The Watchers

by Shane Harris

  Poindexter’s original fear had come true. He had become the center of attention.

  Doubts about TIA’s feasibility also mounted. Technology experts who scrutinized the idea thought it couldn’t avoid rampant false positives. As one skeptic put it during a congressional hearing, a system designed to flag people who rent vehicles and purchase fertilizer might catch “not only Timothy McVeigh,” who bombed the Murrah Federal Building in Oklahoma City, “but most farmers in Nebraska.”

  Poindexter had tried to explain that TIA was more sophisticated than that. It would never be used to label someone a terrorist. Human analysts, law enforcement officials, and judges would have to make that call. TIA was just a tool.

  But civil libertarians wanted to know: Were the current laws for wiretapping and surveillance really so inadequate that the country needed Total Information Awareness to protect itself? This was the response to terrorism? Sweeping up the digital detritus of ordinary people?

  Poindexter wasn’t allowed to give interviews, but the gag order didn’t apply to his number two. As the scandal swelled Popp invited two noted privacy experts over to the office at DARPA, for a firsthand look at Poindexter’s work.

  Barry Steinhardt, the lead technology and privacy expert for the American Civil Liberties union, and Jim Dempsey, from the Center for Democracy and Technology, were curious to hear Popp’s case. They had run into each other a few weeks earlier at a conference in Florida, and Popp could see that they had more questions than he could answer there.

  When Steinhardt finally saw the TIA PowerPoint presentation, he thought that Poindexter had clearly been thinking about privacy protection. But he wasn’t at all satisfied that he’d gone far enough. The presentation lacked details, Steinhardt thought. It hinted at a role for privacy but didn’t spell it out in detail.

  Midway through the meeting, Poindexter popped into the office unannounced. He introduced himself to the visitors, his pipe dangling from his mouth.

  Steinhardt had never met Poindexter. All he’d ever heard came from Ollie North, whom the ACLU had actually defended during his Iran-Contra trial. (The group argued that North’s testimony to Congress was used against him improperly; the same argument had helped keep Poindexter out of prison.) North had painted Poindexter as a technological genius, a man who kept a “supercomputer” in his basement, Steinhardt recalled.

  Whatever his technical bona fides, nothing that Poindexter said during his short visit put Steinhardt at ease. The retired admiral struck him as a modern-day Dr. Strangelove. Popp’s briefing, which lasted several hours, also didn’t convince Steinhardt that TIA was a benign program, much less one that the ACLU should support. Popp could call it research if he wished. But this wasn’t science fiction, Steinhardt thought. This was doable. Poindexter’s team was building something that would gather reams of real information on millions of real Americans.

  Steinhardt was unnerved—by the briefing, by Poindexter, by the rigorous, suffocating security in the building. He was escorted by a DARPA employee whenever he wanted to use the bathroom. Steinhardt and Dempsey left the meeting unconvinced that TIA could do all Poindexter claimed.

  But worse than that, Steinhardt thought that TIA was a harbinger. Left unchecked it led to one awful place—a total surveillance society.

  Poindexter endeavored to carry on. The wind experiments were still on course; another, Noreaster, was slated to begin in March 2003. And even amid the roiling controversy, the TIA network kept expanding, with the NSA adding nodes faster than any other agency.

  The more Poindexter tried to mitigate concerns by explaining the most rational aspects of TIA—the red teaming, the privacy protections, the fact that it was “only research”—the more desperate he sounded. Those jesuitical strains emerged again. People thought he was hiding things, proffering explanations only when asked, only after he’d been caught.

  Poindexter’s defense relied mostly on intermediaries like Hamre, or on Tether, the DARPA director, who insisted on personally giving briefings to members of Congress. But Tether wasn’t versed in the details of every office under his management. He had a hard time explaining the complexities. Whenever Tether went to the Hill, he brought Popp along for backup.

  At times Tether forgot names of specific technologies within the Information Awareness Office portfolio. The office itself and TIA seemed to blur together, into one menacing concept. In those fumbling moments Tether turned to Popp to fill in the gaps. He was mortified. The briefings felt like a shtick. Tether, Popp thought, was just making an awful situation worse.

  With the Hill and the press on watch, Poindexter now sailed under a different kind of all-seeing eye. The Defense Department inspector general launched an investigation. And in January the Senate voted to block all funding for the TIA project until the Bush administration produced a detailed report on how it would affect individual privacy.

  As far as Washington horse trading went, that was the nuclear option. If Congress was really prepared to pull the plug on funding, few avenues of escape remained for Poindexter.

  All those allies he had spent months courting remained conspicuously silent. He had been in talks with the FBI to be the first adopter of a TIA prototype. Now that plan was put on ice. To forestall a showdown on the Hill, Poindexter wrote a detailed review of TIA and the Information Awareness Office. When the hundred plus-page report went to Congress, in May 2003, Poindexter felt he’d given it his best shot. Ever the proud briefer, he considered the paper one of the best he’d written in a long time. He’d turned out every drawer, opened every file. What was left?

  Quite a lot, apparently.

  In the report’s lengthy appendix, which listed every program in the Information Awareness Office’s portfolio, lawmakers found one they hadn’t heard of yet: Futures Markets Applied to Prediction, or, as Poindexter dubbed it, FutureMAP.

  The idea wasn’t so far-fetched. Gather a pool of experts on terrorism, foreign policy, and military tactics and let them place educated bets on when terrorists were next likely to strike.

  “Will terrorists attack Israel with bioweapons in the next year?” To find out, Israel and Middle East analysts would trade contracts in an online “market”; one contract would pay out one dollar if an attack occurred, and the other would pay out the same amount if the attack never materialized. Prices and spreads would indicate the likelihood of each prediction.

  Futures markets like this had been used, with remarkable accuracy, to predict commodities prices, the outcomes of elections, and even Oscar winners. And experience had shown that unless real money was at stake, the predictions were less accurate. The incentive nature of the market compelled people to sharpen their focus and make better calls.

  When news of FutureMAP became public, economists and financial experts praised its application to the terrorist problem; indeed, the Economist magazine’s business intelligence unit had signed on to help run the program.

  But for some lawmakers FutureMAP was a ghastly step too far. As they saw it, Poindexter wanted to make money off terrorism.

  “The idea of a federal betting parlor on atrocities and terrorism is ridiculous and it’s grotesque,” Ron Wyden, a Democratic senator from Oregon, announced in a press release in late July, almost three months after Poindexter’s report had gone to the Hill. Wyden had emerged as TIA’s most visible opponent in Congress. Earlier in the year he had introduced legislation to ban all “data mining” projects at the Defense and Homeland Security departments until Congress had reviewed TIA fully. Wyden had also introduced the legislation that required Poindexter to write the full report.

  Before the FutureMAP debacle DARPA had made desperate attempts to dampen the controversy. The agency even changed TIA’s name to “Terrorism Information Awareness.” The eye-popping pyramid logo was yanked offline, replaced by an innocuous red triangle with a zippy yellow ribbon winding through it. The new logo looked like a piece of corporate stock art, as easily applied to a discount airline as a line of diet foods. It w
as meaningless.

  All attempts to save TIA had failed. “Changing the name doesn’t change the concerns,” a spokeswoman for Wyden declared. The senator was neither a powerful member of his caucus nor a particularly well-known national figure. But with the kind of material Poindexter was handing over, and the public relations tornado he created, a rank amateur could have taken him down.

  Poindexter, who’d been telling Tether that he wanted to resign since the spring, now said he would leave by the end of August. Tether didn’t protest. But when word got back to the Pentagon, rumors started circulating among the press corps that Rumsfeld had fired Poindexter, dumping him over the FutureMAP fiasco. Poindexter’s plan for a graceful exit evaporated.

  On the Hill congressional appropriators began the process of zeroing out his budget. Poindexter was ready to leave, but he knew that this could be the end of TIA too.

  It was time for Plan B.

  “I’m going to turn SAIC loose,” Poindexter told Tether. The intelligence contractor with deep roots at the National Security Agency also had a seasoned lobbying shop. Tether was getting nowhere with skeptical lawmakers on the Hill. It was time to bring in the big guns, Poindexter decided. Tether agreed.

  Poindexter asked Brian Sharkey to make the call, since he had the closest business contacts at the firm. Popp had told Sharkey about the awful Hill briefings, and he urged Sharkey to intervene for his own sake. “You’ve got a $20 million contract on the line,” Popp said. “They’re gonna shut everything down because they don’t understand what we’re doing.”

  Sharkey contacted SAIC’s legislative affairs office in Washington and explained that they needed backup. But the lobbyists rebuffed his request, afraid that if they got on the wrong side of a losing battle members might exact their revenge on other SAIC contracts. It was too risky.

  Sharkey decided to bypass the Washington office and go straight to the top. He called his friend Bob Beyster, SAIC’s legendary founder, chairman of the board, and chief executive, who was based in California. Beyster had built his company by taking chances on employees’ ideas, especially the wild ones that seemed never to have a chance.

  Sharkey had briefed Beyster on TIA already. He explained that the project was on the ropes but that Beyster’s lobbyists were afraid to go up to the Hill and save it.

  That was going to change. Beyster told Sharkey that he’d call the legislative director. Rest assured, SAIC’s team would give Poindexter all the assistance he needed.

  Sharkey and SAIC’s lobbyists fanned out on the Hill, canvassing the key staffers and members of the Appropriations and Armed Services committees. This time they used descriptions of TIA and the Information Awareness Office that Poindexter and Popp had written themselves, like the ones Poindexter used in those early rounds of ally-building talks with the intelligence agencies.

  The presentation was straightforward. Sharkey wanted to show the lawmakers what TIA was, and more important, what it wasn’t. It was tailored for the casual observer, someone who might have read about Poindexter’s super-spy machine in the newspaper but not much else. TIA had pursued two research paths, it said, one using legally obtained, foreign intelligence and the other using “synthetic data”; the words were underlined and in boldface.

  On a slide headlined “TIA is NOT . . .” the notes stressed that “data mining on U.S. citizens” wasn’t occurring, nor was TIA “searching credit card histories /library records/gun ownership records, etc.” TIA wasn’t “developing a Big Brother system to invade privacy,” the slide proclaimed. “Only foreign terrorists need fear its capabilities.”

  While Sharkey and his team mounted the defense, Poindexter called in a heavyweight. Through a DARPA contractor, Poindexter had a connection to William Perry, the secretary of defense under President Clinton. Perry was known for taking a bipartisan approach to national security, a kind of peace-maker for the warrior set. At Poindexter’s request, he came down to Fort Belvoir, to the TIA network hub. There Popp explained Poindexter’s research goals. The team wanted his advice. But more important, they wanted him to make some phone calls.

  Perry obliged. He contacted the staff of two powerful lawmakers with their hands on the Defense Department’s purse strings—Senators Ted Stevens, an Alaska Republican and the chairman of the Appropriations Committee, and Daniel Inouye, a Democrat from Hawaii. Congressional appropriators were in the process of eliminating Poindexter’s budget when Perry intervened. The irony was of a uniquely Washington variety.

  Inouye had chaired the Senate committee investigating Iran-Contra. When the embattled national security adviser appeared before him, Inouye called Poindexter’s testimony, in which he acknowledged withholding information from Congress and even the president, “incredible, mind-boggling, chilling.” Almost two decades later Inouye was effectively passing judgment on Poindexter again.

  Word came back that the appropriators were prepared to make a deal. Sharkey and his team had been making progress with their efforts too. Those in a position to save TIA from the chopping block understood that some of Poindexter’s ideas were worth keeping.

  Popp was relieved. He had planned to remain at DARPA and take over as director when Poindexter resigned. From the signals he’d received via Perry and the SAIC lobbying teams, it was clear he would have to come up with some kind of peace offering. A way for the Information Awareness Office to remain intact but perhaps not with all of its most controversial research. But it must stay at DARPA, he thought. Nowhere else in government had the vision and the discipline to see TIA through properly. Take it away from the original team and the program could fall apart.

  The Hill got the message. Except for the last part.

  TIA had to change homes. Tether in his awkward briefings had failed to make the case that his agency was capable of handling the program. Popp’s sources reported back that the lawmakers were in no mood to let DARPA hold this hot potato. The agency seemed to have strayed well beyond its normal boundaries. They were getting dangerously close to the red zones of policy, of intelligence operations.

  Popp hadn’t planned for this. In his mind he had worked out a way to keep the research alive, but he had to keep his hands on it to do that. It had come to an unthinkable decision, befitting Solomon’s court. Would he let TIA go in order to let it live?

  Perhaps it was best that he didn’t have to decide. In the flurry of backroom conversations and brokering over TIA’s fate, one candidate came forward to take over Poindexter’s programs, to pick up the shattered pieces and rebuild them. Most agencies would have shied away from such a task. But the NSA was no such agency.

  If Poindexter and Popp really thought about it, the NSA was a logical place for TIA to go. They were a signals agency, and their ears were filled with noise. Poindexter knew from his meetings with Mike Hayden, Bill Black, and others at Fort Meade that they were on similar quests.

  But Poindexter had misgivings about Hayden. He’d been picking up signals from the Hill that senior NSA officials were bad-mouthing TIA. Whispering that Poindexter’s idea was far-fetched, that it wouldn’t work. Poindexter had always suspected Hayden was personally behind the rumors. Maybe it was jealousy. Maybe ignorance. He wasn’t sure.

  Still, the NSA was full of kindred spirits. People like Poindexter who put their faith in technology’s power to solve the big problems, to protect the nation. They were true believers. The agency had a special R&D unit that could handle the work and that stood a good chance of eventually building a prototype that could actually be put to use.

  It was called the Advanced Research and Development Activity, and in theory it was a lot like DARPA. Its charter was to field new technology for all the intelligence agencies, just as DARPA was supposed to work for the entire Defense Department. In fact, though, the group was an NSA-owned operation. The director of ARDA reported to the director of central intelligence through the director of the NSA. On paper the line of demarcation seemed inconsequential. In practice it meant everything. ARDA was the agency�
��s property. It was even housed in a building at Fort Meade.

  Popp focused on working out a deal with Congress over which programs to keep and which ones to kill. The decision would not be based on merits. He knew that Congress wasn’t interested in debating the finer points of data mining versus pattern analysis or listening to lengthy explanations about synthetic data worlds. It was time to make sacrifices.

  Popp drew up a simple list, dividing the office’s many programs into two buckets: those that were so controversial as to arouse continued debate and press attention, and those that were less likely to attract unwanted attention. FutureMAP went in the former column, no question. So did the research on identifying people at a distance based on how they walked. useful? Yes. Creepy? Absolutely.

  Popp went through all the programs, applying this brutal calculus. He boiled everything down to a sound bite and asked himself whether it passed that critical “Washington Post smell test.”

  Then it was time for the privacy research. TIA’s critics had largely glossed over this part of the program, and Poindexter’s assurances about selective revelation and audit trails had rung hollow. For Popp it was an obvious choice what to do with this research. Though he might wish otherwise, it belonged in the kill bucket.

  Privacy had become the third rail. It wasn’t just controversial in its own right. The subject created more controversies. Popp thought that if he tried to save the research, people would ask why: What are you doing that requires such strong safeguards? We thought this was just research. You aren’t poking around in off-limits information . . . are you?

  Popp had no reason to know that the NSA already was. His team had unwittingly assisted them by inviting them into the experimental network. And while TIA prepared for its new home, members of Congress were making a connection: John Poindexter and Mike Hayden’s work sounded terribly similar.