It's the normalized search volume for the term »facebook« on Google. It is not too straight forward to interpret but I thought about it for some time and I am still convinced it is a pretty good indicator for Facebook's future.
For example users switching from desktop browsers to apps will no longer go to facebook.com by searching for »facebook« in the address bar and therefore lower the search volume. On the other hand people searching for solutions for issues with Facebook should still track the size of the user base. The same holds for developers interested in Facebook APIs, authors researching Facebook for articles and so on.
You can almost always come up with explanations why the search volume should go down over time, developers already knowing the APIs, Facebook maturing and therefore causing less and less issues for users, users maturing and therefore being better able to deal with issues without searching and so on.
But if you look at similar cases like MySpace [1], Google+ [2], ICQ [3], the German StudiVZ and MeinVZ [4], the search volume on Google was always a really good indicator for the fate of those platforms and I don't have enough convincing arguments to believe that Facebook manages to counter this trend.
Here is a collection of graphs [1] I created some weeks ago. Keep in mind that the graphs are normalized, there are huge differences in absolute search volume, i.e. if you include Facebook in all graphs, most search terms become hardly distinguishable from the time axis.
