Dusting Wikipedia for fingerprints

Last year, someone edited the Wikipedia entry for the SeaWorld theme parks to change all mentions of "orcas" to "killer whales," insisting that was a more accurate name for the species.

There was another unexplained edit: A paragraph about criticism of SeaWorld's "lack of respect toward its orcas" disappeared.

Both changes, it turns out, originated at a computer at Anheuser-Busch, SeaWorld's owner.

Dozens of similar examples of insider editing came to light last week through WikiScanner, a new Web site that traces the source of millions of changes to Wikipedia, the popular online encyclopedia that anyone can edit.

The site, wikiscanner.virgil.gr, created by a computer-science graduate student, Virgil Griffith, cross-references an edited entry on Wikipedia with the owner of the computer network where the change originated, using the Internet protocol address of the editor's network.

The address information already was available on Wikipedia, but the new site makes it much easier to connect those numbers with the names of network owners.

Since Wired News first wrote about WikiScanner last week, Internet users have noticed plenty of interesting changes to Wikipedia by people at nonprofits and government entities, such as the CIA. Many of the most obviously self-interested edits have come from company networks.

Last year, someone at PepsiCo deleted several paragraphs of the Pepsi entry that focused on its detrimental health effects. In 2005, someone using a computer at Diebold deleted paragraphs that criticized the company's electronic voting machines. That same year, someone from inside Wal-Mart Stores changed an entry about employee compensation.

Jimmy Wales, founder of the Wikimedia Foundation, which runs Wikipedia, says the site has a policy that discourages such "conflict-of-interest" editing. "We don't make it an absolute rule," he said, "but it's definitely a guideline."

Internet experts, for the most part, have welcomed WikiScanner.

"I'm very glad that this has been exposed," said Susan Crawford, a visiting professor at the University of Michigan Law School. "Wikipedia is a reliable first stop for getting information about a huge variety of things, and it shouldn't be manipulated as a public-relations arm of major companies."

Most of the corporate revisions did not stay posted for long. Many Wikipedia entries are in a constant state of flux as they are edited and re-edited, and the site's many regular volunteers and administrators tend to keep an eye out for bias.

In general, changes to a Wikipedia page cannot be traced to an individual, only to the owner of a particular network. In 2004, someone using a computer at Exxon Mobil made substantial changes to a description of the 1989 Exxon Valdez oil spill in Alaska, playing down its impact on the area's wildlife and casting a positive light on compensation payments the company had made to victims of the spill.

Gantt Walton, a spokesman for the company, said that although the revisions appeared to have come from an Exxon Mobil computer, the company has more than 80,000 employees around the world, making it "more than a difficult task" to figure out who made the changes.

Walton said Exxon Mobil employees "are not authorized to update Wikipedia with company computers without company endorsement." The company's preferred approach, he said, would be to use Wikipedia's "talk" pages, a forum for discussing Wikipedia entries.

Wales also said the "talk" pages are where Wikipedia encourages editors with a conflict of interest to suggest revisions.

"If someone sees a simple factual error about their company, we really don't mind if they go in and edit," he said. But if a revision is likely to be controversial, he added, "the best thing to do is log in, go to the 'talk' page, identify yourself openly, and say, 'I'm the communications person from such and such company.' The community responds very well, especially if the person isn't combative."

An Anheuser-Busch employee eventually took responsibility for the changes to the SeaWorld page, but only after another user challenged him twice.

Most people using company networks to edit Wikipedia entries dabble in subjects that appear to have little to do with their work, although sometimes they cannot resist a silly dig at the competition.

Last year, someone using a computer at The Washington Post Co. changed the name of the owner of a free local paper, The Washington Examiner, from Philip Anschutz to Charles Manson. A person using a computer at CBS updated the page on Wolf Blitzer of CNN to add that his real name was Irving Federman. (It is actually Wolf Blitzer.)

And The New York Times is among those institutions whose employees have made, among hundreds of innocuous changes, a handful of questionable edits.

A change to the page about President Bush, for instance, repeats the word "jerk" 11 times. "It's impossible to determine who did any of these things," said Craig Whitney, the standards editor of The Times. "But you can only shake your head when you see what was done."

WikiScanner is the work of Griffith, 24, a visiting researcher at the Santa Fe Institute in New Mexico. Griffith said he got interested in creating such a tool last year after hearing that members of Congress were editing their own entries.

Griffith said he "was expecting a few people to get nailed pretty hard" after his service became public: "The yield, in terms of public-relations disasters, is about what I expected."

Griffith, who also likes to refer to himself as a "disruptive technologist," said he was certain many more examples of self-interested editing would come out in the next few weeks, "because the data set is just so huge."

Griffith said he was working on ideas to help users "drill down to the good stuff faster."

Wales said he was considering some changes to Wikipedia to help visitors better understand what information is recorded about them.

"When someone clicks on 'edit,' it would be interesting if we could say, 'Hi, thank you for editing. We see you're logged in from The New York Times. Keep in mind that we know that, and it's public information.'

"That might make them stop and think."