When looking for a needle in a haystack, there is more hay than needles

Well, the WaPo version of this detailed article reads: Surveillance Net Yields Few Suspects, but I think my title is more accurate.

Surveillance [by the NSA] 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.

You mean....when the person is ruled out as a suspect, he is no longer listened to? Sounds like, ya know, a responsible use of limited intelligence resources to me. Of course, when an attack is unexpected--and terrorists generally don't stand up in town square and announce when and where they plan to attack--there will be many false leads.

Even unwitting Americans, they [intelligence officials] said, can take part in communications -- arranging a car rental, for example, without knowing its purpose -- that supply "indications and warnings" of an attack. Contributors to the technology said it is a triumph for artificial intelligence if a fraction of 1 percent of the computer-flagged conversations guide human analysts to meaningful leads.

The data is massive, and computers do most of the listening. I had no idea the computers were this sophisticated:

An alternative approach, in which a knowledgeable source said the NSA's work parallels academic and commercial counterparts, relies on "decomposing an audio signal" to find qualities useful to pattern analysis. Among the fields involved are acoustic engineering, behavioral psychology and computational linguistics.

A published report for the Defense Advanced Research Projects Agency said machines can easily determine the sex, approximate age and social class of a speaker. They are also learning to look for clues to deceptive intent in the words and "paralinguistic" features of a conversation, such as pitch, tone, cadence and latency.

This kind of analysis can predict with results "a hell of a lot better than chance" the likelihood that the speakers are trying to conceal their true meaning, according to James W. Pennebaker, who chairs the psychology department at the University of Texas at Austin.

Remember folks, the needle in this haystack, if they find it, won't just poke your finger. The time has come to realize that intelligence is not as simple as getting a good lead on that red rotary phone from the vinyl suit-wearing buxom blonde and her dashing middle-aged mentor in the field.

posted by Feisty at 03:23 PM

| Access Comments