Came back from a talk by a Facebook Data Team member a little while ago. Don't have time to look it up now, but he cited Bakshy et al 2012 as a paper describing Facebook research in which a proportion of "Likes" from friends were censored for selected users in order to gather information on a social science question. The talk was pretty incoherent on what question they were investigating; I get the impression that it was pretty innocuous, something like "more activity in your feed makes you share more," maybe. Still, it makes me wonder why people are signing up for this fishbowl.