Andrew Lih of the University of Southern California closed the 2010 International Symposium on Online Journalism with research, circulated as a PDF, into whether Wikipedia was still growing.
The data had been hard to come by: the site had expanded so fast that nothing was published between 2006 and 2009, until a large data dump late in 2009. What it showed was a levelling off. Edits flattened out, new article production had been flat since 2007, and the pattern held across most language editions, Russian excepted.
Lih’s explanation was friction rather than exhaustion. The wiki editing system is arcane, and the rules governing submissions and edits had multiplied. Somewhere between an eighth and a quarter of users who start an entry never hit save. To show what that looks like he played videos of people trying to edit and failing; adding or changing anything was difficult for essentially everyone. At the start Wikipedia had a handful of simple rules, neutral point of view among them; by 2010 it had a stack of them, and each one was a barrier to entry.
The Wikimedia Foundation’s position was that the community was stabilising. Lih doubted that a stable community would be enough to keep new material coming, and sketched the alternative: a slow decline in quality, or content that is simply no longer timely. He also noted spam beginning to trickle in.