Fighting Terrorism with Google?
A couple of posts on Dave Farber's 'Interesting people' list set this thought off.
First this one
WASHINGTON - An experimental computer program designed to analyze
intelligence gave U.S. Special Forces a mission recommendation in 2000 that
some say could have prevented the Sept. 11 attacks.
In truth, though, the CIA does study foreign press, but before Sept. 11 made
little use of computers to collect and analyze classified and unclassified
information together, which is what Special Forces began doing in 1997,
enabling them to get a read-out on the terror cells.
This smells of fish to me. Computers can't analyze intelligence (unless a lot of AI breakthroughs have happened in secret). People can analyze intelligence. Computers can aggregate and help them link and inform each other.
If the 'intelligence community' worked with the kind of hyperlinking tools that the rest of us use to help make Google the best way to find anything, maybe they'd have got somewhere.
The idea of how this works isn't hard to grasp - my 7-year-old son got it straight away - but it is hard to map to an insufficiently public space, like an intranet or (in particular) a hierarchically organised intelligence network that is more concerned with 'need to know' and secrecy classification.
Today, David Reed suggested that we harness the public:
An open/transparent world reduces imagination of potential threats. It also increases the reliability of assessing actual threats.
Which tends to synergize with Moynihan's sound-bite: "Secrecy is for losers".
So here's a radical proposal: openly publish most (if not all) of the information collected by the CIA, NSA, ... to public inspection. Figure out how to avoid compromising sources where needed, but get all of it out, efficiently. Use the Internet, because it scales, rather than TV, print, and Radio, which don't.
This will enable all of civil society to become outsourcers of the costly mundane details of threat management, leaving the difficult and specialized functions to experts with specialized resources.
In this world, terrorist's ability to use the leverage of "unknown" threats and rampant paranoia of their targets to amplify their meager efforts would be dramatically reduced.
This would also mean that the emergent properties of Google indexing millions of individual human's links could come into play, as discussed in Cory's article 'How I learned to stop worrying and love the Panopticon'
AltaVista for them, Google for us
But what do they do with all of that data that they collect? Filter it for keywords? Fat chance. The volume of false positives (e.g., people talking about child pornography who aren't child pornographers) far exceeds the volume of actual criminal activity. Even creaky old Lycos gave up on plain-old keyword matching a long, long time ago.
Maybe they manually check it. After all, that approach worked for Yahoo, right? Oh, right, it didn't work. Scratch that.
Then they must use some hybrid approach: human editors and AI (Artificial Intelligence or Almost Implemented, take your pick) working in concert to tweeze out the most relevant material as quickly and efficiently as possible.