Wikipedia and the Oligarchy of Ignorance

In a recent story on Medium called “One Man’s Quest to Rid Wikipedia of Exactly One Grammatical Mistake: Meet the Ultimate WikiGnome,” Andrew McMillen tells the story of Wikipedia editor “Giraffedata”—beyond the world of Wikipedia, a software engineer named Bryan Henderson—who has edited thousands of Wikipedia pages to correct a single grammatical error and is one of the 1000 most active editors of Wikipedia. McMillen describes Giraffedata as one of the “favorite Wikipedians” of some employees at the Wikimedia Foundation, the umbrella organization that funds and organizes Wikipedia along with other projects. The area he works on is not controversial (at least not in the sense of hot topics like GamerGate or climate change); his edits are typically not reverted in the way that substantive edits to such controversial topics frequently are. While the area he focuses on is idiosyncratic, his work is extremely productive. As such he is understood by at least some of the core Wikipedians to exemplify the power of crowds, the benefits of “organizing without organization,” the fundamental anti-hierarchical principles that apparently point toward new, better political formations.

McMillen describes a presentation at the 2012 Wikimania conference by two Wikimedia employees, Maryana Pinchuk and Steven Walling:

Walling lands on a slide entitled, ‘perfectionism.’ The bespectacled young man pauses, frowning.

“I feel sometimes that this motivation feels a little bit fuzzy, or a little bit negative in some ways… Like, one of my favorite Wikipedians of all time is this user called Giraffedata,” he says. “He has, like, 15,000 edits, and he’s done almost nothing except fix the incorrect use of ‘comprised of’ in articles.”

A couple of audience members applaud loudly.

“By hand, manually. No tools!” interjects Pinchuk, her green-painted fingernails fluttering as she gestures for emphasis.

“It’s not a bot!” adds Walling. “It’s totally contextual in every article. He’s, like, my hero!”

“If anybody knows him, get him to come to our office. We’ll give him a Barnstar in person,” says Pinchuk, referring to the coveted virtual medallion that Wikipedia editors award one another.

Walling continues: “I don’t think he wakes up in the morning and says, ‘I’m gonna serve widows in Africa with the sum of all human knowledge.’” He begins shaking his hands in mock frustration. “He wakes up and says, ‘Those fuckers — they messed it up again!’”

Neither the presenters nor McMillen follow up on Walling’s aside that Giraffedata’s work might be “a little bit negative in some ways.” But it seems arguable to me that this is the real story, and the celebration of Henderson’s efforts is not just misplaced, but symptomatic. Rather than demonstrating the salvific benefits of non-hierarchical organizations, Giraffedata’s work symbolizes their remarkable tendency to turn into formations that are the exact opposite of what the rhetoric suggests: deeply (if informally) hierarchical collectives of individuals strongly attached to their own power, and dismissive of the structuring elements built into explicit political institutions.

This is a well-known problem. It has been well-known at least since 1970 when Jo Freeman wrote “The Tyranny of Structurelessness”; it is connected to what Alexander Galloway has recently called “The Reticular Fallacy.” These critiques can be summed up fairly simply: when you deny an organization the formal power to distribute power equitably—to acknowledge the inevitable hierarchies in social groups and deal with them explicitly—you inevitably hand power over to those most willing to be ruthless and unflinching in their pursuit of it. In other words, in the effort to create a “more distributed” system, except in very rare circumstances where all participants are of good will and relatively equivalent in their ethics and politics, you end up creating exactly the authoritarian rule that your work seemed designed specifically to avoid. You end up giving even more unstructured power to exactly the persons that institutional strictures are designed to curtail.

That this is a general problem with Wikipedia has been noted by Aaron Shaw and Benjamin Mako Hill in a 2014 paper called “Laboratories of Oligarchy? How The Iron Law Extends to Peer Production.” Shaw and Mako Hill are fairly enthusiastic about Wikipedia and peer production, and yet their clear-eyed research, much of which is based on empirical as well as theoretical considerations, forces them to conclude:

Although, invoking U.S. Supreme Court Justice Louis Brandeis, online collectives have been hailed as contemporary “laboratories of democracy”, our findings suggest that they may not necessarily facilitate enhanced practices of democratic engagement and organization. Indeed, our results imply that widespread efforts to appropriate online organizational tactics from peer production may facilitate the creation of entrenched oligarchies in which the self-selecting and early-adopting few assert their authority to lead in the context of movements without clearly defined institutions or boundaries. (23)[1]

In the current case, what is so striking about Giraffedata’s work is that, from the perspective of every reasonable expert angle on the question, Giraffedata is just plain wrong. It is not a fact that “comprised of” is ungrammatical or that it means only what Giraffedata says it does. This is not at all controversial. In an excellent piece at The Guardian, “Why Wikipedia’s Grammar Vigilante Is Wrong,” David Shariatmadari demonstrates the many reasons why this is the case (though as usual, read the comments for typically brusque and/or ‘anti-elite’ elitist opinions to the contrary). Even better is “Can 50,000 Wikipedia Edits Be Wrong?” by Mark Lieberman at Language Log, the leading linguistics site in the world, which has been covering this issue—that is, specifically the usage of “comprised of”—since at least 2011. Lieberman wryly notes that “It doesn’t seem to have occurred to Mr. McMillen to check the issue out in the Oxford English Dictionary or in Merriam-Webster’s Dictionary of English Usage, or for that matter in literary history, where he might have appreciated the opportunity to correct Thomas Hardy… and also Charles Dickens.” Bizarrely, Wikipedia itself has a page on “comprised of” that endorses the linguist’s view, rather than Giraffedata’s view.


Drawing the circle just a bit wider, Giraffedata is a linguistic prescriptivist in a world where the experts agree that prescriptivism is ideology rather than wisdom. Prescriptivism itself is an assertion of power in the name of one’s own authority that claims (erroneously) to be based on on higher authorities that do not, in fact, exist. It is, in fact, one of the most persistent targets in writing by actual linguists from across the political spectrum: Lieberman rightly calls it “authoritarian rationalism,” and he and Geoff Nunberg (another of the most prominent US linguists) have an interesting back-and-forth about its fit with general right-left politics.

At another level of abstraction, Henderson’s efforts exemplify a lust for power that entails a specific (if perhaps not entirely conscious) rejection of expertise over precisely the topic he cares about.[2] The development of “expertise” is exactly the kind of social, relatively ad-hoc but still structured distribution of power that the new structureless tyrants want to re-hierarchize, with themselves at top. Does Henderson ask linguists about the rightness or wrongness of his judgment? As Lieberman’s work points out, there are obvious, easily available resources in which Henderson might have checked his judgment; it does not appear even to have occurred to him to do so. As Shariatmadari points out, even in the McMillen article, it becomes clear very quickly that Henderson is aware that the “error” he is “correcting” is not actually a matter even of grammar, but a judgment of taste based on several well-known linguistic fallacies (that synonyms should not exist, or that a word’s origin dictates its current meaning).

None of this is to say that it is “right” or “wrong” to adjust the style of Wikipedia with regard to Henderson’s word choice hobby horse. But here again is another rejection of a perfectly reasonable and even useful form of distributed authority: editorial authority over a written product. Before Wikipedia, and even today, published encyclopedias and other publications had rules called “house styles.” These are guidelines made up provisionally by the publishing house to enforce consistency on their work; some are extremely detailed and some are much looser. The house style for any given publication would dictate whether or not to use “comprised of” in the sense that upsets Henderson so much. It would not be a fact whether “comprised of” is right or wrong, but only a fact within the context of the publication. And this is actually a better account of how language works, or usually works: “this is how we do it here,” rather than “this is correct” and “this is incorrect.” (Wikipedia does have a very detailed Manual of Style, but it largely refrains from guidelines pertaining to usage, unlike the in-house style manuals of publications like The New Yorker or The Wall Street Journal.)

At the next level of abstraction, perhaps the most important one, the Wikimedia Foundation’s endorsement of Giraffedata’s work as among their “favorite” displays a kind of agnotology—a studied cultivation of ignorance—that feeds structureless tyrannies and authoritarian anti-hierarchies. In order to rule over those whose knowledge or expertise challenges you, the best route is to dismiss or mock that expertise wholesale, to rule it out as expertise at all, in favor of your own deeply-held convictions that you trumpet as a “new kind” of expertise that invalidates the “old,” “incumbent” kinds. This kind of agnotology is widespread in current Silicon Valley and digital culture; it is no less prominent in reactionary political culture, such as the Tea Party and rightist anti-science movements.

Thus Henderson’s work connects to the well-known disdain of many core Wikipedia editors for actual experts on specific topics, and even more so for their stubborn resistance (speaking generally; of course there are exceptions) to the input of such experts, when one would expect exactly the opposite should be the case. (As a writer in Wired put it almost a decade ago, “The Wikipedia philosophy can be summed up thusly: ‘Experts are scum.’”) A world-leading expert on Topic A wants to help edit the page on that topic—is the right response to reach out to them and help guide them through the (what should be) minimal rules of your project? Or is it to mock and impugn them for having the temerity to think they are expert in something, in the face of the far more important project that you are expert in? One of its several pages addressing this problem, “Wikipedia: Expert Retention,” notes:

If by “Wikipedia” one means its values as expressed in policy, then it can be said that Wikipedia definitely does not value expertise. Attempts to establish a policy on credential verification have failed. There are competing essays that say credentials are irrelevant and that credentials matter. An attempt to push through a policy to ignore all credentials failed, though it received considerable support.

The culture of Wikipedia has no single commonly held view, as is illustrated in the discussion pages of the above cited essays and proposals. However, the lack of consensus (and indeed doggedly opposed parties) results in a perceived lack of respect for expertise, a deference normally found elsewhere in society. Anti-expertise positions often are not acted against, so they are in effect encouraged. And as they are encouraged, they more than negate any positive regard for expertise, since the latter is only expressed, at present, in the consideration given by individual editors to those whom they recognize as experts. (emphasis added)

This is why it’s important that the Wikimedia Foundation employees pass so quickly over the possible “negatives” in Giraffedata’s work, and choose to single him out for praise. This is exactly—if, I think, unconsciously, what the most persistent members of the Wikipedia community want—the disparagement of existing (or in Silicon Valley terminology, “incumbent”) structures for distributing power in the name of a “democratization” that is actually about people with a significant lust for power that is not patient enough to develop its own distributive structures (that is, to work on developing a house style for Wikipedia, or to, I don’t know, study linguistics). In this way, too much of peer production seems like a marketing sheen placed over a very clear and antidemocratic lust for personal power, much as the 1970s communes were, but writ large and with very central social pillars in its sights.

As Freeman’s work has always suggested, which makes the brute rejection of its reasoning in favor of Hayekian “spontaneous orders” of knowledge (or ignorance), Wikipedia’s structurelessness is very easily seen not as a social miracle of cooperation but as a breeding ground for tyrants.[3] Mako Hill and Shaw: “the adoption of peer production’s organizational forms may inhibit the achievement of enhanced organizational democracy” (22). That they do this in the name of democracy makes them characteristic of the contemporary, digitally-inspired agnotological oligarchy.


[1] It is worth noting that Shaw and Mako Hill rely in part on the so-called “Iron Law of Oligarchy” postulated by the proto-Fascist and Fascist sociologist Robert Michels in the early part of the 20th century. Michels actually thought it applies to all democratic organizations and cannot be prevented, but Shaw and Mako Hill rely on a great deal of post-Michels research that tends to give greater weight to formal methods of preventing oligarchy than Michels did.

[2] “Lust for power” is the usual English translation of the German word machtgelüst, which appears prior to the better-known “will to power” in Nietzsche, and unlike the latter term, is specifically meant to indicate the cathection of desire toward personal power.

[3] Wikipedia founder Jimmy Wales famously traces his philosophical inspiration for Wikipedia to Friedrich Hayek’s 1945 essay “The Use of Knowledge in Knowledge in Society”; Philip Mirowski, our most trenchant critic of neoliberalism, has repeatedly demonstrated the ways in which Hayek’s views specifically advocate ignorance.

This entry was posted in "social media", cyberlibertarianism, rhetoric of computation and tagged , , , , , , , , , , , , , , , , , , , , . Bookmark the permalink. Post a comment or leave a trackback: Trackback URL.


  1. sparkzilla
    Posted February 21, 2015 at 4:46 pm | Permalink

    As an ex-Wikipedia editor and the founder of a crowdsourcing site [1] I agree with many of the points on this article. While I was an editor I was appalled at the rampant abuse of power by unaccountable admins. I also found the content-creation-through-conflict to be very time consuming and painful. After building my own site I realized that the the wiki software causes the problems by allowing powerful users to take control over the page, the rules, and the site. In effect the wiki system gives “tyrants” a place to thrive. Refer to my blog post [2]

    However, it does not have to be this way. Much of Wikipedia’s content, especially news and biographies, can be processed much more efficiently and fairly using a purpose-built content management system. The compilation of sources is not something that needs expert review, especially for news events (which make up much of the “encyclopedia”‘s content). It’s not symptomatic of a “culture of ignorance”, but of a job function that actually doesn’t much expertise. The confusion is caused because Wikipedia editors have two roles: transcriber of sources and fact-checker. Separating these functions would lead to more clarity, but it cannot happen with the current software.

    In our case we divide the page data up so people cannot have veto power over a page. We have a very specific post format and house style to avoid haggling for years over minor details, and to avois harassment and bias we anonymize the relationship between editor and writer while the edits are being assessed. All of these serve to diffuse the accumulation of power by early users or powerful groups. The result is that unlike, Wikipedia, which is written and edited 90% by men (the infamous “Gender Gap”), Newslines is created 80% by women and minorities. Men are welcome too, but they are not excluded.

    It’s also useful to note that Wales et al did not build a custom system to create content, but basically picked up the wiki concept and ran with it. They were able to get the community to adapt to the wiki’s idiosyncrasies, so much so that Wikipedia fans are in deep denial about the shortcomings of the site, even while they suffer its abuses, and ignore how it excludes others. However at a certain point the structure collapses under its own weight, much like the communes you mention.


  2. marknowotarski
    Posted June 11, 2015 at 4:35 am | Permalink

    I had to smile when I saw that an editor is dedicated to removing “comprised of” from Wikipedia. “comprised of” is an essential phrase in patents. It means “consisting at least of”. You almost never see the phrase “consisting of” in patents since courts have held that “consisting of” means “consisting only of”. You almost never want to say “consisting only of” in your patent because the easiest way to get around the patent is add something additional to the invention.

    (Poster is a registered US patent agent)

One Trackback

Post a Comment

You must be logged in to post a comment.