The Watchers (39 page)

Two months after the Syracuse visit, former Navy secretary and 9/11 Commission member John Lehman gave a little-covered speech before the U.S. Naval Institute. Lehman had known Poindexter since the Reagan administration, and he summed up a feeling taking hold not just among Poindexter's supporters but among more detached observers as well.

“In our government bureaucracy today, there is no accountability,” Lehman said. “Since 9/11—the greatest failure of American defenses in the history of our country, at least since the burning of Washington in 1814—only one person has been fired. He is a hero, in my judgment: Admiral John Poindexter. He got fired because of an excessive zeal to catch these bastards.”

The following month Poindexter emerged fully from his brief self-imposed exile. A prestigious conference series sponsored by the McCormick Tribune Foundation asked him to join about forty other luminaries for a meeting on the day's hot issue: “Counterterrorism Technology and Privacy.”

The participants gathered at Cantigny, a sprawling five-hundred-acre estate not far from Chicago, replete with intricate gardens and capped by a handsome Georgian mansion. In addition to Poindexter, John Hamre was there, as was Mike McConnell, the former director of the National Security Agency turned Booz Allen executive and TIA contractor. Spike Bowman attended too. He was the FBI's senior counsel on national security law and a friend of Poindexter's who had attended the speech at Syracuse.

Michael Chertoff, then a judge for the U.S. Third Circuit Court of Appeals, also came. Only six months later Chertoff would step down from the bench when President Bush nominated him as the Secretary of Homeland Security. A pair of key NSA officials also showed up—Joel Brenner, the agency's inspector general, and Vito Potenza, the deputy general counsel. As the agency's top watchdog and its number two lawyer, respectively, both men had been informed about the NSA surveillance program.

Representatives from prominent government watchdogs and security experts also attended. There was Kate Martin, the executive director of the Center for National Security Studies, an authority cited frequently in the national media. Lara Flint represented the Center for Democracy and Technology; her colleague, Jim Dempsey, had met with Bob Popp in his office not long after Safire's
Times
column appeared. And Dempsey's partner that day, Barry Steinhardt of the ACLU, also came.

The gathering at Cantigny erased any doubts about John Poindexter's status. He was a player. He'd been a part of the terror war since its inception. And now, when the brightest minds of the day pooled their collective perspectives, he was at the table.

The discussion lasted for two days, with a dinner and overnight stay at the estate. Passions ran hot. The participants argued about public expectations. Could people reasonably expect the same level of privacy today as they did in an era without the Internet? No one could argue that technology hadn't influenced, and changed, their assumptions. But the government had redrawn the baselines. What were the new limits on its ability to spy on people in the name of protecting them?

Steinhardt and Poindexter hadn't seen each other since that meeting in Popp's office, when he popped in midway through the briefing. Cantigny might not have been the optimal setting for a reunion. During a break in the discussion, Poindexter approached Steinhardt. He'd been itching to tell him something since their last encounter.

“At least I was doing this in public!” Poindexter jabbed. He wasn't hiding from the debate. He'd encouraged it.

Steinhardt admitted that Poindexter was right. He ran TIA in the open. But his indignation was unjustifiable
,
Steinhardt thought. Poindexter had invited ridicule with that preposterous logo and his visions of virtual databases and all-seeing systems. Steinhardt was sure that TIA's gaze would not be limited to “foreign” targets. It would be turned on ordinary criminals—deadbeat dads, men beating their wives, tax absconders. There was a term for this—“mission creep.”

But Poindexter was no longer interested in Steinhardt's critique. He thought that he'd had his chance, and missed it. He scolded Steinhardt for not providing “constructive ideas about privacy and security.”

“The nation faced a grave threat,” Poindexter said. Steinhardt should have tried to help him instead of tear him down.

Poindexter pointed his finger at Steinhardt and got up close to him, so that he was almost poking him in the chest. Steinhardt might think that he'd won, but this wasn't over.

“It's all secret now,” Poindexter said, another reference to TIA. “You forced me to take this underground.”

Steinhardt wasn't entirely sure what he meant. Ever since the program was officially shuttered there had been rumors about some lingering pieces still in operation. But nothing was ever confirmed. The ACLU had helped get language in the defense bill to shut TIA down, but they were unaware of what the classified annex contained.

TIA wasn't gone after all. Steinhardt knew it now.

Looking back, Steinhardt still felt that killing TIA—at least in one form—counted as a win. There was a public debate, and his side prevailed. But had the program not gone away, had it remained in DARPA's purview with Poindexter at the helm, Congress, Steinhardt, and the entire civil liberties community might have kept the admiral under their collective thumb. He would have lived under the microscope. All his funding would have been visible, all his research too. If TIA worked, there'd be another debate over whether to use it. But now, with the program in the black and consumed by the NSA, its onetime watchdogs would lose track of it. For Steinhardt and his colleagues, whose struggles were far from over, a question remained unanswered: Had they won a battle over TIA only to lose a bigger war?

 

The conference wasn't all fireworks. After two days the attendees drew up a lengthy set of principles that they hoped would offer some way forward. At the top of this list was an urgent call to Congress and the administration to face the privacy debate head-on, publicly and honestly. “Government should infringe on privacy only as an imperative to protect the safety of U.S. citizens and resident aliens,” they wrote in a formal report. “The legislative and executive branches share the fundamental constitutional responsibility to protect the privacy and safety of all U.S. citizens and resident aliens—and should act in partnership.”

That wasn't happening. Only a few months before the group gathered the administration gave its first briefing on the NSA's warrantless surveillance program to the full Gang of Eight—the congressional leaders and the top members of the intelligence committees. The briefing itself was prompted by a crisis.

In March, the warrantless surveillance program was up for a required forty-five-day renewal by the attorney general. Since October 2001, John Ashcroft had regularly averred to the legality of the agency's covert activities. But now his deputy, James Comey, had brought to light disturbing details about the breadth and scope of the NSA's spying.

It wasn't warrantless “wiretapping” that had senior lawyers at Justice on edge. Rather, it was the layer of surveillance that often preceded it—the massive sweep of telecom metadata. This included information on Americans collected from U.S. companies inside the country without a warrant.

What started as a program of
targeted
surveillance now included collection and analysis of huge amounts of metadata, unfathomable by human analysts. Jack Goldsmith, the newly installed head of the department's Office of Legal Counsel, had first spotted the trouble when he read John Yoo's memos on warrantless surveillance. Goldsmith found the analysis both indefensible and illogical.

The conclusion was shattering. The attorney general had been signing off on illegal domestic spying ordered by the president of the United States. Now the administration wanted Ashcroft to sign off on the program again.

With the deadline upon him, Ashcroft was bedridden in hospital fighting an acute pancreatic disease. He'd already been complaining that the White House so obscured the NSA's various activities, setting them up in different security “compartments,” that he couldn't see the full picture. The “program” had evolved, and grown new legs.

Comey had taken over as the acting number one while Ashcroft recuperated. He refused to sign the forty-five-day authorization, and a scene of Washington drama for the ages ensued when Alberto Gonzales, the president's lawyer, and White House chief of staff Andy Card showed up at Ashcroft's hospital bedside, pressuring him to sign the authorization instead.

Ashcroft unceremoniously sent them packing. Comey, along with Goldsmith, was prepared to resign if the NSA's activities weren't brought under legal control. Ashcroft would join them, along with FBI director Robert Mueller.

President Bush personally intervened to avert a mass resignation of his entire law enforcement leadership. That spectacle, coming without warning, would have so rocked the corridors of Washington that the NSA's spying would not stay secret for long. It would also have marked an irrevocable downward slide for Bush, who was campaigning for reelection. Such a mass exodus would constitute a vote of no confidence in his leadership of the war on terror.

Tensions were high inside the NSA as well. Brenner and Potenza—the two officials who went to Cantigny—had assured Hayden he could rely on those presidential authorities that Ashcroft was supposed to review. But the White House had forbidden the NSA men from seeing the underlying analysis, prepared by Yoo, that defined the legal supports for the program. As long as they were blocked, Brenner and Potenza could not render a fully informed opinion on whether the president's program was legal. That put Hayden in a tight spot, since he'd spent the past few years telling congressional officials, in his limited briefings, that the NSA's lawyers had signed off on the program.

In the end, Bush acquiesced to certain changes that Justice wanted to get the surveillance activities on legal ground. The chief judge of the Foreign Intelligence Surveillance Court had been rattling cages, upset that information from warrantless interceptions might find its way into legal warrant requests before her court. The program was also modified to keep that from happening.

The upper echelon of power breathed a quiet sigh of relief. Crisis averted. But the surveillance did not stop. And neither did the NSA's quest to overcome another major handicap. Legal obstacles were trivial in comparison with the technological barrier that the agency was struggling with in 2004. And unless the agency's technicians could overcome it, the early-warning system that Hayden had envisioned would never materialize. They needed a big idea.

 

 

 

 

The NSA was coming up with weak links. Stellar Wind and the BAG might have provided new insights into terrorists' social networks, but they had failed to find any blockbuster revelations, much less intelligence that actually had preempted an act of terrorism. This was no early-warning system.

To build one, the agency had to acquire and then analyze information quickly, while it was still fresh enough to produce useful leads. Hayden wanted to engage terrorists in hot pursuit, as they jumped around the global communications grid using disposable cellphones, different phone lines, and ever-changing e-mail addresses to cover their tracks. To catch them before they disappeared, Hayden had to be able to react to new intelligence within minutes, not hours or days.

And that was very hard to do. The NSA could consume seemingly limitless amounts of data, but everything that its electronic vacuums collected went into huge databases. If analysts wanted to look at a day's or a week's catch, they had to retrieve it from storage. By then the information might be outdated. Intelligence in the global terror hunt had a short shelf life.

The agency could monitor individual phone numbers or e-mail addresses in real time, but it could not take an expansive view of an entire communications network and know, at any moment, what was happening everywhere. It could not see all the patterns forming in the noisy ocean of metadata. In that sense, the NSA did not truly have a real-time system for detecting terrorists, at least not before it knew who the terrorists were.

The agency wasn't the only large organization looking to crunch data in real time. Banks wanted that power to detect fraud. Hedge funds dreamed of capitalizing on minute movements of the stock market. And energy companies building seismic models to discover new sources of energy needed this kind of hypercomputing. Throughout the entire history of computers, the answer to dealing with big sets of data was to build very big databases to hold them all, and then to go to work with sophisticated tools. Luckily, mass storage was relatively cheap, and there was no shortage of it. But this approach didn't help information mammoths like the NSA achieve their real-time ambitions.

The techies and computer programmers at Fort Meade, along with their army of contractors, needed another way. And in 2004, with the privacy debate flaring, they hit upon a counterintuitive and elegant idea.

 

Maybe the agency had been thinking about the problem all wrong. What if the response to the constant flow of communications the NSA sucked off the global communications network was to not store them all?

This was a radical notion. The database was the backbone of the NSA's intelligence operations, and the source of its power. (Information was currency in the spy game, after all.) The NSA was essentially a big library. Every record, every bit and byte it collected, was like a book. And the books were stored on vast, deep shelves. The agency could categorize those books and cross-reference names, places, and events like some sophisticated card catalog. But that didn't change the fact that the books were still sitting on a labyrinth of shelves, waiting for someone to retrieve them.