One big graph search
According to the NSA's analysis of a single day's collection, Yahoo was the most collected source, followed by Hotmail, Gmail, and Facebook. Facebook's data was by far the most accurate, however, ranking in at 95.87% attributable (that is, gave verifiable information on a real person). As a point of comparison, the next highest was Gmail, which came in at a measly 6.97% attributable. Facebook's attribution "success rate" is probably due the social network's insistence on non-anonymity and little spam within the service. In fact, the NSA program could be described to be, in effect, one giant Facebook graph search.
As of writing this, there is no official statement on the government's data collection transparency tumblr, ICOnTheRecord. The tumblr was the result of a pledge from the administration to foster more transparency in the collection activities. During a press conference in early August, Obama criticized the methodology of leaks being "released drip by drip, one a week, to kind of maximize attention."
If one Snowden's chief media contacts, Glenn Greenwald, is to be believed, Snowden supplied him with 15-20k secret documents before finally seeking asylum in Russia. Unlike with WikiLeaks' unredacted info dumps, Snowden's media contacts seem to be taking more care to vet and redact parts of these secret documents before going public, thus the relatively slower pace of distribution. This may be yet another train car in a very long line of surveillance revelations.
Sign up for MIS Asia eNewsletters.