by Shane Harris
The military and intelligence agencies had built the world’s most complex computer systems, and yet they hadn’t exploited new, off-the-shelf technology that was putting once unimaginable computing power on desktop machines. Average Americans reaped the benefits. And yet spies and soldiers still worked on proprietary and clunky systems that were plodding a path toward obsolescence. There were smarter, faster ways to do business. And there was an imperative to change. America’s cold war foes were vanquished, but a new breed of “transnational” threats was rising—terrorist groups, so-called rogue nations, and other adversaries that didn’t play by the traditional rules of war. The United States had to outsmart its enemies by predicting their next moves. The Information Systems Office was designing information technologies to do just that.
Leopold wanted to win a support contract under Spinnaker. Typically, DARPA programs relied heavily on outside experts for development, and their own staff of managers turned over at regular intervals to keep the agency infused with fresh thinking. Leopold needed someone to spearhead his Spinnaker team. Would Poindexter be willing to attend a meeting with the DARPA program manager?
He hardly needed to ask. One afternoon, Poindexter headed over to the DARPA offices to meet with Brian Sharkey, a taciturn ex-submarine hunter. As Poindexter listened to Sharkey talk about his ideas, and what he wanted to accomplish, he realized he could have been listening to himself ten years earlier. Sharkey envisioned a collaborative system that would reach across disparate databases and help foresee the next disaster. Senior decision makers were the ultimate customer. The ones overwhelmed by a deluge of information. By noise. Sharkey was proposing much the same technological system that Poindexter had always wanted to build. Sharkey had the notion, and the funding. Poindexter had the vision and the will. And he brought two other invaluable qualities to the table: technological prowess and an intimate understanding of how the national security community worked. And how it often didn’t.
The meeting broke up, and Leopold asked Poindexter if he’d come aboard at Syntek, under contract to Sharkey. As with so many of his most fateful decisions, Poindexter didn’t think for very long. He’d been running a small software consultancy with a friend. But his sole client, an ambitious California-based company, was in danger of being vanquished by Microsoft. Recently, Poindexter had taught himself several complex computer programming languages, and he’d perfected his skills by designing a system to spot minute errors in software code, a task not that dissimilar to finding hidden signals in noisy data. He was more technologically fluent than at any point in his life. He was equipped. And since he had retained his full Navy pension after retiring in 1987, he wasn’t worried about financial survival.
Poindexter agreed to Leopold’s offer. He was back on a path he’d stumbled off years ago.
In January 1996, he joined the staff as a senior vice president. He had a new office. A new team. And Sharkey was gearing up for a new phase of Spinnaker, building on the first phase of research. He was looking for a new name, but the best he’d come up with didn’t exactly inspire: Collaborative Crisis understanding and Management. Poindexter didn’t like it. Mindful that bureaucracies refer to programs almost entirely by their acronyms, he suggested that Sharkey’s choice was . . . unfortunate. A number of his female colleagues apparently had raised the same point already.
“We need to change the name,” Poindexter advised. Why not continue with the sail family? Something easy to remember but also symbolic. Thus, Spinnaker became Genoa, named for the headsail that’s used when moving against the wind.
Sharkey had tried to make some headway on his own, by drumming up interest among the National Security Council staff, which was arguably where Spinnaker would be needed most. But Sharkey was a technician, not a pitch-man. Some who’d heard his early descriptions of the idea thought it sounded rather fanciful—some kind of crystal ball to predict the future. Sharkey hadn’t been able to translate for the staffers in language they could understand.
Poindexter wanted another shot. He called a former secretary, Wilma Hall, a career employee still working on the NSC staff. She was glad to hear from her old boss. It had been a while. Poindexter brought her up to speed on his new project. “Who’s the right person on the staff now to talk to about this?” he asked.
“Oh, it’s Dick Clarke,” Hall said without hesitation. “He’s the Ollie North of this administration.”
Poindexter smiled. The go-to guy. He dialed Clarke’s office and arranged a meeting; Clarke’s assistant was more than happy to make room in the schedule for a former national security adviser, and an admiral.
Like North, Clarke ran the counterterrorism portfolio for the NSC staff. He’d forged a close working relationship with the national security adviser, Tony Lake. Clarke stood out as an innovative, hard-charging, and passionate operator who could navigate treacherous bureaucratic channels without attracting the wrong kind of attention. He’d risen steadily at the State Department, taking over a senior intelligence post in 1985, when North was in his prime at the White House. But when Clarke took over the political and military affairs brief, he ran afoul of investigators over, of all things, improper missile sales through Israel.
In 1992, State’s inspector general chided Clarke for not acting on intelligence reports that Israel had improperly given or sold air-to-air and antitank missiles to third parties, including China, South Africa, Ethiopia, and Chile. The defense secretary, Dick Cheney, also accused Israel of illegally sharing Patriot Missile technology with China. It was the job of Clarke’s department at State to investigate such allegations. But at the time, factions at State, in the Pentagon, and at the CIA were warring internally and with one another over the United States’ Israel policy and what role America’s key Middle Eastern ally should play after the cold war. The furor, or lack thereof, over the weapons deals were emblematic of each camp’s appraisal of the others—one was too indulgent with the Israelis, the other was making life too hard on them. Clarke, the inspector general’s investigation implied, was rooted in the former camp, and he had failed to do his job. He was transferred out of State and landed on the NSC staff, where he was asked to stay when Bill Clinton took the oath of office.
Clarke’s first test on terrorism came quickly. On February 26, 1993, only thirty-eight days into the new administration, Islamic extremists set off a truck bomb in the parking garage of One World Trade Center. They’d hoped to topple one of the Twin Towers into the other, killing tens of thousands, but the massive buildings withstood the blast. Six died.
When the bomb went off, the intricate system of crisis management that Poindexter and North had built was nowhere to be found. In fact, it had been completely dismantled. In the late 1980s, when George H. W. Bush inherited the ghost of Iran-Contra, he directed his NSC staff to get out of the operations business. No longer would the White House direct the terrorist wars. The administration had paid too high a price for Poindexter and North’s zealous pursuits. The crisis management center and its intelligence feeds were shut down. Officials threw out a manual on crisis indicators, meant to carry lessons forward into subsequent administrations. By the time Clarke came to the White House the nexus of the entire government’s counterterrorism mission had disintegrated. And it had never been fully recaptured by another agency.
Clarke could see that firsthand when the bomb went off. No one in the Situation Room alerted him. Indeed, it took a call from his boss, Tony Lake, asking what in the hell just happened in New York to prompt Clarke to contact the Sit Room himself. A young Navy officer on duty answered the phone and hesitantly told Clarke that he wasn’t sure whether he was supposed to call him about these kinds of things. Terrorism was just one of a number of generic “global” issues Clarke had been assigned. No one in the White House was on point for a domestic terrorist attack.
Clarke had managed to cobble together an anemic version of the Crisis Pre-Planning Group from the Reagan years. His Counterterrorism Security Group (CSG) included repre
sentatives from around government, principally the FBI and the CIA. But the members had limited influence with their home agencies, which regarded terrorism as a low priority. Indeed, the CIA’s counterterrorist center was perceived as a dumping ground for career dead-enders, a kind of Land of Misfit Toys that offered little hope for promotion. The FBI, for its part, thought of terrorism as a foreign intelligence concern and therefore out of its lane both legally and professionally. The bureau stuck mostly to bank robbers, mobsters, and drug runners. The CIA and State could handle mad mullahs.
After the World Trade Center bombing, Clarke convened the CSG in the Situation Room, but the group came up empty on leads. The FBI had managed to identify the perpetrators through a painstaking forensic process at the bomb site, but none of the names got hits in the CIA’s database. The bombers didn’t belong to Hezbollah, Islamic Jihad, or any of the other known outfits. The FBI had learned that two of the men appeared at Kennedy airport the previous year without any immigration papers; one was detained when officials saw him carrying a manual titled “How to Make a Bomb.” But he claimed political asylum, and so immigration officers let the man go with an admonishment to show up later at a status hearing.
Clarke was aghast. These guys had appeared out of nowhere, catching the entire national security community by surprise. What cracks had they slipped through? How many signals had been missed in the buildup to their attack? And what else did they have in store?
The 1993 bombing showed Clarke just how weak the government’s early-warning capability had become. He spent the next few years pushing terrorism steadily up the list of threats. Clarke couldn’t give orders to anyone outside the White House, so he had to finagle and fight for his authorities. But still, he managed to find the levers and trigger points, as well as the pools of money, in what passed for the government’s counterterrorism enterprise.
By the time Poindexter called asking for a meeting, Clarke was very interested in the potential of information technology to uncover useful pieces of intelligence and to better connect the far-flung agencies nominally managing the terrorism problem.
“He’s interested in widgets,” Wilma Hall told Poindexter over the phone.
Clarke was also focusing lately on a shadowy Saudi expatriate named Osama bin Laden, whom the CIA had identified as a mere terrorist money-man, but who was clearly emerging as the commander of a troubling new terrorist network.
Poindexter approached the Old Executive Office Building, next door to the White House. It had been a while; aside from a few official visits, the last time he’d walked the floors of the palacelike building that had housed the Crisis Management Center, he was in uniform.
Poindexter walked the halls of the house he’d help build and, though he’d never wished for it, take apart. Old friends and staff who had stayed aboard in career billets stopped to greet him. When he let them know why he’d come calling, it all seemed so obvious. Of course he should be talking to Dick Clarke.
Poindexter knew just where to find Clarke’s office—it was the one that Ollie North used to occupy. Clarke spotted the admiral coming through the door, and he rose to greet him.
“Hello, Admiral! It’s been a long time.” Clarke grasped Poindexter’s hand, and his searching expression gave Clarke his cue. “Remember?” he asked. “I was with you on the trip to Oman.”
The memory quickly filtered back to him. Yes, it was 1984, when Poindexter was talking to Oman and Saudi Arabia about basing a U.S. air squadron. The State Department had sent Clarke as its representative. Poindexter still didn’t remember Clarke clearly, but apparently they had a history. They sat down, and as the conversation turned to Genoa, Poindexter could see they shared a common language as well.
Clarke was no technologist, but he intuitively grasped the concepts Poindexter articulated, now in a much more cogent and realistic form than Sharkey had been able to do. This wasn’t magic. It was science. A systematic, decision-support tool, as the researchers liked to call it. Genoa was meant to help stop a World Trade Center bombing, or to more quickly pick up the pieces after the event. Already, Poindexter explained, the researchers were focusing on terrorism as an obvious application, and they’d studied one attack in detail.
In March 1995, members of a Japanese religious cult called Aum Shinrikyo released plastic bags full of sarin nerve gas in five Tokyo subway trains. A dozen commuters died, and hundreds were sickened by the toxic agent. The Genoa team researched the buildup to that attack, looking for signals that might have alerted observant analysts. They found quite a few.
For starters, the subway attack wasn’t Aum’s first use of chemical weapons. Only nine months earlier seven people had died from a sarin release in the mountain city of Matsumoto. Police blamed a local man who stored pesticides in his home; both he and his wife were sickened. Although investigators found no immediate linkage to Aum, other suspicious indicators preceding the Matsumoto attack went unchecked. In June 1993, the cult purchased a half-million-acre sheep farm in Western Australia. Locals reported seeing dead animals about the property—tests later confirmed they’d been poisoned with sarin. The cult sold the ranch less than a year later, at a loss.
Also in June 1993, noxious fumes emanating from a Tokyo building connected to Aum caused area neighbors to complain to authorities. Residents living near another facility complained almost a year later of peculiar smells. Then, in September 1994, more than two hundred people in seven towns in western Japan came down with rashes and eye irritations from unknown fumes. Less than three months later, an Aum member killed a man with VX nerve gas. And finally, within two weeks of the subway attack, eleven people were hospitalized after inhaling fumes on a train in Yokohama, less than twenty miles from Tokyo. Officials also found three attaché cases containing liquid, fans, vents, and batteries in a Tokyo subway station.
The Genoa researchers determined that all these signals were either known, or knowable, to government investigators and law enforcement officials. Much of the data appeared in news reports. No one had put the pieces together. The researchers concluded that if someone—or some system—had done so, Japanese officials could have stopped Aum from attacking the subway.
In sizing up the cult, the Genoa team focused on its ability to carry out an attack, not the attack itself. What materials would the cult members have to buy? Where would they practice? When would they conduct surveillance on potential targets? These were the questions with which terrorism analysts would fan out into the information space, looking for evidence, discarding false leads, building a hypothesis. Genoa aimed to understand not only the most likely path terrorists would take but other less obvious avenues they might pursue. For instance, if Aum could release sarin in a train, why not in the ventilation system of an office building? Indeed, the cult had tried to release anthrax spores and botulinum toxin, dabbled with cholera and Q fever, and tried to retrieve a strain of the Ebola virus in Zaire.
Genoa could handle queries from the top down: A policy maker comes up with a hypothesis about how a terrorist might attack and then uses the system to look for supporting evidence. But it also worked from the bottom up: An analyst starts by collecting disparate pieces to see what kind of story they tell. The system then constructed a transparent argument, like a sentence diagram, about what was most likely to happen. Analysts and decision makers could look at the diagram on their own computers and then edit it, adding new documents or reports, tweaking theories. They could debate not just each other’s conclusions but the intelligence that led to them. In that sense, Genoa was revolutionary: It promised to flush out the parochial vagaries and biases that infused most intelligence and inevitably helped politicize it.
Clarke would have understood as well as anyone how intelligence often was abused and misused by human analysts. Could machines remedy that? He seemed hopeful.
Poindexter made it clear that Genoa would never replace human analysts. It was an aid, a way to take the heavy lifting, and the drudgery, out of the process, and to give eve
ryone in the system more time to think.
Clarke nodded, affirming that he understood, and that he liked the idea. Poindexter wanted the NSC staff to use Genoa, to put the system through its paces and help work out the bugs. Now he had Clarke in his corner, at least.
“Keep me advised,” he told Poindexter as they wrapped up their discussion. Clarke walked him to the door. “I’m glad to see you’re back.”
There was someone Poindexter needed to meet. After his meeting with Clarke another NSC staffer steered the admiral to the neighboring suite, the Office of Intelligence Programs.
“Mary, meet John Poindexter,” the staffer said, making the introduction to Mary McCarthy, a career CIA officer who had joined the staff in 1996 as the office’s number two. McCarthy had last served as the national intelligence officer for warning, whose job was, broadly speaking, to make sure the intelligence community wasn’t taken by surprise. That she should know Poindexter seemed eminently logical.
McCarthy had never met him. She knew he was an admiral. Of course she knew Iran-Contra. But upon first glance, she thought he could have been an English professor as easily as a former national security adviser. McCarthy also knew he was a Republican, but in her experience so far, partisanship rarely came between professionals. McCarthy, who didn’t hide the fact that she was a Democrat, had worked for plenty of Republicans, including the man who’d helped Poindexter and North build the first pieces of a government-wide counterterrorism operation, Charlie Allen. McCarthy was his deputy when Allen was in charge of warning at the CIA, and she took over his job in 1994. She revered Allen and told people that everything she knew about the business, she’d learned from him.