Sociology's gonna get a lot more interesting -- and more math-like -- in the coming next few years, now that we can quantify mass behavior and social interactions.
I think he might be right, because what we see now is that a lot more social interaction in the broadest sense is conducted electronically and produces huge amounts of data. So, at lest, we get a chance to apply algorithmic techniques to these things. However, we're not very far along. It's more complicated to analyse biographies of criminals to devise effective crime prevention/rehabilitation measures than to figure out buying patterns in supermarkets.
