Thursday, September 11, 2008

Is Wikipedia Production Slowing Down?


Until recently I had assumed that the growth of content on Wikipedia was exponential. It certainly looked that way over its early history, and other kinds of "knowledge publication" like scientific journals (or the Web) have shown consistent exponential growth. But it looks like there was a peak-drop-flattening that started around Spring 2007. I wrote short report on my blog, with some pointers to others who have also seen this "anomaly", but I haven't seen a satisfactory explanation for why this has happened. I'm interested in hearing from anyone who might have a bit of insight about what may have precipitated this change in the dynamics of Wikipedia content production.

8 comments:

Adam said...

Might increasing niche knowledge in an expanding number of niches result in an inverse correlation to the number of people in any given niche with the right personalities for posting it on Wikipedia? It might be worth looking into what niches people -don't- post about on Wikipedia, and what commonalities they share.

Adam said...

Alternately, I just held up a graph of oil prices for the same time frame (1/1/06 - 8/1/07) up to your graph of editors. Not a very accurate method, but there may be an inverse relationship there too. Maybe economic stress simply makes Wikipedia a secondary priority?

Ed H. Chi said...

Adam, those ideas are all somewhat interesting guesses about why this happened. However, it might be very difficult to determine the cause of this effect, since the people who didn't contribute are probably very hard for us to engage at this point to interview or survey.

Peter Pirolli said...

Adam: At first I thought the "niche effect" might make sense. There's evidence of something like that going on more generally, in the sense that some of the pages created earlier get more edits than those created later (assuming that early creations are more general than later ones--which tends to be true). BUT, if you look at some of the additional graphs I plotted on my blog for pages of different "ages" you'll see that the peak-drop-flattening trend happens across pages of different creation dates.

The "oil price" correlation is interesting. There's a general argument that I'll attribute to Yochai Benkler's "Wealth of Networks" that says that participation in these peer-production projects depends on people having "surplus capacity" --i.e., bread-and-butter issues have been taken care of.

Leonardo said...

I used to be an active contributor to english, french and portuguese wikipedias. Along with minor contributions to spanish wiki and sidekick projects(commons,books,wikitionary, etc).

The "oil price" correlation might be right, at least in my case. Also the rantings/feuds among users, I just got tired of it, sometimes the energy I had to waste arguing with obstinated knuckleheads who had no real knowledge about the subjects they were trying to maintain. Also, everything needed to be voted, which is a good thing for sure, but unfortunately leads to deadlocks...

Nonetheless, I don't know if these reasons apply to the rest of the contributor casualties.
Irc wikipedia channels(they're all on freenode in case you don't know) used to be full till last year, now they're getting emptier and emptier... #wikipedia-pt for once, is almost dead.

Ed H. Chi said...

The whole issue about cognitive surplus was talked about by Clay Shirky in his new book, and a portion of the argument was blogged on his blog here. He argues that there is plenty of cognitive surplus around, but of course, he doesn't mention the fact that wikipedia editing is much more cognitively taxing than watching TV.

If there is a way to measure the correlation between economic stress and cognitive surplus, I'd be interested.

Anonymous said...
This comment has been removed by a blog administrator.
Anonymous said...

Forget economic stress. It's nothing but a furphy.
Does it not occur to anyone that the material that appeared on the Wikipedia between 2005 and 2007 was largely "ported" from pre-existing material in various other digital forms?
Once the available content pages were exhausted the rate of growth simply dropped to what we observe now. Maintenance of more volatile content on the Wikipedia also eats into the time available to active contributors.