Interesting, its always a wakeup call to see the amount of real world conclusions that can be gleaned from a (seemingly useless) dataset.
An interesting parallel to human psychology as well: often the people you are quite certain have nothing to teach you are the ones with the most valuable lessons for you to learn.
This is the paper where the "43% of facebook messages are spam" claim originates, something which I am inclined to dispute. Other than that its a great look at user habits. The graphs speak for themselves.
I don't find it entirely impossible. After Startup School I talked to Adam D'Angelo about spam on facebook (largely because I'd never seen any); he said that there is a lot of spam activity, but that internal safeguards prevent most of it from ever being visible. If the researchers were using internal, pre-filter datasets, their claim is reasonable.
Good point. I was concerned that the researches might be counting messages to groups as spam.