How the online "temple of the mind" became the go-to site for looking stuff up: A drama told in the open-source style of Wikipedia
This article was produced under a loose interpretation of Wikipedia's collaborative principles. Where indicated, it was rewritten, corrected, and commented upon by outside participants.
It's easy to forget, but 10 years ago it took serious effort to find decent information on the Internet. Searching the Web was like picking through a vast antique shop: There was the occasional find, and a lot of secondhand junk.
Interested in Russian-made Tupolev airplanes? An obsessive hobbyist had a home page about them. Someone mentioned Volapük at a party? There was a university site with shards of information about the little-known language invented in the 19th century (not to be confused with the French avant-garde rock band of the same name). Then in 1999 came a simple idea from a search engine entrepreneur named Jimmy Wales: a free encyclopedia written by experts donating their time to draft articles and conduct peer review. The content would be copyable and ad-supported. It would be called Nupedia.
The site's team of PhDs moved at a pace that was indeed scholarly: In its first year, they created barely a dozen entries. Wales realized Nupedia needed a feeder site, something less exclusive that would attract writers. On this new site, any visitor would be able to write about whatever he wanted.
The experiment was unpopular with Nupedia's editors and advisory board—the fact that the uncredentialed were encouraged to contribute could, they reasoned, threaten the credibility of the encyclopedia. After less than a week, the new site was spun off into a separate project called Wikipedia—a "wiki" (Hawaiian for quick) is software that allows users to write, edit, and link Web pages on a single shared document. Within a year the new site had 20,000 articles. When Nupedia shut down in early 2003, it was stuck at 24 articles.
Jan. 15 is Wikipedia's 10th birthday. According to its own Wikipedia page, it has 17 million entries in more than 250 languages—including 118,000 in Volapük. The Encyclopedia Britannica has 120,000, and only in English. Free of charge—and of ads—the site is one of the most visited on the Web. It is the first stop for bar wagerers, high school paper writers, oppo researchers, and anyone trying to figure out what the Peace of Westphalia did or when the mortgage-backed security was invented or whom Ringo Starr replaced as the Beatles' drummer. The entry for the Deepwater Horizon oil spill went up on Apr. 21—less than a day after the rig explosion, and before anyone had any idea of its extent. "US Coast Guard has already launched a rescue operation," the article's last sentence read. (Comment)
Even some of the site's staunchest supporters admit they initially doubted it would amount to much. The idea sounds like a utopian social experiment: a record of the world's knowledge, produced and policed by volunteers. Today anyone can still contribute to an entry, and many do. But rather than collapsing into chaos and endless arguments over the exact diameter of the second Death Star (a debate that, in fact, continues), the world of the self-proclaimed Wikipedians has grown into a thriving online society. "Its existence is proof of a radically different way of organizing production," says Yochai Benkler, a professor at Harvard Law School who studies social networks and the Internet.
Exactly what, though, does Wikipedia prove? The site has exponentially outgrown the humble expectations of Wales and co-founder Larry Sanger but has had to weather controversies over inaccuracy and manipulation. Recent years have brought complaints from contributors that the site is abandoning the welcoming ethos that once distinguished it, and voices inside and outside the Wikipedian community are questioning whether radical openness and accuracy can coexist. It remains to be seen whether Wikipedia's collaborative volunteerism can be applied to other realms, whether it can be preserved in Wikipedia itself, and—most fundamentally—whether it can produce something of reliable value for those not involved in the game.
Wikipedia was not the only online encyclopedia to be launched around the turn of the millennium. Along with Nupedia, there was an interlinked database called Everything2 where anyone could submit an article but no one could edit others' submissions. There was also H2G2, a playful miscellany inspired by Douglas Adams' Hitchhiker's Guide to the Galaxy. Before those there was Britannica, which had gone online quietly in December 1993. Microsoft's (MSFT) CD-ROM based Encarta, which was neither collaborative nor originally online, was seen as the primary digital threat to traditional encyclopedias like Britannica because it was given away free with new PCs.
According to Benjamin Mako Hill, a researcher at Massachusetts Institute of Technology who has looked at these early efforts, Wikipedia owes its success to several key decisions. "It's very tempting for people to say, 'We're online, now we're going to do something very different from what people did before,'?" Hill says. But that would leave many contributors confused about what to contribute. Wikipedia didn't do that—the entries, though they're born of a new method of encyclopedia writing, are meant to read like the old Britannica.
Second, Wikipedia was user-friendly. Someone who found an error, or knew something about a subject that didn't have an entry, was a click or two away from taking action. The site's success attracted still more contributors. Soon scholars like Benkler and Clay Shirky were celebrating Wikipedia as a prime example of how the Web could upend assumptions about human motivation and organization. Here were people contributing hours of their time, gratis, to build what Wales called "a temple of the mind." Wikipedia's model bears much similarity to free and open-source software that took off in the 1990s with the release of Linux and Mozilla. But Wikipedia, with its diverse, international body of contributors, couldn't just be dismissed as a cultural oddity or a particular slice of geekdom—at least not after it expanded beyond physics equations and Star Trek recaps and into the rest of humanity's hard-won knowledge.
Still, concerns about the site's accuracy grew along with its influence. As the satirical newspaper The Onion put it in a recent headline: "Wikipedia Celebrates 750 Years Of American Independence." In most widely read entries, falsehoods are quickly corrected, but in others misinformation can survive far longer. In the fall of 2005, John Seigenthaler, a former aide to Robert F. Kennedy and the founding editorial director of USA Today, discovered that a Wikipedia page named him as a suspect in RFK's assassination. More than four months after the false charge was published, the entry was finally corrected. Seigenthaler did not correct the entry himself; he wrote an op-ed for USA Today pointing out Wikipedia's inaccuracies. The incident led Wales to acknowledge that entries were being written faster than they could be reviewed, and to announce a policy where unregistered users would no longer be able to create entries. (Registration is free and instant, and does not require personal information.)
One of Wikipedia's loudest critics is co-founder Sanger, whom Wales laid off when the dot-com bubble burst. For Sanger, scandals like the Seigenthaler episode can't be avoided without changing Wikipedia beyond recognition. "I just don't think it's really even possible given the nature of the project," he says.
In 2006 Sanger started Citizendium, a competing wiki-based encyclopedia with "gentle expert oversight" and contributors who use their real names.
At the same time, analyses of Wikipedia's accuracy have generally found that, while it's occasionally wrong, it's not much more wrong than traditional encyclopedias. A 2005 study in the journal Nature found that in a sample of articles, there were an average of 2.92 mistakes per article for Britannica and 3.86 for Wikipedia. (Britannica objected to the Nature study, calling the methodology "fatally flawed.") Wikipedia, however, has problems Brittanica doesn't. An error corrected in Britannica stays corrected; in Wikipedia, it may not. (By the same token, rapidly changing events can be covered in pace by Wikipedia.) Then there are the tastes of its editors. Popular culture looms large. The entry for the video game Halo, for example, is significantly longer than the one for the Protestant Reformation.
More seriously, Wikipedia has to weed out inaccuracies introduced not by accident or through the ignorance of contributors but intentionally. Sometimes, as in the Seigenthaler case, these are instances of vandalism. Other times it's résumé-polishing. In 2006 aides for Massachusetts representative Marty Meehan removed a Wikipedia mention of his campaign promise not to serve more than four terms—at the time he was on his seventh. Wales has edited his own bio page, deleting mentions of Sanger's role in the creation of Wikipedia.
Thanks to WikiScanner, software that cross-references the Net addresses of contributors, watchdogs have found that computers at ExxonMobil (XOM), PepsiCo (PEP), and Diebold (DBD), among others, have been used to remove unflattering information from the companies' entries. Last year, Wikipedia's arbitration council banned edits from all IP addresses owned by the Church of Scientology after years of dispute and deceptive editing by both adherents and critics of the religion.
Many of these worries apply to businesses that have taken parts of Wikipedia's model and monetized it. TripAdvisor and Yelp have created democracies of taste, empowering travelers and consumers to speak up and create a far greater wealth of opinion than any team of professional reviewers could. Still, the owners of the reviewed hotels and restaurants complain that the sites' anonymity allows for slander without repercussion.
Wikipedia has experimented with concentrating more power in the hands of "administrators," a select group of Wikipedia editors who have the ability to delete and restore oft-manipulated pages and to lock them down for a period of time to prevent further edits. These changes have brought a measure of stability to certain entries. (George W. Bush's is getting vandalized less often. During his Presidency, his picture was swapped with one of Hitler; he was described, briefly, as "the first person ever to masterbate [sic] in a public school more than 38 times.") But Wikipedia discussion boards are full of complaints about entries getting locked down without proper cause. And there seems to be no mechanism for assuring that administrators are knowledgeable about anything beyond the methods and rules of Wikipedia itself. (Comment)
The worry for some who have followed the encyclopedia over its first decade is that as the site matures, it will have a harder time attracting and retaining the volunteers who make it work. A research team led by the computer scientist Ed H. Chi at Xerox PARC has found that Wikipedia's growth, whether measured in entries, edits, volunteers, or even bytes, has been falling off for a few years. The number of active editors and rate of article creation both peaked in 2007. The site, Chi argues, has begun to behave less like a limitless information ecosystem and more like a biological one, a world where actors compete over limited resources. Today, all the obvious entries—Aristotle, electricity, the Magna Carta, the American Civil War—have been written, along with millions more. That leaves fewer new ones to write and increases the chance that a new entry will be deleted on the grounds that it, as Wikipedia's guidelines say, "lacks notability." An entry written by Wales himself about a South African restaurant was deleted on the grounds that it contained "no assertion of importance/significance." (It has since been reinstated.) But the main effect of the new competitiveness is to drive out newcomers, whom Chi found are more likely to have their entries and edits rejected.
Andrew Lih, a Wikipedia administrator and the author of The Wikipedia Revolution, says there was a feeling of euphoria when he joined the site in 2002. "There were only 30,000 or 40,000 entries," he says. "You'd look up Angkor Wat, and there'