Facebook growth, while high, is relatively linear, which is interesting because it seems intuitively like growth (positive and negative) for social network sites should be a textbook case of an exponential relationship. Think about it - you are more likely to use a site if your friends use it (because that's the point), so the amount of new people joining should be proportional to the amount of people that are already there (which is a key characteristic of exponential growth, or decline). Assuming that most of the people that sign up to Facebook actually have friends.
Is there a way to salvage my hypothesis? Perhaps the function is logistic, and the market is near saturation. But Facebook increased its hits by 159% in the last year, which would probably not be characteristic of such a function (*).
I guess the analysis is just too simple at one factor. Perhaps we could say that these factors influence someone's decision to join a social network.
1) The amount of friends you have that use it
2) Your level of technological literacy
3) Whether or not you have regular access to a computer
4) The amount of spare time you have (**)
5) The amount you actually like your friends and want to catch up with them via the internet
So as we see a general increase in 2) and 3) we should also see a general increase in social network users.
Am I missing anything? For the record, I use facebook, because all my friends do.
(*) Because this graph refers to hits, it could be that facebook's recent newsworthy actions increased its hits without increasing its users. But I don't think that most people read a news article and then immediately try to get to the primary source, even if it is readily accessible.
(**) Or are willing to make, provided your work computer doesn't block the site...