How the Wikipedia Sausage Gets Made

Obsessive nerds who learn the arcane rules of editing Wikipedia are making it hilariously biased on some topics.

Tracing Woodgrains has published an enormously long and exhaustively documented study of how some obscure Wikipedia editor has transformed over the decades his obsessive energy into the power to make Wikipedia hilariously bigoted against various highbrow people he doesn’t like, such as the estimable Scott Alexander.

Nobody hates a slightly right-of-center rationalist like a left-of-center rationalist.

Here are some excerpts:

Reliable Sources: How Wikipedia Admin David Gerard Launders His Grudges Into the Public Record

A love story

JUL 10, 2024

… Introduction: Reliable Sources

Wikipedia administrator David Gerard cares a great deal about Reliable Sources. For the past half-decade, he has torn through the website with dozens of daily edits—upwards of fifty thousand, all told—aimed at slashing and burning lines on the site that reference sources deemed unreliable by Wikipedia. He has stepped into dozens of official discussions determining which sources the site should allow people to use, opining on which are Reliable and which are not. …

While by Wikipedia’s nature, nobody can precisely claim to speak or act on behalf of the site as a whole, Gerard comes about as close as anyone really could. He’s been a volunteer Wikipedia administrator since 2004, has edited the site more than 200,000 times, and even served off and on as the site’s UK spokesman. Few people have had more of a hand than him in shaping the site, and few have a more encyclopedic understanding of its rules, written and unwritten.

… Those who master the bureaucracy in behind-the-scenes janitorial battles, after all, define the public’s first impressions of whatever they care about.

Since 2017, when Wikipedia made the decision to ban citations to the Daily Mail due to “poor fact-checking, sensationalism, and flat-out fabrication,” editors have waged an intense, quiet war over which sources to ban, which to give strict scrutiny to, and which to crown as Reliable. Based on the site’s policy, it’s easy to understand why: while editors with a stake in the frame of an article have to acquiesce to determined opponents bearing Reliable Sources—or at least must have long, grinding disputes about what should be emphasized and why—if they can whip a consensus to declare the sources opponents would use unreliable, they can win edit wars before they happen. This extends well beyond simple factual coverage: cite an opinion or even a movie review from one of those sources, and Gerard or other editors sweep in to remove it as having undue weight.

The battle over the Washington Free Beacon,

Where the fine investigative journalist Aaron Sibarium works

a conservative online newspaper that alternates between tabloid-style sensationalism and serious, in-depth investigative journalism provides a good example of how this works in practice: in three sparse discussions (one, two, three), a dozen or so editors opined, for example, that it “doesn’t particularly have a reputation for journalistic credibility,” with one citing two Snopes articles in support but most presenting bare opinions. As a result of those sparse discussions, Wikipedia editors treat the site as generally unreliable. Every citation to it is presumed suspect, and rather than spending time and effort haggling over each, editors are broadly free to remove them en masse after cursory examination. In practice, this means Gerard scanning through dozens of articles in the span of a few minutes, tearing out all information cited to the Free Beacon as presumptively unreliable.

… In other words, whatever Wikipedia’s written policy, the practical day-to-day reality is that Gerard will remove Unreliable Sources en masse with terse explanations and with little consideration for actual content, digging in with elaborate justification when pressed. Given that, it’s worth examining the reliability battles Gerard picks.

Most interesting to me is the case of Huffington Post. See, in addition to volunteering as a Wikipedia administrator, Gerard is the system administrator and owner of the Twitter account for RationalWiki, a left-liberal wiki focused on directing snark and critique towards groups and concepts the authors dislike, from effective altruists to right-wingers to woo. Gerard has edited RationalWiki upwards of 30,000 times. … When it came time to comment about [Huffington Post] on Wikipedia, though, he was rather more enthusiastic, calling the site “a perfectly normal [news organization] on this level” and raising an eyebrow when people wanted to rate its politics section as less than reliable.²

… As of today, Wikipedia treats the Huffington Post as wholly reliable for non-politics content and unclear for political content. …

He regularly makes similar nudges around sites like The Daily Beast (“Generally reliable - not perfect, but a normal news source, editorial processes, etc - no reason not to use it as a source") and Teen Vogue (“Their news coverage is solid - surprising for a fashion magazine, but it's like the surprise when Buzzfeed News turned out to be a good solid RS too”), as well as supporting the removal of any notes of partisanship from Vox.

… In each instance, he is backed up by a vocal contingent of equally opinionated like-minded editors, who go by pseudonyms such as Aquillion, XOR’Easter, or NorthBySouthBaranof. This is the sort of coordination that requires no conspiracy, no backroom dealing—though, as in any group, I’m sure some discussions go on—just the natural outgrowth of common traits within the set of people whose Special Interest is arguing about sources deep in the bowels of an online encyclopedia.

Wikipedia’s job is to repeat what Reliable Sources say. David Gerard’s mission is to determine what Reliable Sources are, using any arguments at his disposal that instrumentally favor sources he finds agreeable. The debate, to be clear, is not between tabloids and the New York Times, a battle the Times cleanly wins. In Gerard’s world, scientists and academics who publish in Quillette or Reason are to have even their opinions discarded entirely, while to cast any doubt on the reliability of the word of Huffington “the truth is not in them” Post and PinkNews is absurd.

From there, it’s simple: Wikipedia editors dutifully etch onto the page, with a neutral point of view, that Huffington Post writers think this, PinkNews editors think that, and experienced Harvard professors who make the mistake of writing for The Free Press think nothing fit for an encyclopedia….

Although [Gerard] cared about intelligence and encouraged people to donate sperm in part to so they could “add a human of higher intelligence to the population,” he was repulsed by conversations about anything to do with race and IQ, a topic a few posters would occasionally raise. When someone discussed the topic at a meetup, he decided that would be the last meetup he went to.

Something not covered in this article is that Gerard has also been involved in the ongoing, and, on the whole, highly successful attempts to ban scientifically accurate coverage of race-IQ questions from Wikipedia as inherently “fringe” “pseudoscience.” For example, Gerard attempted to get the flagship scientific journal of the field of intelligence research, Intelligence, condemned by Wikipedia:

… And in 2020, Gerard finally had the chance to combine his passions: he could reveal the private name of a man he loathed. He jumped at it.

Gerard did not, in fact, always hate Scott Alexander. … To Gerard, Scott’s blog was far too charitable and calm about neoreactionaries, even as he rebutted them, and not nice enough to the social justice left. And to Scott, well, Gerard came off as a particularly obsessive hater who had chosen to repeatedly smear Scott for distorted and fabricated reasons, taking him to task for insufficient charity while providing none, and who would keep doing so until Scott “refuse[d] to ever engage with anyone who disagrees with him about anything at all.”

By 2020, that hatred had deepened and calcified into a core part of Gerard’s identity, and he watched an announcement from Scott in June of that year with eager anticipation: Gerard’s old rival Cade Metz was writing an article about Scott in the New York Times, he was going to use Scott’s real name, and Scott would prefer he didn’t. Scott cited patient care and personal safety as reasons to be circumspect about his name, pointing out that he had received death threats and faced dissatisfied blog readers calling his workplace, and noting that like many psychiatrists, he preferred to be a blank slate to his patients in his out-of-work life and to avoid causing any drama for his hospital.

Finally, Gerard had the opportunity of his dreams: to supply the Paper of Record with a decade of exhaustive notes about everything he hated about Scott Alexander.

Gerard sprung to work on Scott’s Wikipedia page the day after the announcement, quickly becoming the most active editor on the page and its talk section. He started by stripping away most of the page that covered anything other than the New York Times controversy, then carefully and repeatedly guarded the page against articles critical of the NYT’s decision, which had become a news story of its own. When he couldn’t get a response from the National Review removed, he looked for the lines in it that could put Scott in the worst available light and added them to the article (“since the NR is heavily defended as a suitable source in talk”), later restoring them with a quick note: “[I]t’s cited to [a Reliable Source], after all.” As more and more articles came out about the blog and the controversy, particularly an excellent overview in the New Yorker, removing them would have been a Sisyphean task, but Gerard could at least try to turn lemons into lemonade.

A few days after Scott’s announcement, Gerard added an obscure academic paper Scott had written under his own name to the article—then restored it to the page again, and again, and again, and again, and again, and again. The paper had gone largely unnoticed within Scott’s audience and without, and to the average reader his attention to adding it would come across as little but an odd enthusiasm for AI safety. Gerard clung to the article, though: it was his best chance, he figured, to skirt Wikipedia’s policy preferring omission of names subjects prefer to have concealed, and he fought for its inclusion repeatedly in discussions about the page.

In February 2021, after Scott rearranged his life and quit his job in order to minimize the disruption from his name being revealed, then doxxed himself, the New York Times finally published its article. Off of Wikipedia, Gerard was thrilled, bragging about how much he had been able to land in a Reliable Source:

i sent Metz SO MUCH material for that NYT SlateStarCodex article, i can see the ghosts of what i sent

every phrase is firmly backed up by multiple sources - but it was run through the NYT mealymouthed centrist filter

In particular, he noted that he had encouraged Metz to use Scott’s real name. “[I]t isn't the article we wanted,” he noted on his favorite snark page, “and I suspect Cade wanted it stronger too. But it's good enough.”

Good enough indeed, and he quickly got to work fending off critical responses to the NYT article on Scott’s Wikipedia page. After someone pointed out a long list of critical responses from The Hill, Reason, Quillette, Vox co-founder Matthew Yglesias, Harvard psychologist Steven Pinker, journalist Cathy Young, and others, Gerard shrugged: “Zero of those are [Reliable Sources], so we can’t use them.”

Policy, you see. Hands were tied.

This time, though, people were paying attention, and Gerard had a problem: While you can get away with a great deal when people aren’t looking, Wikipedia does not actually want to be known as the site where people spend decades compiling dossiers against their personal enemies. …

After seven years, someone finally saw what was going on.

The ban [on Gerard editing the Wikipedia article of his enemy Scott Alexander] passed.

To the best of my knowledge, David Gerard never responded. He simply shrugged and carried on eliminating Unreliable Sources.

And that’s how the Wikipedia sausage gets made.

Read the whole thing there.