by Shane Harris
“Learn from their example,” she told Poindexter. Don’t make the same mistakes. Think about how people react to names, to perception.
“That’s why I’m here,” he said. “Help me figure out how to avoid that.”
Townsend said that to win over lawmakers, privacy advocates, and individual citizens Poindexter would have to persuade them not only of TIA’s value but that he had truly thought through the privacy protections. He’d have to convince them that selective revelation and privacy appliance weren’t just buzzwords, lip service to the critics. If he couldn’t do that, Townsend warned, “the public would react badly.”
Though she didn’t say it, ultimately, TIA’s fate all depended on Poindexter. Before the public could accept the idea, they first had to accept the man.
Townsend was taken aback by Poindexter’s willingness to listen. She had imagined that he’d do most of the talking. The gentle, almost excessively deferential “gentleman” that she saw before her defied her expectation of an imposing, strong-willed military man toughened by Washington warfare. It was all so disarming.
Poindexter likewise knew Townsend only by reputation. She was an assiduous lawyer and had worked on terrorism cases for years. But he knew nothing of her thinking. Now he could see she was spirited and whip smart. And while he couldn’t say the same for himself, she could cuss like a boatswain’s mate.
Townsend was one of the few senior officials in Washington to have worked counterterrorism for both the Clinton and Bush administrations. Poindexter had a way of gravitating to these careerists, people like Mary McCarthy and Richard Clarke. It was no accident. In his relationships, professional and personal, he made no partisan distinctions. As long as someone was loyally committed to the fight, he didn’t much care how they cast their votes. It didn’t hurt that Townsend was politically more aligned with the current president than the last one, but it didn’t particularly matter to Poindexter either way. That meeting at Coast Guard headquarters was the first of many, and their relationship accrued new value when, several months later, Townsend took a pivotal job at the White House.
As Poindexter shuttled around Washington in the winter of 2002, he left every meeting with the feeling that his audiences understood his approach and were generally supportive of it. At least outwardly. What they said when he wasn’t in the room, he didn’t know. And indeed, many of those supportive officials had deep misgivings about the idea and the man in charge.
For the moment, Poindexter opted for a low public profile. No big speeches. No unveilings just yet. But with him making the rounds of so many chiefs of the spy community, word about his radical proposal to mine private information was going to get around on Capitol Hill. And when it did, Poindexter would find himself in combat, once again, with his oldest nemeses.
Perhaps no one understood that better than Mike McConnell. As a former director of the National Security Agency, and he knew firsthand the extraordinary problems Poindexter would encounter trying to expand the government’s access to data on American citizens.
Poindexter wanted McConnell’s advice. He was a fellow admiral, and Poindexter had followed his career admiringly over the years. During the Gulf War McConnell was the top intelligence officer to then chairman of the Joint Chiefs Colin Powell. Poindexter thought he’d performed superbly, and he’d came to value his opinion when McConnell, after leaving government, attended the Genoa demonstrations and became a major player in the world of intelligence contracting.
The two met in McConnell’s office at Booz Allen Hamilton, a brand name inside the Beltway. McConnell had led the intelligence business there since leaving the NSA in 1996. McConnell was frank in his assessments of what Poindexter was up against. He was heading for a political buzz saw.
Both men could see it. McConnell was concerned that Poindexter would end up creating dossiers of innocent people, and he wasn’t sure that Poindexter appreciated just how disastrously that would play out with Congress and the public. Lawmakers would give Poindexter his biggest grief, but McConnell was prepared to provide “top cover” by talking to senior members and committee staff about TIA, with an eye toward supporting it. McConnell knew how to reach all those key influencers that Townsend had said Poindexter should seek. He would emphasize that TIA wasn’t an operational system (that dirty little word again). McConnell also told Poindexter that he mustn’t portray TIA as a domestic intelligence program.
McConnell could broach the subject of surveillance laws and regulations with far more credibility and less drama than Poindexter. He had enjoyed an impeccable Navy career, and his name counted for a lot in Washington, particularly among senior members of the Bush administration. He could clear Poindexter’s path. That sounded, to Poindexter, like an excellent idea.
But McConnell didn’t want to stop there. He told Poindexter that he should award Booz the entire TIA contract, letting the company effectively take over the research, testing, and construction of the prototype, a soup-to-nuts arrangement.
McConnell and his team arguably possessed the technical expertise and the manpower to handle the entire “system of systems” effort. And he thought he knew how to weather the political controversy. But Poindexter was reluctant to give one company—one man—so much influence over TIA’s development. Business was booming after 9/11. TIA would hardly be the only jewel in Booz’s crown.
“I’d like for you to participate in some way,” Poindexter said, leaving the door open. He could sense McConnell’s disappointment. “I hope that you’ll propose under the BAA,” he added, referring to his broad agency announcement out on the street.
McConnell would have to settle for less than the full project. Yet his ultimate prize was a handsome one. under a contract Poindexter awarded later that year, worth more than $8 million, Booz was tapped to help bring a prototype TIA system to life. McConnell and company would get TIA out of the lab and into the hands of government users, a process that Poindexter dubbed “assured transition.” Poindexter would create the system, but McConnell would spread it around. His connections, and those of his company, would pay off after all.
McConnell’s offer of top cover was not part of the final deal. When Tony Tether reviewed the provisions of the contract, he struck that portion. “We don’t need this,” he told Poindexter.
Poindexter pushed back, arguing that DARPA needed McConnell’s influence. The agency only had one full-time legislative affairs specialist and one PR person. Tether was not persuaded.
Poindexter had worried since coming to DARPA that Tether wasn’t facing up to TIA’s political realities. It wasn’t that he didn’t want to address the controversy, Poindexter thought. He just seemed not to imagine there was one. “They haven’t come up over the hill at us yet!” Tether was fond of joking. But Poindexter knew that if TIA exploded onto the national scene in the wrong light, he’d be left to defend it on his own.
Bob Popp sat at his desk, thinking about a sandwich. It was nearly lunchtime, and he’d spent much of the day reviewing sketches for the Information Awareness Office’s new logo. Every DARPA office had its own custom design that was supposed to convey, elegantly and concisely, its animating ideas. Poindexter had asked Popp to hire an artist, who came up with several conceptual, abstract drawings. Popp thought they would look more at home in a modern art museum than a Defense Department office. “Go look at the other logos,” he told the artist. “They have three letters, with a little leaf, or a lightning bolt.” Nothing exotic.
The artist took Popp’s conservative approach to the other extreme, returning some sketches that used nothing but the office’s initials: IAO. This was harder than it needed to be, Popp thought. As he sat at his desk struggling to make the image work, his secretary came and asked if he wanted lunch from the deli. Perfect timing. Popp handed her some cash and returned to the sketches.
A few minutes later the secretary returned with Popp’s lunch and put his change on the desk. As he reached for the loose bills, he stopped short. A one-dollar bill had l
anded a few inches from one of the drawings. Popp stared at the great seal on the back of the note, a pyramid topped by a shining, all-seeing eye.
“Hey!” he yelled out to his secretary. “Look at that!”
“The eye. Like the letter I,” he said. And the pyramid, an A shape. IA. What could he think of that looked like an O? He thought about it for a few seconds, and then it came to him. “A globe!”
IAO. Information Awareness Office. Popp would take the eye and pyramid of the great seal but have it cast a gaze across a picture of the world. Information awareness. Total information awareness, in fact. It was perfect.
He took five minutes to sketch out a rough version. Then he showed the mock-up to his secretary. “Oh, very clever!” she said. He asked a few colleagues around the office, which by now had grown in size and moved into more substantial quarters. People seemed to like it. Popp had the sketch artist do a more polished job before showing it to Poindexter.
He had the same reaction. What a neat idea. Poindexter had long thought of the perfect information system as a pyramid, with a decision maker at the top able to reach down to the base of the structure for the information he needed. The logo conveyed just the right idea, he and Popp agreed. And it was provocative. It would get the public’s attention, Popp thought. For him, “the public” was the agencies and scientists with which DARPA did business. It was not the man on the street. He never imagined that the general public would see the logo, and so he never dreamed anyone would react unfavorably.
Popp and Poindexter noted that the great seal on the dollar bill incorporated some Latin phrases: Novus ordo seclorum, “New Order of the Ages,” and Annuit coeptis, or “He approves of our undertaking.” Could they work some Latin into the IAO seal? they wondered.
Poindexter had a favorite Latin phrase, and it seemed a perfect choice: Scientia est potentia. “Knowledge is power.” He placed it underneath the all-seeing eye, which in the final version of the logo cast an enormous searching beam of light over an image of the globe.
The people who knew DARPA, they told each other, were going to love this. The logo was published on the office’s Web site in April.
The Information Awareness Office came together in the span of a few months—a breakneck pace for the government, even at crisis tempo. Poindexter added more staff. The TIA network attracted new members every month. By year’s end Poindexter would award tens of millions of dollars in research contracts to more than two dozen companies. Some were big, well-known Washington players, others were smaller shops located far outside the Beltway.
At the beginning of his research experiments Poindexter drew a bright line in the kinds of data he would use. On the TIA Network, which was classified and restricted to intelligence and military agencies, he would tap into real intelligence, collected through foreign operations. This was the daily take of the intelligence community, and as far as Poindexter knew, it was all obtained legally.
But he needed a source of domestic intelligence as well. Something akin to the mounds of corporate and personal information that TIA was meant to mine. Unable to obtain these kinds of databases legally, Poindexter decided to build them. His team constructed a repository of simulated intelligence reports about terrorists, including fake accounts of their daily activities that left transactional footprints. The red team would use these synthetic worlds to run their exercises. In these virtual realities hundreds of thousands of innocent and ordinary electronic people would mix among a few bad actors. It was here that Poindexter would truly discover whether TIA could detect terrorist patterns. Whether it could distinguish signals from noise.
Two paths of research—one foreign, one domestic. Poindexter had no idea that at the NSA, an agency he wanted with him on his quest, the line between those two worlds had practically disappeared.
CHAPTER 16
FEED THE BAG
They called it the BAG. As in black bag. Grab bag. Bag of tricks. And while it was all those things, the acronym that the National Security Agency’s techs used for their terrorist hunting machine stood for something unexpected: Big Ass Graph.
In the late 1990s the engineers and systems gurus at the NSA became enamored of computerized graphs to display huge sets of information. Graphs were simple, and elegant. A set of axes and plotted points. Inasmuch as data was a collection of points—events, people, places—they too could be displayed on a graph in a comprehensible or meaningful form.
That was the idea, at least. Not unlike the data harvesters of Able Danger, who displayed names and events on link charts, or those of Poindexter’s Genoa team, who sought to diagram information as a series of questions and answers, the graph builders of the NSA wanted to turn raw data into visual knowledge. Graphs became their favored method.
The BAG, as its name implied, was big. Enormous. It compressed mounds of data into their linear essence. The BAG could show how swarms of people were connected to different places by displaying each set of data on its own axis. The graph could turn a seemingly random pile of information into a more complete explanation of relationships. Put enough names, phone numbers, and e-mail addresses into a graph, and it might illustrate an entire social network, which in the global terrorist hunt was indispensable intelligence. (The way to roll up a cell was to pick off the members.)
The BAG was the ultimate manifestation of graph theory. And its ultimate aim was to reveal suspicious linkages. Immediately after the 9/11 attacks, with the NSA frantically searching for the next sleeper cell poised to attack, the BAG became one of its favorite tools. But it was not entirely the agency’s invention.
The BAG was created by a computer scientist who perfected his craft far away from the gates of Fort Meade. Dr. J. C. Smart started his technical career at the Lawrence Livermore National Laboratory, about a half hour’s drive east of San Francisco Bay. Smart joined the lab in 1980, after graduating from Northwestern University with a degree in electrical engineering and a specialization in computer science and logic.
Few institutions in the world could offer a bright, enterprising computer geek such a rich history of technological innovation, and a bounding professional playpen, as Lawrence Livermore. The lab played a central role in the evolution of computing in America. A year after its founding, in 1952, it purchased the UNIVAC, the country’s first commercial computer. Year after year it bought the prototypes of the most powerful computers in the world and helped turn a generation of experimental machines into viable, marketable systems that changed the face of science, communications, and business in America.
Livermore, founded at the height of the cold war, had developed nuclear warheads and made early breakthroughs in fusion energy. The line between computer research and warfare was a thin one. In fact, each fed the other. As the nuclear arms race speeded up, the government’s requirement for ever more sophisticated and powerful simulators to help design nuclear weapons fueled the private sector’s construction of new supercomputers. At Livermore, Smart sat at the crossroads of American technology and national security.
Smart worked at the lab for nineteen years. He completed his graduate and PhD work in computer science at the University of California at Davis. In his studies and at work, Smart became an authority on graph theory and its application to national security problems. In 1996, he founded a new center at Livermore dedicated to “information operations, warfare, and assurance.” IOWA, as it was called, put Smart at yet another crossroads—the converging worlds of computer security, espionage, and digital war. Those were the NSA’s sweet spots, and so it was perhaps inevitable that Smart would make the trip east and end up at the agency’s doorstep. He brought the BAG with him.
In 1999, Smart became the technical director of the agency’s signals intelligence program. Four years later, amid the heat of the terror war, he took the same job in the agency’s National Security Operations Center. This was the heart of the agency’s early-warning system, a fitting place for a scientist whose creation was put into service for that purpose.
As was the BAG. The terrorist hunters poured signals into it, in the hopes of finding those hidden connections that would bring down a terrorist network. But if the BAG was a useful tool, it was also a demanding one. For the BAG to tell them things, the hunters had to fill it. Constantly. There was only one source that might satisfy its appetite.
The nation’s phone companies and Internet service providers owned a rich set of details about the people who used their networks. In order to hone their marketing campaigns, companies studied whom their customers called and how long they spoke, and then developed service packages to attract new business. They also used this data to create monthly phone bills. The companies would watch for communications surges at a particular time of day in order to manage the traffic on their networks. All of this data constituted a valuable repository of corporate knowledge. It was also a potential gold mine of terrorism intelligence, and after the 9/11 attacks, the NSA asked the companies to share it.
