That's such a deceiving graph. It started off at 910, then went up to 1010 and then down to 850. It's not that big of a big deal. That graph makes it look like it's the equivalent of the Great Depression.
There's rarely any reason to mark your y-axis from an arbitrary starting point, unless your purpose is to manipulate people (wikipedia admins missing on wikipedia?).
I generally agree, but this instance is an example of the one situation in which it is reasonable, namely, trends. There really should be a zoomed-out view as well for people who aren't familiar with past developments, but it's okay to put the main focus on the small detail.
Confusingly, by default, the graph is on a logarithmic scale. You can change it to a linear one, try it and see; it looks like its growth is constant.
Although I too would like to think we are converging towards having cataloged all knowledge, if Wikipedia's growth is slowing down, I'd bet it's because of all of its bureaucracy.
It's interesting to see that en.wp was the only project that experienced quadratic growth in the number of articles. en.wp and de.wp were pretty close in size until the start of 2006, when en.wp took off and de.wp continued with more or less linear growth.
I'm aware that the english speaking world is larger than the german speaking one, and that a feedback effect exists in large networks with little downtime (ie, no time of the day where contribution dies off, due to the geographical dispersion of english-speaking countries), but I'm still surprised that en.wp was the only project to experience quadratic growth.
A significant portion of articles on any wikipedia are about cities, towns, local governments, local businesses, minor celebrities, landmarks, &c. For major locations like New York or Berlin, there will of course be articles in many languages, but other things that are insignificant outside their locality are likely to only have an article in their native language. Take for example the Catalan wikipedia, which contains a large number of stubs with basic info on locations in rural Catalonia. [1] The English-speaking world is vastly larger in land area and population than the German-speaking world, so I speculate that the difference in size is due to the number of location-relevant articles.
If you play it at logarithmic and slow, you will see actual plateaus and pulses of activity in the smaller languages. That was pretty interesting, to me.
Take it out of log view. It seems like languages quickly get to about 100,000 articles in a very short (relatively time span). I wonder if that's because there are a few hundred thousand universal articles and which they are.
I'm pretty surprised by a number of things here. First and foremost, how popular Polish and Catalan are. Secondly, I thought that there would be vastly more English articles than any other language, and this isn't the case.
Unfortunately, on some of the smaller wikipedias (by users), automatic article creation through algorithmic translation of en.wp is fairly common. The result is a large volume of poor quality articles. Esperanto was one of the wikis that fell victim to this at first.
