Storm Clouds

21. “Storm Clouds”

It was by this point forgotten—few knew it at the time—that, six years earlier, the United States had pursued a classic counterinsurgency strategy in Afghanistan. It was headed up by a three-star general named David Barno, who in October 2003 became commander of US forces in the country, just the second officer to hold the position since the invasion that had toppled the Taliban two years earlier.

Not that Barno had many US forces to command. President George W. Bush and his defense secretary, Donald Rumsfeld, had long ago declared victory in Afghanistan and shifted focus—and resources—to the new war in Iraq.

When Barno arrived, he found himself in charge of just two brigades, a smattering of Special Ops forces, and a handful of close aides, augmented by a small staff of reservists in their forties or fifties. During one lunch hour in the mess hall early on in his tenure, he sat down next to one of these reservists, an overweight, middle-aged master sergeant, and asked how he happened to be there.

The sergeant replied cheerfully that he’d recently had triple-bypass heart surgery, which was a good thing, since the Army wouldn’t have let him come if he’d had a quadruple bypass.

Barno quickly grasped that Afghanistan was not regarded as a high-priority mission.

On his predeployment tour of the country a month earlier, he’d been shocked by how scattershot the effort was. There was no unity of effort: the military did its thing, the embassy did its thing, and their two paths neither crossed nor ran parallel. The military’s headquarters were at Bagram Air Force Base, more than an hour’s drive from the embassy in Kabul; the diplomats and the officers rarely met or spoke. Similarly, the Army had troops in the field; AID had provincial reconstruction teams in a handful of villages, some nearby, but they, too, never coordinated their efforts. The entire enterprise reeked of complacency. The war was considered over. Yet the Taliban were still putting up a fight in the eastern provinces along the Pakistani border and, to a lesser but worrisome degree, in the south.

To Barno’s mind, the setting was ripe for a counterinsurgency strategy.

Barno had been a cadet at West Point, class of 1976, with a heavy concentration in the Social Science Department. An avid reader of military history since his teens, he particularly liked the academy’s course on the history of Revolutionary Warfare. After graduating, he chose to make a career in the Army’s light infantry branch. Over the next three decades, he was never stationed in Europe, never assigned to an armored unit; rather, he deployed as a Ranger to the era’s “moot-wah wars” in Panama and Grenada.

The Army had no doctrine on counterinsurgency when his command in Afghanistan began, so Barno brought with him the textbook that he’d used in the Revolutionary Warfare class as a second-year cadet and reread it thoroughly. He also dipped into his discretionary fund to order several copies of every book about COIN that he could find on the website—including the Marine Corps’s Small Wars Manual, T. E. Lawrence’s Seven Pillars of Wisdom, Lewis Sorley’s A Better War, and John Nagl’s Learning to Eat Soup with a Knife (in the expensive hardcover edition, since the paperback wasn’t out yet)—and distributed them to all his officers, from brigade commander on down to platoon leader.

From these readings, Barno devised a strategy that he called, with a nod to Lawrence, the “Five Pillars.” The central pillar was to secure the population; the others were to defeat the terrorists, help with reconstruction and good governance, seek support from regional powers and international organizations, and order the troops to live alongside the people in their neighborhoods.

In short, it was classic clear-hold-build—a drastic departure from the search-and-destroy tactics that had been pursued previously. Before Barno arrived, American troops would mount their raids, then head back to their base in the outskirts, to no effect; the Taliban would simply melt away during the attacks, then return to the village at night, after the American big guns were gone.

Barno’s plan also called for unified social, political, economic, and military operations. The US ambassador, Zalmay Khalilzad, arrived in Afghanistan just a few weeks after Barno. Khalilzad had been born in Afghanistan to a Muslim family (his father was Sunni, his mother Shiite); he understood the culture, knew some of the tribal elders and warlords, and knew how to cultivate their interests. Barno moved his military headquarters into the Kabul embassy and set up his office just twenty feet from Khalilzad’s. The two struck up a close relationship, consulting and coordinating with each other on every aspect of policy. They also hooked up the four civilian provincial reconstruction teams with military battalions in their areas, partly to give them security but more to get them working jointly toward the same objectives.

The two also brought in specialists: successful corporate executives—some retired, some not—who’d been recruited by Martin Hoffman, a former secretary of the Army, to volunteer for what he called the Afghanistan Reconstruction Group. Led by Louis Hughes, vice president of General Motors, the group included Mitchell Shivers, a managing director at Merrill Lynch, who was assigned to work with the Afghan finance minister, and David Grizzle, a senior vice president at Continental Airlines, who worked with the transportation minister, among others.

By the time Barno rotated out of command in May 2005, support for the Taliban seemed to be waning and support for the provisional government booming. He thought he’d dealt a fatal blow to the insurgency.

Soon after returning to the Pentagon, he briefed Lieutenant General James Lovelace, the Army’s chief of operations. Lovelace said afterward that he’d had no idea what Barno had been doing in Afghanistan. Nor did any of the other senior officers; nor did they care to learn. No one else in the Pentagon, civilian or military, asked Barno for any further briefings. Nor was he rewarded for his creativity: his next posting was assistant chief of staff for installations management, a clerical job supervising the supply depots for the Army’s bases around the world. He retired from the Army one year later and, after a four-year teaching stint at the National Defense University, went to work as a senior fellow at John Nagl’s think tank.

When Barno rotated out of Afghanistan as commander, so did Zal Khalilzad as ambassador (he was transferred to Iraq), and so did the troops from the 25th Infantry Division who had carried out Barno’s plan. Barno’s successor would have had a hard time staying the course if he’d wanted to. It was a moot point, though, because he didn’t want to.

The new commander was Lieutenant General Karl Eikenberry, who four years later would return to Afghanistan as Obama’s ambassador and, among other things, write the secret cable that criticized the proposals for a troop surge and a counterinsurgency strategy.

Eikenberry had also been a West Point cadet, class of 1973. His career as an officer started out much like Barno’s, in the light infantry, but he then took a broader path, serving not only in airborne and Ranger units but also in mechanized infantry. He then moved into the policy world, earning a master’s degree from Harvard in East Asian studies, studying Chinese language and history in Hong Kong and Nanjing, and serving as a military attaché at the US embassy in Beijing, before a tenure at the Army’s Office of Plans, Policy, and Programs in the Pentagon.

In short, Eikenberry knew about COIN as a cadet and junior officer, but he never made it the focus of his interests or ambitions.

His first tour in Afghanistan, from 2002 to 2003, was as the US security coordinator, helping the Afghan army set up its first corps. When he returned as US commander two years later, his natural inclination was to resume where he’d left off. He knocked down four of Barno’s Five Pillars, retaining only the strictly military operations: going after insurgents and training the Afghan army. He disbanded the joint civilian-military teams in the field and sent the corporate volunteers home. He also moved his headquarters out of the embassy and cut off all coordination with its diplomatic corps. The new ambassador, Ron Neumann, didn’t mind; he and Eikenberry didn’t get along, and, in any case, he was under orders from Secretary of State Condoleezza Rice to turn his shop into a “normal embassy”—a reflection of the Bush administration’s deliberate neglect of Afghanistan as it switched gears to Iraq.

By the time Eikenberry’s tenure ended at the start of 2007, Afghanistan was coming apart. The summer before, NATO’s International Security Assistance Force had taken command of the war, except in the area along the eastern border with Pakistan—the scene of the most intense fighting, where American troops retained independent control. This was fine with the European allies, who didn’t want to fight much anyway. The ISAF commander, British general David Richards, saw his job as principally a peacekeeping operation. But as the European troops moved down south, the Taliban—who, it turned out, had never gone away—came out to fight. Suddenly, each of the allied powers—whose leaders hadn’t known they were getting into a war—declared a list of “caveats,” conditions under which its troops could and could not be deployed: one country would defend territory but not go on the offensive; another would keep the peace in the north but not in the south; still another would fly helicopters but not put troops on the ground. All in all, the NATO allies issued eighty-three caveats, strangling the top commanders’ flexibility.

The American military was hardly a model of cohesiveness, either. Eliot Cohen traveled to Afghanistan in April—he’d been hired as Condoleezza Rice’s counselor at the State Department the month before—and was appalled by how fractured the effort was. Some American officers were leading the autonomous combat units in the east; others were reporting to ISAF at Kabul headquarters; then there were the Army Special Forces, covert JSOC teams, and CIA agents scattered all about. And each of these entities was fighting its own war; there was no unified command, even on a strictly military level, much less a socioeconomic one. (General Richards had started to revive some of Barno’s joint civil-military development programs, but the new ISAF commander, a traditional US Army general named Dan McNeil, dismantled them once more.) Cohen told his colleagues that the campaign reeked of “military malpractice.”

Cohen worked closely with Lieutenant General Doug Lute. In November 2008, two months before the end of Bush’s second term, Lute delivered a briefing to Bush, spelling out all the problems that he and Cohen had uncovered: the Taliban’s widening offensives, Karzai’s corruption, the Afghan army’s weakness, and, not least, ISAF’s own internal chaos.

This was the state of the Afghan War when Barack Obama moved into the White House. Lute stayed at his job through the transition and after. His briefing to Bush served as the starting point for Bruce Riedel’s report to Obama. General Dave McKiernan, the commander in Afghanistan whom Gates had appointed in 2008 to replace McNeil, started to unify at least the military side of the campaign. But the basic situation didn’t change much during his tenure, which lasted less than a year.

And so, this was also the state of the Afghan War when Stanley McChrystal took over as commander.

•  •  •

One of the first things McChrystal did upon arriving in Kabul was reach out to Hamid Karzai. McChrystal knew that tensions between the Afghan president and Washington were high-strung and that, more often than not, the problems stemmed from Karzai: his regime’s corruption, his lack of discipline, and possibly his state of mind. But McChrystal also knew that COIN involved working “by, with, and through” the host nation’s government, and in this case, that meant working by, with, and through Karzai.

So McChrystal endeavored to make Karzai his pal and his peer, treating him with the utmost respect. Each time they met, he would assure Karzai that he regarded Afghanistan as a sovereign state and that, therefore, ISAF and its armed forces were all under the command of the country’s duly elected president. Karzai would beam at the kowtowing.

When McChrystal asked Karzai what troubled him most about the ISAF troops’ actions, his reply was instant: they were killing too many Afghan civilians. Dropping bombs on residential areas, smashing down doors of people’s homes in the dead of night, shooting at cars that veered too close to armored convoys—these raids and attacks were damaging, disrespectful, even criminal, and often based on mistaken intelligence.

McChrystal sympathized, and not just for show. Everywhere he went, from the ministries in Kabul to the markets in the villages, he heard the same complaints about the night raids and the bombing runs, even from people who had never personally witnessed raids or bombings. McChrystal knew that this was serious, potentially catastrophic. In conventional warfare, killing innocent civilians fell under the rubric of collateral damage: regrettable but unavoidable by-products of firing or dropping weapons in a populated area. But counterinsurgency warfare was all about protecting the people. Killing civilians in the course of chasing bad guys would breed local outrage, distrust, alienation, and blood revenge from the victims’ relatives; it would spur many people to ally with, even join, the insurgency. As Petraeus’s field manual had put it, “An operation that kills five insurgents is counterproductive if collateral damage leads to the recruitment of fifty more insurgents.”

It wasn’t just the killing that did damage to the war cause; it was the conventional mind-set that dominated the behavior of too many of his men in everything they did. Even when they drove him someplace in an armored convoy, they would barrel down the middle of the road, forcing cars on both sides into the ditch. Looking out the slotted window, McChrystal saw the expressions on the civilians’ faces: the horror and humiliation from suffering yet another indignity at the hands of foreign occupiers. Yet his men thought they were doing the right thing: they’d been taught that self-protection was the primary mission—and protecting the commanding general, a higher mission still. He needed to shift the mind-set, to convince his troops that winning the war meant winning the support of the Afghan people; if the people weren’t convinced that ISAF was there to support them, the war would be lost for sure.

On July 2, well before his assessment team had finished its work, two days before he would hear the team’s first interim briefing, McChrystal issued a document called the “Tactical Directive,” addressed to all American and ISAF military units in Afghanistan.

The main struggle in this war, he wrote, was “for the support and will of the population.” Winning and maintaining that support “must be our overriding operational imperative—and the ultimate objective of every action we take.” This kind of war, he stressed, as many other commanders had stressed in similar documents through the years, “is different from conventional combat . . . We must avoid the trap of winning tactical victories—but suffering strategic defeats—by causing civilian casualties or excessive damage and thus alienating the people.” He allowed, “We must fight the insurgents and . . . defeat the enemy,” but then reemphasized that “we will not win based on the number of Taliban we kill.” Victory will instead depend on “our ability to separate insurgents from the center of gravity—the people.”

He continued, in a passage that he knew would arouse the ire of some of his men, “I recognize that the carefully controlled and disciplined employment of force entails risk to our troops.” They should take steps to mitigate those risks; certainly they should take whatever action they felt necessary for self-defense. “But,” he wrote, “excessive use of force resulting in an alienated population will produce far greater risks. We must understand this reality at every level in our force.”

Then came the hammer: “I expect leaders at all levels to scrutinize and limit the use of force,” especially air strikes, “against residential compounds and other locations likely to produce civilian casualties.” Following this directive, he acknowledged, “requires a cultural shift within our forces—and complete understanding at every level—down to the most junior soldiers.”

Eight weeks later, on August 26, as the final draft of his assessment team’s report was being completed—and a full three months before President Obama decided whether even to pursue a COIN strategy—McChrystal sent all of his troops another document, called “ISAF Commander’s Counterinsurgency Guidance.”

It was an elaboration of his “Tactical Directive” but written in still starker language. Directly below the heading, almost as if it were a subtitle, he wrote, in italics, “Protecting the people is the mission. The conflict will be won by persuading the population, not by destroying the enemy. ISAF will succeed when [the Afghan government] . . . earns the support of the people.”

Key phrases were certain to jump out at most American soldiers and marines:

We will not win simply by killing insurgents . . . This means that we must change the way that we think, act, and operate . . . Every action we take must reflect this change: how we interact with people, how we drive or fly, how we patrol, how we use force . . . This is their country, and we are their guests . . . Large-scale operations to kill or capture militants carry a significant risk of causing civilian casualties and collateral damage. If civilians die in a firefight, it does not matter who shot them—we still failed to protect them from harm . . . Think of counterinsurgency as an argument to earn the support of the people . . . Earn the support of the people and the war is won, regardless of how many militants are killed or captured. We must undermine the insurgent argument while offering a more compelling alternative . . . There is clearly a role for precise operations that keep the insurgents off balance, take the fight to their sanctuaries, and prevent them from affecting the population. These operations are important, but, in and of themselves, are not necessarily decisive.

McChrystal meant to be laying down basic principles; the document was his guidance. “Keeping the right balance over time is critical,” he wrote, “and there is no mathematical formula for it.” But then he supplied just such a formula: “Strive to focus 95% of our energy on the 95% of the population that deserves and needs our support. Doing so will isolate the insurgents. Take action against the 5%—the insurgents—as necessary or when the right opportunities present themselves. Do not let them distract you from your primary tasks.”

Even some COIN enthusiasts thought McChrystal was going too far. It was a provocative enough departure from conventional wisdom when Mao and Galula had described insurgency wars as 80 percent political, 20 percent military. Now McChrystal was splitting it at 95 and 5. And he was expressing this equation well before enough American troops had arrived to carry out a counterinsurgency strategy—well before the president had decided to send enough troops or to adopt such a strategy.

When David Petraeus edited the COIN field manual, he’d toned down some of his writing team’s more exuberant passages, for instance changing “The best weapons don’t shoot” to “Sometimes the best weapons don’t shoot” and inserting several paragraphs throughout the text, making it clear that sometimes a soldier had to go kill bad guys. Petraeus’s final draft didn’t define “sometimes.” Several officers had raised the question at his COIN conference in Fort Leavenworth and afterward: where to draw that line and how soldiers were supposed to recognize the line when they saw it. In the scheme of the “three-block war” (combat on one block, peacekeeping on the next block, and handing out humanitarian aid on the next), how would they know which block they were occupying? Petraeus had tried to capture the fine line in the tone of his prose (he often told officers that setting the right tone was one of a commander’s most vital tasks), but the ambiguity remained. Probably it could never be resolved fully; it was something a soldier felt in his bones after months or years of experience; in any case, it was a shortcoming in the manual that some still criticized.

There was nothing nuanced at all, though, about the tone of McChrystal’s “Counterinsurgency Guidance”: it sounded a stern order for strict restraint; it sent a signal that soldiers could get into serious trouble if they killed a single civilian, whatever the circumstances; it thus had the effect of encouraging not merely restraint but passivity, a dangerous trait in any war.

McChrystal’s impulse was well founded. Before he took command, civilian casualties were soaring, and one reason was that nobody had issued an order to minimize them. His directives marked an attempt, if overwrought, to do that.

•  •  •

Around this time, Sarah Sewall, the director of Harvard’s Carr Center for Human Rights Policy, wrote David Petraeus a note offering to conduct an on-site study of CIVCAS incidents, the military term for civilian casualties.

The two had stayed in touch ever since they cosponsored the COIN field manual conference at Fort Leavenworth three-and-a-half years earlier. Petraeus liked the idea and passed it on to McChrystal. As head of Central Command, which had authority over US military operations in Iraq and Afghanistan, Petraeus could have ordered the study himself. But he knew the art and value of getting buy-in from the crucial players, and McChrystal was the commander on the ground. The study would get a better hearing, and Sewall’s very presence would be better tolerated, if the project appeared to be McChrystal’s idea. So Petraeus forwarded her proposal with as much aloofness as he could muster.

McChrystal recognized what Petraeus was doing, but he liked the idea, too, so he commissioned Sewall to do the study. He was deeply concerned that his officers weren’t following his “Tactical Directive” or his “Counter-insurgency Guidance”; he wanted her study to include an assessment of how closely they were following the orders and some suggestions on how to push through his message more clearly.

Sewall began work in November, though not in earnest until the following February. As she and her team—nearly forty analysts, civilian and military—were flown and driven all over Afghanistan, talking with American and ISAF planners, staff officers, and troops in the field, she realized that McChrystal’s directives, far from being ignored, were instilling fear up and down the ranks. Many of the troops understood the reasoning behind the rules of engagement, but many others didn’t; it had never been explained to them. They would call in air strikes when they knew that insurgents were in a building; their superior officers, also drenched in fear, would deny the request, and everyone involved would walk away from the scene, feeling not just hamstrung but demoralized and vulnerable.

Whenever Sewall dropped in on ISAF headquarters in the course of her work, she could practically smell the toxic sweat beads. McChrystal’s staff formed a tight clique. They were all Rangers, as McChrystal had been, and they all tried to act just like the boss: a groupthink seemed to have taken hold, a shared conviction that they were right about everything and that everyone else was wrong, naïve, or stupid. Sewall understood why McChrystal had tightened the troops’ rules of engagement, and she lauded him for doing so; that was how a commander made his troops understand and internalize his intent. But usually commanders adjusted these rules now and then to accommodate new realities or incorporate lessons learned. The insularity of McChrystal’s staff suppressed the normal tendency to reassess the situation or modify a decision. (Petraeus had a loyal entourage of aides as well, but he’d gathered them from various branches of the Army and even included civilians; his management style welcomed new ideas, to a point.)

Sewall wrote up her findings in a 147-page report. On Tuesday, April 13, 2010, she and a few members of her team briefed McChrystal on their main findings. In one part of the briefing, she told the general that, contrary to his concerns, most of the troops did understand his guidance and how it fit into a counterinsurgency strategy.

McChrystal reacted with skepticism. How could that be true? Too many civilians were still getting killed. Just the week before, an American unit had sprayed a busload of innocent people with gunfire. Those troops didn’t “get it.”

Sewall decided to put it all out there. She told him about the frustrations and complaints she’d heard, the times when an attack or air strike seemed warranted but permission was denied. This was nothing new to McChrystal. He regarded himself as a soldier above all; he frequently visited the troops in the field, even went out with them on patrols; he’d heard the same stories and would respond by trying to explain his guidance more clearly.

She then went one step further. Sarah Sewall, the Ivy League human-rights scholar and the only woman in the room, looked unblinkingly at Stan McChrystal, the peerless professional killer, and said, “General, counterinsurgency is a combination of offense, defense, and stability operations. Don’t forget the offense.”

McChrystal growled, “Don’t tell me how to run my war.”

•  •  •

Meanwhile, the war—McChrystal’s war—wasn’t going well.

The clearest and most disturbing sign was the battle for Marja, a district of roughly eighty thousand people in southern Afghanistan’s Helmand Province, where the Taliban had seized the reins of power two years earlier.

McChrystal saw the battle as a key test of his counterinsurgency plan, and publicly said as much. He’d been preparing for the battle since September, not long after his arrival as commander. Dubbed Operation Moshtarak (the Dari word for “together” or “joint”), it got under way in the predawn hours of Saturday, February 13, 2010, as an advance force of American, British, French, and Afghan troops, most of them marines and Special Forces, poured into the province and occupied key sites, confiscated weapons caches, and began to push out the insurgents. At its full flush, the operation pulled together fifteen thousand ISAF and Afghan troops, the largest American-led offensive since the invasion that toppled the Taliban more than eight years earlier.

In McChrystal’s plan, the assault marked the opening volley of a broader campaign: the “clear” phase in a full-fledged attempt at COIN’s clear-hold-build. He told Dexter Filkins, the New York Times’s war correspondent, “We’ve got a government in a box, ready to roll in.”

He went on to describe an enormous apparatus—more than a thousand police, a new governor, and vast teams of administrators, all Afghans—standing by, all set to swoop into Marja the moment the shooting was over and the insurgents were cleared. As the tens of thousands of America’s surge troops began to arrive in country, the process would start up again in another town, then another, all along the Helmand River, out to the edge of the border with Pakistan. At that point, he predicted, a critical mass of Taliban leaders would abandon their hopes for victory and seek a political settlement, which would soon spark the end of the war.

Filkins’s article about his interview with McChrystal was published on February 12, the day before the offensive was to begin.

Heidi Meyer was among the many political officials who read the Times that day in a state of shock. She’d been working on Afghanistan’s problems for nearly four years: first as a civilian aide inside the US military command, then on the Pentagon’s Afghan policy desk, and now, for the past year, as the US embassy’s team leader for stabilization, the official in charge of helping to stand up institutions of governance in Kabul and out in the districts and provinces. And she knew for sure that there was no such thing as “government in a box,” especially in Afghanistan, which lacked the barest prerequisites for government of any sort.

A few months earlier, Meyer had heard McChrystal’s staff brief the general on the idea of dropping in schools, businesses, a hospital, a governor’s office—replete with adequate supplies and trained personnel—as soon as the fighting was over. The briefer talked casually about doing this in forty-eight “focus districts,” eventually in eighty “key terrain districts,” of which Marja would be merely the first. She told the staffers afterward that the idea was preposterous. This wasn’t the way governance developed, anywhere, and certainly not in Afghanistan, which had no capacity to deliver services so quickly. Its people, including most of its officials, were illiterate; they lacked the skills or technology to draw up a budget or to administer a payroll. This dream, she said, would take a decade or more to realize. Other officials from the AID office told them much the same.

McChrystal knew about the skeptics, but he asked what they would do instead and heard no answer. “Clear” had to be followed by “hold” and “build”; this was the plan on how to do that. His staff officers dismissed the pessimism out of hand. They were Rangers, flush with a can-do confidence, emboldened by their recent tours in Iraq. If the civilian bureaucrats couldn’t get the job done, they’d do it themselves.

And now here they were, actually setting out to do it.

Two days before the offensive began, McChrystal felt confident enough to dispatch the Afghan interior minister, Hanif Atmar, to meet with more than three hundred of the area’s tribal elders, brief them about the impending offensive, and assure them that the fighting wouldn’t last long, that police would be brought in to keep the Taliban out after the military pushed them out, and that the government would follow up with roads, medical clinics, and other vital projects.

McChrystal’s staff figured that Marja would be conquered in a matter of days, at most a few weeks. By the time Sarah Sewall briefed the general on her civilian-casualties report in mid-April, the operation had gone on for two months; the troops had captured a lot of territory, but they hadn’t yet secured it, not to the point where they could safely transfer control to the Afghan police. Even by the end of the year, Taliban fighters still lingered, sniping from rooftops, setting off bombs, beheading collaborators, and knocking on doors at night to warn of the same fate for those who consorted with the infidel occupiers.

The ISAF troops could clear but not hold, and the “build” phase of the plan lay in the incalculably distant future. As Meyer and other civilian specialists had warned, there was no government in the box, no pool of administrators standing by. Some of Karzai’s ministers had nodded their heads when American officers asked if such things were in the works; they’d learned to tell a man with power and money what he wanted to hear. The Independent Directorate of Local Governance, the Afghan agency that funneled much of this money, had recruited a fair number of Afghans to come take the new jobs, but most of the men vanished once the battle dragged on for much longer than expected, once it was clear that Marja was still too dangerous to police and govern.

•  •  •

In the late fall of 2009, as McChrystal was preparing for his battle in the south, Colonel Randy George, brigade commander of the US Army’s Task Force Mountain Warrior, and a career diplomat named Dante Paradiso were engaged in a different kind of counterinsurgency, dealing not with Karzai’s government—which they viewed as fatally corrupt—but with the tribal elders of Nangarhar Province along the craggy border with Pakistan.

On January 21, 2010, their campaign culminated in what should have been rewarded as a triumph. They brought together 160 of the most influential elders of the Shinwari tribe—one of the area’s largest, encompassing about a half million people—to sign a pact denouncing the Taliban, threatening to punish anyone in their tribe who supported the Taliban, pledging to stop growing poppies, and declaring their allegiance to the central government in Kabul, despite grave misgivings over Karzai and his cronies.

In exchange, the elders asked the Americans for more development assistance, the removal of some particularly corrupt local officials, and the right to have a say in which Afghan security forces would be stationed in their province.

George pledged $1 million from his commander’s discretionary fund for development, but emphasized that while the elders could suggest local projects, he, not they, would handle and distribute the money. As for the other demands, he and Paradiso would do their best to persuade their higher-ups but could make no guarantees. That was good enough for the elders; the deal was struck.

It was a remarkable pact. There had been nothing like it, in scope or substance, in the eight years since the Taliban had been overthrown.

And yet the pact was sabotaged by the Afghan president and by the American ambassador.

•  •  •

Randy George had some background in counterinsurgency and the art of dealing with tribal cultures. He’d served three tours of duty in Iraq. The first was as part of the invasion force, but the second was more intriguing: a brigade commander in Kirkuk, where he wound up spending most of his time arbitrating property disputes between local Kurds and Sunni Arabs.

He was in frequent contact with Dave Petraeus, who’d been following his career since the early 1990s, when Petraeus commanded a battalion in the 101st Airborne Division and George was his top lieutenant. George also read the classic counterinsurgency books, including John Nagl’s (the two had been classmates at West Point in the mid-to-late 1980s), and he absorbed their principles. But George also realized that Iraq was very different from Malaya and that the modern American military had much less sway over a sovereign country in the media age than the British army had possessed over one of its colonies in the aftermath of World War II. He gained deeper wisdom about how to apply COIN’s basic principles from reading histories of the Kurds and the Sunnis and from observing the role that history and culture were playing in the tenuous borderline between war and peace that he himself was treading every day.

In Kirkuk, he also met and worked with Emma Sky, the British Arabist who was trying to help American officers understand why the insurgents were fighting. General Ray Odierno had met them both when he was commander of the 4th Infantry Division. When Odierno was promoted to corps commander in the early days of the surge and hired Sky as his political adviser, she suggested that he bring George into his circle, too. Odierno put George to work on a road map for Iraq’s future: what if the surge didn’t work, what if it did, what might happen next in either case, and how would various outcomes affect US interests?

George came away from Iraq with a knack for solving gritty human disputes, and for spinning grand strategic scenarios, in a very foreign land.

He came to Afghanistan about the same time McChrystal did, as commander of an infantry brigade—6,500 American soldiers—spread out across four provinces in the country’s desolate eastern region. Some of his men had been sent to occupy fortified bastions on remote cliffs, positions inherited from the reign of earlier commanders, whose battle plan had been to shoot the Taliban as they crossed over the border from Pakistan. It was a strategy of static defensiveness and, in the end, futility, as the Taliban would simply cross a few mountain paths away.

Early on in his deployment, George wrote a twenty-one-page strategy paper that called for shifting the focus away from “attacking the enemy in remote areas” to “protecting and developing the major population centers” in order to “separate the enemy from the local population” and “connect local populations to a responsive Afghan government.” He also called for a thorough anticorruption campaign. The Taliban were winning the fight for popular support, he observed, by setting up their own courts, which, for all their faults, did at least mete out swift justice while barring bribes—a stark contrast with the official courts, which were both inefficient and corrupt. For the United States and ISAF to tolerate this corruption, he wrote, would only validate the Taliban’s claims to moral authority.

George briefed his plan to McChrystal, who approved it.

Around this time, firefights between the Shinwari tribesmen and the Taliban were intensifying. A Shinwari elder named Malik Niaz appealed to Colonel George for help. Not long after, another elder, Malik Usman, whose brother had been killed by the Taliban earlier in the year, approached George as well.

George and Paradiso learned that Niaz and Usman were rivals (the Shinwari tribe was rife with ancient feuds and factions), and they saw this coincidence as an opportunity to reunite the two elders in a common cause. With the two Americans’ guidance, they drew up a ten-point pledge to band together against the Taliban. George and Paradiso encouraged them to reach out to other elders and get them to sign the pledge, too. Over the next few months, the elders expanded the roster from two to four to sixteen and beyond, culminating in the signing ceremony of January 21, 2010.

The two Americans saw this meeting as a very big deal, maybe the start of an Afghan parallel to the Anbar Awakening, which had sparked the turning point in Iraq three years earlier and which had stemmed from much the same impulse—tribal elders turning to a foreign army, despite their prior antipathy, because they felt more threatened by Islamist militants.

One week after the signing, a story about the Shinwari pact appeared on the front page of the New York Times. Secretary of State Hillary Clinton happened to be in London at a conference of international donors to Afghanistan. She read the story and publicly hailed the pact as a reflection of America’s new strategy.

“This is classic counterinsurgency,” she proclaimed in an interview with CNN. “As General McChrystal said, you’re never going to kill or capture everybody calling himself a Taliban. But you can change the political environment, so that those who continue to call themselves Taliban become more and more isolated, and that’s what we’re seeking.”

Clinton had good reason to believe her comments were in line with broad US policy. In his memo on the new war strategy two months earlier, President Obama had written that “we must focus on what is realistic . . . working with Karzai when we can, working around him when we must”—the latter, in order to improve governance and reduce corruption at the local level. The Shinwari pact seemed a textbook case of working around Karzai when we must.

For precisely that reason, Karzai loathed the pact. It had been conceived, written, and endorsed without his involvement, outside the realm of his control.

Karzai had always opposed plans to build up a base of independent local power because they had the effect—and, he suspected (sometimes rightly), the intent—of stripping away the central government’s powers, which is to say his own power and that of his cronies, whom he’d installed as provincial or district governors, jobs that they treated as cash cows, collecting bribes, extorting farmers, and pocketing revenue. Any plan that empowered people outside this network—for instance, Shinwari tribal elders in Nangarhar Province—threatened to undermine Karzai’s true constituents and thus the foundations of his power. In this sense, it didn’t matter that the Shinwari had pledged to oppose the Taliban or even to support Karzai’s government; the problem was that they were assuming the right to make such a pledge of their own volition.

Karzai complained about the pact to Ambassador Eikenberry and General McChrystal. Dante Paradiso—who was the embassy’s chief representative in the four eastern provinces, heading a team of forty civilian diplomats and aid workers attached to Colonel George’s brigade—expected Eikenberry to stand up for their work. After all, his now-famous critique of McChrystal’s proposal to send more troops and switch to a classic COIN strategy had stemmed from a judgment that Karzai was not a suitable strategic partner. Surely, then, making a deal with influential tribal leaders—real local powers—to band against the Taliban, and even pledge fealty to the central government, would be seen as a step in the right direction.

Instead, Eikenberry stood up for Karzai. The ambassador was also in London when Secretary Clinton touted the Shinwari pact on CNN. The thing he didn’t like was that the pact was news to him. Paradiso had sent the embassy a cable outlining the deal, emphasizing that he and George had not promised to give the elders money outright or to provide them with weapons. Somehow, though, the ambassador never saw the cable. Eikenberry was obsessed with control; he demanded to be—and, just as important, to be seen as—the dominant actor in his realm. He had no patience with minions who tried to make policy or tamper with the delicate machinery of diplomacy on their own, and that’s what he thought Paradiso was doing.

After Eikenberry made his displeasure known, his senior staff followed through. One embassy official complained to a Washington Post reporter that the pact had “really stirred things up.” It was like telling the Shinwari, “Congratulations, you get a pony.” Now, the official said, “other tribes are saying, ‘Why don’t I get a pony?’”

This made no sense to Paradiso. The same thing could be said about any development project. (Why don’t I get a dam? Why don’t I get my sewers cleaned up?) And what would be so bad about giving other tribes a “pony” if they, too, signed a pact against the Taliban? The whole point of the commander’s discretionary fund was to improve battlefield conditions: to pay communities to support the coalition’s cause and to oppose the enemy’s. That’s what Petraeus and MacFarland had done in Mosul and Anbar Province. Meanwhile, Eikenberry’s embassy staff was handing out hundreds of millions of dollars for power plants and construction projects without any trade-offs, guarantees, or accountability.

McChrystal privately supported the pact; he had no problems with what George was doing. But he was in the final stages of preparing the battle for Marja, which he saw as essential to his own broader strategy, and Karzai hadn’t yet granted him permission to move on the city. He needed to coax and massage Karzai to approve it; this was no time to oppose him on a peripheral point.

So Eikenberry won the day. He not only disavowed the Shinwari pact, he issued an edict barring all embassy officials from engaging with any tribes in any setting where a “pact” might be discussed—this, in a country of tribes.

The decision drove a deeper wedge between the embassy and the military. Officers could still reach out to tribes, and often did; the civilians were supposed to lead the way in reforming Afghan governance, but they couldn’t, since governance required dealing with tribes, which they were now barred from doing.

Meanwhile, George and Paradiso were forced to cut off the flow of money to the Shinwari and to retract their part of the deal. Two weeks after the pact’s signing, a violent land dispute had erupted between two factions within the tribe. Some embassy officials pointed to the fight, with some gloating, as proof that the deal was doomed from the outset. Maybe it was; probably it wouldn’t have lasted long or spread far. Still, it was something, which was more than what most districts or provinces had going for them. Even while the Shinwari factions were squabbling, they reaffirmed their commitment to the pact. But after the money stopped flowing and the consultations ended, the elders grew more distrustful toward George and Paradiso as well as toward one another. They saw the cave-in to Karzai as proof that the Americans were complicit with the corruption. The pact collapsed.

•  •  •

It was no wonder that when Sarah Sewall briefed McChrystal on her civilian-casualties report in mid-April 2010, she found the general grumpier than usual.

A few days later, McChrystal and his staff made their ill-fated trip to a NATO conference in Europe, where they were joined by a Rolling Stone reporter named Michael Hastings, who taped some of them making crude and demeaning remarks about Obama, Biden, a few other American officials, and the French foreign minister. It was the toxic atmosphere that Sewall had observed, puffed up to self-destructive levels.

President Obama obtained an advance copy of the article. He read the first few paragraphs and put it down in disgust. He called a staff meeting that night. By the next morning, he knew he had no choice: McChrystal had to go.

He called the general to the White House for a one-on-one meeting on June 23. McChrystal had tendered his resignation the day before. At the meeting, Obama accepted it. That afternoon, he announced the move at a press conference, saying that the behavior depicted in the article, for which McChrystal had taken full responsibility, “erodes the trust that is necessary for our team to work together to achieve our objectives in Afghanistan.” He also emphasized, “This is a change in personnel, but it is not a change in policy.”

The president then introduced the general he’d chosen as McChrystal’s replacement—David Petraeus.

Choosing Petraeus was an exceptionally shrewd move, preempting the otherwise inevitable complaint that Obama was crippling the war effort by removing a perfectly fine commander and replacing him with someone who, competent though he may be, would take months to get up to speed.

Petraeus, of course, was no newbie. He was immersed in the fight, having been CentCom commander for nearly two years. More than that, he was David Petraeus, the best-known, most admired American general in a generation, the author of the COIN field manual, the miracle worker of Iraq. Maybe he could work miracles in Afghanistan, too.

•  •  •

Petraeus took command in Kabul on the Fourth of July, 2010. As his helicopter landed on the grounds of ISAF headquarters, a small group of civilian analysts mingled with the official welcoming party.

The analysts didn’t quite know what they were doing there. McChrystal had invited them on an “opinion leaders” tour of the war, one of several such tours that he’d organized for prominent experts and analysts. Upon their arrival, they’d learned of McChrystal’s dismissal. No one could tell them whether or not the tour was still on.

Still, they weren’t fazed by their odd status. They were by now familiar figures in these settings: Fred and Kimberly Kagan, Steve Biddle, the neocon historian-columnist Max Boot, and Sarah Chayes, a former National Public Radio war reporter who’d moved to Kandahar eight years earlier, learned to speak Pashto, set up some small development projects to wean local farmers away from growing poppies, moved on to join the staff at ISAF headquarters, and was now about to become the Afghan-policy adviser to Admiral Mike Mullen, the chairman of the Joint Chiefs of Staff.

They were all familiar figures to Petraeus, too. He knew most of them from his own opinion leaders’ tours or assessment teams back in Iraq. He was surprised to see them in Kabul, but he also saw an opportunity.

After his opening remarks and other ceremonial obligations, Petraeus approached them and asked if they’d be willing to stay on for a monthlong project. He knew that one of the biggest obstacles in the war was the Afghan government itself. He could design the most brilliant counterinsurgency campaign, but it would be for naught if Karzai, his ministers, and key district governors failed to follow through with sustained security and basic services—which would require thoroughgoing reforms.

He wanted the analysts to draft a detailed plan on how to get these reforms going. He also asked them to keep mum on what they were doing. McChrystal’s assessment team had set off enough of a commotion with its famous sixty-six-page memo, and the Obama administration had long since carved its policy in stone. Petraeus didn’t want anyone in Washington to get the idea that he was reopening long-settled debates.

Though he could not have known it, Petraeus was reopening one of the unsettled debates that had haunted some members of McChrystal’s assessment team, including a few who were now on Petraeus’s secret new advisory group.

Sarah Chayes had raised the issue during that earlier assessment, not as a team member but as one of McChrystal’s staffers who’d been asked to give the team a briefing. Chayes had concluded that the problem with the Afghan government wasn’t that it lacked competence or resources, but, rather, that it was corrupt and predatory. Simply training officials wouldn’t improve matters; it might only make them more effective predators.

Chayes’s briefing had carried weight among several analysts on the team. Most of them had only visited Afghanistan now and then for a few days or weeks at a time; she’d actually been living there, outside the embassy’s gates, for years. They also saw evidence of her argument as they toured the country and talked with local officials and ordinary citizens.

In their report for McChrystal, they’d warned that corruption was pervasive and that the war might be lost without serious reforms. But they had neither the mandate nor the time to delve into the problem. Now Petraeus was asking them to put the issue front and center and to devise a plan for action.

The new team finished its report in early August and briefed it to Petraeus. Afghanistan’s governing apparatus, the analysts concluded, was basically a network of malign actors. Pushing for reform at the top of the network, within Karzai’s central government, wouldn’t work. Better, they suggested, to undermine the network by shutting off its levers of patronage at the bottom and to make this effort part of every brigade commander’s mission. A senior officer should be named to coordinate the campaign, roaming across the country to make sure that the commanders were enforcing the edict. The team recommended that this job be given to H. R. McMaster, who had just arrived in Kabul to serve on Petraeus’s staff.

Petraeus was impressed with their analysis but found most of their proposals impractical. First, he couldn’t simply bypass Karzai. One of Petraeus’s strategic goals was to help stabilize Afghanistan; overhauling all the districts’ local governing boards and spreading power to a new array of warlords was hardly a recipe for stability.

Second, the plan would undermine another one of his strategic goals: protecting the Afghan population. The local officials who were taking bribes and extorting businesses (with or without Karzai’s complicity) were also helping with local security and, in some parts of the country, guarding convoys of ISAF supply trucks as they drove through dangerous areas. If the cash spigot was shut off, they’d let the Taliban attack the supply trucks and maybe join in, attacking not only the trucks but the troops and their local collaborators.

The team’s briefing acknowledged the risk, outlining precisely that scenario as a potential side effect of the plan. But Petraeus saw it as an inevitable and dangerous consequence of their proposal, not a potential side effect. And even if it did work, even if it was possible to devolve power to the local districts, it would take way too much time. Petraeus was under pressure to show fast results in improving security; the plan’s effects would be slow and murky at best.

Steve Biddle, who had also worked on McChrystal’s assessment team and a similar team that Petraeus had appointed in Iraq, was disturbed by the broader implications of Chayes’s analysis. Two years earlier, Biddle had written an article, pointing out a logical hole in Petraeus’s field manual. The manual envisioned America’s role in a COIN campaign as helping the host government fulfill its own best interests by becoming a more legitimate defender of its peaceful citizens’ well-being. But what if the host government had a different view of its best interests? What if it was more interested in catering not to the population as a whole but to a particular sectarian group that was hostile to rival groups? Helping that sort of government perform more effectively might make things worse.

When Biddle wrote the piece, he was thinking of Maliki’s Shiite-dominated government in Iraq; but after hearing Chayes’s briefing, he realized his critique had far greater relevance in Afghanistan. If Chayes was right, if the Afghan government was beholden to a criminal network of patronage, then its leaders would never warm to the notion of reform. The implications were these: either a COIN campaign in this country was futile because America’s “strategic partner” wasn’t really a partner (as Eikenberry had argued in his cables the previous November) or Petraeus would have to persuade, manipulate, and coerce Hamid Karzai into altering his own concept of Afghanistan’s national interests.

This was essentially the conclusion that General Pete Chiarelli and his political adviser, Celeste Ward, had reached about the diverging American and Iraqi interests back in the fall of 2006. And it was the conclusion that Dave Kilcullen had reached more generally in 2008, when he wrote in the interagency guide on COIN that it would be “folly” to engage in counter-insurgency abroad “unless there is a reasonable likelihood that the affected government will introduce necessary reforms.”

Petraeus knew all about this. He’d done quite a lot of persuading and coercing with Maliki in Iraq, and he figured that he’d need to do the same with Karzai. He’d achieved some success with Karzai already. For months, McChrystal had tried to set up Afghan self-defense units in local districts, but Karzai had always blocked him, fearing (not unreasonably) that they would empower warlords and weaken the national army’s authority—his authority. Through a mix of cajoling, arguing, arm twisting, and soothing (and finally agreeing to place the units, ultimately, under the interior ministry’s authority), Petraeus won Karzai’s endorsement for the local units just ten days after taking command.

The team members did inspire Petraeus to issue a document called “Counterinsurgency Contracting Guidance,” outlining new rules for buying goods and services from Afghan companies without fueling corruption or financing insurgents.

But Petraeus recognized that these were first steps, fraught with risk and reversible at any moment.

He also well understood the same broader, more worrisome implication that Biddle had detected: the possibility that, despite his best efforts, COIN simply might not work in a country where the government’s leaders didn’t want to reform—in effect, didn’t want COIN to work.

In the PowerPoint briefing that he delivered to countless congressional delegations at his Kabul headquarters, Petraeus titled one of the slides “Storm Clouds,” which listed all the aspects of Afghanistan that bode ill for the war’s prospects. Topping the list was the Afghan government’s incapacity; cited not far below was the government’s network of criminal patronage.

David Galula’s classic book Counterinsurgency Warfare, which Petraeus was still consulting, contained a chapter titled “The Prerequisites for a Successful Insurgency.” According to Galula, these prerequisites included a weak government, a neighboring country that offers safe havens, and a predominantly rural, illiterate population—precisely the traits marking Karzai’s Afghanistan. Even the country’s geography, it seemed, favored insurgents. As Galula put it, “the ideal situation for the insurgent would be a large land-locked country, shaped like a blunt-tipped star, with jungle-covered mountains along the borders and scattered swamps in the plains, in a temperate zone with a large and dispersed rural population and a primitive economy.”

In other words, an insurgent’s ideal situation was the topography and demography of Afghanistan.

Petraeus saw all this as a daunting stack of obstacles, but surmounting obstacles was his specialty; it was what he did. He’d placed first at Ranger school; he’d survived a bullet wound, a collapsed parachute, and, just a year before taking command of ISAF, a bout of early-stage prostate cancer. More pertinent still, he’d defied the odds in Iraq, with the surge and the spread of the Anbar Awakening. Taming Afghanistan might be his hardest challenge, but “hard” was not the same as “impossible.” Others had failed, but most of them hadn’t seen the war correctly. Petraeus liked to say that once you got “the inputs right”—the right strategy, personnel, tactical guidance, and metrics of success—you were more likely to get the right outcome. The commanders before him didn’t get the inputs right. McChrystal came close, but he went too far on some points, not far enough on others, and, finally, he ran out of time.

Petraeus was confident about inputs; his biggest doubts concerned time. He had a year—six months had already passed in the president’s eighteen-month deadline—to show results. That wasn’t nearly enough time for classical COIN techniques alone, which usually took, as his field manual noted, a decade or longer to play out. (Yes, this war had been going on for almost a decade by the time Petraeus took command, but it hadn’t been waged as a COIN campaign. As several Army officers liked to say, “We haven’t been fighting in Afghanistan for eight years; we’ve been fighting there for one year, eight years in a row.”) Petraeus would have to throw everything he could muster into the war all at once—more COIN, more killing bad guys, more training Afghan soldiers—and hope that all the pieces fell into place, and quickly.

One thing he did early on, both to facilitate this all-out effort and to correct what he saw as overreach, was to revise McChrystal’s signature documents: his “Tactical Directive” and his “Counterinsurgency Guidance.”

Petraeus’s revisions, both issued on August 1, four weeks into his command, recited many of the same basic principles: to “secure and serve the population,” to “live among the people,” to “help Afghans build accountable governance,” and to “reduce the loss of innocent civilian life to an absolute minimum,” in order to avoid alienating the Afghan people, who are “the center of gravity in this struggle.”

But gone were McChrystal’s rigid formulas that had frustrated so many troops in the year since his rules of engagement had gone into effect. Where McChrystal had stressed the need to avoid violence except under limited circumstances (seeing COIN as 95 percent political, only 5 percent military), Petraeus wrote explicitly, “Protecting the Afghan people does require killing, capturing, or turning the insurgents. Indeed . . . we must pursue the Taliban tenaciously.” In his revised “Counterinsurgency Guidance,” he stated the point more aggressively: “Pursue the enemy relentlessly. Together with our Afghan partners, get our teeth into the insurgents and don’t let go.” It was within this context that he summed up his directive: to “fight with great discipline and tactical patience” and “use only the firepower needed to win a fight.”

The new orders were still ambiguous; they still harked back to the unanswered questions at Petraeus’s COIN conference in Fort Leavenworth four years earlier: at what stage is force justified in a counterinsurgency campaign, and how do the troops know when they’ve crossed that line? Still, Petraeus’s guidance set a new tone, one meant to let the troops breathe easier, insisting that they act with an awareness of COIN principles but not to the point where they feared the prospect of court-martial if they accidentally killed a civilian in the course of staving off bad guys.

From that moment, Petraeus stepped up the kill-and-capture side of the operation. In his first three months as commander, US warplanes and drones dropped 1,600 weapons on targets in Afghanistan—nearly half the number that had been dropped in the entire previous year. He also let loose the JSOC Special Ops forces all over the country, to a greater extent than even McChrystal, who’d practically invented JSOC.

The new twist in strategy seemed to have some tactical effect, killing or capturing three hundred midlevel Taliban leaders (including a number of shadow provincial governors and district commanders), as well as killing more than eight hundred rank-and-file insurgents and capturing another two thousand. And the attacks were fairly precise; civilian casualties, caused by air strikes and raids, were down.

Petraeus had long been saying in speeches and congressional testimony that you can’t simply kill your way to victory in these kinds of wars; they usually end with a negotiated settlement, not a surrender ceremony. But he also thought that you did have to kill a lot of insurgents to get them to the bargaining table. He and Karzai had said they would reconcile with any insurgents who pledged to lay down arms and uphold the Afghan constitution. Insurgents weren’t likely to do those things if they thought they were winning the fight.

•  •  •

Hy Rothstein, a retired Special Forces colonel who was now teaching at the Naval Postgraduate School in Monterey, took a fact-finding trip to Afghanistan a few months after Petraeus had taken command. The two had been classmates at West Point and had stayed in touch since. Eight years earlier, in late 2002, Rothstein had written a secret report for the Pentagon, concluding that the US military in Afghanistan had failed to translate tactical success into strategic victory; yes, the Taliban had been ousted from power, but al Qaeda still thrived, and, meanwhile, militant warlords were taking over many villages. The problem, as Rothstein saw it, was that the Americans were putting too much emphasis on bombing, when they should be putting boots on the ground and engaging in “unconventional warfare”—another term for “counterinsurgency,” when that word was still banned from use. (The Pentagon, still run by Donald Rumsfeld, returned the report to Rothstein and asked that he tone it down; he refused.)

Rothstein had made a few trips to Afghanistan since then and had come away shaking his head on each occasion. This time, he was impressed with what Petraeus was doing; finally, a strategy, a coherent plan, was in place.

But Rothstein was concerned about a larger question: Where was this going? Petraeus was achieving some impressive effects (Taliban killed, terrain cleared, local police recruited), but the measure of real, sustained success was outcomes. When the West eventually pulled the plug on this place, outcomes would determine whether it survived on its own.

He came away with the impression that Petraeus was hoping to rack up enough effects, for a long enough time, that at some point they’d spill over into outcomes. Maybe so, but Rothstein saw no sign of a tipping point just yet.

•  •  •

Even Petraeus knew that Afghanistan was a long shot. Then again, Iraq had been a long shot, too. He understood, intellectually, that the two wars and the two countries were very different; in his PowerPoint briefings, he’d often include a slide that read, “Afghanistan Is Not Iraq.” But Iraq was what he knew best; he’d spent three tours there, all in command slots, spanning a total of nearly four years. Afghanistan, by his own admission, he barely knew at all; so it was natural for him to view its problems through an Iraqi prism. But some of his aides and officers found the tendency disturbing. His instinctive reaction to every new challenge was to seek a parallel from his years in Iraq. We solved that this way in Mosul . . . We did this when that happened in Anbar . . . I said this when Maliki threatened to do that . . .

Once, he drew a comparison between Kabul and Baghdad in a conversation with Karzai. Matt Sherman, one of Petraeus’s advisers who had spent considerable time in Iraq and Afghanistan, told him bluntly as they walked away from the meeting, “Don’t talk about Iraq so much,” adding, “It might be a great mental exercise for you to try not thinking about Iraq at all.”

Petraeus nodded and said, “I’m working on it.”

Afghanistan, on its own terms, was a stumper: the levers weren’t working; the effects weren’t producing outcomes. A major obstacle was Karzai himself, and Petraeus couldn’t figure out a way to play him: he was clearly shrewder than Maliki had been; his hold on power relied on more networks that Petraeus couldn’t disrupt.

In October, Karzai threatened to kick all foreign-owned private security guards out of the country. In a rare show of unity, Petraeus and Eikenberry went to see him together and urged him to reverse his decision; otherwise, they said, every economic-development project would shut down, as the aid workers and business executives couldn’t stay without reliable protection. Karzai insisted that the Afghan police could do the job; the two Americans expressed their doubts.

Karzai threw a fit. He told them, “I have three main enemies”—the Taliban, the United States, and the international community—and “if I had to choose today, I’d choose the Taliban!”

At the last minute, Karzai backed down. But a few weeks later, in an interview with the Washington Post, he went on the rampage again, demanding that the Americans cut back on their military operations and end the Special Ops forces’ nighttime raids altogether.

When he’d first assumed command, Petraeus told himself that if he ever thought the mission simply could not be accomplished, he would tell the president, and that would probably mean he’d have to resign. If Karzai really did forbid the nighttime raids, he knew he’d have to push that button.

Petraeus didn’t want to take that step, so he played what leverage he had. He phoned Ashraf Ghani, one of Karzai’s top national-security advisers, and said, “Your president has put me in an untenable position. Please take note of that word. I chose it carefully.”

Ghani rushed to ISAF headquarters to assure Petraeus face-to-face that Karzai wasn’t serious. A few days later, the Afghan president backed off again, for a while.

It wasn’t the first or last time that Karzai made trouble or howled in protest or denounced the Americans or threatened to go join the insurgency. Eikenberry was right: Karzai was not a suitable strategic partner. But the problem wasn’t Karzai personally. He had multiple constituencies: the Americans and NATO, for their money and security; his patronage network, for extending his power across the country; certain elements of the Taliban, with whom he’d have to make a deal at some point; and ordinary Afghans, for their complicity, if not quite warm support. The Americans and NATO, he could toy with; he’d learned that they were invested in this mission at least as much as he was, and so they weren’t going away. (Petraeus’s “untenable” threat to do just that was an exception, and all the more potent as a result.) His other constituents Karzai needed to placate more consistently, and they didn’t like nighttime raids; they didn’t like foreign soldiers, period. Whatever Karzai’s mental state, he wasn’t behaving much differently from the way that any Afghan president would have behaved in his situation.

Petraeus’s frustrations, in short, stemmed not so much from Karzai as from the nature of Afghanistan itself: its primitive economy (which impeded the rise of an educated, entrepreneurial class); its vastly scattered, rural population (which a weak central government could rule only through a corrupt patronage network); and its long border with a state whose leaders were assisting the insurgency (which limited the success of any fight confined to Afghan territory).

A glance at a map, a few villages, or some key passages from Galula would have revealed that Afghanistan and COIN made an unlikely match.

Petraeus had glimpsed a bit of this himself five years earlier in the fall of 2005. He was about to complete his stint as commander of the mission to train the Iraqi army, when Donald Rumsfeld called and asked him to stop off in Afghanistan on his way home to assess the security situation. The Joint Chiefs of Staff were beginning to call the global war on terror the “long war,” a seamless campaign that could stretch on for decades. Noting Afghanistan’s widespread illiteracy, its lack of infrastructure and investment, the poor shape of its armed forces, and the signs of a rising insurgency, Petraeus told Rumsfeld, “Of all the long wars, this one is going to be the longest.”

Now here he was, engulfed in its tidal waves himself.

At Petraeus’s urging, and with President Obama’s authorization (as well as, more often than not, the Pakistani military’s compliance), the CIA stepped up its drone strikes against insurgent leaders basking in sanctuaries across the border. But despite an impressive hit rate (and relatively few civilian casualties), US intelligence reported that the campaign was doing little to alter the militants’ behavior; they were still able to retreat to safe havens at will. It was another case of effects without outcomes.

To combat corruption, Petraeus set up a special command and put his favorite one-star general, H. R. McMaster, in charge. McMaster linked up with the intelligence community and special FBI agents. He and his team interviewed hundreds of Afghan officials and citizens, mapped out the networks of criminal patronage from the most powerful ministries to the most remote district governors, and drew up a list of nine objectives and fifty-four tasks they had to accomplish. He compiled evidence of massive malfeasance by some of Karzai’s top officials and cronies. They included the chairman of Kabul Bank, who’d spent over $150 million of state deposits on personal villas in Dubai; the Afghan surgeon general, who’d stolen tens of millions of dollars’ worth of drugs from the main military hospital; and officers in the Afghan air force, who were moving narcotics around the country on behalf of drug lords.

His efforts resulted in the firing of a few dozen police chiefs and local government officials, but not much more. No higher-ups were prosecuted; a few were taken into custody briefly, until Karzai set them free.

•  •  •

Petraeus was rotated out of Kabul on July 18, 2011, just as Obama’s troop cuts were about to get under way. He hadn’t made as much progress as he’d pledged toward accomplishing the mission. Probably no one could have.

He was also exhausted: dark circles under his eyes, his back slightly bent when he walked, his concentration a bit less steely than usual. This was, to say the least, normal for a twice-injured fifty-eight-year-old cancer survivor wrapping up his sixth consecutive command in a span of nine years, the last three of those commands and the last four of those years deployed abroad in combat. But Petraeus would heatedly deny that he was at all tired, challenging anyone who commented on his signs of fatigue to a five-mile foot race at dawn.

A few nights before his farewell ceremony, Petraeus sat in his office with a small group of aides and visitors, watching videos of his days back in Mosul as commander of the 101st Airborne Division, then more videos, archival films of the 101st of yore, the Screaming Eagles from World War II. It was an evening of unwinding, an immersion in nostalgia and glory.

They watched no videos from his year in Afghanistan.