The NSA and You

THE NSA AND YOU….The Washington Post’s big story on Sunday about the NSA’s domestic spying program provides answers to some questions and clues about others. Here’s a Q&A style roundup using excerpts from the story:

Q: How many serious suspects has the program identified?
A: Fewer than 10 U.S. citizens or residents a year, according to an authoritative account, have aroused enough suspicion during warrantless eavesdropping to justify interception of their domestic calls.

Q: How many Americans has the NSA listened in on to identify those suspects?
A: Two knowledgeable sources placed that number in the thousands; one of them, more specific, said about 5,000.

Q: How many calls does the NSA vacuum up in order to identify those thousands of targets?
A: The program has touched many more Americans than that. Surveillance takes place in several stages, officials said, the earliest by machine. Computer-controlled systems collect and sift basic information about hundreds of thousands of faxes, e-mails and telephone calls into and out of the United States before selecting the ones for scrutiny by human eyes and ears.

Successive stages of filtering grow more intrusive as artificial intelligence systems rank voice and data traffic in order of likeliest interest to human analysts. But intelligence officers, who test the computer judgments by listening initially to brief fragments of conversation, “wash out” most of the leads within days or weeks.

Q: What information is the NSA collecting in order to decide whose conversations to listen to?
A: “We debated a lot of issues involving the ‘metadata,’ ” one government lawyer said. Valuable for analyzing calling patterns, the metadata for telephone calls identify their origin, destination, duration and time. E-mail headers carry much the same information, along with the numeric address of each network switch through which a message has passed….the FISA court, as some lawyers saw it, had no explicit jurisdiction over wholesale collection of records that do not include the content of communications.

Q: Really? The NSA’s computers aren’t data mining the actual contents of conversations?
A: For years, including in public testimony by [General Michael Hayden, the deputy director of national intelligence], the agency has acknowledged use of automated equipment to analyze the contents and guide analysts to the most important ones. According to one knowledgeable source, the warrantless program also uses those methods.

Q: So is the NSA targeting people who are found in al-Qaeda rolodexes? Or what?
A: Machine selection would be simple if the typical U.S. eavesdropping subject took part in direct calls to or from the “phone numbers of known al Qaeda” terrorists, the only criterion Bush has mentioned. That is unusual. The NSA more commonly looks for less-obvious clues in the “terabytes of speech, text, and image data” that its global operations collect each day, according to an unclassified report by the National Science Foundation soliciting research on behalf of U.S. intelligence.

Q: Is the program legal?
A: The minimum legal definition of probable cause, said a government official who has studied the program closely, is that evidence used to support eavesdropping ought to turn out to be “right for one out of every two guys at least.”….Michael J. Woods, who was chief of the FBI’s national security law unit until 2002, said in an e-mail interview that even using the lesser standard of a “reasonable basis” requires evidence “that would lead a prudent, appropriately experienced person” to believe the American is a terrorist agent. If a factor returned “a large number of false positives, I would have to conclude that the factor is not a sufficiently reliable indicator and thus would carry less (or no) weight.”

I have to confess that after reading this story I’m not entirely sure how much new information it contains. It provides some figures about how many conversations are being intercepted and how many suspects have been identified, although even that’s a little hazy since the program might have identified suspects beyond just those who earned themselves a domestic warrant. It also provides some confirmation of the “common channel” theory I blogged about earlier, which suggests that the NSA’s data mining is focused more on metadata than on the content of domestic communications ? though one source suggests content is also being data mined.

Beyond that it’s hard to say more. The article combines information specifically about the warrantless wiretap program with public information about general NSA capabilities in a way that’s hard to tease apart. The most likely short explanation, I think, is that NSA has created patterns of possible terrorist activity and is actively mining communications metadata from domestic calls to find targets that fit those patterns. Data mining of the content of domestic calls may also be involved, though that’s less clear. Feel free to shoot this guess down and provide your own speculation in comments.