Just how does the NSA piece together all that metadata it collects? Thanks to "newly disclosed documents and interviews with officials," The New York Times today shed light on how the agency plots out the social activity and connections of those it's spying on. Up until 2010, the NSA only traced and analyzed the metadata of emails and phone calls from foreigners, so anything from US citizens in the chains created stopgaps. Snowden-provided documents note the policy shifted later in that year to allow for the inclusion of Americans' metadata in such analysis. An NSA representative explained to the NYT that, "all data queries must include a foreign intelligence justification, period."
During "large-scale graph analysis," collected metadata is cross-referenced with commercial, public and "enrichment data" (some examples included GPS locations, social media accounts and banking info) to create a contact chain tied to any foreigner under review and scope out its activity. The highlighted ingestion tool in this instance goes by the name Mainway. The NYT article also highlights a secret report, dubbed "Better Person Centric Analysis," which details how data is sorted into 164 searchable "relationship types" and 94 "entity types" (email and IP addresses, along with phone numbers). Other documents highlight that during 2011 the NSA took in over 700 million phone records daily on its own, along with an "unnamed American service provider" that began funneling in an additional 1.1 billion cellphone records that August. In addition to that, Snowden's leak of the NSA's classified 2013 budget cites it as hoping to capture "20 billion 'record events' daily" that would be available for review by the agency's analysts in an hour's time. As you might expect, the number of US citizens that've had their info bunched up into all of this currently remains a secret -- national security, of course. Extended details are available at the source links.