Did you hear that a study in the scholarly journal Nature declared that Wikipedia is more accurate than Britannica? Take that, you British dullards!
Well, there are several problems with that declaration. For starters, the study was not really a study, it was a piece of journalism that did not go through the same rigors of peer review as other articles published in Nature.
But much more importantly, the "study" was RIGGED!!! In the sample of 50 scientific topics covered by both Britannica and Wikipedia, Jim Giles, the author of the "study" found that experts counted 162 errors in Wikipedia and 123 in Britannica. Now, I'm no mathematician, but I have the feeling that 162 is greater than 123. So how do you derive that Wikipedia "is more accurate" from those numbers? Easy, you downplay, explain away, or completely dismiss some of Wikipedia's errors so that you don't have to count them. That's exactly what the author of the "study" did.
By the same token, you exaggerate mistakes in Britannica. In the articles on the Acheulean industry, the expert consulted by Giles found ONLY ONE mistake in the Britannica article and SEVEN mistakes in the Wikipedia article. So if you count Britannica's one mistake as twenty mistakes, and discard six of Wikipedia's mistakes, then yeah, Wikipedia, while still not perfect, is more accurate than Britannica. That's a result that can be arrived at only by very careful manipulation of the data!
For much more details, go to Nicholas Carr's blog: http://www.roughtype.com/archives/2006/02/community_and_h.php