by Steve Coll
In Washington, Kilcullen’s latest report reinforced the case that the Afghan war was spiraling in the wrong direction. At the White House, about once a quarter, National Security Adviser Steve Hadley invited the N.S.C.’s Afghanistan staff to his office for an informal roundtable discussion. Doug Lute, the two-star general and “war czar” coordinating both Iraq and Afghan policy, would attend these sessions, along with John Wood, the senior N.S.C. director for Afghanistan, and his colleagues, including Paul Miller, Christa Skerry, and Marie Richards. Pakistan was a perennial topic. So was Afghan governance. Hadley would throw out a question, such as “How do you build a police force in a country as broken as this?” The conversation would range widely. Hadley listened but rarely revealed his own opinion.9
Since Doug Lute’s arrival at the White House in mid-2007, Iraq had consumed almost all of his attention. President Bush was committed to the primacy of Iraq; he was more personally invested. In the interagency committees, while trying to turn the Iraq war around, Lute had developed a circle of confidants, civilians with an analytical bent, including David Gordon and Eliot Cohen, and, at the Pentagon, Jim Shinn, who had made a fortune in the software industry before joining government, and Mary Beth Long, the international lawyer who worked on counternarcotics. They had bonded over Iraq’s unrelenting crises. By mid-2008, however, the surge of troops into Baghdad that Bush had authorized eighteen months earlier under Petraeus’s command had reduced the violence there. There was now greater space to consider Afghanistan.10
Lute was a large, affable, humorously skeptical general who had risen high because of his facility for operations and process. He had grown up in Michigan City, Indiana, an Eagle Scout and president of his high school class. He won acceptance to West Point one year behind the class Petraeus dominated (marrying the superintendent’s daughter, among other achievements). Lute chose the armored cavalry upon graduation in 1975. He spent much of his early career with a cavalry regiment in Germany tasked to hold off the Soviets at the outbreak of World War Three. He took a master’s degree at Harvard, taught international politics at West Point, fought in a cavalry unit during the first Gulf War, worked on the Balkans during the 1990s, and landed as a military assistant to Chairman of the Joint Chiefs Hugh Shelton in 2000.
General John Abizaid, then leading Central Command, recruited Lute as his J-3, or chief operations officer, just as the Iraq war turned into a bloody grind. Of Lebanese descent, Abizaid was the first Arab American to wear four stars. Lute traveled continually with him to both Iraq and Afghanistan. If Petraeus increasingly led the school of generalship in the Army that advocated for counterinsurgency, a doctrine that typically required long time lines and state building, Abizaid was the most credible voice of an opposing school of thought. He believed counterinsurgency carried out by large numbers of highly visible American forces in Muslim societies, over an extended period of time, was unlikely to succeed because locals would revolt before there was time for the United States to achieve the expansive political and economic objectives that counterinsurgency theory promoted. Repeatedly, Abizaid warned colleagues about Iraq and Afghanistan: The clock is ticking on the American presence. The Petraeus network within the Army sometimes dismissed such thinking as defeatism. Abizaid considered his warning to be realism. He had burned his convictions into Doug Lute’s head by the time the Bush White House recruited Lute as war czar—or, as he was sometimes referred to less provocatively, a senior coordinator.11
In May 2008, Lute arranged for a Gulfstream V jet to carry him, Cohen, Gordon, Shinn, and Long to Afghanistan. They heard much to be discouraged about. Road building was behind schedule and the roads that had been completed weren’t being defended successfully. Battlefield intelligence collection seemed disconnected from strategy—the system was a massive vacuum cleaner of stray facts, feeding into conclusions that were “not rigorous,” as Long put it. The chain of command was a fractured mess. As an operations man, Lute was especially appalled by the incoherent command structure that had grown up in Afghanistan as a result of ad hoc compromises with N.A.T.O. and within the American military.
In Kandahar, the group spent hours poring over battle maps of Kandahar and Helmand as local N.A.T.O. commanders inventoried all the units deployed in their areas. These included American “black” or covert Special Operations units, “white” or Green Beret–style American Special Forces, British forces, Dutch forces, U.S. Marines, and multinational Provincial Reconstruction Teams. There were also small international training units embedded with the Afghan army; these officers reported through their own chain of command to Kabul. By the time they were done, they had identified more than half a dozen distinct N.A.T.O. or American armed forces operating in southern Afghanistan—only one of which the top N.A.T.O. general on the scene in Kandahar commanded. This patchwork was the legacy of Rumsfeld’s reluctance to deploy in Afghanistan, N.A.T.O.’s fractured system of military federalism, and the insistence of U.S. Special Operations Command and the U.S. Marines that they operate independently, which Rumsfeld and then Gates had accepted.12
American officers told them about a soldier wounded in the foot near Herat, in western Afghanistan. If he had been struck above the waist, he likely would have bled out and died because of the lengthy time required for Spanish commanders, who were deployed in the region, to request permission from the N.A.T.O. chain of command to dispatch a rescue helicopter. That episode became a touchstone for Gates. The “bleed out” time in Afghanistan—the time a wounded soldier had to wait for specialized help to arrive—was about two hours, on average. Gates pushed to reduce it to one hour.
The study team spent long hours on the flight home, over wine and baguettes, discussing what they had seen. They agreed that Afghanistan and Pakistan had to be understood as a unified theater and set of problems. Back in Washington, Lute briefed the interagency deputies committee about the command problem. Besides the half a dozen overt military commands, “No one in this room has good visibility on the C.I.A. with partnered Afghan intelligence units,” Lute said. That summer, Lute sent a memo to Bush warning that the Afghan war could fall apart during the next six months, as the president departed office. Bush wrote back almost immediately. He wanted to work on the problems even though the time was short. Hadley told Lute, “I want to launch a very deliberate look at this.”
The National Security Council met on September 17. Bush told his war cabinet he wanted a review similar to the one that had produced the recommendation for a troop surge into Baghdad, a decision in which Bush took pride. That launched the last and most ambitious of his administration’s strategy reviews on Afghanistan.
During the next two months, Lute convened sixteen sessions lasting a total of more than forty hours in Room 445 of the Eisenhower Executive Office Building, a secure conference room. Deputies of roughly Lute’s rank in the national security bureaucracy staffed the study from the Pentagon, the State Department, C.I.A., the Office of the Director of National Intelligence, the Office of the Vice President, and other agencies. Typically there were about two dozen people in the room. The review went topic by topic—poppy and heroin, corruption, Afghan security force training, N.A.T.O.’s command structure, Pakistani sanctuaries. Typically, an intelligence analyst such as Peter Lavoy of the National Intelligence Council or Tony Schinella of the C.I.A.’s military analysis unit would open a session with an in-depth presentation. Schinella’s team updated the C.I.A.’s top secret District Assessment maps for the review; in the years ahead, the agency would revise the maps every six months, in part because they were so popular. After their briefings, the intelligence analysts would usually leave and the political appointees and generals from the Pentagon would debate what should be done.13
The N.S.C. staff under Lute also produced memos concerning basic questions about the war, ones that might seem worrying to be asking after seven years of involvement: Why was the United States still in Afghanistan? What did it want to achieve? Was it necessar
y to crack down on corruption to achieve America’s strategic goals? Was it necessary to invest in reconstruction and governance to succeed at counterterrorism?
The apparent success of the Baghdad surge, skillfully promoted in the media by Petraeus, had created a kind of intellectual bubble in Washington around his ideas. Diplomats, generals, and aid workers toted thumbed copies of Learning to Eat Soup with a Knife, a history of counterinsurgency by an Oxford-educated protégé of Petraeus’s, Lieutenant Colonel John Nagl. Philip Zelikow, a historian and counselor to the State Department, while working on the Iraq war, had coined a phrase that now became a kind of mantra in Room 445: Clear, hold, build. Zelikow meant that a counterinsurgency campaign should seek to clear enemy-held areas by military force, hold them, and build popular consent through reconstruction and better governance. Zelikow was no longer in government. He did not actually believe that the formula he had developed for Iraq policy would work in Afghanistan, because of the problems of Pakistan sanctuaries, weak governance, and poverty. Yet the consensus in Room 445 was that the United States needed to resource a more full-bodied counterinsurgency strategy in Afghanistan, which would involve more reconstruction aid and training of Afghan security forces. They believed the central government in Afghanistan still had a chance to establish itself. The Taliban were on the upswing but did not threaten a takeover of Kabul or the overthrow of Karzai’s regime. The idea was to carry out counterinsurgency primarily through Afghan security forces, but to give the forty thousand or so Western soldiers then in the country a new focus.14
The support for a larger counterinsurgency mission in Afghanistan—more money, more troops, and more diplomats—introduced a logic problem that would bedevil American strategy in the war for the next several years. The purpose of the American intervention in Afghanistan was to defeat Al Qaeda, which remained dangerous, everyone at the table agreed. However, Al Qaeda was primarily in Pakistan. So why carry out counterinsurgency against the Taliban in Afghanistan? The answer the review arrived at was that only by creating conditions in Afghanistan that precluded a full Taliban comeback and takeover could Al Qaeda be prevented from restoring itself in a sanctuary similar to what it had enjoyed in Afghanistan on the eve of September 11.
“Al Qaeda is more disrupted than at any time since October 2001, but the organization is damaged, not broken,” as Peter Lavoy summarized some of the thinking at a N.A.T.O. briefing that autumn. “The international community cannot afford to let pressure off al Qaeda because it has demonstrated an ability to reconstitute itself in the past, and could easily reverse-migrate back to Afghanistan if the Taliban were to regain control. . . . The consequences of failing in Afghanistan and permitting al Qaeda to shift its center of gravity to Afghanistan would pose a threat to all nations inside their borders.”15
The argument made sense, but as a rationale for putting tens of thousands more American and European soldiers in harm’s way it had an indirect, speculative quality. Al Qaeda had established itself in Pakistan and had no need of Afghanistan. And seeking to defeat the Taliban in a counterinsurgency campaign was a much different kind of war from the C.I.A. and Special Operations covert campaign against Al Qaeda or the supposed peace enforcement operations sold to N.A.T.O. governments as their militaries rotated to the south. The C.I.A.’s operating directives still tasked operators and analysts to focus on Al Qaeda; they had no orders from the White House to help defeat the Taliban, according to what agency leaders told their British counterparts at the time, to the latter’s dismay, given the rate that British soldiers were taking casualties in Helmand inflicted by the Taliban.16
On October 15, Lute, Cohen, and Shinn flew back on their Gulfstream jet to Afghanistan and Pakistan to conduct a week of field interviews before writing up the strategy review. They traveled to the border between Pakistan and Afghanistan at Torkham. Lute’s Marine aide arranged for a kosher TV dinner to be brought along for Cohen, who was fairly certain he consumed the first kosher meal ever transported there. The big questions were, first, were N.A.T.O. and the United States winning or losing? A second was trying to figure out what the enemy’s game plan was. The intelligence remained poor; Cohen concluded that American analysts were much better at counting things than extracting strategic insights. A third big question was what to do about Pakistan, at a time when “everyone was getting a darker and darker view of Pakistan,” as Cohen put it.
At the embassy in Kabul, they realized there had been a major turnover in personnel even in the months since their spring visit. The delegation from Washington knew more than the people briefing them. After one desultory session, Cohen and Lute headed to the men’s room. Standing at the urinals, they remarked simultaneously, “We’re fucked.”17
Lute facilitated the final discussions in Room 445 and also weighed in with his own analysis of the Afghan war’s fractured command structure. He created a PowerPoint deck with a slide titled “Ten Wars,” referring to what his final count had determined were ten distinctly commanded military forces in Afghanistan. How could they bring greater coherence to the effort? What if they adopted a more complete counterinsurgency approach inspired by their experience in Iraq?
The core intelligence resource for the review was a National Intelligence Estimate on Afghanistan and Pakistan that Peter Lavoy had overseen across the summer and autumn. It acknowledged that the security situation in Afghanistan looked “bleak,” as a classified summary for N.A.T.O. governments put it. “The Afghan government has failed to consistently deliver services in rural areas. This has created a void that the Taliban and other insurgent groups have begun to fill in the southern, eastern and some western provinces. The Taliban is mediating local disputes in some areas, for example, offering the population at least an elementary level of access to justice. . . . The Taliban have effectively manipulated the grievances of disgruntled, disenfranchised tribes.”
The N.I.E. endorsed Kilcullen’s assessment of rising professionalism in Taliban guerrilla operations, presumably aided by the Pakistan Army. The guerrillas were “demonstrating more sophisticated infantry, communications and command and control techniques. Their marksmanship is more precise, and their explosives more lethal than in previous years.”
The N.I.E. predicted that even if the Afghan National Army and the Afghan National Police could be funded and trained into a professional force of several hundred thousand men, that project “would be insufficient if Pakistan remains a safe haven for insurgents.” Equally, solving the problem of the Pakistani sanctuary abetted by I.S.I. was “necessary but insufficient” to win in Afghanistan—that would require “addressing the severe governance, development and access to justice gaps.”
The estimate rendered a split verdict on I.S.I.’s culpability and Kayani’s role. “Pakistan now identifies both al Qaeda and the Taliban as existential threats,” the estimate concluded, but the Taliban that bothered Kayani was the Pakistani branch, not the Afghan one. To influence Afghanistan, I.S.I.’s Directorate S still “provides intelligence and financial support to insurgent groups” attacking N.A.T.O. and Indian targets inside Afghanistan.18
Briefing the findings to N.A.T.O., Peter Lavoy said that I.S.I.’s aid to the Taliban was explained by three factors. First, “Pakistan believes the Taliban will prevail in the long term, at least in the Pashtun belt most proximate to the Pakistani border.” Second, Pakistan “continues to define India as the number one threat” and “insists” that India seeks to undermine Pakistani security through covert operations inside Afghanistan. Third, Pakistan’s high command believes “that if militant groups were not attacking in Afghanistan, they would seek out Pakistani targets.”19
The world outside Room 445 changed dramatically during the two months that the review went on. The study began as Lehman Brothers failed, touching off the worst financial crisis and recession since the Great Depression. The collapsing economy lifted Obama’s presidential prospects, and by October it seemed likely that the deputies would han
d off their findings to a Democratic administration. Before the election, members of Lute’s staff met advisers to both the Obama and McCain campaigns informally at the Army-Navy Club. They discussed with them the Afghan war’s deterioration and the policy questions it raised. Obama had made clear during his campaign that he favored more effort in Afghanistan, but his specific views about troop levels and military doctrine were unclear.
Lute completed the final top secret report, about forty pages in length, about ten days after Obama was elected president. “The United States is not losing in Afghanistan, but it is not winning either, and that is not good enough,” read the first sentence. The report made ten recommendations about fighting the Taliban, including the dispatch of more troops, a more ambitious effort to train and equip Afghan forces, tackling corruption and finding an approach to counter drug operations that would starve the Taliban of finance without alienating farmers—a form of drug policy for which N.A.T.O. had been searching without success for at least four years. At the Pentagon, Gates authorized preparations to send two more American brigades to Afghanistan, if the next president chose to do so. The review also urged that Afghanistan and Pakistan be organized as a single theater. A theme of the review, at least for some of the analysts involved, was the “need to re-Americanize the effort,” as Eliot Cohen put it. It had become a N.A.T.O. war, but it was not going to succeed that way.20
—
Bush also prepared to hand off to Obama the expanding C.I.A. drone campaign in the tribal areas that he had approved in July. In the four months between that decision and the completion of Lute’s strategy review, the C.I.A. had carried out twenty-three drone strikes on Pakistani soil, more than in the previous four years combined. Mike D’Andrea now oversaw what amounted to an undeclared air war in the Federally Administered Tribal Areas waged from the Global Response Center on the Langley campus. C.I.A. mission leaders tracked drone feeds on the Response Center’s big screens while Air Force pilots and sensor operators flew the missions more than two thousand miles away at Creech Air Force Base in Indian Springs, Nevada. The first strike under the new rules killed seven foreigners in Azam Warsak village in South Waziristan, according to a ledger of attacks kept by the F.A.T.A. government. In North Waziristan, the C.I.A. fired missiles at a dozen Haqqani-linked targets, some of them long ago mapped and surveyed in Miranshah and contiguous Dandey Darpa Khel, where the Haqqanis ran their large madrassa. The first two dozen strikes killed 40 foreigners but also 178 “locals,” according to the Pakistani government count. On September 8, 2008, missiles demolished a North Waziristan home belonging to Jalaluddin Haqqani. He was absent, but Al Qaeda’s latest chief in Pakistan, Abu Haris, was reportedly killed. Eight women and five children also died, according to the F.A.T.A. records. They record ten other cases of civilian casualties between September and December of 2008. In all of 2008, at least one child was killed in a third of all C.I.A. drone strikes, according to the available figures—a shocking percentage, if accurate.21