Swarming Information

To net terrorists, abandon bureaucracy.

The now infamous failure of national security agencies to share information that might have prevented the Sept. 11 attacks led Congress, the White House and the organizations themselves to attempt major changes. But for the most part, those reforms have been plagued by familiar problems: turf wars, slow-moving bureaucracy and a perception that the compartmented nature of intelligence work precludes collaboration.

Lessons from a rare example of information-sharing success and a spontaneous grass-roots response to disaster could hold a novel recipe for transforming the intelligence community.

In 2003, CIA veteran John Brennan took on one of the biggest challenges of his career: getting members of the so-called intelligence "community" to work together on the terrorist threat by standing up the all-source, all-agency Terrorist Threat Integration Center. To succeed, he had to get warring national security fiefdoms -- the CIA, the Defense Intelligence Agency, the National Security Agency, the FBI, the State Department and others -- to share intelligence and personnel.

Brennan linked his center's computer to the databases from all the three-letter agencies, giving his analysts direct access to more intelligence than any other federal organization in history. The center's real success came from those analysts. Although they came from organizations that have battled to control turf and information, inside the center they worked side-by-side without incident, sharing information and collaborating on analysis.

"Within [the center], I never had an instance of institutional rivalry," recalls Brennan, who retired from TTIC -- since renamed the National Counterterrorism Center -- in August 2005. "Your badge designating you as CIA or FBI or DoD quickly becomes irrelevant as you come together to work in a collaborative environment."

"I wish I could take credit," says Brennan, now a private consultant to government counterterrorism programs. "But if you put people together, give them a very important mission and give them more information than they've ever had access to, chemistry develops."

Brennan might be modest - but he's also on to something. "If you move below the executive level to the analysts who do the work . . . they can figure this [information-sharing] stuff out on their own," says Randy Pherson, a former career CIA analyst who now trains FBI analysts and consults with the Homeland Security Department's intelligence operations. "The little guy knows more about what he's trying to do than the big guy."

High Stakes

Indeed, high stakes seem to bring out the best in people. Less than two hours after Hurricane Katrina hit the Gulf Coast 0n Aug. 29, for example, volunteers had set up a "wiki" -- an open, collaborative online tool -- to be a central repository for information to help survivors find aid and volunteers to provide it.

As officials struggled to straighten out their bureaucratic hierarchies, the Hurricane Katrina Help Wiki (www.katrinahelp.info) quickly became a more comprehensive source of information than any government outlet, helping tens of thousands of people use dozens of largely uncoordinated public and private efforts to save themselves and their pets, find food and shelter, and locate loved ones. Thousands contributed information updated postings and made corrections.

To be sure, the wiki was an even more radical experiment than Brennan's center: There was no Brennan to run it. The handful of organizers were facilitators, not directors. They couldn't order people to participate or point to a White House directive for authority. They had little control over the site or its contributors. Yet the result was stunning.

Such self-organizing collaborations -- "swarms," as they've come to be called -- are in some cases more effective, efficient and resilient than other organizations, researchers have found. Responsibility is shared, participants feel more invested and oversight is spread to the edges of the group instead of hoarded at the top. Still, their success is something of a mystery.

"The system works, but sometimes you can't really explain why it works and how it works," says Eric Bonabeau, a Cambridge, Mass.-based researcher and consultant to the public and private sectors, who has studied the phenomenon. "Its behavior is the result of myriad interactions." Bonabeau has co-authored two books on the subject, Intelligence Collective (Hermès Sciences, Paris, 1994) and Swarm Intelligence (Oxford University Press, 1999).

For the system to work in government and business, managers must let go, he says, and few executives (in government, especially) are willing to embrace a solution they can neither completely control nor explain, even for intractable problems. "Managers would rather live with a problem they can't solve than with a solution they don't understand," Bonabeau says.

A government executive's aversion to the swarm approach might be as prudent as it is conservative. Throwing a group of people into a room and telling them to solve a problem holds little appeal to someone who might have to testify before Congress if the effort fails.

On the flip side, removing "ownership" is part of what might make swarms work, particularly interagency collaborations such as Brennan's center. "Ownership implies hierarchy implies authority, which implies command over resources," says former CIA analyst Pherson. "Most of that doesn't work for [information-sharing]."

Sharing intelligence isn't the only information challenge facing the national security community. Despite efforts by the FBI, Homeland Security and others, there is no effective system for passing data and analysis among local, state and federal agencies involved in counterterrorism, nor for sharing it with the private sector.

The most recent attempt, the top-heavy federal Homeland Security Information Network, is collapsing under its own weight, according to news accounts. Created as a system for informally sharing tips, questions and information at all levels -- from city cops to federal intelligence analysts -- the network now is used mainly by DHS to broadcast information that state and local recipients say is of little use.

Communities of Interest

What better to replace a failing top-down bureaucratic methodology than a networked anti-bureaucratic one? Pherson has hatched just such an approach. He calls it "Agile Global Intelligence Network of Networks."

In Pherson's schema, the thousands of people who should participate in HSIN -- local law enforcement officers, security staff at factories and power plants, bank security personnel, state homeland security officials, emergency managers and more -- would form communities of interest, using something as simple as an e-mail listserve.

For example, a security official for a Houston oil refinery might belong to a community for refineries, another for the Houston area and perhaps one for port security. It would be up each official to join the most useful groups.

The result, predicts Pherson, would be an overlapping network of networks that would quickly and efficiently circulate information not only up and down the local-state-federal ladder, but around and among users, public or private sector. The system is smarter about sharing, and protecting, information, Pherson says, because "you're imposing human brain filters at every step of the way."

In November 2004, Congress assigned the director of national intelligence to create a networked "trusted environment" in which intelligence agencies can share information and work together. The effort, has been slow in starting up, and its first act -- creating what amounts to an online phone book of the intelligence world so specialists can find their counterparts -- was both necessary and alarmingly rudimentary.

Could self-organizing collaborative communities then grow organically from that? Skeptics say no.

"Folks in the intelligence community aren't in the same gene pool as folks who are doing self-organizing, joining things," says Linda Millis, head of the Markle Foundation's National Security Program. Her group has proposed a complex information-sharing system, known as the SHARE Network, which served as the basis for legislators' instructions to the director of national intelligence.

With two decades of intelligence experience, Millis reflects the thinking of many seasoned veterans and senior officials. And perhaps they are right -- although Brennan's success argues otherwise. If doubters are correct, then even the most sophisticated network in the world won't fix the information-sharing problem. And who will explain that to a Congress and America next time the system fails?