Edit This Page
Is it the end of Wikipedia?
Nov 5, 2009
18 Min read time
Can you trust Wikipedia? Most of us have stopped asking and simply bookmarked it. That makes sense when you consider the alternatives: you can explore the first dozen or so Google search results, or you can go straight to the occasionally erroneous Wikipedia entry, typically culled from the very same search results. If you are looking for fast, up-to-date information, it is Wikipedia or Google (not Wikipedia or Britannica), and Wikipedia wins on speed.
Wikipedia still has its critics, skeptics who doubt its merits as a reference source. But even they cannot deny the tremendous social innovation unleashed by Wikipedia-the-project. Every professional conference—on topics ranging from entrepreneurship to journalism to philanthropy—now includes the mandatory, impassioned plea for the industry to adopt The Wikipedia Model, as if it were a set of Lego pieces that could be ordered from eBay and assembled in a newsroom or on the trading floor.
The enthusiasm may not always be well-informed, but it is understandable. From the start, Wikipedia was an improbable outcome. According to a popular techie quip, it works in practice, but not in theory. Think about it: a bunch of strangers—and not the world’s most sociable strangers—leveraged the power of the Internet to create a highly functioning, über-productive community that voluntarily creates usable (and frequently used) knowledge for others. How much money would you have been prepared to bet against that result a decade ago?
In his first book, The Wikipedia Revolution: How a Bunch of Nobodies Created the World’s Greatest Encyclopedia, Andrew Lih leads readers through the site’s exciting history. He is well-suited to the job. In July 2003 Lih joined the then-two-year-old encyclopedia, and within a few months became one of its administrators. (That a novice could move up so quickly illustrates how badly Wikipedia needed talent in its early days.) Since then, he has emerged as a leading explainer of Wikipedia to the masses, offering sharp, informative commentary on the site’s development. Given Wikipedia’s noble mission to democratize access to knowledge, as well as its unfortunate capacity to attract controversy, The Wikipedia Revolution—strikingly readable and largely free of jargon—is sorely needed.
In addition to engaging description and smart history, Lih tries to tackle a larger story: what makes Wikipedia work? Here, unfortunately, he is much less successful.
• • •
Wikipedia owes its existence to the nerdy culture of the first online discussion systems, such as Usenet; the unlikely success of moderated social sites, such as Slashdot, which was so influential in Wikipedia’s early days that the project was briefly known as “the Encyclopedia That Slashdot Built”; and the emergence of a tight community around WikiWikiWeb, the first application to use the wiki style of editing. The dotcom recession also helped by freeing a supply of superbly talented technologists from the burdens of paid employment.
Wikipedia’s editing philosophy is about simplicity, and two of the cofounders, Larry Sanger and Jimmy Wales, get much of the credit for it. From the earliest days, the community relied on three core rules:
1. Neutrality: entries cannot be partisan and should reflect fact rather than opinion (this “neutral point of view” principle is the site’s only nonnegotiable policy).
2. Verifiability: entries should quote facts that can easily be checked.
3. No Original Research: entries should not contain original and unpublished thought.
At first, decisions were made quickly, and every mention of Wikipedia in the media was met with extensive (and often self-congratulatory) discussion on Wikipedia’s mailing lists. Lih feels nostalgic for those heady days of experimental fluidity, which turned Wikipedia into a magnet for geeks who wanted to test their own tools and approaches, and improve the site in the process. “If there was ever a project that had unhammered nails, thousands and thousands of them, it was Wikipedia,” Lih writes. “And best of all, Wikipedia welcomed anyone with Internet access to start hammering.” The old guard did not object much; “don’t bite the [newcomers]” quickly became an influential policy.
Why do Wikipedians spend countless hours improving the site, often doing mundane, repetitive tasks they would never do for money?
Concerned that excessive bureaucratization might stifle creativity and hamper growth, the early “Wikipedians” adopted a “people first, rules later” philosophy. But as the community expanded and conflicts multiplied, rules and guidelines piled up. By 2006 they constituted one fourth of the site and were one of its fastest-growing areas. Moreover, the ad-hoc rules were often flawed, sometimes even contradictory. Lih points to the dispute over the name of Gdańsk, a Polish city whose “German name (and former official appellation)” is Danzig. Gdańsk /Danzig was the subject of a heated editing war between pro-German and pro-Polish factions. After nearly two years and more than 8000 words of debate, the community settled on a fix that historians had already figured out: use “Danzig” for years between 1308 and 1945 and “Gdańsk” before 1308 and after 1945.
One solution to this disagreement (and many others) would have been to multiply entries: why not have a Gdańsk entry and a Danzig entry? There are, after all, enough electrons to go around. Something like the alternative idea developed at the now-defunct GNE (a recursive acronym for “GNE’s Not an Encyclopedia”—an inside joke in programming circles), which was a collection of unedited articles, “a library of opinions, an attempt to build a comprehensive documentation of all human thought,” with editing itself identified as a source of bias. Wikipedia decided early on to reject this split-the-difference approach. Having a neutral point of view required having a point of view, and the salutary policy was to push contributors to agree on a common statement.
Still, the initial spirit was largely anarchic, which was bound to cause some troubles, as when sensitive celebrities and VIPs discovered errors in their biographies. Wales became a contact point for all high-profile complaints and often stepped in to address them. The German edition of Wikipedia found an elegant solution to this problem by introducing so-called “flagged revisions” of articles. Anonymous users—the majority of visitors to the site—would see an approved (“flagged”) version of the article rather than the latest version. In August 2009 the English edition of Wikipedia announced that it would be adopting the flagged revisions policy for some articles.
• • •
Aside from describing Wikipedia’s history, Lih offers some suggestive reflections on the roots of its remarkable growth and its distinctiveness as a project. After all, the million-dollar question about Wikipedia—the one that foundations and businesses are desperately trying to answer—is not how it works but why it works. Lih’s forays into philosophy, psychology, and sociology, however, are too brief and shallow to be of much use.
Much could have been said on these topics. Two of Wikipedia’s co-founders found each other on philosophy-related mailing lists. Indeed Sanger has a philosophy PhD (his Ohio State doctoral thesis is titled “Epistemic Circularity: An Essay on the Problem of Meta-Justification”), while Wales almost completed a PhD in finance. They came to the project with assumptions about human cooperation that appear to be rooted in philosophy, economics, and evolutionary psychology (among other disciplines), but those ideas are poorly articulated in the book.
Lih does point out that Sanger and Wales were heavily influenced by Ayn Rand’s Objectivism (shades of Alan Greenspan), according to which, reality exists independent of consciousness and life’s great purpose is the rational pursuit of self-interest. Wales’s fascination with Rand was so deep that he even named his daughter after a protagonist in one of Rand’s books. But Lih does not explain the steps from Objectivism to an encyclopedia that “could detail what is true in the world without judgments.” After all, didn’t the Encyclopedia Britannica (or Diderot’s Encyclopédie, for that matter) aim to check judgments at the door and detail only “what is true in the world”? And isn’t that the aim of the new computational search engine, WolframAlpha? How does Objectivism enter the picture?
Maybe it doesn’t. While Sanger and Wales present themselves on blogs and at new-media conferences as mavericks with ideas, they did start off with some rather conventional plans for an online encyclopedia. And their decision to switch over to the anyone-can-edit mode—they are still debating which of the two came up with this scheme—may have been a stroke of luck rather than a product of a theory of cooperation or philosophically-rooted convictions about the virtues of self-interest. Linking Wikipedia to Objectivism may simply be an effort at lending some ex postgravitas to the project (or, more likely, its founders).
Lih also could have told us more about the puzzling psychology of Wikipedians. Who are those people? What makes them so addicted to “wikicrack,” to spending countless hours improving the site, often doing mundane, repetitive tasks that they would never do for money? Lih relies on the work of Harvard law professor Yochai Benkler to address the puzzle. Benkler’s studies of “peer production” draw on the thought of Russian anarchist Piotr Kropotkin, who believed that cooperation is as important in the evolution of species as competition and that “mutual aid” is essential to human survival. Lih does not mention that Rand and Kropotkin are not exactly intellectual soulmates. Lih also does not explain how these two diverging philosophies—one prizing egoism, the other altruism—could live happily together in one site. “Wikipedia is the obsessive-compulsive’s dream come true. It has a bottomless pit of source material with which to indulge one’s pet peeves or obsessions,” Lih offers. But we already knew that. What we (still) do not understand is why some people find deleting commas on Wikipedia more rewarding than playing solitaire or browsing Gawker. Is the public-benefit aspect important? The pleasures of a complex cooperative activity? The unusual possibility of being cooperative from home? Lih leaves us wandering.
To be sure, he does offer some fresh insights about Wikipedia itself. For example, he compares the transformation of this initially small project into a digital metropolis to the process of urban planning outlined in Jane Jacobs’s The Death and Life of Great American Cities. There is, indeed, much to be said about the similarities between Wikipedians and Jacobs’s “self-appointed public characters,” the people who hang out and help produce the “social structure of sidewalk life.” But Lih does not push this comparison far enough: he clings, so to speak, to the “life” side of the equation and never ventures into the darker “death” side. Surely, the increased bureaucratization of Wikipedia would destroy some of those public characters that he admires so much. And how will that affect the project?
Experts are forced to engage in pointless debates with Wikipedia’s bureaucratic guardians, many of whom are persuaded only by hyperlinks, not cogent arguments.
Other aspects of Wikipedia’s history are vividly described but lack discerning intellectual treatment. Lih sheds little light on the “routinization” of the charismatic and ultimately benevolent authority of Jimmy Wales, how that authority evolved into a vast bureaucratic apparatus with a Kafkaesque system of rules. And while Lih notes user ambivalence toward voting, he leaves it largely unexplained. The attitude seems to have grown from an earlier Wiki culture developed by Meatball Wiki, one of the projects preceding and inspiring Wikipedia. The meatballers saw voting as an unnecessary distraction. “Don’t vote on everything, and if you can help it, don’t vote on anything,” read one page on the site.
Wikipedia’s elders adopted those views, realizing that voting could be easily gamed and should not be used often. Instead they settled on a kind of enlightened autocracy: ordinary users would express their views on an issue, after which the more powerful administrators would interpret the vox populi and make a decision. Most of the time, consensus would emerge early on, and the decision was easy; however, as Wikipedia began attracting relatively diverse crowds of editors, achieving consensus grew more difficult. Voting opportunities were further reduced as articles became higher-ranked on Google. A high Google rank means more exposure, which led to more vote-rigging. No longer would there be “votes for deletion,” merely “articles for deletion,” which Wikipedians would discuss. A disinterested administrator would gauge the consensus and make a final decision.
For a site that wants to democratize and revolutionize access to knowledge, such a conservative stance on voting seems puzzling and worth studying in detail, but Lih does not explore this incongruity. There is no guarantee that a more democratic Wikipedia would survive, but it would be interesting to investigate why users so quickly and confidently opted for consensus- rather than voting-driven decision-making.
• • •
“Wikipedia approaches its limits,” ran a striking August 2009 headline in the usually sober Guardian.
With infinite storage and lots of free labor, the very notion of “limits” seems misplaced. However, the limits alluded to in the Guardian are more editorial than logistical. The low-hanging fruit is disappearing—Wikipedians can write only so many biographies of Seinfeld characters—and getting new content onto the site is not as easy as it used to be. A recent study by Palo Alto Research Center (PARC) found that Wikipedia’s key growth indicators—the number of new pages and new editors—have floundered for the past two years, while coordination and editing costs have ratcheted up. Today’s Wikipedians waste a growing portion of their editing time on bureaucratic infighting rather than creating new content. According to the PARC study, Wikipedians also exhibit increasing resistance to new content, especially that contributed by occasional editors.
Some of these concerns were predictable. The sum of human knowledge is expanding but still finite; tools to curate it are improving but still imperfect. As Wikipedia has accumulated a wealth of data—its English version contains more than three million articles—opportunities for making novel contributions have diminished. Wikipedia was bound to hit a knowledge constraint at some point, and it may have already done so. The PARC study suggests that article growth peaked in 2007-2008 and has been declining since.
Some obstacles to continued evolution are fundamental. The set of practices operative on Wikipedia is very loosely defined, which was a boon to early growth but is perhaps now an impediment. When Wales and Sanger started the site, they aspired to create a modern encyclopedia that would continue the tradition of Diderot’s Encyclopédie and Encyclopedia Britannica. “Wikipedia is an encyclopedia” became one of the defining principles of the project; eventually it also led to numerous squabbles over what exactly a digital encyclopedia should be. Adhering to the spirit of Diderot and Britannica’s authors required establishing a certain threshold of importance that encyclopedia articles needed to pass. Paper publishers had fixed numbers of pages to work with and had to compromise. With server space cheap, Wikipedia did not face the same challenges. Some Wikipedians—now known as “inclusionists”—viewed this lack of physical constraint as an opportunity. Others—“deletionists”—thought that filling the infinite space with trivia would distract Wikipedians from curating information that truly matters and dilute Wikipedia’s credentials as a reference resource.
Most inclusionists were not extremists: they did not favor articles on that morning’s breakfast. But in a vast range of cases, they thought that limited and imperfect information was better than nothing. Deletionists disagreed, and to resolve the many borderline cases, the community had to find an objective and quantifiable metric for discrimination. Neither cash nor file-size could do the job, so they settled on the principle of notability: “a topic is presumed to be notable if it has received significant coverage in reliable sources that are independent of the subject.”
Wikipedians are 80 percent male, more than 65 percent single, more than 85 percent without children, and around 70 percent of them are under the age of 30.
The principle did not end the debates, but shifted them onto the interpretation of the qualifiers: “significant,” “reliable,” and “independent.” Do mentions in popular blogs such as BoingBoing or TechCrunch provide “significant coverage”? If they do, can those sources be viewed as “reliable” on all subject areas? Can film criticism published in Cahiers du Cinéma count as “independent of its subjects?”
Most such questions had to be answered on a case-by-case basis, and, gradually, those precedents led to the emergence of hundreds of guidelines that could later serve as shortcuts in dispute-resolution. For example, having articles published in MathSciNet is not a guarantee of notability (MathSciNet falls under the insufficient category of “review publications that review virtually all refereed publications in that discipline”), while notability is assured if one has been elected a fellow of the Royal Society or received a Linguapax Prize. Similarly intricate guidelines establish the notability of diplomats, porn stars, athletes, victims of criminal acts, postal codes, irrational numbers, music ensembles, court cases, and even boulevards (a boulevard “heavily lined with commercial or other major non-residential development that serves as the main road within a suburb or some other heavily-developed area” could be notable; roads that simply have “Boulevard” in their names . . . probably not).
In many cases, however, notability cannot be determined even by following thoughtfully developed guidelines. It is, for example, much harder to verify the notability of a figure from the 1920s than from the 1990s. Most of the important characters of that earlier era are gone from public memory, and newspaper archives from those days cannot be accessed easily online (where Wikipedians spend most of their research time). Given that the flow of articles on Wikipedia far outweighs the attention span of its editors, the latter often have to make the same tough choices that print editors do: why waste a day improving one hard-to-nail-down article when one can improve a hundred?
This creates enormous knowledge gaps in Wikipedia and further alienates well-informed subject experts, particularly those who may know much that is hard to verify online. The subject experts—who usually have great demands on their time, as well—are forced to engage in pointless intellectual debates with Wikipedia’s bureaucratic guardians, many of whom are persuaded only by hyperlinks, not cogent arguments. This raises participation costs—busy experts quickly give up and leave the site—and creates tension between “experts with an attitude” and “verifiability freaks,” which partially explains those gaps.
Early on, Wikipedia’s utter disregard for intellectual elites was not obvious, because there was more than enough work for users of all backgrounds: even placing a comma in the right place makes the project better.
But this also means that the bar for improving the encyclopedia is set very low. Thus the standard criticism of Wikipedians: they are obsessed with popular culture and less equipped to document the high-brow. The 711-word entry on nouvelle vague filmmaker Claude Chabrol, for example, is much less impressive than the 1867-word article on Transformers-director Michael Bay.
However, high-brow entries also suffer because Wikipedia’s economics of knowledge creation are fundamentally unsound. As long as an hour of research yields less “Wikipedia value” than an hour spent planting one hundred commas, few enthusiasts will do the intellectual heavy-lifting. Besides, one cannot learn much about Chabrol from a cursory Google search. Thus, the real tragedy of the Wikipedia method is that it reduces intellectual contributions to such granular units that writing a 2000-word entry on Chabrol in one sitting feels like painting the ceiling of the Sistine Chapel. And if you do go to such lengths to improve the site, you do not want the bureaucrats—who may know nothing about Chabrol—to judge your contribution. There is something unappealing about the value system of a project that prizes, say, movie reviews quoted from college newspapers over elaborate entries in the authoritative Schirmer Encyclopedia of Film, simply because the latter does not have an easy-to-link Web site.
The Google fetish, it should be noted, is not ideological, but practical. Since Wikipedia’s editors are bombarded with editing tasks—one study estimates three new edits every second—they cannot investigate every entry thoroughly. They are constrained by what can be discovered readily—by Google. But most human knowledge, probably, still lies outside of Google’s reach.
If the search-giant’s book-digitization efforts succeed, the conditions in which Wikipedians operate will be radically transformed. Yet such changes would not necessarily translate into new or better Wikipedia articles. Wikipedia’s potential lies in harnessing the “wisdom of crowds”; however, those crowds are only as wise as they are diverse. The individuals who compose the crowd need to bring different sets of expertise to the project. But in Wales’s own words, Wikipedians are “80 percent male, more than 65 percent single, more than 85 percent without children, around 70 percent under the age of 30.” This homogeneity, too, may explain the persistence of certain knowledge gaps.
Some might disagree. As long as the information is available online, Wikipedia, they say, will eventually organize it according to the dictates of good judgment. But despite all of the notoriety guidelines clogging the site, judgment is lacking in the world of Wikipedia. There is virtually no sense of relative importance: improving an article about a prominent historical figure is as important as writing the biography of a soap opera character, as long as both are deemed notable. One does not have to be a natural-born elitist to see that relying on this simplistic binary will inevitably keep the focus on the frivolous, which is never in short supply.
Wikipedia’s current troubles may be the products of its remarkable success. In just a few years, it has become the go-to source for a first-cut at any topic. That has raised public expectations about coverage: people now expect important topics to be covered, and they expect the coverage to be good (as judged by experts). At least in its current shape, however, the Wikipedia model is having trouble meeting those demands. It needs a major upgrade, and the Wikipedia community has brought in management consultants to help figure out how to move forward.
Thinking about the encyclopedia’s future would be impossible without realizing how it got where it is now, and Lih’s book provides real illumination of that path. But the history only takes us so far, and in deciding where things should go next, we still need the book that Lih did not quite pull off.
November 05, 2009
18 Min read time