Nascent

Science articles? Is that really a fair test? 50 biographies, or geography/history articles, might have shown a bigger difference.

I can understand why science articles were chosen (no subjectivity about whether something is accurate or not), but given the nature of Wikipedia, and its highly technical audience, strength in science is to be expected.

The link for "accompanying Nature editorial" appears to be a subscriber only link.

You're right, Christopher. Thanks for pointing that out. I'm trying to see if I can do something about it.

It's true that the outcome would probably have been different had they sampled history articles. But note that even this sample of science articles contained a good number of biographies, and Wikipedia performed on par with EB on those.

Also, had they sampled articles about information technology, current events, political controversies or popular culture, the outcome might have been more in Wikipedia's favor.

Wikipedia would very much like to correct any errors that Nature's referees found. All of the listed articles (except those with 0 errors) have been labelled as problematic. Is there any way that Nature's editors can publish or summarize the specific errors found?

See also Wikipedia's page on this Nature review.

One disappointing aspect of this comparison, for me, was that it was not controlled for article length. Errors per word might have been a more meaningful statistic.

Wikipedians have been very interested in this comparison, and are working hard to find the errors and improve the articles in question (along with all of the other 865,000 articles).

Some have put together a "quick table comparing the EB and Wikipedia file sizes for the articles. In Wikipedia's case, we selected the displayed text [...], removed all tables/external links/see also/references/etc, and we used the versions prior to the Nature article's publication. One result of interest: our 2nd largest article on the list, Robert Burns Woodward, had zero errors. That same article is EB's smallest article, and has 3 errors.

"Result: WP averages about 1 error every 2 kilobytes, and the average article size is 6.80 KB. Britannica averages about 6 errors every 2 KB, and the average article size is 2.60 KB."

You may find it interesting to read about Wikipedia's responce to your article here.

Steven, an excellent question. We're trying to see if we can publish the full list of errors found by our reviewers, or least send them to you (and to Britannica if they want). We'll post an update here as soon as we have a firm answer.

As for the Wikipedia page on the Nature study, wow! :)

Christopher, the Nature editorial is now open to non-subscribers too.

I'd like to thank all those involved with this comparison. It's very interesting to see such an excellent critique of our articles, especially when viewed alongside those of a professional Encyclopedia organisation.

I hope that most of the errors are simple, anonymously-introduced ones that we can quickly correct. If you can make the reviews available to us then we would be more than grateful, but even without those details we appreciate your investigations.

Regarding Maeve's comments about article size, unfortunately the comparison he/she refers to is erroneous. The "Robert Burns Woodward" article is not Britannica's shortest article — the person doing that comparison apparently didn't notice that Britannica breaks up long biographies into sub-articles, and hence only counted the introductory paragraph.

On the other hand, the counts here (errors/word) do count the "entire" article, and the Wikipedia articles currently have about twice as many words, on average, as Britannica.

One problem, of course, is that we don't know which versions of the articles were compared by Nature, and furthermore some topics have entries in multiple places. However, there are still some things that mystify me — Nature claimed that it compared articles of "approximately the same length". However, for example, the Wikipedia article on the West Nile virus has been over 1000 words for a year now, while the Britannica article of the same title is only 250 words, almost 5 times shorter. How did Nature compare these two?

More detailed information about the review process used, and a full list of the errors identified, is now available from Nature.com here (Word file).

For those who prefer a PDF version, I've put one here.

If you can't "trust" wikipedia, you can still use it. The references are usually elaborate and often peer reviewed. So, if you're worried that you can't cite wikipedia as a reference in your work, then you can sure site one of its references!

In regards to the comparative accuracy of wikipedia to encyclopedia Britannica the cause of misinformation in either source will mainly be at the result of human error or personal inference on any given topic. The Achilles heal of the wiki is the publics belief that there is a malevolent force out there who’s aim is to provide others with misinformation. Likewise those who are in support of Britannica use the armament that a sentence printed on paper and published in a book is infallible. Time and again throughout history cartographers have falsified maps going to print replacing the names of some remote region with their own name or that of a loved one. Further more it needs to be recognized that sources of information and the setting in which they are distributed ultimately have only marginal impact on the accuracy of well intended information. For example a recent anecdote has come to my attention where a public school teacher and a student got into a heated debate in the classroom with the instructor insisting that a kilometer was of greater length than a mile. In a letter to the students parents the teachers closing paragraph read…
“Although he was correct, Alex’s actions show a blatant disregard for authority, and a complete lack of respect for his school. In the future Alex would be better off simply accepting my teachings without resistance”

While this comment is staggering in of its self it makes it easy to loose focus on the real issue which is of all the facts that any person knows a certain percentage of them are in reality incorrect despite societies best effort to educate.