The Numbers of Wikipedia

Abstract

• Felipe Ortega's Wikipedia analysis is profiled.
• As reported in media, the number of active editors is stalled -or stabilized.
• More women do contributions than men!

Last Febrary I was in Madrid and attended to a conference dedicated to Wikipedia, held in Medialab Prado. Medialab is a diy/open lab, which regularly organices talks and workshops on vanguard topics. That day, Miguel Vidal and Felipe Ortega (University Rey Juan Carlos) were invited. Miguel is one of the oldest Spanish Wikipedia editors. Miguel explained how Wikipedia is organized and how to contribute  (PDF) -there is life beyond the "edit" button!

Miguel's talk was great, but I was amazed by the second conference, maybe because it showed many graphics :) Felipe Ortega has been working with Wikipedia data many years. In November 2009, Wall Street Journal profiled his analysis in the article Volunteers Log Off as Wikipedia Ages. 

«Volunteers have been departing the project that bills itself as "the free encyclopedia that anyone can edit" faster than new ones have been joining, and the net losses have accelerated over the past year. In the first three months of 2009, the English-language Wikipedia suffered a net loss of more than 49,000 editors, compared to a net loss of 4,900 during the same period a year earlier, according to Spanish researcher Felipe Ortega, who analyzed Wikipedia's data on the editing histories of its more than three million active contributors in 10 languages.».

Controversy spread widly. Jimmy Wales, Wikipedia founder, quickly responded. In an interview to Telegraph, Wales denies site is 'losing' thousands of volunteer editors. So Felipe's conference seemed fun. Some basic data:

• There are 3,203,546 registered editors in the 10 biggest Wikipedias.
• Until October 2009, there were 296,387,800 editions in the English Wikipedia.
• The editor's mean time in 346,9 days.

Important issues are clearly illustrated on the presentation graphics.

• Slides 14 and 15 shows that the number of active articles remain constant, in every language analyzed, even if Wikipedia popularity was raising. Same to active discussions.
• According to size, there are languages with two article "populations": short articles, counting around 100 words, and large articles, with ~350 words. Short articles are most common in English and Dutch Wikipedias. However, large articles are most common in Spanish Wikipedia (see slide 16).
• Discussions play a major role in English Wikipedia: 80% of articles have discussions. Just the opposite of Spanish Wikipedia, because only 20% do have debates.

This numbers shows that every Wikipedia develops its own culture. Felipe also made somes comments relative to the results about Wikipedia user profiles.

• There is a high level or participation. 65% of users are readers, 25% are casual editors and 10% are regular editors.
• A fascinating data: Women are move involved than men in Wikipedia: 82% of women are editors, compared to 41% of men.
• There is a high amount of total editors, but the most part of editions are done by a small community of highly active editors: 90% of revisions are done by 5% of editors.
• 36% of users have high school studies. The second biggest group are college graduates with 30%. 5% are doctorates.

About content quality:

• There is an increasing number of reverted editions.
• Featured articles average 1,000 days to obtain the distinction.
• Featured articles have between 10 and 200 times more editors than normal articles.

Felipe said experienced editors are an invaluable resource for Wikipedia and must be preserved, but it's a difficult task, as they tend to burn out, and in the long term they become inactive.

To have a higher level of user involvement, Felipe thinks Wikipedia should be promoted in high school and colleges. He mentioned usability studies being done to simplify edition and offer more information about editorial workflow. In my opinion, Wikipedia needs a far better user experience. To do small editions and corrections is easy, but longer contributions have a step learning curve. Some examples:

• Everything's a wiki! But discussion pages would be better organized as comment threads, just like blogs. Users must learn how to add dates, times and names, it doesn't make sense.
• Template promotion. English Wikipedia articles are better structured than Spanish ones, and template use is more widely used. Still, template use should be encouraged through easier edition workflows (specially at creation time).

I'm sure Wikipedia numbers will benefit from every effort put in user experience.

If you want to read more, take a look to Felipe Ortega site at URJC. Other graphics and tools are hosted at WikiXRay project.