Nature says Wikipedia "ain't as bad as all that."

Back on December 15, of two-aught-aught-five, Nature Magazine wrote an article discussing their comparison of Wikipedia and the Encyclopedia Britannica. They selected 50 science articles from both resources to review, looking for "actual errors, omissions or misleading statements." 42 articles were actually reviewed (8 of the 50 were not reviewed).

Guess what? For the 42 articles reviewed, Wikipedia faired pretty well: just under 4 errors per article (162 total errors) for Wikipedia where EB chalked up just under 3 errors per article (123 total errors) . It should be noted that these were strictly science articles, not history, music, art, literature or other disciplines.

Personally, I'm not surprised that Wikipedia did so well when stacked up against EB. What does surprise me is that both publications had this high an error rate.

42 articles were reviewed. Wikipedia currently has 900,720 articles (as of this writing, 1/4/2006 at 23:15 EST). At Nature's published error rate of 3.86 errors per article, there are about 2,316,669 errors on Wikipedia. EB, at their rate of 2.93 errors per article and with 120,000 articles, has 351,428 errors.

But, as with most statistics, what does this tell us? Not so much, really. Per article, Wikipedia has 1/3 more errors than EB. At the same time, Wikipedia has 19 times as many articles. What we don't see is length and depth of articles, structure and clarity, ease of cross-referencing, cycle time to fix "bad" articles, shear numbers of articles, etc.

To me, there are three things Wikipedia has going for it that EB doesn't have: fleetness in responding to errors, lot's of cross-referencing in the form of links to on-site and off-site references and just plain more articles. EB seems to still be paper-based, or at least paper-focused, and has many fewer articles and a much smaller "staff." It just doesn't seem to be as quick-of-foot in responding to problems as does Wikipedia. This assessment may just be personal bias on my part.

There's clearly one thing that EB has going for it that Wikipedia still can't hope to touch and probably never will: all of EB's articles are created and maintained by professional writers and subjected to professional review by boards of experts. Though this leads to only a slight improvement in accuracy, the overall quality of language use, structure, style and clarity is more consistent; EB is better written and is only slightly less error prone.

I've looked at some Wikipedia articles and their EB counterparts. Generally, the Wikipedia articles seemed better linked. This wasn't a scientific study; it was just me perusing some articles and subjectively seeing more links on the Wikipedia pages. The EB articles read better, were more concise and "denser," something that I like.

As I've said before: Wikipedia is a good jumping off point for your research. Its highly linked nature will allow you to get to additional materials pretty quickly. More so than at EB, I think. However, as I've also said before, whether you use Wikipedia, the venerable Encyclopedia Britannica, or any other reference, cross-check your references for accuracy. Never rely on a single source. It will come back to haunt you.

There are a number of links available for the Nature article. BBC.com published an article, as did C|Net's News.com (here) and O'Reilly.com's O'Reilly Radar (here)

Last, here's a link to an article where Wikipedia responds to outside reviews. They get much deeper into the statistics of the review process.