There's a quiet revolution underway at the CIA and its sister agencies. A new generation of analysts, determined to drag their Cold Warera colleagues into the world of Web 2.0 information-sharing, have created Intellipedia, a classified version of Wikipedia they say is transforming the way U.S. spy agencies handle top-secret information by fostering collaboration across Washington and around the world. Rolled out in 2006 to skeptical veterans at CIA headquarters in Langley, Va., Intellipedia has grown to a 900,000-page magnum opus of espionage, handling some 100,000 user accounts and 5,000 page edits a day, according to the CIA and the Office of the Director of National Intelligence.
Its advocates claim Intellipedia is not just a sign of change at the agency, but that it is also producing results. The first time chlorine was used in an improvised explosive device in Iraq, someone created a wiki page asking what intelligence officers and others in the field should do to collect evidence of the usage. "Twenty-three people at 18 or 19 locations around the world chimed in on this thing, and we got a perfectly serviceable set of instructions in two days," says Tom Fingar, who headed the National Intelligence Council from 2005 to 2008. "Nobody called a meeting, there was no elaborate 'Gotta go back and check with Mom to see if this is the view of my organization.' " Last year traffic on Intellipedia became so heavy that the Office of the Director of National Intelligence had to find extra money to upgrade its servers. (Read "The CIA Scandals: How Bad a Blow?")
Intellipedia's godfather is CIA analyst D. Calvin Andrus, who wrote a paper in 2004 titled "The Wiki and the Blog: Toward a Complex Adaptive Intelligence Community." For decades, the U.S. intelligence system had been structured to answer static Cold Warera questions, like how many missiles there are in Siberia. What the U.S. needed after Sept. 11, Andrus argued, was something that could handle rapidly changing, complicated threats. Intelligence organizations needed to become complex and adaptive, driven to judgments by bottom-up collaboration, like financial markets or ant colonies or Wikipedia. (See the top 10 Secret Service code names.)
Sean Dennehy, 39, and Don Burke, 43, used the Andrus paper to push the idea of an intelligence-community wiki on their superiors at the CIA. They didn't get very far until the then newly organized Office of the Director of National Intelligence concluded that the idea had potential and even then it faced stiff cultural resistance. "There's been pushback throughout the whole process," says Burke. Initially, analysts who were asked to participate said they were too busy or just preferred the old, proprietary databases managed by individual agencies.
One of the biggest hurdles was convincing security-minded spies that the system would be safe from outsiders. To assuage them, Intellipedia was built into the existing secure and classified networks known as Intelink, which connects the 16 spy agencies in the U.S. as well as the U.S. military, the Department of State and other agencies with access to intelligence.
After three years, Intellipedia is humming. It operates in three spheres: unclassified, secret and top secret, with top secret being the most active, boasting 439,387 pages and 57,248 user accounts. Intellipedia is largely managed by volunteers and patrolled by "shepherds" who keep track of individual pages in their areas of expertise. Like Wikipedia, articles are created instantly a page on the Mumbai terrorism attack last November was up within minutes of the news breaking but authorship must be clear; there are no user names to hide behind. (See pictures of two days of terrorism in Mumbai.)
Just how far the changes will go remains to be seen. The intelligence community's other ventures into social media have been less successful. Last September, the Director of National Intelligence rolled out a social-networking site called A-Space, with linked video and photo programs. A-Space has some 8,619 accounts, all of them top secret, but insiders say it is troubled and slow to get off the ground; at one point it was suspended because particularly sensitive intelligence was misused. New efforts at tagging and instant-messaging have also been slow. (Read "Six Ways to Fix the CIA.")
And Intellipedia's boosters concede that their wiki is still largely an adjunct to the work of America's intelligence analysts. No finished intelligence product for decision makers is generated from Intellipedia National Intelligence Estimates (NIEs) are still written the old-fashioned way, authored and circulated for peer review and consensus. When Tom Fingar tried in 2006 to produce an NIE on Nigeria using Intellipedia, he failed because it generated a stream of information rather than a formal thesis and was taken over by the traditional system.
But Greg Treverton, director of the Rand Corp. Center for Global Risk and Security, says the problem isn't that Intellipedia can't produce NIEs but that decision makers rely too heavily on such reports to begin with. "There's much too much concentration on finished intelligence," Treverton says. "Intelligence analysis should be a sense-making exercise, a process of working on problems and trying to get sharper at them. Intellipedia is ideal for that. If you slice it at any given time, you are saying, 'Here is the best state of understanding at the moment.' "