by Steve Coll
TWENTY-NINE
Dragon’s Breath
Datta Khel, a town in North Waziristan, lies about twenty-five miles southwest of Miranshah, toward the border with Afghanistan. Near its market and bus depot is an open area suitable for an assembly. On the morning of March 17, 2011, the day after Raymond Davis flew home, about thirty-five maliks—tribal leaders approved and subsidized by the government of Pakistan—had gathered for a jirga, to resolve a feud over a chromite mine. Two tribes, the Manzarkhel and Maddakhel, were removing chromite, but there was a question of who owned what. Khasadars, or local police, paid by the Pakistani government, were in attendance. On the other side of the world, in the C.I.A.’s Global Response Center, targeting analysts watched the meeting on video transmitted from armed drones above. The agency’s unmanned air force had been unusually active over North Waziristan in recent days. On March 11, drones struck a “suspected vehicle boarded by militants,” as a ledger of drone strikes kept by the local government recorded. Two days later, a strike blew up a “state car” traveling across North Waziristan. On March 16, the day I.S.I. arranged for Davis to be released, a drone destroyed a target near Datta Khel, reportedly killing up to five people. The next morning, from the evidence available, it appears that C.I.A. analysts watching from Langley tracked a Taliban suspect to the jirga and then decided to kill everyone present.1
Malik Jalal, a wealthy tribal leader in the region, was about two miles away from Datta Khel that morning, but he could see drones in the air. Watching the machines hover and unleash death suddenly was a feature of daily life in Waziristan by 2011. Being attacked by a drone is not the same as being bombed by a jet. With drones, there is typically a much longer prelude to violence. Above Miranshah and its neighboring towns, drones circled for hours, or even days, before striking. People looked up to watch them, hovering at about twenty thousand feet, capable of unleashing fire at any moment, like dragon’s breath. Predator and Reaper drones emit what, on the ground, sounds like a flat, gnawing buzz. (Locals sometimes refer to a drone as a bangana, a Pashto word for “wasp.”) Their missiles make a whoosh-like sound when released, then streak off their rails to Earth to explode. That morning, Jalal watched missiles fly and heard several explosions. He drove to the site and saw many body parts scattered on the ground as ambulances rushed to and fro, tending to survivors.2
Angry protests erupted across Pakistan after TV networks headlined the attack and portrayed it as an atrocity against civilians. A few Taliban may have been present at the assembly, Pakistani officials told reporters, but the majority were not anti-American fighters. Kayani issued a rare public statement condemning the C.I.A.: “It is highly regrettable that a jirga of peaceful citizens, including elders of the area, was carelessly and callously targeted with complete disregard to human life.”3
Cameron Munter, the U.S. ambassador in Islamabad, challenged Mark Kelton, his station chief. Yes, I.S.I. whipped up anti-American feeling after drone strikes and spread false reports of civilian deaths, but this one looked indiscriminate on its face. The jirga “had all the signatures” of a terrorist meeting, Kelton assured him. Since 2008, the C.I.A. had increasingly relied on “pattern of life” analysis to identify targets inside Pakistan. If visibly armed individuals behaved and spoke on cell phones in ways that confirmed their participation in groups carrying out terrorism or cross-border attacks against N.A.T.O., they might be struck, even if their identities as individuals were unknown.
“It was the timing that mattered,” Munter argued. Pasha had just gone to great lengths to release Raymond Davis. Now it looked as if the C.I.A. was humiliating the I.S.I. director. “When you kick somebody in the teeth a day after they have done something for you, you are going to pay a price.”
Kelton was unmoved. Davis did not enter into his calculations, he said. The United States was at war and the enemy was on the battlefield. If you can find the enemy, you strike them.4
The number of drone strikes on Pakistani soil more than doubled during 2010 from the previous year, from an estimated 54 to 122, or more than 2 per week.5 To suppress the Haqqani network as U.S. forces poured into Afghanistan, the agency conducted an air war against Afghan Taliban sanctuaries in North Waziristan. This expansion beyond Al Qaeda broadened the target set.
The drone program stretched Kayani’s already fraying tolerance of the Plan Colombia–inspired model for U.S. counterterrorism in Pakistan. During 2010, while Holbrooke had dangled the promise of a transformational “strategic dialogue,” Kayani was willing to put up with the C.I.A.’s attacks on his own citizens. The general was realistic. He knew drone technology was too sensitive for the United States to transfer to Pakistan, and in any event, his country lacked the satellite networks to operate unmanned aerial vehicles. The C.I.A. killed Haqqani and other militants who generally accommodated the Pakistani state and therefore were not high on the general’s list of internal enemies. Yet the agency also killed militants like the Mehsud brothers of the Pakistani Taliban that threatened Pakistan’s stability. Kayani could accept this mixed picture, or he could try to challenge the C.I.A. and reduce its presence in Pakistan, at the risk of damaging or breaking the alliance and its financial aid flows. By the time of the strike on the chromite jirga, he was moving toward change. He wasn’t looking to eliminate all drone strikes inside Pakistan or go to war with the C.I.A. Kayani did want to regain control of counterterrorism strategy from the C.I.A. and reduce domestic resentment toward the Pakistan Army over its cooperation with America. There was hardly a member of the national parliament, even from the secular-leaning Pakistan Peoples Party, who would vote publicly in favor of the current high-tempo C.I.A. drone war. Under Kayani’s theory of working from the “fundamentals,” above all public and political opinion, it seemed obvious that something had to give.6
Cameron Munter did not oppose the drone program, particularly when it targeted international Al Qaeda leaders, but by the spring of 2011 he led a faction within the Obama administration that questioned whether D’Andrea and Panetta had lost perspective. The strikes in Waziristan were too visible, too indiscriminate, and had created a dangerous political dynamic within Pakistan, Munter argued by secure video from Islamabad whenever the principals met to discuss the program.
Panetta, a genial man with a sharp tongue, hammered Munter in these meetings. The C.I.A. director argued that signature strikes had eliminated important Al Qaeda figures. Also, the agency needed to strike Taliban targets while they were inside Pakistan because once militants crossed over into Afghanistan they dispersed and operated more stealthily. In Waziristan they were easier to identify.
Munter countered that the C.I.A. had moved beyond targeted counterterrorism and was now running the world’s most expensive artillery system. The agency could lob missiles at Haqqani formations in Waziristan a couple of times a week, but those attacks couldn’t possibly be decisive in the Afghan war. What they were guaranteed to do was to mobilize yet more Pakistani volunteers to the fight and undermine the Pakistan Army in the eyes of the public. Munter felt isolated. He recognized that his arguments were dismissed by some in Washington as clientitis—too much sympathy for Pakistani viewpoints. “This is a never-ending war,” the C.I.A. officers he butted heads with told him. “Whose side are you on?”
When Munter’s interventions blocked a particular strike, C.I.A. colleagues would turn the knife: “Those sixteen Uzbeks you saved yesterday? They’re going to try to kill American troops.” Yet Munter was also more convinced than ever that he was on the right side.7
After the mess at Datta Khel, Obama ordered a suspension of C.I.A. drone strikes. None took place in Pakistan for a month. Obama did authorize an exception to his freeze if the C.I.A. located a “high-value target.” When drones struck South Waziristan in mid-April, a debate erupted in the White House about whether the C.I.A. had violated Obama’s order.8
As the agency’s operations in Waziristan came to resemble an air war, they changed life
on the ground. The strikes spurred militants to try to identify spies who might have betrayed them. Around North Waziristan’s main towns, Miranshah and Mir Ali, which took the brunt of the C.I.A. attacks, paranoia spread. The Taliban blamed local maliks who had long presided over the area’s economy—smuggling, arms dealing, mining, and government contracting. Taliban gunmen seeking control of local rackets executed maliks and their family members in the hundreds. In local bazaars, the Taliban distributed D.V.D.s of their socially superior victims confessing that they had spied for C.I.A. drone operators. The confessions included elaborate narratives about how the agency supposedly distributed “chips,” or homing beacons, to local spies. The spy would toss a chip over a neighbor’s wall or into a Taliban jeep to guide drone missiles to it. The men also confessed that the C.I.A. had given out special pens with invisible ink, which were used to mark Taliban vehicles for destruction. The Taliban tortured their prisoners, so the confessions could hardly be taken at face value. The Taliban also had a powerful motive to force the maliks to admit to spying, because that would complicate the position of the victim’s tribe and family as they considered revenge, as spying for the United States was widely regarded as criminal. No chips were ever discovered or photographed by local journalists. Although homing beacons are common in police and espionage, and may well have been used by some paid C.I.A. agents in Waziristan, there was an air of dark hysteria in the torture-induced confessions of the Taliban’s victims and the rumors they spawned.9
Kayani had tolerated the C.I.A.’s drone program not only because it eliminated declared enemies of the Pakistani state, but also because Pakistan Army officers deployed to Waziristan informed him that the strikes were for the most part popular among maliks and tribes opposed to the Taliban. The missile strikes encouraged fence-sitters to stay away from the militants, his generals reported, which made it easier for the army to establish some control as it set up permanent bases across the Federally Administered Tribal Areas for the first time in Pakistani history.
That spring of 2011, a Pakistani brigadier deployed to South Waziristan, where drone strikes were “almost a daily routine” around the capital of Wana, surveyed locals about the C.I.A. operations. They told him, “People were fed up [with] Taliban atrocities and wanted them to be eliminated.” One elder said, “Local people wish that drones can carry more missiles.” During the brigadier’s two-year tour, he investigated four strikes and found only one case of a civilian death, involving a man who was “guiding the Taliban about the routes that were safe to travel.”10
The C.I.A. reduced the number of mistakes it made gradually after the rough start of its escalation in 2008 and 2009. In the last year of the Bush administration, by one independent count, children died a third of the time in drone attacks. In Obama’s first year in office, the figure was 20 percent. By 2012, it would be 5 percent. Smaller drone missiles, more precise targeting technology, more experienced operators, and greater emphasis from the Obama White House on preventing innocent deaths caused civilian casualties to drop. Yet all it took was one mass casualty attack like the one at Datta Khel to revive the imagery of callous atrocity.11
The video feeds transmitted from above North Waziristan documented one problem clearly: Pakistani forces still could not or would not stop Haqqani fighters from crossing into Afghanistan. It was not unusual for C.I.A. targeting analysts to watch from above as Taliban commanders loaded vehicles in Miranshah or its environs with men and weapons and then drove right through several Pakistani military checkpoints. C.I.A. officers at the embassy in Islamabad would occasionally show these videos to Kayani and Pasha, to reinforce the point that the border was open, American troops were being hit as a result, and that something had to change. The Pakistanis typically counterpunched by providing coordinates inside Afghanistan where anti-Pakistani militants operated, challenging the Americans to do something about that problem.12
The American leadership in Kabul—Chris Wood, the C.I.A. station chief; Ryan Crocker, the ambassador; and Petraeus, at I.S.A.F. headquarters—were united in advocating that disrupting the Haqqanis and the Taliban’s sanctuary in Quetta and Karachi was essential if the troop surge was to be successful. To execute his war of attrition against the Taliban in the time Obama had provided, Petraeus regarded drones as a tool to decimate the Haqqani network while the 101st’s Second Combat Brigade hit Taliban infrastructure in the green zone and the Marines cleared Helmand. The Haqqanis had about four thousand fighters, according to American estimates. Drones might be provocative but they were more effective than Pakistani ground forces and less provocative than a U.S. invasion or conventional air war would be.
Hamid Karzai egged them on. “I’m not asking you to go to war in Pakistan, but there’s no pressure on them,” he pleaded to the Americans he met. “You are getting played on this.” Another time, Karzai confided, “Do you know why I’m always blowing up in public about civilian casualties, night raids? I’m trying to push you off your flawed approach. I’m trying to shock you into changing your policy on Pakistan.”
When congressional delegations visited Kabul, they usually took a briefing from the C.I.A. station chief. Wood delivered what he jokingly called the “Chris Wood Power Hour,” in which he argued that Pakistani sanctuary was the single most important problem in the war.
“We either address the sanctuary and we win the war, or we don’t and we lose the war,” Wood told visitors repeatedly during his 2011 tour in Kabul. “It’s that simple.” He had run operations in Islamabad, knew Pakistani officers and the military well. He did not buy the arguments that I.S.I. had rogue officers or retirees operating independently—an I.S.I. colonel was no more likely to go rogue than an American colonel commanding a tank unit. Nor did he believe the army was so weak or domestic politics so fragile that Pakistan could not withstand the backlash if it carried out a decisive crackdown on militants in Waziristan, Quetta, and Karachi.13
Wood believed that greater pressure on Kayani and Pasha—perhaps private threats of personalized travel sanctions against the generals, or threats to delegitimize the Pakistan Army internationally, as a sponsor of terrorism—might succeed. Yet Wood’s views never attracted a critical mass of cabinet-level allies. For one thing, Pakistan controlled supply lines to the Afghan war. They could squeeze the Americans if pressed too far. Recognizing this vulnerability, Petraeus ordered planning for air bridges to Bagram and Kandahar from Qatar and Central Asia that could resupply I.S.A.F. adequately in case an all-out confrontation with Pakistan closed the land routes. The secret planning for a raid on the house in Abbottabad, Pakistan, where Osama Bin Laden might be holed up, and the uncertainty of what backlash might result if an attack there went ahead, and potentially went wrong, added urgency to the planning for alternative routes. But full independence from Pakistani supply lines would take time and involve huge expense, at least until the number of American troops that needed to be fed and watered fell back toward fifty thousand. This was a legacy of the logistics bargain struck with Musharraf amid the shock of September 2001.14
—
After John Podesta declined the position, Hillary Clinton selected Marc Grossman as Richard Holbrooke’s successor. Initially she asked him if he would prefer to be ambassador to Kabul or the special representative to Afghanistan and Pakistan. Grossman said that was up to her. When Ryan Crocker, formerly ambassador to Iraq and Pakistan, emerged as a candidate for Kabul, Grossman accepted Holbrooke’s position. Obama asked Crocker to try to repair the frayed relationship with Karzai and to negotiate a long-term strategic partnership agreement with Afghanistan. Grossman would handle the secret talks with the Taliban. He was a career foreign service officer who had risen to become ambassador to Turkey during the 1990s. He also served as assistant secretary of state for Europe during the Kosovo crisis, the last war Bill Clinton had to manage. After George W. Bush took office, Grossman worked closely with Colin Powell and his deputy, Richard Armitage, as undersecretary of state for
political affairs. Then he retired and went into business as a Beltway consultant. He was a tall man who wore wire-rimmed glasses and dressed like a civil servant. He had credibility with both Democrats and Republicans. In marked contrast to Holbrooke, he was quiet, methodical, a planner, not a jazz-inspired improviser.
“I can’t be Dick,” he told Clinton. “Dick was Dick. I do things more modestly, systematically.”
There were those at the White House who wondered if he really wanted the job or if he had accepted it merely from a sense of duty. Grossman bristled at any such suggestion. He might not be flashy but he was dedicated, and he had put his own reputation on the line in a project fraught with political risk. At the Obama White House, it was certainly a relief to have someone less freewheeling than Holbrooke in charge of the sensitive matter of probing the Taliban in negotiations. Grossman did understand that his mandate was ambitious, whatever its probability of success. He was to finish what Holbrooke had started. He might help end the war.
“The time has come to jump over the inter-agency’s hesitation, which will be never-ending,” Grossman wrote to Clinton on March 24, 2011, in a memo titled “A-Rod on Deck,” which adopted Holbrooke’s code name for Tayeb Agha. They had come to the point where they had to “see what A-Rod can do in a real negotiation,” Grossman wrote.15
Grossman hoped to agree with Tayeb Agha on a series of confidence-building measures that, once finalized, would lead to peace negotiations including Karzai’s government, perhaps in the summer of 2012. Grossman’s hope was that the Taliban would initially issue a statement denouncing terrorism, and then they would agree on a political office for the movement, and then the United States and the Taliban would exchange prisoners. In the meantime, Grossman would organize commitments of support to Afghanistan from international allies at a series of conferences. That diplomacy would provide a basis for U.S. support to Kabul if talks with the Taliban failed. If the talks succeeded, countries such as Pakistan, Qatar, and Turkey could provide a circle of support for the peace negotiations. There was also the possibility that the Taliban would raise the subject of U.S. troop withdrawals from Afghanistan as part of the negotiation. The White House signaled that it would be open to putting that question on the table in the right circumstances, as part of a larger discussion of postsettlement security arrangements in Afghanistan.16
