NSA and technology

Precisely what the Bush administration’s National Security Agency wiretap program is, technically speaking, remains unknown to all but perhaps a few people in Washington. Certain aspects of it are becoming clearer by the day, however, and no doubt more will emerge in congressional hearings. In the meantime, the more that emerges, the less the program looks like Big Brother than it does a powerful antiterrorism tool that scans millions of communications. That tool could be abused in theory, but instead already appears to be responsible for successes in the war on terror.

The latest hint of this came Friday, courtesy of Homeland Security Secretary Michael Chertoff, who described to Roll Call’s Mort Kondracke what sounds like a large and mostly automated data-mining operation. “It’s hard to talk about classified stuff,” he said, but then hypothesized about “a large volume of data, a large number of numbers you’re intercepting” and of “culling through literally thousands of phone numbers” or possibly e-mails or Internet I.P. addresses. The former federal judge and head of the Justice Department’s criminal division defended the program vigorously, calling its surveillance methods “the most significant tool in the war against terrorism.”

That follows President Bush’s distinction of the program from more familiar wiretap operations, as well as remarks by the principal deputy director for national intelligence, Gen. Michael Hayden, who said at a Dec. 19 White House press conference that the program involves “a quicker trigger. It’s a subtly softer trigger. And the intrusion into privacy is significantly less. It’s only international calls. The period of time in which we do this is, in most cases, far less than that which would be gained by getting a court order.”

If data-mining is what’s going on here, the majority of it might be automated — meaning that NSA analysts would only be examining a tiny fraction of the total data the NSA receives. Last week, in a little-noticed interview with Reason magazine, self-styled NSA whistleblower Russell Tice spoke hypothetically of “a broad brush system, a vacuum cleaner that just sucks things up” in which “a machine does 98.8 percent of your work. What you come out with from a haystack is a shoebox full of straw. Once you have that, you have people that can look at it” — after “hundreds of thousands if not millions of communications were processed.” Which would shed some light on why the Foreign Intelligence Surveillance Act would need to be circumvented for those millions of calls and e-mails.

Interestingly, Mr. Tice foresees a legal distinction between this type of activity and traditional surveillance. “[A] lawyer — and I think lawyers are going to be arguing semantics in this case — the argument could be made, well, if a machine was doing the looking and sucking in, it doesn’t matter because that’s not monitoring until a human looks at it.”

It’s not semantics; this would be an argument to take seriously. In fact, such a distinction is probably what President Bush alluded to with his seemingly arcane distinction between “monitoring” and “detecting” programs in his early January defense of the NSA program. “Monitoring” is an active wiretap; “detecting” is the less intrusive, “softer trigger” Gen. Hayden spoke of.

As it happens, though, the government’s data-mining abilities in that regard are open to question. Speaking on NPR last week, telecommunications and Internet engineer Phil Karn said the U.S. government “still doesn’t have — as far as we know — the technology that can listen to all that voice and turn it into a transcript and then scan that transcript for keywords… they still have to hire human linguists to sit there and listen to it and translate.” Barring some new unknown technology, this would limit the program to e-mail and Internet communications.

About all we know at this point is that sinister plots or eavesdroppers in vans are not what the NSA surveillance program is made of.