by Steve Coll
That autumn in Washington, meeting at the State Department, Tomsen led a new interagency Afghan working group through a secret review of U.S. policy. Thomas Twetten, then chief of the Near East Division in the CIA’s Directorate of Operations, attended for Langley. Richard Haas, from the National Security Council, participated in the sessions, as did delegates from the Pentagon and several sections of the State Department.3
An all-source intelligence analysis, classified Secret, had been produced as a backdrop to the policy debate. The document assessed all the internal government reporting about U.S. policy toward Afghanistan from the summer of 1988 to the summer of 1989. It laid out the splits among American analysts about whether Pakistani intelligence—with its close ties to the Muslim Brotherhood–linked Islamists—supported or conflicted with U.S. interests.4
Influenced by the McWilliams critique, members of the Afghan working group looked for a new policy direction. They were not prepared to give up completely on the CIA-led military track. The great majority of Afghans still sought Najibullah’s overthrow, by force if necessary, and U.S. policy still supported Afghan “self-determination.” Military force would also keep pressure on Gorbachev’s reforming government in Moscow, challenging Soviet hardliners in the military and the KGB who remained a threat to both Gorbachev and the United States, in the working group’s view. But after days of debate the members agreed that the time had come to introduce diplomatic negotiations into the mix. Ultimately, Tomsen finalized a secret new two-track policy, the first major change in the American approach to the Afghan war since the withdrawal of Soviet troops. The new policy still sought Najibullah’s ouster, but it also promoted a moderate, broad-based successor government.
On the first track of the new approach, the State Department would open political negotiations aimed at “sidelining the extremists,” meaning not only Najibullah but anti-American Islamists such as Hekmatyar and Sayyaf as well. American diplomats would begin talks at the United Nations with the Soviet Union, with Benazir Bhutto’s government, and with exiled King Zahir Shah about the possibility of a political settlement for Afghanistan. The State Department could now honestly argue that U.S. policy was no longer the captive of Hekmatyar or Pakistani intelligence.
At the same time the CIA would continue to press the covert war to increase rebel military pressure against Najibullah. The use of force might coerce Najibullah to leave office as part of a political settlement, or it might topple him directly. The CIA would continue its collaboration with Pakistani intelligence and would also bypass ISI channels by providing cash and weapons directly to Afghan commanders fighting in the field. Tomsen hoped to overtake the moribund, discredited Afghan interim government with a new commanders’ shura to be organized with American help, made up of rebel military leaders such as Massoud, Abdul Haq, and Ismail Khan. By strengthening these field commanders, the Afghan working group believed, the United States could circumvent the Islamist theologians in Peshawar and their allies in ISI. The new policy pointed the United States away from the Islamist agendas of Pakistani and Saudi intelligence—at least on paper.5
TOMSEN FLEW TO Islamabad early in 1990 to announce the new approach to Pakistan’s government. Oakley arranged a meeting at the Pakistan foreign ministry. Milton Bearden had rotated back to Langley the previous summer; his successor as Islamabad station chief, known to his colleagues as Harry, attended for the CIA. Harry, a case officer from the old school, had a pleasant but unexpressive face, and he was very difficult to read. He was seen by his State Department colleagues as closed off, unusually secretive, and protective of CIA turf. Pakistani intelligence also sent a brigadier and a colonel to take notes. Tomsen had invited ISI in the hope that they would accept and implement his initiative. He described the secret new American policy in a formal presentation that lasted more than an hour. The Pakistanis expressed enthusiasm—especially the diplomats from Pakistan’s foreign ministry, led by Yaqub Khan, who had long advocated a roundtable political settlement involving King Zahir Shah. Even the Pakistani intelligence officers said they were in favor.
Tomsen planned to fly on to Riyadh to make the same presentation in private to Prince Turki at the headquarters of Saudi intelligence, and from there he would go to Rome to open discussions with the aging exiled Afghan king. But it took only a few hours to learn that the chorus of support expressed at the foreign ministry had been misleading. After the presentation, Tomsen and Oakley were talking in the ambassador’s suite on the Islamabad embassy’s third floor when the CIA station chief walked in.
“Peter can’t go to Rome,” Harry announced. “It’s going to upset the offensive we have planned with ISI.” The chief explained that with another Afghan fighting season approaching, the CIA’s Islamabad station had been working that winter with Pakistani intelligence on a new military plan to bring down Najibullah. Rebel commanders around Afghanistan planned to launch simultaneous attacks on key Afghan cities and supply lines. The new offensive was poised and ready, and supplies were on the move. If word leaked out now that the United States was opening talks with King Zahir Shah, it would anger many of the Islamist mujahedin leaders in Peshawar who saw the king as a threat. The CIA chief also argued that Islamist mujahedin would not fight if they believed the king was “coming back.” Hekmatyar and other Islamist leaders would almost certainly block the carefully planned offensive. Tomsen was livid. This was exactly the point: The new political talks were supposed to isolate the Islamist leaders in Peshawar. But they discovered that Harry had already contacted the CIA’s Near East Division in Langley and that Thomas Twetten, the division chief, had already complained to Kimmitt at State, arguing that the opening to Zahir Shah should be delayed. Bureaucratically, Tomsen had been outflanked. “Why are you so pro-Zahir Shah?” Twetten asked Tomsen later.
Tomsen flew to Riyadh and met with Prince Turki to explain the new American policy—or, at least, the new State Department policy—but Rome was out for now. It was the beginning of another phase of intense struggle between State and the CIA, in many ways a continuation of the fight begun by McWilliams.6
What did it matter? At stake was the character of postwar, postcommunist Afghanistan. As Tomsen contemplated Afghanistan’s future, he sought a political model in the only peaceful, modernizing period in Afghan history: the decades between 1919 and 1973 when Zahir Shah’s weak but benign royal family governed from Kabul and a decentralized politics prevailed in the countryside, infused with Islamic faith and dominated by tribal or clan hierarchies. Tomsen believed the king’s rule had produced a slow movement toward modernization and democratic politics. It had delivered a national constitution in 1963 and parliamentary elections in 1965 and 1969. By appealing to Zahir Shah as a symbolic ruler, the State Department hoped to create space in Afghanistan for federal, traditional politics. After so many years of war, it obviously would not be possible to return to the old royalist order, but wartime commanders such as Massoud and Abdul Haq, whose families had roots in traditional political communities, might construct a relatively peaceful transition. The alternative—the international Islamism of the Muslim Brotherhood, enforced by Pakistani military power—promised only continuing war and instability, Tomsen and his allies at State believed. CIA analysts, on the other hand, tended to view Afghanistan pessimistically. They believed that peace was beyond reach anytime soon. Pakistani influence in Afghanistan looked inevitable to some CIA operatives—Islamabad was relatively strong, Kabul weak. There was no reason for the United States to oppose an expansion of Pakistani power into Afghanistan, they felt, notwithstanding the anti-American rhetoric of ISI’s jihadist clients.
Tomsen might possess an interagency policy document that committed the CIA to a new approach to the Afghan jihad, but he had yet to persuade CIA officers to embrace the policy. Some of them found Tomsen irritating; he had a habit, perhaps unconscious, of coughing up light laughter in the midst of serious conversation, including during solemn, tense interactions with key Afghan commanders or Pakistani g
enerals. Some of the Afghans seemed to recoil at this, CIA officers observed. Tomsen sought to strengthen his position inside the embassy by building a partnership with Oakley, but the ambassador was an elusive ally, embracing the envoy and his views at some points but denouncing him disrespectfully in private at other times. More broadly, the CIA operated in Pakistan largely in secret and with great autonomy. The Islamabad station was connected to Langley with a separate communication system to which diplomats did not have access. In the station and at headquarters most CIA officers regarded Tomsen’s new policy as a naïve enterprise that was unlikely to succeed. They also saw it as an unwelcome distraction from the main business of finishing the covert war. As for postwar Afghan politics, the CIA’s Twetten felt that the Afghans “were going to have to sort it out themselves… . It might get really messy.” The United States should not get involved in picking political winners in Afghanistan or in negotiating a new government for the country. There was nobody capable of putting Afghanistan back together again, Twetten believed, including Massoud.7 Still, the CIA had a mission backed by a presidential finding: to support Afghan “self-determination,” however messy, through covert action and close collaboration with Pakistani and Saudi intelligence. The CIA’s Near East Division officers said they had no special sympathy for Hekmatyar or Sayyaf, but they remained deeply committed to a military solution in Afghanistan. They were going to finish the job.
A Pakistani military team traveled secretly to Washington to lay out an “action plan” for an early 1990 offensive. The plan would include support for a new conventional rebel army built around Hekmatyar’s Lashkar-I-Isar, or Army of Sacrifice. Pursuing its own agenda, Pakistani intelligence had built up this militia force, equipping it with artillery and transport, to compete with Massoud’s irregular army in the north.8 Hekmatyar’s army was becoming the most potent military wing of the ISI-backed Muslim Brotherhood networks based in Pakistan—a force that could operate in Afghanistan but also, increasingly, in Kashmir.
The CIA station in Islamabad helped that winter to coordinate broad attacks against Afghanistan’s major cities and roads. Some of the planning involved ISI, but the CIA also reached out through its secret unilateral network to build up key Afghan commanders, including Massoud. If dispersed rebel units—even some at war with one another, such as those loyal to Hekmatyar and Massoud—could be persuaded to hit Najibullah’s supply lines and cities at the same time, they might provide the last push needed to take Kabul. The CIA and Pakistani intelligence remained focused on the fall of Kabul, not on who would take power once Najibullah was gone.
Harry, Gary Schroen, and their case officers met repeatedly during that winter of 1989–1990 with officers in ISI’s Afghan bureau to plan the new offensive. Harry met face-to-face with Hekmatyar. The CIA organized supplies so that Hekmatyar’s forces could rocket the Bagram airport, north of Kabul, as the offensive began.9
Massoud figured centrally in the CIA’s plans that winter. Schroen traveled to Peshawar in January to talk to Massoud on a secure radio maintained by his brother, Yahya. Schroen asked Massoud to cut off the Salang Highway as it entered Kabul from the north. If Massoud’s forces closed the highway while other ISI-backed rebel groups smashed into Khost and Kabul from the east, Najibullah might not be able to resist for long, the CIA’s officers believed. Massoud negotiated for a $500,000 cash payment, and Schroen delivered the money to one of Massoud’s brothers on January 31, 1990.
But Massoud’s forces never moved, as far as the CIA could tell. Furious at “that little bastard,” as he called him in frustration, Harry cut Massoud’s monthly stipend from $200,000 to $50,000. The Islamabad station sent a message to the Panjshir emphasizing the CIA’s anger and dismay.
All across Afghanistan the CIA’s offensive stalled. The mujahedin seemed uncoordinated, unmotivated, and distracted by internal warfare. They did not capture any major cities; Najibullah remained in power in Kabul, unmolested.
AS SPRING APPROACHED, the CIA station began to pick up reports from its unilateral Afghan agents that Pakistani intelligence was now secretly moving forward with its own plan to install Hekmatyar in Kabul as Afghanistan’s new ruler. The CIA’s informants reported that a wealthy fundamentalist Saudi sheikh, Osama bin Laden, was providing millions of dollars to support ISI’s new plan for Hekmatyar. The Islamabad station transmitted these reports about bin Laden to Langley.10
On March 7, 1990, in downtown Kabul the conspiracy erupted into plain view. Afghan air force officers loyal to Najibullah’s hardline communist defense minister, Shahnawaz Tanai, swooped over the presidential palace in government jets, releasing bombs onto the rooftop and into the courtyard, hoping (but failing) to kill President Najibullah at his desk. Defecting armored forces loyal to Tanai drove south from the city, trying to open a cordon for Hekmatyar’s Army of Sacrifice, which hurried toward Kabul from the Pakistani border.
With help from Pakistani intelligence Tanai and Hekmatyar had been holding secret talks about a coup attempt for months. The talks united a radical communist with a radical Islamist anticommunist. The pair shared Ghilzai Pashtun tribal heritage and a record of ruthless bloodletting. Tanai led a faction of Afghanistan’s Communist Party, known as the Khalqis, who were rivals of Najibullah’s faction.11
According to the CIA’s reporting at the time, the money needed to buy off Afghan army units and win the support of rebel commanders came at least in part from bin Laden. These reports, while fragmentary, were consistent with the agency’s portrait of bin Laden at the time as a copious funder of local Islamist causes, a donor rather than an operator, a sheikh with loose ties to Saudi officialdom who was flattered and cultivated in Peshawar by the recipients of his largesse, especially the radicals gathered around Hekmatyar and Sayyaf.12
During the same period that the Tanai coup was being planned—around December 1989—Pakistani intelligence reached out to bin Laden for money to bribe legislators to throw Benazir Bhutto out of office, according to reports that later reached Bhutto. According to Bhutto, ISI officers telephoned bin Laden in Saudi Arabia and asked him to fly to Pakistan to help organize a no-confidence vote in parliament against Bhutto’s government, the first step in a Pakistan army plan to remove her forcibly from office.13
That winter, then, bin Laden worked with Pakistani intelligence against both Najibullah and Bhutto, the perceived twin enemies of Islam they saw holding power in Kabul and Islamabad. If Bhutto fell in Islamabad and Hekmatyar seized power with Tanai’s help in Kabul, the Islamists would have pulled off a double coup.
Did bin Laden work on the Tanai coup attempt on his own or as a semiofficial liaison for Saudi intelligence? The evidence remains thin and inconclusive. Bin Laden was still in good graces with the Saudi government at the time of the Tanai coup attempt; his first explicit break with Prince Turki and the royal family lay months in the future. While the CIA’s Afghan informants named bin Laden as a funder of the Hekmatyar-Tanai coup, other accounts named Saudi intelligence as a source of funds. Were these separate funding tracks or the same? None of the reports then or later were firm or definitive.14
It was the beginning of a pattern for American intelligence analysts: Whenever bin Laden interacted with his own Saudi government, he seemed to do so inside a shroud.
Hekmatyar announced that he and Tanai had formed a new Revolutionary Council. But within hours of the first bombing attacks in downtown Kabul it became obvious to wavering Afghan commanders that the coup would fail. Government troops loyal to Najibullah routed Tanai’s defectors in Kabul. Tanai himself fled to Pakistan where he and his cabal were sheltered by Pakistani intelligence. Hekmatyar’s Army of Sacrifice never penetrated the capital’s outskirts.
It remains unclear exactly when the CIA’s Islamabad station learned of the Hekmatyar-Tanai coup attempt and whether its officers offered any comment—supportive or discouraging—to Pakistani intelligence. Many CIA operatives felt that Pakistani intelligence officers “never were honest with us on Hekmatyar,” as Tho
mas Twetten, then number two in the Directorate of Operations, recalled it.15 At a minimum, ISI’s officers knew when they planned their coup that the CIA was creating a helpful context by organizing attacks on Najibullah’s supply lines. But the CIA also kept secret from Pakistani intelligence the extent and details of its unilateral contacts with Afghan commanders such as Massoud. The agency did not inform ISI, for instance, about the $500,000 payment it made to Massoud on January 31, just five weeks before the coup attempt. The coup’s timing is also swathed in mystery. Tanai may have moved hurriedly, ahead of schedule, because of a military treason trial under way that winter in Kabul that threatened to expose his plotting.
In the aftermath Massoud stood his ground in the north. The CIA might be angry at him for failing to hit the Salang Highway that winter, but what was he supposed to make of Hekmatyar’s plot to take Kabul preemptively, a conspiracy so transparently sponsored by Pakistani intelligence, the CIA’s intimate partner in the war? Massoud had ample cause to wonder whether the CIA, in making its $500,000 payment that winter, had been trying to use his forces in the north to help install Hekmatyar in Kabul.
Massoud told Arab mediators that he still hoped to avoid an all-out war with Hekmatyar. He did not want a direct confrontation with ISI, either. Massoud made plain his ambition to assume a major role in any future government in Kabul. He expected autonomy for his councils in the north. He did not aspire to rule the Pashtun areas of Afghanistan directly, however; he knew that was impractical for any Tajik leader. How much Massoud would be willing to negotiate with Pashtun leaders remained open to question. Peter Tomsen hoped his National Commanders Shura would provide a vehicle for such compromise beyond the control of Pakistani intelligence. One thing was certain: Massoud would not stand idly by while Hekmatyar seized power in the capital.