Wikipedia, the online encyclopedia for which anyone may write and edit, is now in its sixth year and has nearly 1.3 million articles in English. Recently, Wikipedians from around the world gathered in Cambridge, MA, to discuss, among other things, how to make the enormous online encyclopedia more accurate, more organized, and easier to use. Author and Web expert Lawrence Lessig refered to the conference, known as Wikimania 2006, as “the Woodstock of the 21st century.”
Technology Review asked Jimmy Wales, the encyclopedia’s cofounder, to update us on his project. His message: Wikipedia is akin to rock-and-roll music in the 1950s: many people are skeptical of it because it’s unconventional. It’s also not perfect, but with the help of its software engineers, administrators, and contributors, it will get better, he says. Given time, Wales says, Wikipedia will be as significant as Elvis, doing for the reference book world what the king of rock-and-roll did for music.
Technology Review: Larry Sanger, one of your early collaborators on Wikipedia, once described Wiki entries as not necessarily the truth, but at best, a kind of “received truth,” or the consensus view on a subject. Do you agree, and do you think this is problematic?
Jimmy Wales: I don’t think “received truth” has ever been clearly defined. When you have thoughtful, reasonable individuals discussing how to present things in a way that’s satisfactory to a broad range of people, and when you use scholarly standards, such as requiring authors to cite sources, the end result isn’t very different from a traditional reference work.
TR: Sometimes Wikipedia entries are not the received truth. They’re not even accurate, by the best lights of the consensual view.
JW: Wikipedia is a work in progress. Mistakes are made during the editing process – sometimes before they have time to be corrected. There are errors in any large-scale human product. I think people have the wrong idea of how accurate traditional reference works are. In the study done by the journal Nature last December, experts seriously looking for errors found about three errors per article in the Encyclopedia Britannica.
The real question is: Does our process weed out errors? The study that hasn’t been done, but would be worth doing, is comparing 100 random Wikipedia articles to themselves two, three, and four years ago, to see the trajectory. I think you’d see dramatic improvements in almost every case. In some cases, you may find we had it better a year ago, which means something went wrong with our system.
TR: The Nature study also found about four errors per Wikipedia article compared with Britannica’s three. Why is Wikipedia wrong more often than conventional encyclopedias?
JW: Because it’s new. If you look at the best articles on Wikipedia – the work that has had the most attention, diverse contributors, and healthy dialogue – they’re significantly better than conventional encyclopedias. In areas where we’re not as good, we’re improving in that direction.