A well-built man with a strong jawline and light brown hair pulled back from a balding scalp, Cummings worked to change the culture of FBI investigations. Instead of arresting a would-be terrorist as soon as agents had sufficient evidence, as was the protocol before 9/11, FBI men and women under Cummings left targets in the wild longer to be monitored and tracked. This allowed agents to gather as much information as they could about possible terrorists and their associates.18 Cummings pressed FBI agents to find not only the suspected terrorist but also the web of people linked to the suspected terrorist.
But increasingly what Cummings and his agents discovered was that the threats weren’t coming from a network of people—the FBI, a few years after 9/11, had become less concerned about terrorist cells—but instead from young men acting alone. The threat had shifted from an organized group to a lone wolf, Cummings believed. With the intelligence apparatuses of several nations focused on Al Qaeda in Afghanistan and Pakistan, it had become nearly impossible for the organization to train a group of terrorists and then send them to the United States or Europe without being intercepted. Simply, a 9/11-style attack was no longer within Al Qaeda’s capability. The best the organization could do was inspire someone already in the West to carry out a terrorist attack—an attack Al Qaeda’s leadership would likely know nothing about until it happened—and then claim credit once the smoke had cleared. That’s why the FBI became so obsessed with the possibility of a lone-wolf attack; the Bureau now believed that at any time, in any community, someone could radicalize and become a terrorist, with a bomb, a gun, even with household chemicals.
In light of this new theory, the main concern at the Bureau became how to identify these lone wolves before they struck. To assist with this, the FBI came up with a kind of radicalization spectrum, running from sympathizer to operator. All operators were sympathizers at one point, the spectrum theory goes, but not all sympathizers become operators. “We’re looking for the sympathizer who wants to become an operator, and we want to catch them when they step over that line to operator,” Cummings said. “Sometimes, that step takes ten years. Other times, it takes ten minutes.” The FBI tries to identify those who might take this step by scrutinizing Muslims who are espousing radical beliefs, expressing hatred of the United States or its foreign policy, or associating with others who are doing one of those two things. The FBI obtains some of this information through tips or by monitoring radical forums and chat rooms online. But the majority of this information comes from the street level, from informants.
Throughout the FBI’s history, the numbers of informants the Bureau employs has been a closely guarded secret. Periodically, however, these figures have been made public. A Senate oversight committee in 1975 found the FBI had 1,500 informants. In 1980, officials disclosed there were 2,800. Six years later, following the FBI’s push into drugs and organized crime, the number of informants ballooned to 6,000, according to the Los Angeles Times.19 That number grew significantly after 9/11. For example, in its fiscal year 2008 budget authorization request, the FBI disclosed that it had been working under a secret November 2004 presidential directive demanding an increase in “human source development and management,” and that it needed $12.7 million for a program to keep tabs on its spy network and create software to track and manage 15,000 informants.
The FBI’s use of informants today is unprecedented. In addition to the roster of 15,000 informants that the Bureau maintains—many of them tasked with infiltrating Muslim communities in the United States—for every informant officially listed, there are as many as three unofficial ones, known in FBI parlance as “hip pockets.” Informants can be doctors, clerks, imams. Some might not even consider themselves informants. But the FBI regularly taps all of them as part of a domestic intelligence apparatus whose only historical peer might be Cointelpro, the program the Bureau ran from the 1950s to the 1970s to discredit and marginalize groups ranging from the Ku Klux Klan to the Communist Party to the Reverend Martin Luther King Jr.’s civil rights organization.
To manage this comprehensive system, the FBI uses a computer program known as Delta, which allows agents to search the ranks of informants using specific parameters—among them age, ethnicity, nation of origin, and languages spoken. “The idea behind Delta was to make it more efficient not only to document information, but to manage information and incorporate elements of oversight,” Wayne Murphy, the FBI’s assistant director of intelligence, told the news media when Delta was first announced to the public in July 2007.20 Effective informants today move around the country doing the FBI’s bidding, and Delta has made this fluent movement possible. An FBI informant who can look and speak the terrorist part can move from case to case, jurisdiction to jurisdiction, state to state, earning tens of thousands of dollars at every stop. It’s not uncommon for informants to make $100,000 or more on a case, plus a “performance incentive” of potentially tens of thousands of dollars if the case results in convictions, and then move across the country to do it all over again in some other city. Delta streamlines the horse-trading of informants among FBI handlers by cataloging them and providing detailed information about their case histories. Using Delta, FBI agents who need an informant can search the database and find candidates—just as a corporate recruiter might use LinkedIn while searching for software engineers to hire.
In addition to dramatically changing the way the FBI tracks and uses informants, Delta also put a stop to an older, rampant practice of agents hoarding the best informants for job security. Peter Ahearn, the retired FBI special agent who oversaw the Joint Terrorism Task Force in Western New York, explained to me how after Delta was implemented, agents could no longer guard their informants—in the past, agents had treated informants like their personal pets, to be let out of their cages only when necessary and never to be shared with other agents—because there was now a digital clearinghouse of snitches. If agents in a particular city needed an informant for a certain task, all they had to do was load Delta and see which informants might be available for transfer. “I could sit down in front of a computer and type in ‘one-legged Somali,’ and I’d find we’ve got one in Kansas City, and I could call up the handler and ask if I could borrow the guy,” Ahearn told me, exaggerating for humor.21
The FBI’s extensive and better-organized use of informants represents one great change for the Bureau in the post–9/11 era. But another—and perhaps more jarring—change involves data mining. Before 9/11, due to security concerns and an antiquated computer system, most FBI agents couldn’t even search the Internet from their desks, let alone track terrorists. In fact, on September 11, 2001, FBI agents were forced to send photographs of the hijackers by express mail because the Bureau’s computer system didn’t allow them to email images. Former FBI director Louis Freeh, who retired in June 2001, was so technology averse that he refused to use email. U.S. Senator Charles E. Schumer has described the Bureau’s outdated computer system and Luddite culture as the FBI’s “greatest failure” under Freeh.22
That the FBI was behind the times technologically is an understatement, and so using data for intelligence purposes represented a giant leap forward for the Bureau. To assist in taking that leap, in 2005, FBI director Mueller tapped Philip Mudd, a former CIA analyst and top-level briefer under CIA director George Tennant. Mudd had risen to second-in-command of the CIA Counterterrorism Center, which oversees all clandestine operations involving Al Qaeda and other terrorist groups. Mudd’s move to the FBI was a very unusual one for a CIA man. While interagency cooperation has increased significantly since 9/11, strong rivalries and prejudices still exist among federal law enforcement agencies. The CIA, FBI, DEA, Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF)—each is suspicious of the other. The FBI views the CIA as a group of bluebloods whose job is made easy by not being bound by the U.S. Constitution. The CIA, in turn, views the FBI as a ragtag group of desk-jockeying accountants and lawyers who hate to get their hands dirty with fieldwork. The CIA and FBI view DEA and AT
F agents as underachievers, and, in turn, DEA and ATF agents believe the CIA and FBI are filled with ineffectual snobs. In short, the agencies don’t trust each other, and any agent switching teams is received with intense suspicion. “There’s always been this competition and distrust,” Dale Watson, the FBI’s former assistant director for counterterrorism, who did a detail assignment at the CIA in the mid-1990s, told me. “No one at the FBI trusted the CIA enough to share a lot of information, and vice versa.”23 Because of this mistrust, there were significant misgivings about Mudd when he arrived as deputy director of the FBI’s new National Security Branch, whose creation President George W. Bush ordered in 2005 to consolidate the Bureau’s counterterrorism, counterintelligence, and intelligence capabilities under one department. Among seasoned FBI agents, there was a belief that Mudd would try to remake the Bureau into a domestic CIA. In many ways this has come true, as Mudd has altered the way the Bureau operates, pushing agents to increase their intelligence-gathering capability, in particular by increasing the number of informants they have on the streets providing them with information.
To coordinate intelligence gathering and informant recruitment, Mudd took over a program called Domain Management, which the FBI had created to track immigrants from China and other countries who were suspected of being involved in industrial espionage—mainly the theft of intellectual property from corporations and universities. Mudd expanded Domain Management to use commercially available data, as well as government data from I-9 “Employment Eligibility Verification” immigration forms, to pinpoint the demographics of specific ethnic and religious communities—say, for example, Iraqis in central Los Angeles or Pakistanis in the Washington, D.C., suburbs.* In February 2006, shortly after taking his position at the FBI, Mudd demonstrated Domain Management to high-ranking agents. He displayed a map of the San Francisco Bay area, which highlighted the places where Iranian immigrants were living. That was where, Mudd said, the FBI was “hunting.” 24 Domain Management could tell FBI agents with precision where Muslims lived in San Francisco—as well as nationwide—allowing them to direct resources and informant recruitment to specific neighborhoods.
The FBI officially denies that Domain Management works this way. Its purpose, Bureau spokespeople have said, is simply to help allocate resources according to threats. But FBI agents have told me that with counterterrorism as the Bureau’s top priority, agents often look for those threats in Muslim communities—and Domain Management allows agents to understand those communities’ locations and demographics. One former FBI official jokingly referred to Domain Management as “Battlefield Management.”
Some FBI veterans have criticized Domain Management as unproductive and intrusive—one agent told Mudd during a high-level meeting that the program pushed the Bureau to “the dark side.” This tension has its roots in the stark difference between the FBI and the CIA. While the latter is free to operate internationally without regard to constitutional rights, the FBI must respect those rights in domestic investigations. Mudd’s critics inside the Bureau saw the targeting of Americans based on their ethnicity and religion as going a step too far. For his part, Mudd brushed off the criticism as coming from old-school agents unwilling to adapt to a rapidly changing world. “There’s 31,000 employees in this organization and we’re undergoing a sea-change,” he told the New York Times during his first year at the FBI. “It’s going to take a while for what is a high-end national security program to sink down to every officer.”25
But internal FBI documents obtained by the American Civil Liberties Union in December 2011 suggest that Mudd’s critics had reason to be concerned. The documents show that since 2005, the FBI has used its community outreach programs—which had previously been operated out of the FBI’s Office of Public Affairs—to gather intelligence on people not suspected of having committed crimes. In other words, no criminal predicate existed to justify an investigation. Many of the activities documented through this intelligence gathering were religious ones protected by the First Amendment, such as where and when Muslims worshipped. The FBI then fed the information collected into Domain Management for future analysis. In a March 2008 memorandum, for example, FBI agents wrote about an outreach effort to a Pakistani American community organization in San Francisco.26 In the memorandum, the FBI agents documented religious activity and the identities of the organization’s officers and directors. The year before, in 2007, agents were present at a mosque outreach meeting in San Jose, California, that was attended by fifty people representing twenty-seven local Muslim communities. A resulting FBI memorandum, which was included in three case files, identified each of the fifty participants by name and affiliation and then analyzed the demographics of the attendees.27
Historically, the FBI’s community outreach programs were designed to build trust between federal law enforcement and local communities—to make it easier for the FBI to investigate crimes by having a public more willing to volunteer tips and information. But under Domain Management, community outreach programs now serve as Trojan horses for intelligence-gathering agents, giving them cover as they harvest information under the guise of community engagement. The FBI then uses this information to determine where to assign informants and agents and what screws to turn when trying to win cooperation from would-be informants in Muslim communities. Some FBI agents under Domain Management are assigned full-time to recruiting informants, and these agents often use immigration violations, evidence of crimes, and embarrassing information, such as about extramarital relationships, to coerce Muslims into becoming informants, who in turn tell the FBI under duress about other Muslims they should consider targeting for scrutiny.
It was a cloudy winter’s day in February 2011 when I arrived at the FBI Academy in Quantico, Virginia, a sandstone fortress of a building on a 385-acre Marine Corps base west of Interstate 95. I had asked J. Stephen Tidwell to help me understand how and why the FBI employs Domain Management and its thousands of informants. Now executive director of FBI National Academy Associates, a nonprofit that organizes training sessions at the FBI Academy for local law enforcement, Tidwell retired in 2010 as an executive assistant director of the FBI. While at Bureau headquarters, he authored the Domestic Investigations and Operations Guide with Arthur Cummings, and before that oversaw a large and controversial intelligence-gathering operation that recruited informants from and spied on members of Muslim communities in Southern California.
Tidwell arguably knows as much about FBI counterterrorism operations as anyone, and on that February afternoon, he drove me in his black Ford F350 through Hogan’s Alley, a ten-acre recreation of a town at the FBI Academy crowded with houses, bars, stores, and a hotel, which the Bureau uses as “a realistic training ground” for its new agents.28 The FBI jokingly refers to Hogan’s Alley—which gets its name from a nineteenth-century comic strip—as “a hotbed of terrorist and criminal activity,” and agents who work sting operations learn their craft here. At one end of the town is the Biograph Theater, named for the Chicago movie house where FBI agents gunned down John Dillinger in 1934. Tidwell pointed to the model cinema and laughed. “Dillinger, Biograph Theater, Chicago,” he said. “See, the FBI has a sense of humor.”
A former West Texas cop, Tidwell is a barrel-chested man with close-cropped brown hair that is slowly graying. Wearing khakis, a blue sweater, and an oxford shirt, he drove me back to the main FBI Academy building and continued the nickel tour. In one of the hallways, he stopped at and pointed to a plaque hanging on the wall, which commemorated John O’Neill. “John understood the threat Al Qaeda posed long before anyone else at the Bureau did,” Tidwell said. We then walked to the office of FBI National Academy Associates, which is tucked into a corner of the FBI Academy’s main building. Not far from Tidwell’s neatly kept, windowed office is a place where visitors can buy FBI Academy souvenirs such as T-shirts and coffee mugs.
Many current and former FBI agents I’ve spoken to have offered negative comments about Domain Management and its creator,
Philip Mudd, drawing a caricature of the former CIA analyst as a soulless purveyor of the dark arts whose evils have infected the Bureau. Tidwell isn’t one of them. In fact, he’s one of Mudd’s most vocal supporters. I asked him whether he believed Domain Management’s obvious intrusion into minority communities, with maps created according to demographic and religious data, was worth whatever benefits could be achieved for criminal investigations and intelligence gathering. Tidwell leaned back in his chair and crossed his legs, placing his left foot on top of his right knee, as he thought about how to answer the question. “I don’t think it’s useful to think of Domain Management strictly in the way you are,” he told me. “Let’s imagine we’re out in a field to investigate a report that there’s been a murder. We’re looking for the body, and it’s all woods and brush except for a large barn in the middle of the property. One person suggests that we divide up the property into sections and have agents walk it inch by inch until we find the body. Another suggests that we get up on the roof of the barn and look for the body from that vantage point. But a third person says neither of those plans is the most effective way to find the body. He instead points to the sky, where birds are circling. He says, ‘Let’s search the ground those birds are flying over.’ That’s what Domain Management does.”
The Terror Factory Page 4