There was this discussion the other day about how to make a message spread throughout a "social network" efficiently (where to start spreading the message).
This article here shows how this kind of sophisticated research would actually not even really be needed to control what the public thinks.