The Imagineers of War

by Sharon Weinberger

  Back at DARPA, Lee was not anywhere near hunting down insurgents; he was still working to get his ideas off the ground. When he arrived at the agency, he faced, as he put it, “a steady stream of not necessarily trustworthy defense contractors just one after another coming through my little office with ideas.” It was, as he described it, a “turbulent time.”

  Lee had a fortunate alternative to the slick defense executives lining up outside his office. Dugan had assigned him a group of military officers working at DARPA on a short-term basis, a sort of professional internship called the Service Chiefs Fellows Program. Normally, the officers toured military laboratories and did not do much substantive work, but Dugan wanted them to actually create a project with Lee. Soon, Lee and the fellows brainstormed a contest based on the DARPA Grand Challenge, only instead of racing robotic cars, contestants would use social media in something resembling a national treasure hunt. The fellows proposed having teams compete to locate red weather balloons that DARPA would release across the United States. Lee was not sure about the idea: having people hunt for balloons sounded a little odd, even for DARPA, but Dugan encouraged him. “That idea might be stupid, but that’s what you came up with yesterday, so you’re going to execute,” he recalled her telling him.

  Like the Grand Challenge, the Network Challenge, as it was called, was a contest, but on a smaller scale. DARPA offered a $40,000 prize to the first team that could, on a specific day, identify the locations of the ten red weather balloons placed across the United States. The idea was that teams would use social media to help locate the balloons. The contest would test the teams’ ability to leverage a network, figuring out how to motivate people to participate while weeding out possible fake sightings, and to do it quicker than other competitors. On December 5, 2009, the day of the challenge, Lee’s biggest fear was that no team would identify all the balloons, undermining the point of the challenge. In the end, it took only nine hours for a team from MIT to win. They beat the competitors by using a sliding scale of financial incentives that rewarded not just those who spotted balloons but those who recruited others who successfully spotted balloons. Alex “Sandy” Pentland, an MIT computer science professor who headed the winning team, called the task “trivial.”

  Pentland had reason to be self-confident. He had already established a reputation as one of the nation’s leading “big data” scientists. Long before Google Glass, Pentland had written about wearable sensors that would record everything the user sees, hears, and experiences. His specialty was sifting through data to predict patterns of human behavior, an area he called “social physics.” Pentland’s team had created a novel financial incentive system based on the assumption that people’s actions are dictated not purely by profit but by intangible benefits that come from exchanges that strengthen someone’s position in his or her social network. “If you look at the models for incentives, or for management through the army, through companies, through economics, they’re all about individual incentives, and they ignore the social fabric. What I just said about red balloons was that it wasn’t about economics; it was about the social fabric,” Pentland said.

  Pentland theorized that someone’s position in the network—his or her social standing—was the primary motivator. In his calculation, people act to make their social fabric stronger, not necessarily just to earn a bit of money. “I give you a favor. Maybe in the future you’ll give me a favor. That’s what drove this thing,” he said. “That’s a very different way of thinking about things. Instead of paying attention to individuals, you pay attention to relationships.”

  Taking what was learned from the Network Challenge and moving it into a formal DARPA program was the next step, and Lee again had a bit of good fortune. Randy Garrett, an NSA official involved in big data, had recently moved over to DARPA. At the NSA, Garrett had been a key official for RTRG, the program that had helped track and kill insurgents in Iraq. Garrett had also been working on creating a data cloud that would allow analysts to search through real-time data as it was vacuumed up by intelligence agencies. This cloud would include “essentially every kind of data there is,” Garrett said. There were some obvious parallels between the work of the NSA and the Network Challenge. Garrett’s NSA work had focused on integrating large streams of data in real time to spot something of interest, such as insurgents. The Network Challenge did roughly the same thing using social media data and red balloons. The national security establishment has an enormous amount of real-time data at its fingertips, and the biggest source of that, of course, is the NSA, which intercepts millions of calls a day around the world, in addition to various forms of Internet traffic, from e-mails to Skype calls. Afghanistan, after ten years of war, was one of the NSA’s top targets for cell phone interception. Now some of that data was about to be made available to DARPA.

  “Someone made the observation and brought it to my attention that it might be possible for DARPA to get direct near-real-time access to several hundred data intelligence feeds from the theater in Afghanistan,” Lee recalled. “I thought that was very interesting. Most of the data feeds were classified only at the secret level. Some were even unclassified. One immediate question was, what might be possible if we did large-scale data mining on all of those feeds?” Lee began to contact all the experts he knew in data mining, including Werner Vogels, the chief technology officer at Amazon, who “provided a lot of framing for how we would approach this problem because it’s very similar to the kinds of data mining that Amazon does on its customers.”

  What Lee eventually formulated was a data-mining program based on the latest predictive analysis work being done in the commercial sector, but using military data from Afghanistan. “For example, we were trying to understand if the price of potatoes at local markets was correlated with subsequent Taliban activity, insurgent activity, in the same way that Amazon might want to know if certain kinds of click behaviors on Amazon.com would correlate to higher sales of clothing versus handbags versus computers,” Lee said.

  Big data was about to be enlisted in a program to predict whether a village in Afghanistan was being taken over by the Taliban or when insurgents might plan the next attack. More important, big data was going to take DARPA back to war.

  —

  In February 2010, just two months after Lee’s red balloon contest, Dugan did something that no DARPA director had done since the Vietnam War: she traveled to a war zone to see what the agency might be able to contribute. General Michael Oates, the head of the Joint Improvised Explosive Device Defeat Organization, the Pentagon’s bomb-fighting agency, invited Dugan on a three-day tour around Afghanistan. Military personnel expressed surprise to see her. “You’re from DARPA,” she recalled their general reaction. “We call you when we have three- to five-year problems.”

  When Dugan got back to Washington, D.C., she assembled the office directors and their deputies and gave them a month to come up with ideas for technologies DARPA could contribute immediately to the war in Afghanistan. Dugan already had her own ideas for projects in Afghanistan as well. At RedXDefense, the family-run company Dugan founded, she had developed a theory of bomb detection with a copyrighted slogan called the “Bookends.” The “books” were the weapons that insurgents used, while the “ends” were the terrorist organizations that built and placed the bombs. Her theory was that defeating IEDs required identifying bomb makers and bomb-making facilities, rather than trying just to detect the bomb. (It was not a particularly unique theory: the Pentagon’s bomb-fighting agency’s own slogan was “Defeat the Network.”) After her return from Afghanistan, Dugan developed a slide briefing summarizing her ideas:

  BIG BREAKTHROUGH…Bookends suggests that fighting in the books is wrong…[The] [o]nly thing that works there is humans/dogs…What if the key to boosting performance in the books is simply that we get more eyes on target?

  More noses?

  Dugan described several proposed DARPA programs for Afghanistan in that same briefing. One, called More Noses, wa
s a plan to send several hundred dogs outfitted with sensors and GPS trackers. Normally, when bomb dogs smell explosives, they are trained to sit down, alerting their handler to a possible threat. With Dugan’s proposed program, hundreds of dogs would fan out over a specific area in Afghanistan and operate off leash, sniffing out possible bombs. When a dog sat down, that is, sensed a possible bomb, the sensors the dogs wore would send a signal back to the person monitoring the data remotely. More Noses equipped dogs with sensors; More Eyes, another new program, equipped people with sensors. The people, or Afghans to be specific, would be given smart phones, which they could use to send back information about possible threats. More Eyes, according to Dugan, would use the “newest social networking” techniques to create “a civilian populace reporting capability.” More Eyes, together with More Noses, would create an “offense” system to track down IEDs.

  By April, DARPA had identified about a dozen projects that could have an immediate effect on the war, and then Dugan narrowed those down to a final list. The technologies ranged from a blast gauge that would go in soldiers’ helmets to detect exposure to possible blast waves from IEDs to an imaging sensor, called the High Altitude LIDAR Operations Experiment, that could be used to create three-dimensional maps of Afghanistan. Dugan’s priority, however, was a new program based on Lee’s big data work, called Nexus 7, which would help predict insurgency in Afghanistan. In August, Dugan met with the chairman of the Joint Chiefs of Staff and laid out DARPA’s plan for Afghanistan. The data-mining project, her briefing noted, would “sequester [a] team of the Nation’s leading researchers in large scale computation techniques and social science.” She called Nexus 7 “the potentially big win.”

  Key members of the Nexus 7 team came from Sandy Pentland’s Human Dynamics Lab at MIT, drawing on the same ideas that drove his team’s win with the balloon contest and applying them to an entire society. Pentland described his contribution as informal, providing more of an intellectual framework than nuts-and-bolts work. “When Nexus 7 started up, some of my students went and joined it,” he said. “My role was to cause people to realize that there was something different that could be done, something qualitatively different than anything they’d ever done before.”

  Forty years earlier, another MIT scientist, Ithiel Pool, had promised DARPA he could use science to help the Pentagon understand the dynamics of insurgency. Dugan might not have realized it, but DARPA had just created a new version of Simulmatics’ 1960s-era “people machine.” Pentland called his version “computational counterinsurgency.”

  —

  In the summer of 2010, Nexus 7, a data-mining program named after a humanoid robot in the movie Blade Runner, was launched with a former NSA official as its head. Similar to his work at NSA, Garrett’s goal with the DARPA program was to “actually build this big data aggregated environment, a cloud, and then see how you would use it.” The program was a direct carryover of work started at the NSA, according to one scientist involved in creating Nexus 7.

  In budget documents, Nexus 7 was obliquely described as a program that combined data analysis and forecasting with social network analysis. “For the military, social networks provide a promising model for understanding terrorist cells, insurgent groups, and other stateless actors whose connectedness is established not on the basis of shared geography but rather through the correlation of their participation in coordinated activities,” the description stated. “Nexus 7 supports emerging military missions using both traditional and non-traditional data sources for those areas of the world and mission sets with limited conventional Intelligence, Surveillance and Reconnaissance.”

  DARPA brought in the researchers and contractors working on Nexus 7—about two dozen computer scientists, social scientists, economists, and counterinsurgency experts—and took over the tenth floor of the agency’s headquarters for a brainstorming session. The meeting included big data gurus, like Sandy Pentland, to provide technical advice, and L. Neale Cosby, a retired army officer, who was there to help provide an operational perspective. The question, said Cosby, was, “how can we take all that data that comes in and streams, minute by minute, second by second, into [the NSA at] Fort Meade and other places and use that data to make sense in assessing the actual security of a village in a place like Afghanistan?”

  The direct relationship between the NSA and DARPA was one of the hallmarks of Nexus 7, but it was also the most problematic, because working with data from the NSA required navigating a maze of legal and statutory requirements that often prevent sharing and aggregating data among government agencies. As for why DARPA wanted the NSA data, Cosby invoked the famed bank robber Willie Sutton: “Why rob a bank? Because that’s where the money is.” The NSA was the bank; it had all the data.

  Dugan believed she could avoid the scandal of Total Information Awareness, which had dragged DARPA into a national privacy debate in 2003, by restricting the work to a war zone. Total Information Awareness wanted to find the terrorists that might be lurking in the United States, which was bound to concern privacy advocates, while Nexus 7 was focused on Afghanistan. More important, she also made the entire program secret. DARPA workers one office down had little idea what was going on when the group of young computer scientists set up shop in the agency’s headquarters. “We had to have cover stories to tell people if various Beltway people came to visit me in my office and they were walking through this pandemonium,” Lee said.

  The Nexus 7 program was different in many respects from a normal DARPA project. Typically, DARPA contracts work to universities or businesses, but the core of Nexus 7 was Peter Lee, who created the program and ran it from his office. “Nexus 7 turned out to be a bunch of desks, laptops, and secure computers literally in the hallway outside of my office,” Lee said. “It was just a zoo.”

  Not all the work was done at DARPA, however. The agency recruited David Kilcullen, an Australian counterinsurgency expert who had advised American government officials, including General Petraeus. Kilcullen by 2010 had moved to the private sector and headed a company called Caerus Associates, selling his services to government clients. Nexus 7 meshed well with Kilcullen’s belief that metrics, ranging from the cost of transportation to the price of exotic vegetables, could be used to gauge a population’s susceptibility to insurgency.

  DARPA was creating an intelligence program far more ambitious than anything John Poindexter had attempted with Total Information Awareness ten years prior. In those days, the work was focused on predicting large-scale terrorist events—plots that might require complex, long-term planning. Nexus 7 was in the weeds, looking at patterns of daily life, to make specific predictions on the ground in Afghanistan. “We were really using the latest research in quasi-experimental design, in machine learning, and data mining literally on hundreds of intelligence feeds to make inferences about what would happen next,” Lee said.

  According to Dugan, Nexus 7 started making its “first discoveries”—or meaningful predictions—eighty-two days into operation. Over the weekend that those results came in, Dugan briefed the marine general James Cartwright, the vice chairman of the Joint Chiefs of Staff, to get an official green light for Nexus 7 and its personnel to deploy to Afghanistan. Cartwright, a technology enthusiast, embraced her ideas and her approach. His response, said Dugan, was “Go and go faster.”

  Before Nexus 7 made it to Afghanistan, its creator, Peter Lee, abruptly left DARPA after less than a year to become the head of research at Microsoft. On the day he left for Seattle in September 2010 to start his new job, the Nexus 7 team, some members as young as their mid-twenties, was departing for Afghanistan. “I should have been with them,” he said regretfully.

  DARPA would eventually deploy more than a hundred people across Afghanistan, working on Nexus 7 and other technology programs. “It was the first operational deployment from DARPA since the Vietnam War,” Dugan later recounted. The program also became Dugan’s top priority as she shuttled back and forth to Afghanistan with General Cartwr
ight. In congressional testimony in 2011, Dugan did not use the Nexus 7 name, but simply described “a 90-day Skunk Works activity” that involved scientists and counterinsurgency experts working on “crowd sourcing and social-networking technologies.”

  Nexus 7 went from inception to execution in just a few months, and it was not without hiccups. General Stanley McChrystal, the head of the International Security Assistance Force in Afghanistan when Nexus 7 started, was interested in the data-driven work promoted by DARPA. But in 2010 he was forced to resign after a Rolling Stone magazine profile, which depicted him and his staff as mocking senior White House leaders. General Petraeus returned to Afghanistan to take over, but he was not enthusiastic about Nexus 7. A disastrous meeting between Petraeus and Dugan in Afghanistan almost brought it to a halt. DARPA’s proposal for algorithms did not sit well with a general who believed he wrote the book, metaphorically and literally, on counterinsurgency.

  At that point, however, Nexus 7 had support from General Cartwright, and soon the DARPA team, or what Dugan called the “DARPA army of technogeeks,” started showing up in Afghanistan. They were young and had no military experience, and the culture shock soon became apparent. Military officials in Kabul were reluctant to share intelligence with computer scientists just out of graduate school, and the intelligence they did provide was not nice and neat, like consumer data. Once in Afghanistan, the analysts began to gather up as much intelligence as they could: phone records from the NSA, radar feeds from the military, and intelligence reports. But much of the data that came into Nexus 7 was qualitative, rather than quantitative, which was not easy to plug into a computer program. Even when the data was quantitative, like from radar, it rarely covered the exact same place over time.