The Watchers (45 page)

News reports indicated that McCarthy had failed a polygraph test about whether she'd had any contact with journalists. But contacts didn't equal a specific disclosure. McCarthy was hardly the only senior CIA official being questioned with a lie detector about talking to the press. And clearly other intelligence agency employees had leaked information about other programs, yet they hadn't suffered McCarthy's fate.

Some of her friends thought that Goss was looking to scare the spy workforce. If he was willing to ax someone who'd dedicated much of her career to fighting terrorism, then he would come after anyone.

If those were his intentions, they backfired. Goss and his staff's hard-charging management style had alienated the CIA lifers. Leakers or not, they knew how to run the traps at Langley. Two weeks after he dismissed McCarthy, Goss found himself sitting in the Oval Office with President Bush, who announced that Goss was leaving too.

Bush needed to reassert control over the CIA. He needed someone with a smoother style but still a firm hand. He needed someone he could trust.

The president turned to Mike Hayden, who took over the following month.

 

Since leaving government for the second time Poindexter had come to rely on his friends more than ever. Many of them, particularly Fran Townsend, still exerted tremendous influence in the administration. Some were at the peak of their power; some were in transition. Still others were about to embark on second lives of their own. While the admiral enjoyed his resurgence, and watched with quiet delight as his ideas were absorbed into the mainstream of American spy craft, one of his close allies came forward with an unexpected message: John Poindexter had it all wrong. His system of total information awareness could never catch terrorists.

Jeff Jonas had come a long way since Carmel. After meeting Poindexter at the Highlands Forum, his cheat-catching software had made the leap from Las Vegas to Washington. Poindexter's research team had put NORA through its paces with the experimental network. But the software required a lot of hand-holding. The TIA researchers weren't sure whether Jonas's program had flaws, whether they weren't expert enough to use it, or whether it was just ill suited to the task of catching terrorists—as Jonas had suspected from the beginning.

Poindexter ultimately decided that NORA was very good at sorting out aliases and finding connections among groups of people in discrete environments, like a casino floor. But it wasn't quite up to the task of finding bad guys in the vast unknown.

Jonas thought he'd tried to tell Poindexter that in their first meeting. But Jonas was open-minded by nature. He had come under Poindexter's spell—the White House photos on the wall, the high-level support for his idea from the Pentagon. Jonas was a patriot at heart. Poindexter became a catalyst for that emotion.

But Jonas had moved on since then. He'd made a lot more friends in Washington. With his compelling personal story and decidedly unstuffy charm, Jonas became a darling of the policy wonk crowd. He once threw a cocktail party for his new friends at the Ritz-Carlton, not far from the Pentagon, and entertained them with a professional “mentalist,” who performed card tricks and sleights of hand. Jonas was unlike anyone they knew. And he was ceaselessly interested in whatever new idea someone wanted to plant in his ear.

After meeting Poindexter, Jonas had fallen in with the civil libertarian set. He'd never imagined that data privacy could be such a contentious issue, but he came to see the problem differently. The people he had spied on, if he could even call it spying, handed over their personal information willingly to the casino. But this new surveillance the government had enacted—this wasn't voluntary. And, he thought, it wasn't accurate.

In early 2005, Jonas attended a conference at the Heritage Foundation, a conservative think tank in Washington. During the lunch break he found himself standing in line for a sandwich next to a guy he'd never met, but who'd also grown up in the Bay Area.

Jim Harper was a conservative, of the libertarian variety. He had carved out a niche as a digital privacy maven, running a Web site called Privacilla that came at the issue from a free-market, pro-technology perspective. He viewed himself as a counterpoint to left-wing activists, whom he regarded as hostile to capitalism. These were the ACLU types and advocates like Marc Rotenberg, who'd been one of Poindexter's earliest critics.

Harper felt out of place at Heritage, which he regarded as a den of “terror warriors.” He thought that these were the sort who, after 9/11, claimed the attacks had given the government permission to throw off law and custom. He thought that people had gone “tribal.”

At lunch, Harper looked around the room for someone to talk to. That's when he spotted the guy who didn't look like anyone else. The one in the black pants and T-shirt.

Harper knew a bit about Jonas's story; he'd even come up in discussion that morning. People knew him as the Vegas data guru who had tried to help catch terrorists. Harper also knew that Jonas had written a paper with a Heritage fellow and wondered if he was just the go-along-to-get-along type with the terror warriors.

As they got to talking Harper saw that Jonas was open to new ideas. So over lunch, and in e-mails in the weeks that followed, they struck up a conversation that Harper was eager to have: Does pattern analysis of data really catch terrorists?

Harper believed that it didn't. Jonas agreed with him. And so another relationship blossomed. Harper suggested that they write a paper together, taking this flawed theory head-on. He knew that Jonas had worked with Poindexter, and he believed that killing TIA, while perhaps a pyrrhic victory, was still a victory for his side. Jonas didn't want to whack Poindexter in print. Harper didn't much care either way.

Perhaps owing to their shared geographical history, he and Jonas hit it off quickly. They traded coy e-mails. “Where's the love?” Harper would ask when he hadn't heard back on a recent round of edits to the paper.

Over the next several months, they hashed out their ideas, first in an outline, then in a draft. As they moved toward publication, Jonas emerged not only as a skeptic of Poindexter-style pattern analysis, but as an outright opponent. In December 2005 he spoke before a meeting of the Homeland Security Department's Data Privacy and Integrity Advisory Committee. Harper was a member, along with other prominent experts from academia, government, and think tanks, and they met periodically to discuss privacy with senior members of the department, including the secretary, Michael Chertoff.

The group gathered at a hotel a few blocks from the White House. Harper joined a team of questioners sitting before Jonas's panel, committee-style. He had a canned question at the ready, which would give his writing partner an opening to voice the ideas they'd been mulling over in print.

“I found it remarkable,” Harper said, “that you, as an expert in this field and somebody who works for a company that would sell this product if it were viable, tells us that predictive data mining is not viable. I think that is an important thing to hear from you.”

Jonas obliged. “While there's a large market for data mining, I'm suggesting that using data mining to predict the intent of a terrorist, where you're basically putting the finger on a trigger that is going to cause government scrutiny upon an individual, is where I see it as a bit problematic.”

He didn't use the name, but Jonas had implicated TIA directly. “I consider my life's work in the area of knowing who is who and who's related to who,” he said. But this kind of knowledge had limitations. The chances of finding one meaningful signal in the noise were almost nil. “One in a million things happen millions of times a day,” Jonas said. “And bad guys don't leave as many transactional footprints.”

Jonas had just turned his back on Poindexter. He declared that no one could detect terrorists by looking for patterns, not without ensnaring many more innocent people than guilty ones. When their paper was published a year later, Jonas and Harper warned against embracing technology. “Though data mining has many valuable uses, it is not well suited to the terrorist discovery problem,” they wrote. “It would be unfortunate if data mining for terrorism discovery had currency within national security, law enforcement, and technology circles because pursuing this use of data mining would waste taxpayer dollars, needlessly infringe on privacy and civil liberties, and misdirect the valuable time and energy of the men and women in the national security community.”

Poindexter read the paper and found it both flawed and misguided. He objected strenuously to any characterization of TIA as “predictive,” and he took pains to tell people why it wasn't “data mining.” He knew the difference, though most people didn't. Data mining was something that marketers used to sell more products—more beer and diapers. But Poindexter was chasing a deeper level of insight. He was after patterns and the ability to detect anomalies without any known starting point. He wanted to see the unseen. Not the future, just the present as it truly existed. He thought that Jonas's technology—which was more akin to data mining—was too simplistic. Just like his critique. While Jonas and Harper hadn't singled anyone out, the message was implied. Poindexter's grand vision would never work. The guru of data mining, Jonas, had said so.

The paper was only ten pages long, but it was a watershed. It became a pivot point in the debate over TIA, NSA surveillance, and the whole concept of using data to detect and preempt terrorism. Critics from the left and the right, anyone who'd harbored a grudge against TIA or even a sneaking suspicion that its powers were overstated, seized upon this new theory. Technologists without a dog in the political fight also thought highly of it, and many were inclined to believe Jonas because he was an expert and also had worked with Poindexter. Now he declared that this powerful, mystical, and artful science the man had employed could never do what he imagined.

Poindexter didn't take it personally. He moved on, confident that he could solve the problem eventually.

Jonas's admiration for Poindexter never dimmed. He still considered him a friend, and an unparalleled mind. But they had declared a war of ideas, and there could be no truce.

Elsewhere, a real war was on. And Poindexter's friends, along with his adherents, were arming their weapons.

 

 

 

 

In the summer of 2006, the United Kingdom's warning system was blinking red with indications of new terrorist plots. Home Secretary John Reid, who, like some of his American colleagues, had been accused of exaggerating the danger, declared that Britain faced “probably the most sustained period of severe threat since the end of the Second World War.” And most of the sources of this new wave, British officials concluded, traced back to Pakistan. There, in mountainous regions outside the government's reach, Al Qaeda's senior leadership had regrouped, after eluding U.S. forces in Afghanistan years earlier. They had recruited new operatives and were busy plotting more strikes on the West.

That's when twenty-five-year-old Ahmed Abdulla Ali appeared on British security's radar. Agents kept close watch on Ali, whom they feared had connections to the resurging terrorist network in Pakistan. Britain was still reeling from a coordinated bomb attack on its subway and bus lines less than a year earlier, pulled off by Muslim men living in the United Kingdom. Now they worried that Ali, who was born in Britain, had joined the growing ranks of homegrown radicals and could be searching for targets in a brazen new attack.

It wasn't hard under the circumstances to arouse the government's suspicion and for security officials to follow up on it. British law allowed for intrusive and secret surveillance of terrorism suspects as well as for preliminary detention before the government ever brought official charges. In June, after Ali flew back home from a trip to Pakistan, security officials covertly searched his luggage. What they found only heightened the government's concern: a container of powdered Tang, the neon-orange flavored drink, and a large number of batteries. Fine powders could be used as accelerants in bombs, and batteries were a common triggering device.

Officials began a massive surveillance campaign. They followed Ali on foot. They saw him using public phone boxes, mobile phones, and anonymous e-mail accounts. Undercover officers observed Ali as he paid cash for a £138,000 flat in the London borough of Waltham Forest. Then they bugged the apartment with a camera and microphone. On August 3, they watched him and another man construct disturbing-looking devices out of drink bottles. If Ali and his associate were building bombs, they were certainly small ones. Investigators could presume that they were meant to be hidden. Three days later, on August 6, officers followed Ali to an Internet café, and they watched him spend two hours researching airline timetables.

 

Back in Washington, senior intelligence and security officials had been watching developments across the pond since late June. The Brits had alerted the Americans, and for weeks no one was precisely sure of the suspects' intentions. But by August multiple streams of intelligence had helped bring the plot into sharper relief. Both sides were convinced that they were dealing with a terrorist cell that planned to go after civilian airliners.

Michael Jackson, the deputy secretary of Homeland Security, was used to a daily flow of threat intelligence. It arrived with the ferocity of a fire hose. Each day he and his boss, Michael Chertoff, examined a grid of known and suspected threats displayed as a collection of dots. Each dot represented a person or potential threat that the government was monitoring. It could be a known target of surveillance or a tip from a nosy neighbor about the suspicious guy next door. Sometimes the dots disappeared after a day or two. The target went silent or the tip turned out to be a bogus prank. But occasionally connections formed between the dots, an indication that intelligence had linked them together in some way. And sometimes clusters of dots formed. When that happened, Jackson and Chertoff knew to pay closer attention.