Cognitive Distortions

Cognitive Distortions

How the culture wars came for Wikipedia’s articles about human intelligence.

Shuichi Tezuka
Shuichi Tezuka
23 min read

In June 2018, the book Modern American Extremism and Domestic Terrorism published a false and defamatory statement about a living person, copied from a piece of Wikipedia vandalism. The statement was a fabricated quote about Linda Gottfredson, a psychologist known for her opposition to racial affirmative action policies, which compared her views in this area to Nazism. In an article published in the Critic in October 2020, I reported that an original source for this quote does not exist, and that the vandalism had remained in the Wikipedia biography for nearly two years—long enough for an academic book to repeat the hoax.

This was not an isolated case. Wikipedia hoaxes about topics including recently deceased people, South American wildlife, Middle Eastern cuisine, and hair products have all come to be regarded as “accepted knowledge” if they remained on the site for a sufficient period of time, and in many cases the false information was repeated in reputably published books or newspapers. The influence of Wikipedia hoaxes has also extended to papers published in academic journals. For example, a Wikipedia hoax describing a fictitious medical condition known as “glucojasinogen” went on to be discussed in several medical journals as though it were a real condition.

Damaging as they may be, hoaxes by individual pranksters are usually removed from Wikipedia as soon as they are discovered. Something far more potentially harmful, both to Wikipedia and to public knowledge as a whole, is when several users work together across multiple Wikipedia articles to discredit a body of scientific research, and do not allow their misrepresentations to be corrected by any of the site’s other users. This is what has occurred over the past two years on Wikipedia’s articles related to human intelligence.

Intelligence and psychometry

The study of human intelligence falls within a broader field known as psychometry, which refers to the measurement of psychological traits. Intelligence research is among the most replicable bodies of research in the social sciences: while many areas of psychology have been affected by the replication crisis during the 2010s (including some other branches of psychometry), a 2019 paper states that within intelligence research, “there is no replication crisis about key empirical findings.” Human intelligence is also among the most socially important areas of psychology, as Quillette described in a 2018 article, because of the large impact that a person’s intelligence may have on his or her life.

Before 2020, Wikipedia’s articles related to psychometry and human intelligence were mostly consistent with this field’s published literature, although many of these articles were somewhat outdated because there have never been many Wikipedia users with the necessary knowledge and interest to keep them updated. Under normal circumstances, Wikipedia articles increase in quality over time as more people contribute to them. However, for reasons that will be explained, the recent trend in articles related to human intelligence has been for Wikipedia’s coverage to become steadily more divorced from its source material. (In this article, when I refer to Wikipedia, I am referring specifically to the English-language version of the site.)

Over the past two years, there has been a collective decision by several members of Wikipedia that “scientific racism […] has infiltrated psychometry” and that the field must no longer be trusted. This assumption is explained in an FAQ created in May and June 2021:

Psychometry is a field where people who advocate scientific racism can push racist ideas without being constantly contradicted by the very work they’re doing. And when their data did contradict their racist views, many prominent advocates of scientific racism simply falsified their work or came up with creative ways to explain away the problems. See such figures as Cyril Burt, J. Phillipe [sic] Rushton, Richard Lynn, and Hans Eysenck, who are best known in the scientific community today for the poor methodological quality of their work, their strong advocacy for a genetic link between race and intelligence, and in some cases getting away with blatant fraud for many years.

In my 2021 paper, ‘Cognitive Creationism Compared to Young-Earth Creationism,’ for the inaugural issue of the Journal of Controversial Ideas, I describe how a large portion of attacks on intelligence research—and on psychometry or behavioral genetics more broadly—arise from attempts to discredit the small sub-area of intelligence research that deals with differences between racial or ethnic groups. The FAQ item quoted above was created for Wikipedia’s “Race and intelligence” article, which has been the epicenter of the recent changes to how Wikipedia covers topics related to human intelligence. That article was the first to be changed based on this new set of assumptions, and the changes have subsequently spread outward to Wikipedia articles about a broader range of topics. The following five cases are not the only articles affected, but are examples of the overall trend.

Case 1: Recent human evolution

One of the more recent articles to be affected by these changes is the “Recent human evolution” article, which discusses how humanity has evolved over the past 50,000 years. This article is not obviously related to either psychometry or race, so it is an example of far-reaching influence of the assumptions originating from those topics.

In March of this year, around six paragraphs of text were removed from this article, which had been cited to articles published in the New York Times, with edit summaries stating that these Times articles are unreliable sources. (When looking at an edit in a Wikipedia page’s history, the material in the left column but not the right column is what was removed, while the edit summary is the explanation for the change shown in the upper right.) Wikipedia users decided that these New York Times articles could not be cited because their author, Nicholas Wade, is also the author of the book A Troublesome Inheritance, which had been criticized for its speculative statements related to human intelligence. For example, Wade’s book proposes an evolutionary explanation for high average IQ scores among Ashkenazi Jews, an idea that will be discussed in more detail later.

In the rejected New York Times articles, Wade was not presenting any controversial hypotheses, but simply reporting the findings of various genetics studies published in mainstream journals. But it was decided that the views Wade had expressed in A Troublesome Inheritance discredited all of his writings, regardless of what they were about or where they had been published. One of the Wikipedia users who supported classifying these Times articles as unreliable sources explained the basis for this decision:

A reliable source is a source you can rely on. If we, as editors, have to “pick and choose” Wade's good stuff from the bad stuff, that means we cannot rely on him. When Wade is “the only science journalist to cover a study”, how do we know that in this case, he did not misrepresent it?

Wikipedia’s policy regarding reliability of sources, also known as RS policy, provides an objective set of criteria for determining what sources are reliable. According to this policy, the most important criteria for determining a source’s reliability are its age and the reputation of its publisher, with recent articles from reputable newspapers or academic journals being considered reliable. It is a tortured interpretation of RS policy to say that reputably published articles are unreliable because of the viewpoints they present. As for the argument that reputably published sources are unreliable because of views their authors have expressed in unrelated publications, this interpretation of RS policy is applied virtually nowhere else at Wikipedia. But as the following sections will show, on articles related to human intelligence it has become commonplace.

Case 2: Nations and IQ

The concept of a national IQ refers to the average IQ score (or the score on an international standardized achievement test, such as the TIMSS or PISA) of an entire country. One sub-area of intelligence research concerns efforts to measure these national scores, along with investigation into their possible causes and consequences. Among the more prominent studies in this area is a 2010 paper proposing that lower-scoring nations have their IQ scores depressed by the burden of infectious diseases in early childhood, and that this has detrimental effects on these nations’ economic development. This study was discussed in the 2011 annual letter from the Bill & Melinda Gates Foundation:

The huge infectious disease burden in poor countries means that a substantial part of their human potential is lost by the time children are 5 years old. A group of researchers at the University of New Mexico conducted a study, covered in The Economist, showing the correlation between lower IQ and a high level of disease in a country. Although an IQ test is not a perfect measure, the dramatic effect you see is a huge injustice. It helps explain why countries with high disease burdens have a hard time developing their economies as easily as countries with less disease. (Gates 2011)

In November and December 2020, around 12 paragraphs of text were removed from Wikipedia’s “Nations and IQ” article. This material had been cited to three books published by Cambridge University Press and MIT press, and to seven papers published in the journals Personality and Individual Differences, Intelligence, and Frontiers in Computational Neuroscience. The justification given for these removals was similar to the one given on the “Recent human evolution” article: that the authors of most of these sources held unacceptable views about race and IQ, which precluded citation of their research related to national IQs. It also was argued that the material must be removed because national IQs are “thinly disguised claims of differences between races.”

In this case, more than a third of the article was removed, including about 75 percent of the scholarship it had cited from the past decade. The article had previously cited eight academic sources published since 2012, and the removals eliminated six of these sources (four papers and two books). In place of what was removed, material was added that was cited to Superior: The Return of Race Science, which is a non-academic book written by a journalist, as well as to two academic books published in 2006 and in 1974.

Aside from causing the Wikipedia article to become even more outdated than it had been before, these removals also are contradicted by Wikipedia’s policy concerning the coverage of fringe theories—if this material really can be considered fringe, as the users removing it have argued. One of Wikipedia’s policies for covering such theories is known as “parity of sources,” which stipulates that a theory should be covered insofar as its coverage can be based on sources of similar recency and quality to those that criticize it. But in this case, most of the material sourced to recent academic books and papers, which discussed international test score comparisons in positive terms, was removed in favor of criticism that relied on older or journalistic sources. A more significant example of this unusual approach to sourcing will be discussed later.

Case 3: The Flynn Effect and dysgenics

The Flynn Effect is the name for a gradual increase in performance on IQ tests over the course of the 20th century, which has been equivalent to a gain of about three IQ points per decade. This phenomenon is named after James Flynn, who brought wide attention to it in a pair of papers published in the 1980s. Many possible causes have been proposed, but the most widely accepted are that it is the result of improved education and nutrition, as well as increased skills at the type of analytical thinking that IQ tests require.

This increase in performance has been offset by another trend in the opposite direction. When discussing humans, the term “dysgenics” usually refers to the hypothesis that the genetic basis of intelligence is declining due to the most intelligent people tending to have the fewest children, as fictionally depicted in the movie Idiocracy. With the ability to examine polygenic scores for intelligence and educational attainment, which directly measure these traits’ genetic basis, recent research has indicated that such a dysgenic trend is indeed occurring. This finding is described in a 2020 literature review published in Nature Human Behaviour:

One active and politically sensitive area of research is the relationship between education and fertility. Genetic variants associated with education are also associated with a lower number of children born, resulting in declines in the average EA PGS [Educational Attainment Polygenic Scores] in the 20th century.

From 2021 to the present, Wikipedia has seen a steady series of removals of content about the Flynn Effect, which had been cited to papers published in the journals Intelligence, Perspectives on Psychological Science, Personality and Individual Differences, and Behavioral and Brain Sciences, and to a book from Stanford University Press. Some of the edit summaries state that this material was removed because it was “profringe,” the term for the inappropriate use of Wikipedia to promote fringe theories, although this removed material had nothing to do with race or ethnicity. The person removing it mentioned in another edit summary that they were removing material that originated from “racial hereditarians”—in other words from researchers who have, in unrelated publications, supported what Wikipedia’s members consider unacceptable ideas about race and IQ. As before, the assumption is that if a researcher has supported those unacceptable ideas, their reputably published writings about other human intelligence topics should not be cited either.

The Wikipedia member responsible for these removals has also made a similar set of changes to the “Dysgenics” article. In January of this year, all discussion about the existence of a real dysgenic trend among humans was removed from that article, along with all citations of 21st century studies showing it to be real. In the same set of edits, a statement was added to the article saying, “genetic studies have shown no evidence for dysgenic effects in human populations.”

In a discussion about the changes to the “Dysgenics” article, Wikipedia user Ferahgo the Assassin—a postdoctoral researcher in behavior genetics—pointed out that the “no evidence for dysgenic effects” statement contradicts every recent major textbook and literature review about this topic, including The Cambridge Handbook of Intelligence and Cognitive Neuroscience (Cambridge University Press, 2021), An Introduction to Statistical Genetic Data Analysis (MIT Press, 2020), and the paper from Nature Human Behavior quoted above. While a few of the site’s members acknowledged the validity of Ferahgo’s objection, the person rewriting the “Dysgenics” article ignored it. Since that time there have been a few attempts by other people to make this aspect of the Wikipedia article consistent with the field’s source literature, but the person who rewrote the article has undone these attempts, again describing these sources as “profringe.”

Commenting on these attempts, another member of Wikipedia described the sources rejected from the “Dysgenics” article as “the same old garbage from the white supremacist science crew in a very slightly different wrapper.” On this article, most of the rejected sources are by authors who have never discussed race and IQ in any of their publications. “White supremacist science crew” is a particularly ironic description of Kathryn Paige Harden, the author of the Nature Human Behavior paper, who is known for her opposition to race research. However, there is indeed some overlap between the researchers who have studied dysgenics and those who have studied race and IQ, and this overlap is evidently enough for Wikipedia members to reject all sources that support a dysgenic trend in human intelligence, no matter who their authors or publishers are.

Case 4: Ashkenazi intelligence

Although they comprise only about 0.2 percent of the world’s population, the Jewish people account for a large portion of its top achievers in domains of intellectual success. For example, they have won between a fifth and a quarter of the world’s Nobel prizes, and comprise over half of its chess champions. Ashkenazi Jews are particularly noted for their high achievement, including their high average performance on IQ tests. In his textbook IQ and Human Intelligence (Oxford University Press, 2011), Nicholas Mackintosh gives the following summary:

[I]t has long been known that Ashkenazi Jews have an unusually high average IQ (see Chapter 1); some of them also have the misfortune to suffer from a number of diseases, such as Tay Sachs disease, caused by the possession of two copies of particular recessive genes. One suggestion is that the two are linked: while homozygotes with two copies of the genes develop the disease, heterozygotes with only one copy develop higher than usual intelligence (Cochran et al., 2006). (Mackintosh 2011, p. 285)

Aside from its scientific importance, this topic of research is also an important part of the rebuttal to antisemitic explanations for Jewish achievements. In 2006, Steven Pinker wrote in the New Republic that “Jewish achievement is obvious; only the explanation is unclear. The idea of innate Jewish intelligence is certainly an improvement over the infamous alternative generalization, a worldwide Jewish conspiracy.”

In spring of 2006, a Wikipedia article about Ashkenazi intelligence was created. Over the following 12 years, various members of Wikipedia proposed a total of six times that this article should be deleted—once in 2007, a second time in 2007, once in 2008, once in 2011, a second time in 2011, and once in 2017. All of these deletion attempts failed for basically the same reason: that despite Wikipedia users’ discomfort with having an article about this topic, the topic of Ashkenazi intelligence was too widely discussed for Wikipedia to avoid covering it. A search for “Ashkenazi intelligence” at either Google Scholar or Google Books confirms this, with dozens of results from mostly high-quality journals and academic publishers.

In October 2020, Wikipedia’s “Ashkenazi Jewish intelligence” article was nominated for deletion the seventh time. The argument presented for its deletion was more or less the same one that had been made in every previous deletion proposal:

[O]ur article is some sort of pseudo-academic jaunt through fringe literature as promulgated by the IDW-sorts and the evo-psychs. Meanwhile, nary a hint is here that the true context of this is antisemitism. The article is here to wave a flag: such discussions of race and intelligence cannot possibly be race realist in the WP:NONAZI sense because look at who benefits at this article? *wink*, *wink*

This seventh attempt employed a tactic that had not been used in the other six. Rather than directly arguing for Wikipedia to cease covering the article’s topic, this deletion proposal suggested that the most effective way to address the nominator’s complaint would be to delete the article and then recreate it in an improved state. This argument succeeded where every previous deletion attempt had failed, and Wikipedia’s article about Ashkenazi Jewish intelligence was deleted on October 19th, 2020.

After the article’s deletion, this stated plan to recreate it turned out to be a false promise. Instead, references to high average IQ among Ashkenazi Jews were subsequently removed from every other Wikipedia article in which this topic had been discussed, including the “List of Jewish Nobel laureates” article and the general “Ashkenazi Jews” article, with edit summaries stating that the various papers, articles, and books discussing this topic were no longer reliable sources. Among the many sources rejected with this justification were papers and articles published in the Journal of Biosocial Science, Mens Sana Monographs, Commentary, and the New York Times, and the book Abrahams Children by Jon Entine (Grand Central Publishing, 2007). Following the final removal of this material in March 2021, Wikipedia no longer covers the topic of Ashkenazi intelligence.

Case 5: No source required

One of these changes has been the subject of particular controversy. It had its origin on Wikipedia’s “Race and intelligence” article, which is about the differences in average scores between ethnic groups on IQ tests, and other tests of cognitive ability such as the Armed Forces Qualification Test and mental chronometry tests.

This change began with the addition of the following statement, first in the article’s introductory section and later in the body of the article: “there is no scientific evidence that these differences in test scores have a genetic component.” This replaced a sentence that had been in the article for the past decade: “there is no direct evidence that these differences in test scores have a genetic component.” [Emphasis added.] This sentence was modified without replacing any of the sources that it cited, although all of these sources supported what the sentence had said originally, not the new phrasing.

This distinction between “no direct evidence” and “no scientific evidence” is small but important. “No direct evidence” was an accurate statement, because research in behavioral genetics is not yet able to directly test the hypothesis that group differences in average test scores have a genetic component. Doing so would require a comparison of average polygenic scores between ethnic groups, but as explained by FAQ item 3.5 of the Social Science Genetic Association Consortium, such comparisons between different ancestry groups are currently unable to produce a reliable result. Consequently, most studies that argue in favor of a genetic component to group differences have made their case based on indirect evidence, such as measuring the correlation between IQ scores and gene-based measures of ancestry.

When this change to the article was first made, a few other Wikipedia members objected that “no direct evidence” was what the article’s sources said, and that no source had been provided to support the new phrasing. The user making the change replied with an explanation for why a source was not required:

The fact that the phrase “no direct evidence” occurred in someone’s paper is irrelevant. […] We don’t say “There’s no direct evidence that the Universe was created in 7 days around 4000 years ago;” or “There’s no direct evidence that homeopathy can work better than modern medicine.” When talking about fringe theories, one says “no evidence,” not “no direct evidence.”

The “no scientific evidence” statement, still cited to all the sources that had supported the original phrasing of “no direct evidence,” was then copied to several other Wikipedia articles, including “Intelligence quotient,” “Heritability of IQ,” and “Racial achievement gap in the United States” (in a section of the last article titled, “supposed genetic factors”). The statement has remained in these articles for the past two years, and has been uncritically repeated in posts at Substack, Quora, and Reddit, all of which include links to the Wikipedia articles from which it is being quoted.

In the two years since this sentence was altered and then copied into other articles, discussions about it have followed a predictable pattern. Various members of Wikipedia have periodically objected that the statement is not supported by its sources, only to be told in response that this objection is invalid because the hereditarian hypothesis (that is, the hypothesis that group differences in average IQ include a genetic component) is considered a fringe theory. The unstated implication is that the need for Wikipedia to take a strong stand against what it considers “fringe” supersedes the requirement for its articles to reflect what their sources say. A more detailed summary of the objections to this statement’s sourcing, and of the tactics that those responsible for the statement have used to prevent it from being changed, was presented in October 2021 to Wikipedia’s Arbitration Committee, its equivalent of a supreme court. In response, some members of the Arbitration Committee acknowledged that there was a problem, but they ultimately declined to make a decision about the issue.

These objections have also been answered by haphazardly adding more sources to the “no evidence” statement, without regard for whether or not these sources actually support it. One of these subsequently added sources is a paper published in 1970, meaning it can only be informative with respect to whether or not evidence existed for this hypothesis half a century ago. Another is Intelligence: Knowns and Unknowns, a report published by the American Psychological Association in 1996. This report states that genetic interpretations of group differences have no “direct empirical support”—in other words no direct evidence, as the sentence in the Wikipedia articles had said before it was changed.

Of all the sources added for this statement, the one that has been most persistently objected to is Earl Hunt’s textbook Human Intelligence (Cambridge University Press, 2011), because this source says close to the opposite of what it is being cited to say. Hunt’s book explicitly states that there are both genetic and environmental influences on group differences in IQ scores, and that there is circumstantial evidence that these differences include a genetic component. Ironically, two months before this book was added as a source for the “no evidence” statement, its position in favor of a genetic contribution had been given as a justification for removing material cited to the book.

There has been a pervasive sense of despondency among the various Wikipedia members who have tried to change these parts of the affected articles so that they match what their sources say. After one such unsuccessful attempt, a member of Wikipedia known as Stonkaments commented,“Sadly, this whole episode has greatly harmed my estimation of the accuracy and neutrality of the Wikipedia project more generally, especially on articles that are likely to be a sensitive subject for the identity politics mob.” When another member known as Literaturegeek was asked whether he thought this overall situation was hopeless, Literaturegeek replied, “Yes, it is hopeless.”

The basis for the changes

It is tempting to assume that these changes are consequences of the George Floyd protests that began in May 2020, and the consequent rise in the influence of Critical Race Theory. In fact, the examples discussed above began in April 2020, about a month before Floyd’s death. Still, it is possible that this timing is not truly a coincidence, and that the social environment in spring and summer of 2020 inevitably produced these sorts of effects.

The event that precipitated most of these changes was a discussion that occurred in March and April 2020, which concluded that Wikipedia should classify the hereditarian viewpoint as a “fringe theory,” giving it the same status as the view that the Earth is flat or that vaccines cause autism. Over the past two years, nearly every removal of content from intelligence-related Wikipedia articles has cited this discussion or its conclusion as justification. More specifically, it has been argued that the removals are justified by the authority of the sources presented in that discussion. There have been a few later attempts to re-examine the “fringe” decision, which found that this conclusion accurately represents the views of Wikipedia users, but without examining whether or not the sources on which it is based are sufficient to justify the resulting changes.

The sources that were the basis for these removals—a total of 10 sources—are the single most important example of older sources superseding more recent sources, in contravention of Wikipedia’s “parity of sources” principle. Four of these 10 sources (The Science and Politics of IQ, The Mismeasure of Man, and a pair of statements made by anthropology organizations in the 1990s) were over 20 years old. A fifth was a past edition of Psychology: Themes and Variations by Wayne Weiten, and this discussion deliberately cited an outdated (2004) edition of the book—every subsequent edition published from 2007 onwards takes a more nuanced position with respect to the cause of group differences. A sixth source was The Trouble with Twin Studies, which was published in 2015 but is generally not taken seriously by psychologists because it is an attempt to discredit the entire field of behavioral genetics. Finally, the seventh and eighth (a 2017 editorial published in Nature and a 2019 statement from the American Association of Biological Anthropologists) do not actually mention intelligence or IQ, and only argue against racial discrimination or racial essentialism.

Even if one looks narrowly at just the material removed from Wikipedia’s “Race and intelligence” article, and ignores all of the subsequent removals from other intelligence-related articles, the removed material was mostly cited to sources that were newer and higher quality than the 10 that these removals were based on. For example, this change removed five paragraphs of text that cited—among other sources—three books from Oxford and Cambridge University Press, and papers from American Psychologist, Perspectives on Psychological Science, and Psychological Review, all of which were published within the past 15 years. This removal eliminated most of the article’s explanation of how within-group heritability and between-group heritability are statistically related to one another, which is the single most central concept of the race and IQ debate. A smaller removal of material cited to Hunt’s Human Intelligence eliminated the only mention of another central point: that the hereditarian hypothesis does not depend on “race” being a biological category, only on it being correlated with biological variation.

The following table lists the articles that have experienced major removals based on the “fringe” decision, as of July 2022, including one article not otherwise discussed here (Spearman’s hypothesis). The second column shows the number of sources that have been entirely removed from articles due to their viewpoint, or because of viewpoints their authors have expressed in other publications, and does not include those that were removed for other reasons such as age or lack of relevance. Links to the removed material are provided in each case. The numbers in the third column were calculated by comparing the size of the articles in kilobytes before and after the major removals, so this column does not include the various cases in which Wikipedia users tried to add new sources and had their changes immediately reversed. This table is a very conservative estimate of the removed material, because it is confined to only the seven articles where these removals were the largest.

Article

Number of sources removed as “fringe”

Total portion of article removed

Time of main removals

Race and intelligence

14 (2 + 8 + 1 + 1 + 2)

About 11KB / 10%

April-May 2020

Ashkenazi Jewish intelligence

25

18.7KB / 100%

October 2020

Nations and IQ

9

About 10KB / 35%

November-December 2020

Dysgenics

12 (5 + 2 + 5)

About 3.5KB / 32%

January 2022

and May 2022

Recent human evolution

5

7.6KB / 7.2%

March 2022

Flynn effect

6 (2 + 4)

About 3KB / 5%

February 2021 and May 2022

Spearman’s hypothesis

16

11.6 KB / 55.1%

May 2022

In May 2021, Ferahgo the Assassin attempted a similar analysis, listing all of the sources rejected up to that point due to the “fringe” decision. According to Ferahgo’s analysis, members of Wikipedia had rejected a total of 45 sources published in 18 journals, but her analysis does not include any of the more recent removals that have occurred since the beginning of this year. Including these later removals (and limiting the analysis to the seven articles in the table above), the “fringe” decision—and the 10 sources upon which it was based—have been the basis for removing a total of about 85 reputably published sources, along with about 65 kilobytes of article text.

Like the false promise to recreate the Ashkenazi Jewish intelligence article in an improved state after its deletion, the argument for classifying the hereditarian view as a fringe theory included a similar promise that the change would not produce this outcome. When another user expressed concern that this proposal was an attempt to reject reputably published sources supporting the view, describing the likely outcome as “a horrifying precedent,” the initiator of the discussion denied there was any intention of doing that. Even at the time, this denial was unpersuasive, because during the same discussion the individual making this promise described these sources as “white supremacist sources.” Predictably, the same person began the process of removing these sources from articles within 10 days after the discussion’s conclusion.

Bias and its real-world effects

Wikipedia describes itself as having a scientific point of view or an academic bias—or, to quote the second linked page, “Wikipedia should present articles with a balance that is supported by reliable peer-reviewed sources that exercise proper editorial control and are based on accepted scientific method—mainstream science by definition.” A number of researchers (myself included) have found that Wikipedia has a left-wing bias, with Allsides listing five studies that have reached this conclusion. In most topics at Wikipedia there is no contradiction between these two types of bias, but the recent history of articles related to human intelligence shows what happens when the ideas presented in a field’s academic literature are incompatible with what leftists see as acceptable.

The result is for Wikipedia to assume that a scientific field has been infiltrated by racism, as asserted by the aforementioned FAQ. This assumption has led to decisions that the majority of recent scholarly sources must be excised from the “Nations and IQ” article; that the “Dysgenics” article must contradict all recent textbooks and literature reviews about its subject; and that pre-2000 sources, and outdated editions of books, provide the authority to reject a large portion of the recent scholarship about several human intelligence topics. The latter judgment has applied both to articles that relate directly to race or ethnicity, and to articles about other intelligence topics such as the Flynn Effect.

The process by which these changes have spread outward, from the “Race and intelligence” article to Wikipedia articles about many other topics, is an example of how moral judgments about science cannot remain confined to a single idea for long. Whenever a scientific concept—or the source literature supporting a scientific concept—is rejected as incompatible with racial justice, this judgment will inevitably affect a broadening range of ideas over time. In my Journal of Controversial Ideas paper, I described this process in more detail in the section titled, “Demands for Purity and the Domino Effect.”

For the time being, the professional psychology literature still takes a nuanced approach to all of these subjects, as can be seen from how they are covered in modern textbooks such as Hunt’s Human Intelligence (2011), Mackintosh’s IQ and Human Intelligence (2011), Ritchie’s Intelligence: All That Matters (2016), Haier’s The Neuroscience of Intelligence (2017), and Warne’s In the Know (2020). But judging from the way Wikipedia hoaxes are regularly repeated in published books and papers, it is clear that journalists and academics often take their cues from how topics are presented at Wikipedia. Can this nuance in the professional literature persist now that it has disappeared from Wikipedia’s coverage?

Most academics do not openly admit to being influenced by Wikipedia, but the site’s influence on academia appears to be substantial. A 2017 study titled “Science is Shaped by Wikipedia: Evidence From a Randomized Control Trial” found that when Wikipedia articles were created about chemistry topics, these topics received significantly more coverage in the academic literature than a control group of chemistry topics that were not the subjects of Wikipedia articles. The study’s authors conclude that “[b]ecause our work goes beyond correlation to establish causation, we can conclude that Wikipedia doesn’t just reflect the state of the scientific literature, it helps shape it." [Emphasis in original.]

The influence of Wikipedia has also been examined with respect to a few other topics. First, an analysis by the IMS Institute for Healthcare Informatics found that Wikipedia is the most-used source of medical information by both doctors and patients, despite the site containing a disclaimer saying that it does not give medical advice. (Their reliance on Wikipedia as a medical source helps to explain how Wikipedia hoaxes such as “glucojasinogen” have made their way into medical journals.) Second, coverage of the Gamergate controversy has closely followed its coverage at Wikipedia, with dozens of newspapers and academic books copying or closely paraphrasing Wikipedia's articles about the topic. Finally, a 2021 study found that the coverage of cities at Wikipedia shapes their prospects of being visited by tourists, and consequently also shapes their economic futures.

The most obvious reason for this trend is that Wikipedia is the top Google result for most topics, and it is never paywalled, so people researching a topic who are rushed or lazy may use Wikipedia as a source of information because of its convenience. Wikipedia also may affect public and academic discourse in a more subtle way, due to the widespread perception that the view presented there is the “mainstream” one. For example, Google’s “featured snippets” usually include quotes from Wikipedia articles, and YouTube includes quotes and links from Wikipedia alongside some high-profile videos, to provide what is assumed to be a mainstream context.

The original purpose of Wikipedia was to reflect the current understanding of the topics that it covers, not to exert an influence over fields to enact social change. The fact that it performed the first function so well for most of its existence, and came to be regarded as a trustworthy source, is what has made it such an effective tool for those who wish to use it for the latter purpose. While Wikipedia may ultimately prove successful at undermining research about topics related to human intelligence, it also may undermine its own reputation in the process. Formerly trusted institutions have begun to lose society’s trust as these institutions have surrendered to “woke” ideologies, as Quillette has previously described in the case of the New York Times, and Wikipedia will not necessarily be immune to this effect.

Science / Techsocial scienceWikipediaIntelligence ResearchLong Read

Shuichi Tezuka

‘Shuichi Tezuka’ is a pseudonym.