostly invisible outside specialist mailing lists, while the reputational verdict was total and immediate.
Tooby also supplied the number that makes the triage defensible. Working from citation indexes, he found that MacDonald’s three books had been cited by anyone other than MacDonald twice in seven years, three counting one he happened to know the indexers had missed. By the ordinary measure of academic impact, the books did not exist. Pinker and Tooby were not ignoring an influential body of work. They were ignoring an uncited one.
The measure was wrong. Citations track circulation among scholars. They do not track circulation among people who are not scholars and do not care to be. Over the following fifteen years The Culture of Critique found a readership on the racialist right, MacDonald aligned himself publicly with white nationalists, and the alt-right carried the book to an audience that no citation index counts. The refusal to engage became part of the case for the book. If the professors call him an antisemite and will not answer his evidence, the reader is invited to conclude they cannot.
Nathan Cofnas (b. 1988) put that argument to Michael Schulson of Undark in 2018. He said the establishment had rejected MacDonald and declared him unrepresentative of evolutionary psychology, and asked why nobody had pointed out what was wrong with the work. MacDonald was off the radar of mainstream thinkers, he said, and influential anyway. He had been interested in the books since he was a teenager, and an academic mentor had discouraged him from writing about them.
In March 2018, Human Nature published “Judaism as a Group Evolutionary Strategy: A Critical Analysis of Kevin MacDonald’s Theory.” Cofnas granted the question. He told Undark there was nothing wrong in principle with trying to understand why any group behaves as it does, and that certain questions should not be closed because they recall a history we dislike. He said MacDonald could be rejected for scientific reasons.
Then he opened the books. The charges were misuse of sources, a double standard applied to Jews and non-Jews, and a complicated explanation preferred where simpler ones were available. On the Frankfurt School, MacDonald had repeatedly treated the absence of an explicit condemnation of Jewish nationalism as evidence of approval of it. On Boas and his students, evidence that supported ordinary explanations was converted into evidence for the elaborate one. Jewish participation in liberal movements counted; Jewish participation in conservative movements did not.
Every one of those charges is checkable by a reader with a library card. That is the difference between the two operations. A denunciation invites an argument about the man. A source comparison creates a problem he has to answer.
He answered. MacDonald produced seventeen thousand words at Unz and on ResearchGate, disputing the readings and defending the passages. He ended by suggesting that Cofnas, who is Jewish, had a motive beyond scholarship, and he said the same thing to Undark: there is an ethnic motive. Asked about the double-standard charge, MacDonald explained that when he evaluates Jews he looks to see whether they are Jewishly motivated, and that you cannot find Jewish motives in non-Jews. Pinker had identified that circle in 2000 without reading the trilogy. Cofnas produced it as a quotation from the author in 2018.
The readership was the other surprise. Papers in the field ordinarily draw a few hundred downloads. Cofnas’s drew more than thirty thousand in its first fortnight and passed fifty thousand. Jane Lancaster, the editor of Human Nature, said she was astonished, that the paper had matched a year of downloads in a month, and that the worry about legitimizing MacDonald never came up. Not responding, she said, does not make it go away.
Set that beside three citations in seven years and the attention argument fails. The demand was enormous.
Edward Dutton wrote a defense of MacDonald’s hypothesis. Lancaster rejected it as unsubstantiated. Todd Shackelford took it for Evolutionary Psychological Science, saying he wanted a journal where riskier work appeared, and that if MacDonald is wrong we should see why he is wrong. Cofnas was one of the two anonymous peer reviewers on the paper defending the theory he had just attacked, and both reviewers recommended publication.
Robert Boyd, asked to comment on Dutton’s paper, said he would not touch the topic with a ten-foot pole, that the data are poor, the priors strong, and every discussion instantly moralized. Pinker read the Dutton paper only after Undark asked him to, then wrote to the editor to say he was disappointed and that MacDonald’s theory and Dutton’s defense were both extraordinarily weak.
The MacDonald case connects to Richard Lynn (1930–2023) through Cofnas’s own biography. Undark reported that Cofnas and Dutton met at the London Conference on Intelligence and that both had ties to Lynn’s Ulster Institute for Social Research, where Cofnas published a book in 2012 and later withdrew it. He was inside that world.
Three cases, then. MacDonald, admired in territory adjacent to Cofnas’s own, audited at the citation level and rejected. Lynn, the founder of the research program Cofnas works within, described to that program’s own adherents as unreliable. Jason Arday (1985–2026), celebrated by Cambridge and the progressive press, audited by the same method in July 2026.
The political location of the target moves through the full range. Cofnas has an auditor’s temperament, distrusts institutional reputation as a proxy for quality, and goes to the point where a claim touches its source.
Auditing MacDonald was safe in one direction, since the mainstream had already written him off, and it earned Cofnas standing with people who otherwise had no use for him. It was costly in the other, since the world that reads MacDonald concluded that a Jewish critic of MacDonald proved MacDonald’s thesis.
What the MacDonald case adds to the Arday case is the symmetry. MacDonald was classified before he was read. Arday was classified before he was read. In the first instance the classification was hostile and, on the evidence, accurate. In the second it was admiring and, on parts of the record, unearned. In neither instance did the classification tell anyone whether a given sentence matched the source it cited.
Which brings the argument to Ghent, where Cofnas was suspended on August 20 pending a preliminary disciplinary investigation, five months after rector Petra De Sutter defended his appointment on the ground that dissident opinions may exist and should be tested by other researchers.
The institution always has the same two options. Argue about the man who opened the book. Or open the book.
