The Hive

Can thousands of Wikipedians be wrong? How an attempt to build an online encyclopedia touched off history’s biggest experiment in collaborative knowledge
The Cathedral
Source: Wikimedia Foundation

Wales was looking for someone with good academic credentials to organize Nupedia, and Sanger fit the bill. Wales pitched the project to Sanger in terms of Eric S. Raymond’s essay (and later book) “The Cathedral and the Bazaar.” Raymond sketched two models of software development. Under the “cathedral model,” source code was guarded by a core group of developers; under the “bazaar model,” it was released on the Internet for anyone to tinker with. Raymond argued that the latter model was better, and he coined a now-famous hacker aphorism to capture its superiority: “Given enough eyeballs, all bugs are shallow.” His point was simply that the speed with which a complex project is perfected is directly proportional to the number of informed people working on it. Wales was enthusiastic about Raymond’s thesis. His experience with MUDs and Web rings had demonstrated to him the power of the bazaar. Sanger, the philosopher, was charier about the wisdom-of-crowds scheme but drawn to the idea of creating an open online encyclopedia that would break all the molds. Sanger signed on and moved to San Diego.

According to Sanger, Wales was very “hands-off.” He gave Sanger only the loosest sketch of an open encyclopedia. “Open” meant two things: First, anyone, in principle, could contribute. Second, all of the content would be made freely available. Sanger proceeded to create, in effect, an online academic journal. There was simply no question in his mind that Nupedia would be guided by a board of experts, that submissions would be largely written by experts, and that articles would be published only after extensive peer review. Sanger set about recruiting academics to work on Nupedia. In early March 2000, he and Wales deemed the project ready to go public, and the Nupedia Web site was launched with the following words:

Suppose scholars the world over were to learn of a serious online encyclopedia effort in which the results were not proprietary to the encyclopedists, but were freely distributable under an open content license in virtually any desired medium. How quickly would the encyclopedia grow?

The answer, as Wales and Sanger found out, was “not very.” Over the first several months little was actually accomplished in terms of article assignment, writing, and publication. First, there was the competition. Wales and Sanger had the bad luck to launch Nupedia around the same time as Encyclopedia Britannica was made available for free on the Internet. Then there was the real problem: production. Sanger and the Nupedia board had worked out a multistage editorial system that could have been borrowed from any scholarly journal. In a sense, it worked: assignments were made, articles were submitted and evaluated, and copyediting was done. But, to both Wales and Sanger, it was all much too slow. They had built a cathedral.

The Bazaar

In the mid-1980s, a programmer named Ward Cunningham began trying to create a “pattern language” for software design. A pattern language is in essence a common vocabulary used in solving engineering problems—think of it as best practices for designers. Cunningham believed that software development should have a pattern language, and he proposed to find a way for software developers to create it.

Apple’s Hypercard offered inspiration. Hypercard was a very flexible database application. It allowed users to create records (“cards”), add data fields to them, and link them in sets. Cunningham created a Hypercard “stack” of software patterns and shared it with colleagues. His stack was well liked but difficult to share, since it existed only on Cunningham’s computer. In the 1990s, Cunningham found himself looking for a problem-solving technique that would allow software developers to fine-tune and accumulate their knowledge collaboratively. A variation on Hypercard seemed like an obvious option.

Cunningham coded and, in the spring of 1995, launched the first “wiki,” calling it the “WikiWikiWeb.” (Wiki is Hawaiian for “quick,” which Cunningham chose to indicate the ease with which a user could edit the pages.) A wiki is a Web site that allows multiple users to create, edit, and hyperlink pages. As users work, a wiki can keep track of all changes; users can compare versions as they edit and, if necessary, revert to earlier states. Nothing is lost, and everything is transparent.

The wiki quickly gained a devoted following within the software community. And there it remained until January 2001, when Sanger had dinner with an old friend named Ben Kovitz. Kovitz was a fan of “extreme programming.” Standard software engineering is very methodical—first you plan, then you plan and plan and plan, then you code. The premise is that you must correctly anticipate what the program will need to do in order to avoid drastic changes late in the coding process. In contrast, extreme programmers advocate going live with the earliest possible version of new software and letting many people work simultaneously to rapidly refine it.

Over tacos that night, Sanger explained his concerns about Nupedia’s lack of progress, the root cause of which was its serial editorial system. As Nupedia was then structured, no stage of the editorial process could proceed before the previous stage was completed. Kovitz brought up the wiki and sketched out “wiki magic,” the mysterious process by which communities with common interests work to improve wiki pages by incremental contributions. If it worked for the rambunctious hacker culture of programming, Kovitz said, it could work for any online collaborative project. The wiki could break the Nupedia bottleneck by permitting volunteers to work simultaneously all over the project. With Kovitz in tow, Sanger rushed back to his apartment and called Wales to share the idea. Over the next few days he wrote a formal proposal for Wales and started a page on Cunningham’s wiki called “WikiPedia.”

Wales and Sanger created the first Nupedia wiki on January 10, 2001. The initial purpose was to get the public to add entries that would then be “fed into the Nupedia process” of authorization. Most of Nupedia’s expert volunteers, however, wanted nothing to do with this, so Sanger decided to launch a separate site called “Wikipedia.” Neither Sanger nor Wales looked on Wikipedia as anything more than a lark. This is evident in Sanger’s flip announcement of Wikipedia to the Nupedia discussion list. “Humor me,” he wrote. “Go there and add a little article. It will take all of five or ten minutes.” And, to Sanger’s surprise, go they did. Within a few days, Wikipedia outstripped Nupedia in terms of quantity, if not quality, and a small community developed. In late January, Sanger created a Wikipedia discussion list (Wikipedia-L) to facilitate discussion of the project. At the end of January, Wikipedia had seventeen “real” articles (entries with more than 200 characters). By the end of February, it had 150; March, 572; April, 835; May, 1,300; June, 1,700; July, 2,400; August, 3,700. At the end of the year, the site boasted approximately 15,000 articles and about 350 “Wikipedians.”

Presented by

Marshall Poe is the founder of MemoryArchive, a universal encyclopedia of memories.

Join the Discussion

After you comment, click Post. If you’re not already logged in you will be asked to log in or register with Disqus.

Please note that The Atlantic's account system is separate from our commenting system. To log in or register with The Atlantic, use the Sign In button at the top of every page.

blog comments powered by Disqus


A Stop-Motion Tour of New York City

A filmmaker animated hundreds of still photographs to create this Big Apple flip book


The Absurd Psychology of Restaurant Menus

Would people eat healthier if celery was called "cool celery?"


This Japanese Inn Has Been Open For 1,300 Years

It's one of the oldest family businesses in the world.


What Happens Inside a Dying Mind?

Science cannot fully explain near-death experiences.

More in Technology

More back issues, Sept 1995 to present.

Just In