by Steve Coll
Bhutto’s aides released to the press the various warnings she had written down, to be publicized in the event of her death. On January 3, Musharraf held a press conference in Islamabad to defend himself against Bhutto’s accusatory ghost. His government announced it had obtained phone intercepts indicating that Baitullah Mehsud, emir of the newly declared Pakistani Taliban movement, had organized Bhutto’s assassination. Musharraf wore a suit and tie and sat at a table by himself. “I have been brought up in a very educated and civilized family, which believes in values, which believes in principles, which believes in character,” he said. “My family is not a family that believes in killing people.”
Musharraf said Bhutto had ignored warnings that campaign rallies had become too dangerous, and he blamed her for standing recklessly in the open air to greet her supporters. As for Bhutto’s suspicions that Pakistan’s deep state might have been involved, Musharraf dismissed such thinking as a “joke.” With apparent sincerity, he added, “No intelligence organization of Pakistan is capable of indoctrinating a man to blow himself up.”21
In February, the Peoples Party swept to power and Asif Zardari, Bhutto’s erratic widower, became Pakistan’s president, at the time perhaps the world’s most unlikely leader of a nuclear-armed nation outside of North Korea. He told Patterson that the United States was Pakistan’s “safety blanket” and assured her that he believed there was no such thing as a “moderate Taliban.” He added, “I am not Benazir, and I know it.”22
Although it was not yet obvious to everyone in the Bush administration, Musharraf was finished politically. He would cling to the office of president for a few more months before resigning. He later went into exile. The twin pillars of the Bush administration’s reformed policy in Pakistan—Musharraf and Bhutto—were gone.
SEVENTEEN
Hard Data
Toward the end of 2007, a team of career C.I.A. analysts at the Directorate of Intelligence, who specialized in assessing wars, carried a set of highly classified, color-coded maps of Afghanistan to the White House, to unfurl them for decision makers at the National Security Council. The map project was called the District Assessments. It attempted to describe political control and security in each of Afghanistan’s 398 official administrative districts. The C.I.A. unit collected and aggregated about three dozen independent indicators for each district. Some of the data could be objectively counted—the number of roadside bombings or assassinations during the last six months, for example. Other judgments, such as the extent of Taliban influence on a rural population, might require more qualitative measurement, but these assessments could still be informed by data such as whether the Taliban operated checkpoints on the roads or were reliably reported to run informal courts in local mosques. After all the intelligence was organized, the C.I.A. team gave each Afghan administrative district one of four ratings. The spectrum ranged from districts controlled by the Karzai government to districts controlled or contested by the Taliban. There was also a category (and a distinct color on the map) for districts judged to be under “local control,” meaning that they were run by independent warlords or drug barons not aligned with either side—or were controlled by nobody at all, essentially ungoverned.
In the closed world of secret intelligence, most analytical products wound up in locked cabinets, having had little impact. But every now and then a bestseller broke through. The C.I.A.’s District Assessment maps of Afghanistan proved to be such a blockbuster, one of the most popular top secret analytical products the agency had ever distributed. Tony Schinella, a courteous academic with a black mustache, would soon rotate in to supervise the project; he would work on it for almost a decade, becoming one of the most influential analysts of the Afghan war in the government. Schinella had analyzed the Balkan conflicts of the 1990s, and then, after September 11, the wars in Afghanistan and Iraq. He had earned a doctoral degree at the Center for International Studies at the Massachusetts Institute of Technology. Political scientists at M.I.T. had advanced the use of data and mathematical modeling in search of insights into war and peace. The Afghan maps Schinella and his colleagues created followed this methodology.1
The impetus for the maps’ initial creation was a new focus on the Afghan war within the Bush administration, following the late 2006 strategic review carried out at the White House. In mid-2007, Bush and his national security adviser, Stephen Hadley, had recruited Major General Douglas Lute to the White House to coordinate support for the dual American war efforts in Iraq and Afghanistan. In time, Lute would become Washington’s most important policy adviser on Afghanistan, but in his early days at the White House, he spent at least 90 percent of his time managing the fiasco in Iraq. Lute was among those at the White House who were enthusiastic about the District Assessments project. Rare is the general who does not love a map, and with the distractions of Iraq, here was a concise way to track progress or the lack of it in Afghanistan. Intelligence consumers in diverse agencies seemed to appreciate the C.I.A. maps’ mesmerizing colors, unfurled in cascades of red and green.
In the emerging era of Big Data, the C.I.A. maps were an exemplar of the power of data visualization. They also provided a way to capture changes in the war’s progress empirically. American military commanders in Afghanistan often rotated annually and ambassadors moved on every two or three years. In 2007, Condoleezza Rice hired Eliot Cohen to become her counselor at State. He was a Harvard-educated political scientist who favored bow ties. Rice asked him to spend as much time in the field as possible to offer independent assessments. He found in Afghanistan that there was a pattern of briefings at military headquarters, whether at Kabul or in the regions. The commanders starting a rotation would say, “This is going to be difficult.” Six months later, they’d say, “We might be turning a corner.” At the end of their rotation, they would say, “We have achieved irreversible momentum.” Then the next command group coming in would pronounce, “This is going to be difficult. . . .”
At the U.S. embassy in Kabul, State Department reporting seemed to some of Rice’s advisers to have become detached from reality. Bill Wood, the ambassador, was more optimistic than her senior analysts. David Gordon, a political scientist who served as Rice’s director of policy planning, considered Wood “a solid citizen” but noted that he “was of the view that things were becoming stabilized” when the evidence was hard to find. Eliot Cohen commissioned memos objecting to an oft-quoted claim, usually emanating from I.S.A.F. commanders, that more than 75 percent of the war’s violence occurred in just 10 percent of Afghan districts, mainly in Helmand and Kandahar. He didn’t think the United States knew enough about what was happening in all of Afghanistan’s 398 districts to be so precise.
The C.I.A. District Assessments purported to let numbers tell the war’s story, but it did not require access to secret information to see the trends. Early on after his arrival at State, before the District Assessments became popular, Cohen used unclassified United Nations maps dating back to 2002 to try to shock Rice and other principals into recognizing how badly Afghanistan was deteriorating. The U.N. maps depicted Afghan districts as green if they were safe for aid workers. (On the C.I.A. maps, green depicted government control but did not necessarily indicate that the area was safe for international travelers.) Cohen unfurled annual maps that showed Afghanistan’s green zone was shrinking fast.2
The indicators reflected on the C.I.A. maps came from all sources, including military field reports, C.I.A. agent reporting, diplomatic reporting, allied governments, the United Nations, and published journalism. The C.I.A. took final responsibility for each district’s rating, but the team’s analysts solicited data and dissents from other intelligence agencies. There were reasons for competing analysts to be skeptical. Paul Miller, the former C.I.A. analyst who moved to the National Security Council to work on Afghanistan just as the maps were being produced, noted that the color-coded scheme didn’t have a category for “unknown.” He thought the agency didn’t
own up to its intelligence gaps; he would have guessed that a large portion of the map should have been painted as “unknown.” The officers who produced the maps heard this criticism regularly; they argued that it was misplaced because as troop and civilian deployments to Afghanistan rose after 2007, the United States had more reach in its field reporting than at any time since the Taliban’s fall.3
The District Assessments had an antecedent in the Vietnam War, a project known as the Hamlet Evaluation System, undertaken by the Pentagon, starting in 1967. Civil Operations and Revolutionary Development Support, or C.O.R.D.S., oversaw that work. American provincial and district advisers working on the pacification campaign for C.O.R.D.S. filled out monthly questionnaires about all the hamlets in their areas, grading security indicators on a scale of “A” (most aligned with the South Vietnamese government) to “E” (heavily influenced by the Viet Cong) to “V” (totally under Viet Cong control). Later they also tried to measure development and governance indicators such as health care and education. The problem was that it was unrealistic to expect the Americans—mainly military officers—who were scattered around violent sections of South Vietnam to accurately collect such metrics. One issue was that “you know what your boss wants to hear . . . that we were winning the war,” as Ron Milam, an adviser who worked on the project, put it. To make the scores look better, hamlets under Viet Cong control were sometimes removed from the calculations or blended with hamlets with better scores to make the picture look more encouraging.4
As part of the Pentagon’s effort to convince the American public that the Vietnam War was going well, U.S. briefers created sanitized, unclassified versions of maps derived from the Hamlet Evaluation System. They used them in Saigon to impress visiting members of Congress, European ambassadors, newspaper columnists, and other opinion makers. The maps became a symbol of “the inclination of successive administrations in Washington to manipulate information about the war for political advantage,” as the historian David Elliott put it.5
The C.I.A. analysts who worked on the District Assessments forty years later were post-Vietnam skeptics well aware of that history of delusion and public dishonesty. If anything, in the judgment of Pentagon commanders and some White House aides, the C.I.A.’s analysts were inclined toward overwrought pessimism about the war in Afghanistan and the deterioration of Pakistan after 2007. In any event, the C.I.A. maps showed starkly that the Taliban were marching toward control or the ability to contest control of about half of the country, district by district. They did not provide a basis for official optimism.6
That did not prevent generals in Kabul from being optimistic. During a 2007 secure videoconference with General Dan McNeill, then top American commander in Afghanistan, Robert Gates, Rumsfeld’s recently arrived successor as secretary of defense, asked, “Dan, I’m trying to get a sense if we are making progress. Are we making gains in quelling the insurgency? If we are winning, by what measure?”
McNeill defended the war effort. Since arriving in Afghanistan, he had repeatedly heard dark forecasts about the war’s slide from C.I.A. analysts and others at State and D.I.A. He thought the pessimists were overstating the evidence. The general’s take was that the war effort in the south had almost collapsed late in 2006, but that fight-back by the Canadians and more aggressive tactics by N.A.T.O. had stabilized the situation. Fundamentally, McNeill argued, the war hadn’t changed. They had to be patient. Privately, McNeill figured it would take up to two decades to put Afghan forces in a position where they could defend the country adequately on their own.
When the general signed off, Gates turned to his aides, incredulous. As a former director of analysis of the C.I.A., he knew “a thing or two about taking the Agency with a grain of salt,” he told them. Yet the C.I.A. analysts’ “assessments that the situation is dramatically declining doesn’t comport with what I’m hearing” from McNeill or the U.S. embassy in Kabul. “My sense is that we are not getting this right and that the situation is going sideways on us.”7 The C.I.A. maps revived a dilemma familiar from Vietnam: Could verifiable facts about a deteriorating war lead policy makers to understand the futility of the country’s policies?
—
Throughout 2007, the U.S. government’s classified information systems were replete with candid reports about rising violence in Afghanistan and failing governance by the Karzai administration. The problem was not a paucity of information; it was Washington’s ability to recognize the pattern. Provincial Reconstruction Teams scattered around Afghanistan regularly sent to Washington lengthy written assessments, drawing on their access to local conditions. The P.R.T. report from Kandahar, the Taliban’s birthplace, was particularly bleak. The Canadian military offensive late in 2006 had raised the local population’s expectations, but N.A.T.O.’s “failure to consolidate victory” had proven to be “costly.” The inability of the United States to crack down on the Taliban’s sanctuaries in Pakistan had given rise to conspiracy thinking among Kandaharis. American passivity toward Pakistan was interpreted as “proof of a plot to prolong instability in Afghanistan, or in the most extreme version, proof of a secret alliance between the West and the Taliban.”8
The major political challenge “is the lack of good governance,” the report warned. Exploitative since 2002, the Karzai regime was sinking into systematic racketeering across the south. “Real power is wielded by a select few and standard practices would be considered corrupt in the West. Ahmed Wali Karzai, brother of President Karzai, exercises far more power than his position as head of the (advisory) Provincial Council should convey.” Fifty thousand students were forced out of schools by violence during 2006 and virtually all schools in the province had by now closed. The rise of suicide bombing attacks against rolling convoys had caused American and N.A.T.O. forces to ease their rules of engagement so that they now fired rounds into vehicles suspected of an attack. Inevitably, the snap judgments of exposed soldiers on the front lines led to mistakes and innocent deaths. “These deaths, while not easy to avoid, do much to undermine our credibility with the population.”9
“Corruption by police and government officials is having a corrosive impact on government credibility while boosting that of the Taliban,” noted a separate report about Kandahar from the U.S. embassy in Kabul that summer. Despite the killing of hundreds of Taliban volunteers, “insurgents have increased their presence in many districts.” The Taliban struck soft targets such as the Afghan police, a force that was “very weak, poorly controlled, and virtually unsupported” by N.A.T.O. or the United States. Afghan police earned just $70 per month in Kandahar for work that was “high risk and low reward.” Taliban night letters and assassinations intimidated villagers, farmers, and mullahs preaching publicly about the war; they increasingly adopted a stance of neutrality to avoid death, and, if they happened to work for Canada or the United Nations, they hid such information from their neighbors.10
There were also “strong indications that some district leaders and Chiefs of Police are engaged in supporting the trafficking of narcotics,” the U.S. embassy in Kabul reported. “These same officials have extracted bribes from farmers to keep them off the list of fields to be targeted under the governor-led eradication program.” There were also “strong indications” that Abdul Raziq, the local ruler and prized American ally in Spin Boldak, on the road to Quetta, “controls large-scale narcotics trafficking.”11
Graeme Smith, the Canadian journalist who lived in Kandahar during this period, worked with Afghan colleagues to conduct an informal survey of Taliban fighters in the region. Thirty-seven of the forty-two Taliban they reached belonged to tribes that had lost out during the reigns of the Karzais and Gul Agha Sherzai. These Taliban also had only the thinnest understanding of their international enemies. They did tend to grasp that the United States was a world power, “a direct equivalent of past empires that sent crusaders to the Middle East.” Canada puzzled them, however. Some did not even understand that it was a countr
y: “It might be an old and destroyed city,” one respondent said. This was not an aspect of the war’s asymmetries that could be coded easily onto a data map.12
Regular travel to Afghanistan and Pakistan, sometimes in bow tie and flak jacket, left Eliot Cohen with the impression that Rice and Bush were getting “an awful lot of happy talk from people who should know better.” The Afghan war might be better than Iraq, but there were many signs that things were not well.
On one of his field surveys, Cohen flew with a colonel over a contested Afghan valley. “Tell me the mechanics of this war,” Cohen asked.
“Clear the valley,” the colonel replied.
“What do you mean?”
“You walk through a valley until you get into a firefight and then you keep shooting until it stops.”
“That’s a little troubling,” Cohen said.
“It’s a valley-by-valley war, sir.”13
—
Between 2002 and 2006, the C.I.A.’s operations in Afghanistan had concentrated largely on the secret Afghan militias built up by the Omega teams along the Pakistan border to hunt for Al Qaeda members, cooperation with Karzai and a few key governors such as Sherzai and Asadullah Khalid, and support for the N.D.S. The demands of the Iraq war, instability on the Seventh Floor at Langley, and short tours by Kabul station chiefs of erratic quality limited the agency’s effectiveness. In 2007, as the Bush administration authorized much larger aid budgets for Afghanistan, the C.I.A.’s engagement expanded. The agency’s financial support for N.D.S. increased. The agency operated about a dozen bases outside Kabul Station and commanded a private air force of planes and helicopters. Its two main priorities remained locating Al Qaeda cells operating across the border in Pakistan’s tribal areas and the hunt for senior Al Qaeda leaders, particularly Bin Laden and Zawahiri. It was baffling and not an advertisement for C.I.A. capabilities and focus that the pair remained at large. Yet the agency was now drawn increasingly into the conflict against the Taliban as well.14
