Global Neighborhood Watch

Groups of concerned citizens have reduced street crime. Can they do the same for terrorism?

The year is 2010. A salesman, traveling by train from Dulles International Airport to Union Station in Washington, hears a beep emanate from his mobile phone. He's startled, because the sound indicates that a chemical sensor in his briefcase detects the presence of penthrite somewhere in the train car. Penthrite, one of the world's strongest explosives, is used to manufacture a sophisticated form of C-4, the plastic explosive that Richard Reid hid in his shoes when he boarded an American Airlines flight in December 2001.

The salesman quickly scans the train car and spies a beat-up-looking backpack under a seat at the far end. He realizes that his mobile phone has gone off like this before, and nothing dire has transpired. But at the next stop, a woman boards and stands next to the backpack, and a pager-like device strapped to her waist also emits a beeping noise.

As each rider's sensor detects penthrite, it alerts an agent in the National Counterterrorism Center, the U.S. government's fusion point for all terrorism intelligence. Seeing two alarms go off, the agent calls the salesman and sends a text message to the woman, asking them to describe, independently, what they see. How big is the backpack? Where is its owner? What is he wearing? The agent then enters their observations into a powerful computer. The machine quickly churns the information and looks for meaningful patterns, which, hopefully, will reveal whether there's a real attack in the offing.

This vaguely Orwellian futuristic scenario is how two network theorists imagine a country might enlist its citizens in fighting terrorists. They call it Global Neighborhood Watch. Just as a traditional neighborhood watch deputizes people living on the same block to prevent local crime, the global watch would turn participants into mobile intelligence gatherers, feeding data from chemical sensors or simply with their own eyes into a sophisticated, government-run system that would create hypotheses about what that data means.

This widely distributed, and largely autonomous intelligence network, is the brainchild of two academics who've spent their careers studying how and why people organize themselves into groups-Bill Mc-Kelvey, a professor at the Anderson School of Management at UCLA, who conducted some of the earliest categorizing of organizational forms in society and business, and Max Boisot, a senior research fellow at the Sol C. Snider Entrepreneurial Research Center at the University of Pennsylvania's Wharton School. They are inspired by a fairly unshakable piece of conventional wisdom in the post-Sept. 11 era, one which, in their view, is off the mark. Namely, that stopping terrorists before they act requires human beings to connect the dots, or data, between people, organizations, places and actions that ultimately spell out what terrorists are plotting to do.

No work reinforced the importance of dot connection to intelligence analysis more than The 9/11 Commission Report, which called for a national intelligence director to ensure that the government's 15 intelligence agencies are constantly connecting.

But McKelvey and Boisot urge caution. "Joining the dots, whilst a problem, is not the problem," they wrote in an unpublished paper they have shown to a small number of current and former intelligence officials in Washington. That's because connected dots reveal not only information, but patterns, which are far more variable and confusing, but are necessary to developing probable explanations of what's going on. On their own, dots don't describe what is happening, or what might happen. But as more dots are connected, the number of patterns they create grows exponentially. Therefore, McKelvey and Boisot wrote, the real problem isn't only collecting dots, but "finding the computational capability to process and corroborate trillions of possible patterns." The professors set their train scenario in the future, because today, they say, the computational power and the social willingness to create such a pattern recognition network don't exist.

From McKelvey and Boisot's perspective, intelligence agencies are drowning in dots. Global Neighborhood Watch, while possibly becoming a giant dot collector, would marry the filtering and cognitive power of human beings with the computational power of advanced technology-something the professors call a "socio-computational" approach to intelligence analysis.

People have an innate, and largely mysterious, ability to rapidly filter out irrelevant or confusing signals and quickly identify signs of potential danger-backpacks left unattended on crowded train cars, for instance. Only a few such signals can rapidly generate trillions of possible patterns, so computers are needed, both to find a few key patterns, based on what's in their memory about the most relevant signals of a potential act of terrorism, and then to generate hypotheses about what's happening, so officials can act.

In the case of the train riders, two sensor firings establish a link between events. The National Counter-terrorism Center agent, with the aid of the computer, uses that information to ask the watch members various questions, which add context. Is the owner of the backpack nearby? If so, what does he look like? Union Station is situated near Georgetown University Law Center. Could he be a student? Working back and forth like this, McKelvey and Boisot theorize, the center is more likely to devise a scenario than it could working on its own.

The intelligence center is up and running now. But the professors fear that with its mandate to fuse data from 15 agencies, it merely adds a layer to the top of those hierarchical silos. What's needed, they argue, is a network-oriented approach, since, after all, terrorists operate in that fashion.

"Our 'distributed' socio-computational approach gets around silo thinking," McKelvey and Boisot wrote. Silos extract information and meaning from data and then pass it up to the next level. "That is, dots [data] collected at the base get 'joined' or linked . . . by intelligence analysts in the middle of the hierarchy before being 'assessed' " at the top, they wrote.

By contrast, the distributed neighborhood watch, and its attendant computing power, focuses less on collecting many dots and more on establishing meaningful relationships and patterns from them. "Assessment does not thereby disappear, but it now operates across different levels," the professors wrote. The watch members help find the most promising patterns, and the intelligence center and government officials select the ones upon which to act.

McKelvey and Boisot readily admit the global watch's most obvious drawback: "It requires ordinary citizens to take on the role of secret agents and to snoop on other citizens' neighbors." In 2002, then-Attorney General John Ashcroft proposed a nationwide snooping program called the Terrorism Information and Prevention System-Operation TIPS-that was ridiculed and effectively dismissed. But McKelvey and Boisot aren't Washington insiders, and so likely feel more emboldened to offer up controversial and, some might say, heretical ideas.

Boisot argues that the government and the public should discuss what it would take to compel people to join citizen-based anti-terror groups, not just in the United States, but especially in countries where Islamic fundamentalism is spreading, perhaps to the chagrin of the citizenry there. With traditional neighborhood watches, which have operated for years and contribute to a reduction in crime, citizens have agreed that policing their own neighborhoods, while diminishing personal privacy, offers substantial bene-fits, he says. Similarly, societies must feel a shared responsibility for stopping terrorism and decide whether there are shared benefits in doing so, even if they don't end up reporting what they see to the government, Boisot contends.

McKelvey and Boisot are waiting to see how the intelligence establishment reacts to their ideas. Boisot says he has sent the paper to a senior CIA official known for encouraging dialogue among diverse groups in the production of intelligence analysis. A CIA spokeswoman said the official wouldn't be able to comment on McKelvey and Boisot's work.