Ghost Wars

by Steve Coll

  Tenet signed off on a compromise: The CIA would secretly buy its own airworthy Mi-17 helicopter, maintain it properly in Tashkent, Uzbekistan, and use CIA pilots to fly clandestine teams into the Panjshir.

  The helicopter issue was a symptom of a larger problem. By the late summer of 2000 the CIA’s liaison with Massoud was fraying on both sides. On the American side, the most passionate believers in Massoud were in the Counterterrorist Center, especially in the bin Laden unit. Officers with the unit who worked out of the Islamabad station were seen by their colleagues as “slightly over the top,” recalled one American official. Massoud’s intelligence network cooperated on collection and planning, but it became increasingly clear that Massoud did not intend to launch a snatch raid against bin Laden.

  The CIA’s CounterNarcotics Center reported that Massoud’s men continued to smuggle large amounts of opium and heroin into Europe. The British reported the same. They could all readily imagine the headlines if their operation was exposed: CIA SUPPORTS AFGHAN DRUG LORD. The Counterterrorist Center’s view of Massoud’s strategic importance to the United States was “not embraced,” recalled one American official involved. “There was much gnashing of teeth and angst and clucking and hand-wringing.”

  For their part, Massoud’s aides had hoped their work with the CIA would lead to wider political support in Washington and perhaps military aid. They could see no evidence that this was developing. Instead they were badgered repeatedly about an attack on bin Laden. “We never thought of capturing bin Laden alive in that type of Hollywood operation,” recalled one of Massoud’s intelligence aides. “It was never a consideration for people who knew the real situation in Afghanistan.” The Northern Alliance’s few shaky helicopters could barely clear the mountain passes. They had no air cover. Their forces were not very mobile on the ground. Bin Laden usually was surrounded not only by his own bodyguard but by hundreds if not thousands of Taliban soldiers. One of Massoud’s aides likened the mission urged on them by the CIA to a game of chess in which they would have to capture the king without touching any other piece on the board.40

  Massoud and his men respected many of the individual CIA officers they dealt with but increasingly felt frustrated by the agency’s policies and tactics. Massoud’s men asked their CIA counterparts, as this intelligence aide recalled it: “Is there any policy in the government of the American states to help Afghanistan if the people of Afghanistan help you get rid of your most wanted man?” America’s decision to abandon Afghanistan after the Soviet withdrawal was never far from their minds. But the CIA officers could make no such promise. The most they could say was that bin Laden’s capture “would definitely influence policy in Washington,” creating goodwill toward the Northern Alliance.

  This was not enough. Massoud’s men could easily imagine—and discussed among themselves many times—mounting a joint operation with the CIA to assassinate bin Laden by sniper fire, bombing, or a commando raid if this would result in a new American policy recognizing the Northern Alliance. But the CIA was not permitted to engage in that sort of military planning, and the agency had been unable to deliver any change in U.S. policy toward the Afghan war, either.41

  29

  “Daring Me

  to Kill Them”

  BY THE LATE SPRING OF 2000, Richard Clarke and his White House counterterrorism group had grown frustrated by the quality of intelligence reporting on Osama bin Laden’s whereabouts. The CIA’s unilateral human sources and its liaisons with Pakistan, Uzbekistan, and Massoud had generated volumes of fragmented hearsay but nothing solid enough to warrant missile strikes or a snatch operation. Clarke and his aides brainstormed for new ideas. Could they find a way to place a beacon on one of bin Laden’s aircraft so they could track the plane with bin Laden aboard and shoot it down in flight? Could they erect an enormous phony television tower near the Afghan border and use long-range spy cameras to watch for bin Laden? Clarke and his aides observed Pentagon Special Forces train British and French teams that planned to capture fugitive Balkan war criminals. Could one of these teams be inserted into Afghanistan?

  Clarke asked his longtime acquaintance in the national security bureaucracy, Charles Allen, who ran all of the CIA’s intelligence collection efforts, to work with Admiral Scott Fry, head of operations at the Joint Chiefs of Staff, on fresh approaches to the bin Laden problem. Clarke and his aides continued to hope the Pentagon would come up with a plan to use American commandos in Afghanistan. Their detailed tracking maps of bin Laden’s travels from Kandahar to Kabul to the eastern Afghan mountains seemed to offer a way forward. Clarke and the bin Laden unit at CIA felt they had established that it was highly probable, for instance, that bin Laden would return again and again to Tarnak Farm near the Kandahar airport. Wasn’t there a way to put reliable American eyes on that compound, equipped with secure communications that could be linked to missile submarines? Could a Special Forces team be provisioned to lie buried in the sand flats near Tarnak for a few weeks, ready to call in a strike whenever bin Laden turned up? As he pushed for answers, Clarke summoned the direct authority of President Clinton. In February 2000, National Security Adviser Sandy Berger had submitted a long memo to Clinton describing all the ongoing efforts to capture or disrupt bin Laden. Clinton had scribbled his dissatisfaction about the results in the margin. A savvy bureaucrat, Clarke photocopied the president’s scrawl and used it as a cudgel at interagency meetings.1

  Several years later a number of people involved in these highly classified discussions claimed credit for the idea of sending Predator reconnaissance drones to Afghanistan to search for bin Laden. Despite the confusion of competing recollections, it seems clear, in a general sense, that Clarke, Fry, Berger, Allen, Black, and officers in the CIA’s bin Laden unit jointly conspired, amid persistent squabbling among themselves, to launch the Predator experiment. Allen recalled that CIA senior management were at first reluctant, and that it was “a bloody struggle.” They hoped to solve the primary problem that had dogged their hunt for bin Laden since the winter of 1999 when they had stared day after day at satellite pictures of the Arab hunting camp in western Afghanistan, unable to develop enough confidence to fire missiles. Satellite and U-2 reconnaissance photography could identify fixed targets such as buildings, homes, and training camps with high precision, but these systems could not single out mobile targets or individual faces. In the case of the hunting camp, Clinton’s counterterrorism group had been forced to rely on identifications provided by the CIA’s Afghan tracking team. They had not been able to look directly at live photographs or video of bin Laden to develop a consensus within the national security cabinet that the risks of a missile or bombing attack were justified. The Predator, they hoped, could bridge these intelligence gaps.2

  The CIA and the Pentagon had each experimented with unmanned reconnaissance drones since the early 1980s. In the first years of the agency’s Counterterrorist Center, Dewey Clarridge had sought drones to help search for American hostages in denied areas of Beirut and rural Lebanon. As early as 1987 the CIA secretly adapted kit airplanes manufactured in California to carry cameras in a highly classifed project called the Eagle program. Clarridge hoped to operate the drones out of a hotel room in Beirut. The agency bought special wooden propellers made in Germany to help the drones fly quietly. Clarridge also experimented with arming the drones with small rockets that could be fired by remote control, but the rockets selected proved wildly inaccurate.3 In the same period, and sometimes in cooperation with the CIA, the Pentagon’s laboratory for experimental security technology, the Defense Advanced Research Projects Agency, funded prototypes of a long-endurance, unmanned drone called Amber. This was an extraordinarily lightweight (815 pounds) wasplike drone invented by Abraham Karem, the former chief designer for the Israeli air force. A lively engineer with unbounded imagination, Karem immigrated to the United States in the late 1970s and started an experimental aircraft company in California. The Amber prototypes he produced flew longer and better than a
ny drone to date. But Karem’s company went bankrupt amid bureaucratic battles in Washington. The Pentagon tended to invest in large, fast, complex drones that resembled pilotless fighter jets. These were very expensive, technically sophisticated, and politically unpopular. The CIA preferred smaller, lighter, cheaper drones that could take pictures and intercept communications in situations where satellites or high-flying spy planes did not offer enough coverage. Its experiments were easier to fund, but many at the Pentagon and in Congress dismissed the smaller prototypes as clunky toys of marginal value.4

  The Predator had gasped to programmatic life in the early 1990s as an awkward bastard child of the Amber. A large defense contractor bought up Karem’s assets, including his designs, and the U.S. Navy pitched in funds for more prototypes. The CIA’s director of espionage operations in the early Clinton administration, Thomas Twetten, held a review of the agency’s own secret drone projects, all still in experimental stages. When he listed options for CIA director James Woolsey, the director’s eyes lit up. Woolsey had met Abe Karem in Israel, and he also knew about Amber. “I know the guy” who can get this done,Woolsey told Twetten. The pair flew to California and tracked Karem down at the defense contractor who had bailed him out. They were selling prototypes to Turkey. Woolsey declared that he would take five on the spot for the CIA. The only problem was that the nascent Predator—long and ungainly—sounded like “a lawnmower in the sky,” as Twetten recalled it. The CIA managers told Karem he had to silence the motor, and he agreed.5

  From the CIA’s first purchases Predator operations required close cooperation between the agency and the Pentagon. This was never easy. The Air Force howled when it learned Woolsey had bought Predators in secret. The CIA chafed as it tried to sort out budgetary and operating rules with the Air Force. There were times when it seemed that the Predator’s chief innovations lay in its ability to generate table-thumping, vein-pumping bureaucratic agitation inside secure Virginia conference rooms. Ultimately the CIA arranged for Air Force teams trained by the Eleventh Reconnaissance Squadron at Nellis Air Force Base, Nevada, to operate the agency’s clandestine drones. First in Bosnia and then in Kosovo, CIA officers began to see the first practical returns on their decades-old fantasy of using aerial robots to collect intelligence.6

  The Predators deployed secretly to Bosnia in 1995 were designed to loiter over targets for twenty-four hours and could fly as far as five hundred miles from their home base at an altitude of up to twenty-five thousand feet. They were extraordinarily slow—their average speed was just seventy miles per hour—and they were so light that they sometimes drifted backward in the teeth of headwinds. A Predator’s “pilot” sat with several enlisted “payload specialists” inside a sealed, unmarked van near the runway of the drone’s operating base. (In its Balkans operations, the CIA flew Predators secretly out of Hungary and Albania.) At first the Air Force recruited pilots for the drones who had been grounded from normal flight by medical disabilities. Generators and satellite dishes surrounded the flight van. Inside, the pilot toggled a joystick before a video screen that showed the view from the Predator’s nose. Radio signals controlled the drone’s runway takeoff and initial ascent. Then communications shifted to military satellite networks linked to the pilot’s van. The Predator’s nose carried a swiveling Sony camera similar to those used by TV station helicopters that report on freeway traffic. It also could carry radar imaging and electronic intercept equipment.7

  In the first flights over Bosnia the CIA linked its Langley headquarters to the pilot’s van.Woolsey emailed a pilot as he watched video images relayed to Virginia. “I’d say, ‘What direction for Mostar? … Is that the river?’ ”Woolsey recalled. “And he’d say, ‘Yeah. Do you want to look at the bridge? … Is that a guy walking across the bridge? … Let’s zoom further, it looks like he has a big funny hat on.’ ”8

  There were serious glitches. Pilots struggled to learn how to fly such a light, awkward plane from satellite-delayed television images. After tugging their joysticks, it would take several seconds for the plane to respond. There was no adequate system to control ice on the Predator’s wings. The drone was not stealthy and could be targeted by antiaircraft fire. And after Bosnia there were debates about the Predator’s ultimate mission.

  One camp favored using the drone only for traditional intelligence collection: taking pictures and verifying reports from human agents on the ground. But others argued that the Predator could be a powerful weapon if it was integrated into what military officers sometimes called “the kill chain.” The Air Force had long struggled to develop weapons systems that could accurately track and attack isolated mobile targets such as cars and trucks. Its new airborne sensor and command system, known as J-Stars, could follow moving vehicles on a battlefield and identify, for example, whether the vehicles had wheels or tank tracks. But the J-Stars system could not make a close-up identification of a human face or a license plate number. The Predator’s cameras might provide this ability if the drone’s roving eye could be connected in real time to the larger Air Force command network. In that case the Predator might hover over a moving vehicle, transmit a running image of its license plate to CIA officers or Pentagon commanders in Virginia, tag the truck with a laser beam, and hold the beam on the target while a bomber swooped in to drop computer-aided munitions directly onto the truck. Or possibly the Predator itself could be armed with a remotely fired air-to-ground weapon if the technical problems of weight and missile velocity could be solved. As early as 1995 the Navy fashioned tests to link the Predator’s roving cameras to cruise missile submarines submerged offshore. In the Kosovo conflict of 1999 the Air Force secretly equipped Predators with laser target finders and satellite links that would make drone-guided bombing operations possible for the first time, although no such attacks were actually carried out.9

  All of this history—all of these unresolved questions about the Predator’s purpose and value—shaped debate among CIA officers, White House aides, and Pentagon brass as they considered how to use the drone in the hunt for bin Laden in the summer of 2000. The Predator was cheap by the lavish standards of Pentagon weapons programs, but at about $3 million per drone, each one lost would take a bite out of the CIA’s pinched budgets. Influential skeptics such as Thomas Pickering worried about the intelligence community’s built-in bias for “a near-term technical solution, rather than the long-term buildup” of reliable sources and recruits. Jim Pavitt feared that funds allocated to the Predator would inevitably come at the expense of money for human intelligence—HUMINT, in Washington’s acronym vernacular. Richard Clarke replied with his usual bluntness: “Your valuable HUMINT program hasn’t worked for years. I want to try something else.” Cofer Black, at the Counterterrorist Center, sided with Clarke while trying not to offend Pavitt. Frustrated at the hand-wringing and endless argument, Clarke enlisted Sandy Berger to formally order the Predator to Afghanistan. Berger did.10

  Then they argued more about the scope of the Predator’s mission. Clarke was intrigued by the idea of linking the Predator’s camera to the cruise missile submarines lurking secretly in the Arabian Sea. He pushed for a lethal operation in Afghanistan, not one that would solely take pictures. Berger was interested, but officers at the CIA were skeptical about the submarine proposal. There were too many unknowns. It would take too long to get munitions to the target even if the Predator saw bin Laden. “The Agency was very clear,” remembered a White House official. “They wanted to do an initial period of testing… . They didn’t want to hardwire it to the submarines” or to some other bombing plan. This official recalled “some skepticism” at the CIA “that you could get that kind of clarity” from the drone’s cameras to justify a missile launch.11

  Black advocated arming the Predator itself with an air-to-ground missile so it could fire instantly if it located bin Laden. But State Department lawyers objected, arguing that an armed drone might violate the Intermediate Nuclear Forces Treaty, which banned the United States from acquiring ne
w long-range cruise missiles. Was an armed Predator the same as a cruise missile? While the lawyers debated, Black and the Counterterrorist Center, now officially in command of the nascent mission ordered by the White House, proposed a different kind of experiment.12

  In the Balkans and in Iraq, Predator pilots and their support equipment (the pilot’s van, satellite dishes, and generators) had been parked at air bases in friendly neighboring nations. The operations were sensitive and clandestine, but the host governments were not unduly frightened about exposure. Here the situation was different. As the planning developed in the early summer of 2000, Uzbekistan agreed to allow secret Predator flights from one of its air bases for a limited period of time, but Islam Karimov’s government was adamant about secrecy. The agency’s officers feared that even the small cluster of vans and satellite dishes necessary to pilot a Predator would attract unwanted attention among Uzbek soldiers and officers. The cooperation between the CIA and the Uzbeks was so secret that many people in Karimov’s own government still did not know about it.13

  To address this problem the CIA proposed to experiment with a new stage in Predator operations. Improvements in communications systems now made it possible, at least in theory, to fly the drone remotely from great distances. It was no longer necessary to use close-up radio signals during the Predator’s takeoff and ascent. The entire flight could be controlled by satellite from any command center with the right equipment. The CIA proposed to attempt over Afghanistan the first fully remote Predator flight operations, piloted from Langley. The drone itself would be housed and recovered at hangars on a remote Uzbek airfield, but it would be flown with a joystick propped on a table inside a CIA operations center in Virginia.