You can't - the government's argument is that you can use the metadata to usefully augment other intelligence/investigative techniques without undue violation of the privacy of, well, everybody whose metadata you've collected. If you know a terrorist has used a particular phone number you can use the historical metadata record to try to identify the terrorist's potential associates. Ex-NSA head Michael Hayden claims that's the way the data is used.

http://www.npr.org/blogs/thetwo-way/2013/06/09/190092800/bus...

Given the very limited number of hard, verifiable facts about these programs, whether you believe or disbelieve this explanation is almost a matter of faith.

Imagine 3 burner sim cards used by three individuals used to call each other with call durations of 5 seconds. Such a scenario would probably light up on the NSA's dashboard like a Christmas tree.

In the investigation of the Hariri assasination such analysis was employed by painstakingly looking into millions of phone calls in Lebonan. If I remember correctly, the UN investigation team identified 4 rings each color coded (red, blue, yellow, and purple) each ring would only call phones in their ring and the ring leaders talked to each other. So you had nodes that talk to each other and one single node would talk to one other node outside of each group. They also correlated the timing of the calls and locations in relation to where the Hariri convoy was on the day of the assassination (eg. a green call made as Hariri left his palace.)

I can't remember the details but the report was a very interesting read.

http://www.time.com/time/world/article/0,8599,2089420,00.htm...

http://www.cbc.ca/news/world/story/2010/11/19/f-rfa-macdonal...

One of the major advances in the last decade or so in intelligence gathering has been social-network/cohort analysis. Phone metadata makes it possible to identify relationships between people. Let's say you know two "terrorists". If you can identify, say, 3 other people who are all friends with those two then you can tentatively identify those folks as possible terrorists. If you obtain other information such as actual phone conversations, email, etc. then you might be able to confirm that, which then gives you the ability to find their friends, and so forth.

Potentially it's an effective way to identify folks who might deserve more investigation. Though it's highly questionable how effective it is as a means of stopping crime or terrorist attacks.

Basically, you're looking for connections to already known groups of terrorists. If you find that terrorists X and Y communicate with each other, and that immediately after, X always calls Z, W, and Q, then you might want to engage in deeper surveillance of Z, W, and Q. If you find that Q always calls P, M, and N, and you also find that Y regularly calls N, that might suggest that N is somebody important within the terror network.

This excellent article was on the HN front page a few days ago:

http://kieranhealy.org/blog/archives/2013/06/09/using-metada...

With no other information, you probably can't. But imagine the metadata as a graph; phone numbers are nodes, edge weights correspond to the length and frequency of calls between the two phone numbers. Clique (http://en.wikipedia.org/wiki/Clique_(graph_theory)) -like entities in that graph are people who are all in close communication, so if you know that one phone number is a suspected xyz, the others in the clique are probably of interest too.

"Nobody is listening to your telephone calls" because the NSA has voice recognition. The machine listens to the call, and saves the transcript as metadata. :)

With call length? No idea. Called numbers? Now we are getting somewhere.

But it probably also includes location, if only the cell tower used.