Now that StackOverflow is a relatively mature site (approaching 300,000 questions), I think it would be quite interesting to take a look at its growth model - in other words, how the total number of questions has varied over time. What might be even more curious is the growth model for some of the more widespread tags such as c# and java.
The model of growth for the English Wikipedia is widely considered to be approximated well by a logistic function, though it has largely been near-exponential so far.
Logistic Growth Model ~ 1/(1+e^-x)
Of course, the nature of StackOverflow is quite different to that of Wikipedia. Although it was designed to in part be a wiki site, it is not often used in a very collaborative manner. It may well be logistic, exponential, quadratic, or one of a number of others...
As far as an explanation goes, here are a few points to consider:
- The number of users has increased rapidly since the beta period. More users mean more questions. I wouldn't want to guess what model user growth follows, but it's certainly much steeper than linear.
- Much of the answering (in particular the accepted answers) would seem to be done be a "core group" of the top 100 or so members (Stackers?).
- Many of what may be called common, basic, or "newbie" questions on StackOverflow have already been answered thoroughly (in many cases a number of times). This discourages the creation of new ones, or at the least they get closed soon after appearance.
So, has anyone yet analysed the statistics for StackOverflow? Is the raw data even available so that we might do so ourselves? Any info in this respect would be worth looking at in my opinion.
