I love Wikipedia, too, and often cite it in this blog and elsewhere. For a certain kind of information it is both comprehensive and comprehensible, a ready public domain source that everybody can turn to. But in truth, it’s a good bit less than its partisans like to claim.
For instance, when an article in the journal Nature reported that peer-reviewed fact-checking found Wikipedia with 4 errors in 42 articles and Britannica with 3 per 42, Wired’s headline called it “a toss up” and said Wikipedia “is about as accurate in covering scientific topics as Encyclopedia Britannica.”
Well, no.
What those numbers actually show is that Wikipedia had a much higher error rate than Britannica: 10 percent of Wikipedia articles had errors while 7 percent of Britannica’s did. If I’m wrong 10 times and you’re wrong 7 times, you whipped me pretty good.
Wikipedia is probably better by now. For one thing, its original Dodge City, anything-goes format has been changed to guard against bias or animus introduced by anonymous contributors. Some articles are now sealed against revision by what Wikipedia calls its “community,” because – as in all communities – some members are assholes.
And those who are not assholes nonetheless represent a narrow slice of humanity. Though triumphalists see the web and society as largely congruent, they’re not. When you look at activist internet users, the gap grows much wider.
For instance, who are the people who selflessly contribute all that information to Wikipedia? Their own data show more than 80 percent of them are male, more than 65 percent are single, more than 85 percent are childless and around 70 percent are under age 30. “We are mostly male computer geeks,” says founder Jimmy Wales.
Now comes news that non-profit Wikipedia is spending $600,000 on a handful of consultants, advertising executives and others to help it orient its operations for the future.
I think that’s great, and I hope all these changes make Wikipedia better. But it’s worth noting that restricting access, insisting on editing and engaging professionals is a lot like what other encyclopedias and media companies have been going for a long time.
UPDATE This just in: Wikipedia is exploring ways to flag the articles that are most trustworthy.
"Wikipedia with 4 errors in 42 articles and Britannica with 3 per 42"
ReplyDelete"What those numbers actually show is that Wikipedia had a much higher error rate than Britannica"
http://en.wikipedia.org/wiki/Statistical_significance
Does this suggest that the margin isn't significant or the sample too small?
ReplyDeleteMy main point is that it was used to promote equivalency, which I don't think it supports
If the margin isn't statistically significant, the data have to be interpreted as basically equivalent. You can't complain about newspaper headlines that say 'McCain-Obama tied in latest poll' when one of them had 47% and the other 46%. In fact you should complain if it says one has a lead.
ReplyDeleteOn a side note, I do agree that the margin was too small to make any real judgement.
Howard, the key point about the Nature "study" is that it was rigged to favor Wikipedia. See Nicholas Carr's assessment of what the survey actually concluded:
ReplyDeletehttp://www.roughtype.com/archives/2006/02/community_and_h.php
Oh, and as long as we're talking about inaccuracies, you might want to fix that "founder" Jimmy Wales thing. (See the open letter from Larry Sanger to Jimmy Wales, for details.)