on march 16, 2004, the Pakistani army surrounded a large, fortress-like house in South Waziristan along the Pakistan-Afghanistan border.1 The house belonged to a twenty-nine-year-old charismatic tribal leader, Nek Muhammad Wazir, and two other more senior militants, Haji Muhammed and Nar-ul-Islam.2 In this tribal region, long known for its violent revolts against the authority of the central government in Islamabad, Nek Muhammad was an unusual figure. Aside from being a young leader in a society where age confers political power, he looked to some observers like a young Che Guevara, with a scruffy beard and long, flowing, black hair.3 Like many revolutionaries before him, he was not shy of journalists or television cameras. Among his followers, he was reputed to be so fearless in battle that he earned the Pashto nickname “Bogoday” (the stubborn one).4 His death—the first targeted killing by a Predator drone in Pakistan—would set in motion a chain of events that would lead to international condemnation of the use of drones for targeted killings and a crisis in US-Pakistani relations.
Born near the market town of Wana, Nek Muhammad was a shopkeeper in the local bazaar and a petty car thief before joining the Taliban at the age of 18.5 Once the US invasion of Afghanistan began, a number of Arab and Chechen militants affiliated with al Qaeda and the Taliban sought refuge across the border in Pakistan. Nek Muhammad offered them sanctuary at inflated rents, making him a wealthy powerbroker in the region.6 Aided by Uzbek militants from the Islamic Militant Union (IMU) and local tribes, he launched a series of attacks on US military outposts across the border in Afghanistan.7 This drew the attention of policymakers in Washington, who began to press Pakistani president Pervez Musharraf to hand over the al Qaeda–linked militants and to pacify the unstable border regions threatening America’s occupation of Afghanistan.
Musharraf was initially reluctant to confront Nek Muhammad and his allies because he knew that a military operation in South Waziristan would enrage the local tribes. The region had long been out of Islamabad’s formal control, and the modus vivendi that existed between the government and the tribes relied on neither overly interfering in each other’s affairs. Yet two assassination attempts and a fatwa calling for his death from al Qaeda’s second in command, Ayman al-Zawahiri, changed his mind.8 In late 2003, Musharraf ordered 80,000 troops into South Waziristan and demanded that Nek Muhammad and his allies turn over all of the al Qaeda–linked militants that he was sheltering, especially Tahir Yuldashev, the powerful head of the Islamic Movement of Uzbekistan (IMU).9 Weeks of intense fighting and bombardment from helicopter gunships did little to break the resistance of Nek Muhammad and his allies. Instead, they launched scattered attacks on the Pakistani Frontier Corps troops in a grinding battle for control over South Waziristan, a campaign that eventually brought the Frontier Corps to Nek Muhammad’s compound in March 2004.
The tense standoff quickly turned into a disaster for the Pakistani Frontier Corps: Nek Muhammad’s allies in the Ahmedzai Wazir tribe had set a trap for them. In a sudden attack, fifteen Frontier Corps troops and one Pakistani army regular were killed and another fourteen soldiers were taken hostage.10 Dozens of trucks, armored personnel carriers, and pieces of artillery were burned.11 Across the region, fighting raged, with ambushes of military vehicles and house-to-house fighting in Wana.12 Recovering from their setback at Nek Muhammad’s compound, the Pakistani Frontier Corps drew back, but continued their cordon operation elsewhere, while, undaunted, Nek Muhammad continued to launch periodic rocket attacks on their outposts. Amid growing concerns over civil casualties and calls for the people of South Waziristan to resist Islamabad’s imposition of direct rule, Musharraf ordered his troops to pull back in April 2004. In an embarrassing turn of events for the Pakistani military, General Safdar Hussain, the commander of the operation, was forced to meet Nek Muhammad and his allies to personally negotiate a ceasefire. General Hussain was photographed presenting flowers to the young firebrand and drinking tea with him, effectively enshrining Nek Muhammad as the most powerful man in South Waziristan.13 The Shakai peace treaty signed between the Pakistani military and Nek Muhammad even referred to him as a “mujahid” or holy warrior.14 Nek Muhammad was keenly aware of how he had turned the tables on the military, noting that Hussain was forced to come to his madrassa so “that should make it clear who surrendered to whom.”15
Only moments after the peace ceremony, Nek Muhammad told the assembled journalists that he would continue to wage jihad and swore his allegiance to Mullah Omar, the leader of the Afghan Taliban. He also immediately backtracked on his promise to turn over all foreign fighters to Islamabad.16 Although the government had offered generous terms—including amnesty for all foreign fighters and reparations for civilians in South Waziristan—Muhammad resumed attacks on US positions in Afghanistan.17 He also began a brutal campaign killing all of the local leaders (maliks) who had supported the government’s assault.18 Soon, Musharraf ordered his troops back into South Waziristan and fighting resumed with tribes and foreign militants. For Islamabad, the situation was growing more dangerous. Widespread resentment of the Pakistani Army’s attacks on Waziristan pushed many tribes toward Nek Muhammad and his allies. Opposition to Islamabad’s rule was beginning to coalesce into what was called the Tekrik-i-Taliban (Pakistani Taliban) movement, which would later launch a string of terrorist attacks and bring the violence of the frontier closer to the capital.
Furious at the betrayal and humiliation, Islamabad was willing to consider previously unacceptable options to eliminate Nek Muhammad. The CIA seized the opportunity: it proposed regular overflights by armed Predator drones over South Waziristan. This offer was partially due to necessity—export restrictions forbade the Bush administration from giving drones directly to Pakistan—but it was also a long-sought goal by some within the CIA to have the ability to directly strike militants in ungoverned territories, especially along the border within Afghanistan.19 For many in the CIA, it was clear that the militant movements in Pakistan were a vital source of arms and fighters for the Afghan Taliban. Pakistan’s intelligence service, the powerful Inter-Services Intelligence (ISI), was willing to grant the United States the right to fly armed drones over the region, but it imposed strict conditions in a secret deal with the CIA. The United States could only fly drones over approved spaces and at approved altitudes—designated “flightboxes”—and the ISI had to approve each drone strike before it happened.20 These attacks would never be acknowledged by the Pakistani government and would be conducted solely under the CIA’s authority. This granted plausible deniability to the Pakistani government for, as President Musharraf allegedly remarked, “in Pakistan, things fall out of the sky all the time.”21
On June 17, 2004, Nek Muhammad was on his mobile phone in the outside courtyard of the house of his friend Sher Zaman Asrafkhel, surrounded by a number of men and boys having dinner.22 Above them came a buzzing sound and then a blinding flash.23 A drone strike killed five of his companions, including two children, and severed the left hand and leg of Nek Muhammad.24 In shock, he was rushed to the hospital in Wana, where his last words allegedly were, “why aren’t you putting a bandage on my arm?”25 The Pakistani army claimed credit for killing Nek Muhammad, crediting night vision helicopters with tracking him down, but within days it was reported that the United States was behind the strike. Pakistan never held the US government to account for civilians killed in the strike, including the children.26 Nek Muhammad’s grave in Waziristan was accompanied by a handmade sign which said only “he lived and died like a true Pashtun.”27
The killing of Nek Muhammad was the first targeted killing in Pakistan, but it was not the first time the United States had conducted a targeted killing with a drone outside a declared war zone. In 2002, the United States had struck a convoy in Yemen and killed Qa’id Salim Sinan al-Harithi, one of the planners of the USS Cole bombing in 1998; Kamal Darwish, a naturalized US citizen; and four others. At the time, that targeted killing—initially denied, then confirmed by senior US officials—was an exception. In Pakistan, targeted killings by drones soon became the rule. Between 2004 and 2015, the United States conducted 400 drone strikes in Pakistan, killing between 2,276 and 3,614 people, according to data collected by the New America Foundation.28 The use of drones to kill militants had become an almost routine event; by 2010, the Obama administration was conducting a drone strike in Pakistan almost every three days.29
The expansion of the targeted killing program in Pakistan was gradual but steady. At first, the bargain that the United States struck with the ISI held, and Pakistan retained final approval over any prospective drone strike target. Throughout 2004–2005, US officials shared detailed images of training camps with their ISI counterparts and, in the words of the one senior CIA counterterrorism official, “every one of those shots was with Pakistani approval.”30 Yet as evidence mounted of Pakistani’s complicity in hosting the Taliban and other militant groups targeting US troops in Afghanistan, Washington’s patience wore thin.31 The United States had evidence that the ISI was working with a variety of militant networks, including the Taliban, the Haqqani network, and even al Qaeda, and was actively resupplying Taliban forces battling US troops in Afghanistan.32 In January 2008, the Bush administration decided to abrogate the secret deal of 2004, that had led to the death of Nek Muhammad, and began launching drone strikes without Islamabad’s formal consent—at will and with changed targeting practices.33 The CIA moved from personality strikes—in which the identity of the victim is known with a high degree of certainty—to signature strikes—based on the patterns of activity present among a target population. It was rumored that the threshold for likely success of a strike was lowered from 90% to 50%, although this has been disputed by some US officials.34
For the most part, Pakistan stood by as the US drone campaign ramped up throughout the Federally Administered Tribal Area (FATA), a semi-autonomous tribal region in its northwest that was home to some of the most dangerous Islamist groups. It did not contest the use of its airspace by shooting down US drones. In reality, they were playing, in the words of US officials, a “double game.”35 The Pakistani military and ISI secretly allowed the CIA to launch drones from some of its airbases and occasionally even suggested targets. While Islamabad publicly pretended that it condemned drone strikes, it was more than happy to see Washington take out its most bitter enemies. Prime Minister Gilani privately told US officials “I don’t care if they do [drone strikes] as long as they get the right people. We’ll protest in the National Assembly and then ignore it.”36 The killing of Baitullah Mehsud, the leader of the Pakistani Taliban and a spiritual successor to Nek Muhammad, was widely applauded within the security and intelligence establishment in Pakistan.37 Privately, some Pakistani officials supported expanding drone strikes even if it caused civilian casualties. President Zadari told US officials to “kill the seniors. Collateral damage worries you Americans. It does not worry me.”38 But when drone strikes killed large numbers of civilians, they were publicly condemned by the Pakistani government and blamed on the CIA. The wary, often hypocritical relationship between Washington and Islamabad was built upon a tacit understanding that targeted killings by drones were useful against their joint enemies, even if they did not have the same list of enemies.
Although there were expectations that President Barack Obama would scale back the targeted killing program once he came into office, he instead increased the tempo and scale of the attacks. Only three days after his inauguration, he authorized two drone strikes in Waziristan.39 The strikes killed fourteen people, including some civilians, but reportedly missed their high-value targets.40 Although President Obama was reportedly angry about the civilian casualties, he nevertheless authorized the drone program to continue.41 Senior administration officials, including chief counterterrorism advisor John Brennan and CIA Director Leon Panetta, backed the drone strikes as the only way to keep pressure on the dangerous militant networks and to enable the Obama administration to end the war in Afghanistan.42 As the drone strikes ramped up in 2010–2011, the Obama administration remained silent on the legal rationale for the program, even refusing to acknowledge in the courts that the drones program existed.
Behind closed doors, it was an entirely different story. The Obama administration undertook a program to institutionalize the targeted killing program, even producing an internal “kill list” with clearly specified criteria about who could be added to it.43 Soon, this process had settled into a bureaucratic routine. The Obama administration compiled dossiers on possible drone strike targets and debated the merits of striking each person in a weekly meeting dubbed “Terror Tuesday.”44 Over one hundred people from the Pentagon, CIA, State Department, and other government agencies would participate in these secret teleconferences. Led by President Obama’s chief counterterrorism adviser John Brennan, participants debated whether the target was considered “high value,” whether the threat that they posed was “imminent,” and how connected they were to al Qaeda and the Taliban.45 President Obama presided over the discussions of potential targets, signing off on approximately one third of all drone strikes before they happened.46 In Pakistan, because there were so many drone strikes, President Obama approved only the riskiest strikes.47 The CIA-led targeted killing in Pakistan—once considered an exception when Nek Muhammad was in the crosshairs—had now become routine. Targeted killings were, in the words of Panetta, “the only game in town in terms of trying to disrupt the al Qaeda leadership.”48
Yet the truth was that drones were increasingly targeting neither al Qaeda nor its leadership. By 2012, the targeted killing program had expanded beyond al Qaeda to take on an array of different militant networks inside Pakistan, including the Afghan Taliban, Pakistani Taliban, and the Haqqani network. Moreover, it was not only targeting leaders. The expansion of drone strikes from personality strikes to signature strikes had left Predator and later Reaper drones targeting low-level operatives or “foot soldiers” in the words of US officials. An estimate by Peter Bergen suggested that only 2% of all drone strikes were directed against leaders of militant organizations.49 As he later put it, the drones program had “increasingly evolved into a counterinsurgency air platform, whose victims are mostly lower-ranking members of the Taliban (Pakistan) and lower-level members of al Qaeda and associated groups (Yemen).”50 One mid-ranking Haqqani network member remarked in 2010 that “it seems that they want to kill everyone, not just the leaders.”51 The definition of an “imminent threat” had begun to slip as well. The US program was increasingly targeting individuals who posed a speculative rather than real or evident risk to US personnel or interests.52 The United States also expanded the geographic reach of the targeted killing beyond Pakistan to include new targets in Yemen, Somalia, and Libya, with unconfirmed reports that additional strikes had made in Mali and the Philippines. The dramatic expansion of the program suggested to UN investigator Ben Emmerson that the CIA’s drone program had slipped beyond the control of the White House.53
By 2012, the consequences of this expansion of targeted killings had begun to show. There was a growing international backlash against the United States for engaging in targeted killings with drones outside declared war zones. UN Special Rapporteur Christop Heyns attacked the US targeted killing program as a form of “global policing” and argued that the frequency of targeted killings risked loosening the restraints on the use of force by other actors.54 Human rights groups such as Amnesty International, Human Rights Watch, and others condemned the US drone strikes for producing excessive civilian casualties. Moreover, despite the complicity between the ISI and CIA over drone strikes, the US relationship with Pakistan began to break down. Islamabad began to worry over growing popular protests over the drones program and unrest in the tribal regions as militants capitalized on public discontent over drones to recruit more operatives. Enraged by the murder of two Pakistanis by CIA contractor Raymond Davis and the killing of Osama bin Laden by US Special Forces in Abbotabad, Pakistan expelled the US personnel responsible for the drones program from their airbase in Shamsi. The Pakistani government publicly blasted the United States for its “illegal and counter-productive” use of drones and argued that it was fueling the insurgency against the central government.55 By June 2012, the Pew Foundation found in a poll that 74% of Pakistanis considered the United States an enemy.56
In May 2013, at a speech at the National Defense University in Washington, DC, President Obama addressed the growing controversy over the drones program and defended the practice of targeted killing. Acknowledging the moral dilemmas that drone technology produced, he insisted that the program was legal and effective, but nevertheless promised to cease signature strikes, to raise the standard for targeting individuals, and to provide more oversight over the program.57 While US drone strikes in Pakistan declined in tempo, the targeted killing program never stopped there or elsewhere. Since assuming office in 2017, President Donald Trump has continued many of President Obama’s targeted killings policies and even ratcheted up the strikes, launching 238 between January 2017 and November 2018.58 He also swept aside many of the self-imposed limits on the strikes designed to reduce civilian casualties.59 Other countries—such as the United Kingdom, Nigeria, Iraq, and Pakistan—began to follow the US example in using targeted killing. What had begun as a way to strike a dangerous enemy in Pakistan and to control risks for US pilots had become a new and possibly dangerous global practice at odds with much of international law. Was it something about drone technology that made targeted killing not only permissible but actually a normal, everyday occurrence? In other words, did the ability to kill at a distance make killing more likely?
The term “targeted killing” is a relatively new one, having been brought into common usage over the last twenty years. The practice of targeted killing began with Israel’s response to the al Aqsa intifada in September 2000. The intifada began as a series of violent protests over Israeli prime minister Ariel Sharon’s visit to the al Asqa mosque, but escalated into a general protest over Israel’s occupation of the West Bank and Gaza. Facing growing unrest and suicide attacks, Israel began to use helicopters and manned aircraft to launch attacks against militant Palestinian groups in the Occupied Territories. Over the period 2000–2005, Israeli airstrikes killed over 203 Palestinian militants and 114 civilians.60 Although many Israeli airstrikes targeted specific individuals, Israeli officials maintained that these killings did not constitute assassinations, but were rather “targeted killings.”61 According to their legal arguments, targeted killings were different from assassinations in three ways. First, they were directed by government-controlled military forces in an active armed conflict. Unlike assassinations, which were typically conducted by spies or hired guns and involved subterfuge or deception, these killings were authorized and conducted through a military chain of command. Second, Israel was in a state of war against Palestinian militants, so the rules prohibiting assassinations in peacetime did not apply. If killing was legally and morally permissible in combat, there was no reason not to engage in targeted, and presumably precise, forms of killing in wartime.62 Third, Israeli authorities argued that militants were combatants, not political leaders, so their killing did not constitute assassination in the formal sense.63 In a war against terrorists who embed themselves in the civilian population and conduct suicide attacks, Israeli officials argued, governments had the right to conduct targeted killings as a form of active self-defense. Israeli officials later admitted that they made a concerted effort to legitimize the concept of targeted killings to permit the elimination of militants in the Occupied Territories.64
Initially, few states backed Israel’s position on targeted killings. Even sympathetic US officials claimed that they remained opposed to the practice and did not accept the distinction between targeted killings and assassinations.65 US Ambassador to Israel Martin Indyk remarked in July 2001 that the “United States government is very clearly on the record as against targeted assassinations. They are extrajudicial killings and we do not support that.”66 European governments were similarly opposed, arguing that these attacks were both illegal and counterproductive. At the same time, the US government was privately wrestling with the possibility of killing Osama bin Laden after he issued his fatwa urging all Muslims to kill US citizens and their allies. The Clinton administration quietly commissioned a series of internal legal findings that attempted to get around the ban on US officials engaging in assassinations, which had been in place since 1976.67 Aside from political indecision, there was resistance against targeted killing from the intelligence community and the military. CIA director George Tenet did not believe that he had the authority to “pull the trigger,” in part because the order to fire weapons should only come through a military chain of command.68 Military officials were equally concerned about using lethal force in a region where the United States was not formally at war.69 No one wanted to pay the political, and potentially legal, price if a targeted killing resulted in civilian casualties.70
The rapid advance of drone technology had begun to make targeted killings in places like Afghanistan feasible, if not easy, for the United States. The Israeli targeted killings had been conducted by manned aircraft, supported by drones, and launched at targets that were only a few miles away in the Occupied Territories. The United States had to be able to strike at a much greater distance at targets half a world away in Afghanistan. This produced an array of technical and logistical obstacles, including how the United States would pre-position drones for strikes in Afghanistan, how they would maintain communication with the drones, and how they would be sure at that distance that they were getting the right target. The chief technical problem was that there was a time lag between identifying a target and launching a missile. Under a program called “Afghan Eyes,” the United States had deployed Predator drones to find Osama bin Laden and other al Qaeda operatives. Their efforts had succeeded. Unarmed Predator drones had detected bin Laden a number of times and recorded video footage of a tall man in white flowing robes surrounded by a security detail in an al Qaeda compound, which some CIA analysts were convinced was Osama bin Laden.71 But the United States was not ready to strike him. Since the Predator drones were not affixed with Hellfire missiles, US officials would have to call in airstrikes from manned aircraft or use cruise missiles, both of which would not arrive for some time after the target was spotted. This meant that the bin Laden could escape and, even worse, that civilians could be killed if they moved into his place in the meantime.
By January 2001, the United States had managed to get around the technical difficulties of affixing Hellfire missiles to Predator drones and began to test them on dummy targets in the Nevada desert.72 Neither the Pentagon nor the CIA was convinced that the Predator was sufficiently precise or reliable to be used in a targeted killing on a real battlefield.73 In its first year in office, the Bush administration was equally unwilling to cross the line against assassinations.74 The Bush administration slowed down the decision-making process, leaving no consensus over whether the United States had the right to take the shot if it had bin Laden in its sights.75 The administration was divided over who had the right to authorize the killing, who pulled the trigger, and who paid for the missiles.76 The technology was ready, but the policy for using Predator drones for targeted killing remained confused and uncertain.
The September 11 attacks produced a sea change in US policy as the Bush administration authorized previously unacceptable lethal actions to destroy al Qaeda.77 While the US government did not formally abandon its legal ban on assassination, it embraced Israel’s arguments about the distinction between assassinations and targeted killing, although the extent of explicit coordination between US and Israeli policymakers on this point remains unclear.78 Secretary of Defense Donald Rumsfeld became convinced that “the techniques used by the Israelis against the Palestinians could quite simply be deployed on a larger scale.”79 The Bush administration won Congressional support for an Authorization to Use Military Force (AUMF) against those responsible for the September 11 attacks, which it subsequently interpreted as blanket approval for worldwide operations against militant groups even loosely affiliated with al Qaeda. The AUMF became a core part of the expansive legal rationale for drone strikes under the Bush, Obama, and Trump administrations, essentially greenlighting military action worldwide against any force presumably related to al Qaeda.80 This would be interpreted later to include the Taliban, Islamic State, and other Islamist groups less directly related to al Qaeda. A series of executive orders signed by President Bush authorized new techniques and expanded the capacity of the CIA and branches of the military to conduct manhunts, interrogations, and targeted killings. On the grounds of self-defense, the United States would have the right to strike at al Qaeda and associated forces worldwide. The Bush administration granted the CIA an almost unlimited authority to kill terrorists, using a combination of manned aircraft, drones, special forces, and other means.81
Initially, a relatively small number of high-ranking al Qaeda operatives—estimated between seven and two dozen—were pre-approved for targeted killing if the CIA, military forces, and even some contractors could locate them.82 According to former Deputy Secretary of State Richard Armitage, the Bush administration also issued a legal finding that killing terrorists worldwide did not constitute assassination and hence did not violate the long-standing legal ban on this activity.83 The United States did not accept that its embrace of targeted killing produced a generalizable right of targeted killing for all states, as it emphasized its unique situation in a global war against al Qaeda.84 This assertion of a unique right to worldwide targeted killing for the United States was not widely accepted. In 2012, former CIA director Michael Hayden remarked that “right now, there isn’t a government on the planet that agrees with our legal rationale for these operations, except for Afghanistan and maybe Israel.”85
Yet the law was not a real obstacle to the creation of a targeted killing program with drones. In the post–September 11 period, both the CIA and Pentagon overcame their squeamishness about targeted killing and ordered that missiles be attached to many Predator drones already in service. The technology was now ready to be used. With the technical issues resolved, Predator pilots were now circulating over Afghanistan with Hellfire missiles, ready to either strike targets directly or call in support from manned aircraft if needed (fig. 3.1).86 The technical problems posed by distance had been resolved and the United States alone now possessed the ability to precisely target and kill individuals from cockpits located in trailers thousands of miles away. The only question was whether this capability would be taken off the “hot” battlefields of Afghanistan and Iraq and expanded to places where al Qaeda existed but the United States was not officially at war.
Figure 3.1 MQ-1 Predator unmanned aircraft, armed with AGM-114 Hellfire missiles, over southern Afghanistan, 2008.
The first US targeted killing—of Sinan al-Harithi, a prominent al Qaeda leader in Yemen—occurred in November 2002. Although he had been placed on the CIA “kill or capture” list in 2001, al-Harithi was tracked down in the deserts of Yemen through a combination of human intelligence, aerial surveillance by Predator drones, and efforts by the National Security Agency (NSA) to track his cell phone use.87 The Predator drone strike—launched after consultation with the CIA and lawyers affiliated with US Central Command (CENTCOM)—killed al-Harithi, Darwish, and several companions.88 The Yemeni government consented to the killing, but only on the condition that it remain a secret. Both Yemen and the United States initially claimed that the deaths were due to a car bomb, but this fiction was shattered when Deputy Secretary of Defense Paul Wolfowitz admitted that the United States had killed al-Harithi.89 The Yemeni government was furious that the promise of secrecy was broken.90 Despite the Pentagon taking credit for the strike, the United States was reluctant to concede that this targeted killing created a precedent or represented a change in US policy. The State Department continued to argue that its opposition to Israeli targeted killings during the intifada remained intact, despite Wolfowitz’s admission of US responsibility for targeted killings in Yemen.91
Inside the US government, the success of the al-Harithi killing was widely noted. The armed Predator program had already been expanded and developed for use in Iraq, although the United States would not resume targeted killings until the Nek Mohammed strike in 2004. But the most important shift was not in technology, but rather geography. With the al-Harithi and Mohammed killings, the United States was asserting a right to use targeted killings against al Qaeda and affiliated targets in countries where it was not at war. The use of Predators for targeted killing in declared wars like Afghanistan was uncontroversial: the United States had a clear legal right to strike targets in that country. It made no legal difference whether those strikes were conducted with manned or unmanned aircraft. But off the “hot battlefields” the law was different. Such drone operations are technically conducted under conditions of peacetime, and there is no presumptive right to use violence in such territories except under exceptional circumstances, like pursuing a terrorist across a border.92 A different set of laws and human rights standards would apply to these circumstances, leaving the United States with no right to strike on the grounds of self-defense unless an imminent threat could be established.93 The CIA would also not be considered a “lawful combatant” under international law.94 Aware of the distance between accepted international law and their expansive interpretation of self-defense against the al Qaeda, both the Bush and the Obama administrations kept the program cloaked in secrecy and denied the jurisdiction of US and international courts.95
The al-Harithi targeted killing also illustrated the persistent bureaucratic divisions between the CIA and the Pentagon that would come to dominate the drones program. The ad hoc nature by which Predators were armed and later deployed for targeted killings left different agencies inside the US government in charge of different parts of the program. Today, the United States does not have a single targeted killing program, but rather two: one run by the CIA and another run by JSOC. These programs operate in parallel. CIA-controlled drone strikes in Pakistan, Yemen, and Somalia are conducted under the intelligence agency’s legal authority (Title 50), which enables it to conduct armed, covert action to influence political and economic conditions worldwide.96 In recognized, officially declared war zones, such as Afghanistan and Iraq, drone operations are officially conducted by the US military under a different authority (Title 10). JSOC operates under the same legal authority as other military forces, but its reach is worldwide and its targeted killings with drones can occur outside of recognized battlefields. This means that strikes outside declared war zones—for example, in Yemen—could be conducted by either the CIA or JSOC, or some combination thereof.
The CIA’s drones program evolved organically from attempts to kill bin Laden and some of his top deputies. Although early CIA operations such as the al-Harithi attack often required the use of US Air Force drone pilots, the CIA has been fiercely opposed to surrendering control over its targeted killing program to the military.97 Over time, the CIA’s use of drones produced a change in the organization’s orientation, away from intelligence gathering toward so-called “kinetic” or direct military action. Initially, the CIA’s paramilitary wing, the Special Activities Division, wanted control over the CIA’s drones. Yet the CIA leadership ultimately decided to give control over its drone fleet to the Counterterrorism Center (CTC).98 The CTC was led by “Roger,” a longtime CIA official with an unusual reputation in Washington as a chain-smoking, irascible convert to Islam.99 A controversial figure because of his responsibility for the CIA’s secret prisons and use of water-boarding against some senior al Qaeda officials, he nevertheless was able to build up a substantial reservoir of support for the CIA’s drone programs on Capitol Hill, which allowed the program to grow substantially.100 One estimate suggested that the CTC has expanded from only 300 employees in 2001 to over 2,000 within a decade of the September 11 attacks. 101 The CTC also commands substantial support from the CIA’s analytic division as “targeters” for the drone strikes.102 Under pressure to adapt its personnel structure, the CIA even created a career track for those specializing in targeting individuals for drone strikes. Although the CIA still uses drones for surveillance and non-lethal operations, the shift in its orientation toward paramilitary action has gradually turned the organization into “one hell of a killing machine.”103
Although the CTC sets targets and commands the operations of drone fleets, it does not directly fly them. At Creech Air Force base in Nevada, there were a number of US Air Force units specifically designated for CIA use either for reconnaissance or targeted killings. One unit, the 17th Reconnaissance Squadron, flew between forty-five and eighty aircraft designated for CIA use and were responsible for more “kills” through targeted killings than any other unit in the US Air Force.104 This unit was also responsible for flying many of the missions outside of declared war zones, such as those in Pakistan, Yemen, and Somalia. The fact that US Air Force pilots are flying CIA-commanded missions and the air force remained the owner of the drones themselves was not widely known, in part because the CIA’s imprimatur acted as a shield to block Congressional and public scrutiny of the program.105 Inside the government, the CIA is referred to as a “customer” of the US Air Force in that it sets objectives and mission parameters to be completed by trained drone pilots.
By contrast, the JSOC drones program is located in a military chain of command and reports to the US Special Forces Operations Command (SOCOM). JSOC’s drone fleets are assigned to different regional combatant commands and work in different theaters in support of US military objectives. In theory, this should mean that JSOC is governed by traditional military rules of engagement and the law of armed conflict. In practice, JSOC is even more secretive than the CIA, and its compliance with traditional military rules of engagement and the law of armed conflict is uncertain.106 JSOC’s permissive rules of engagement were based on the Bush administration’s executive order authorizing worldwide operations against al Qaeda. JSOC has been accused of torture and abuse of detainees in secret prisons in Iraq and elsewhere.107 The scale of the abuses at JSOC-run prisons in Iraq was so severe that other government agencies, including the CIA, barred their personnel from participating in interrogations in those prisons.108 JSOC operations are so highly compartmentalized and rapid that it is unclear whether all of them are subject to appropriate legal oversight. It also has limited Congressional oversight over its budget or activities.109
Under President Obama, JSOC expanded its kill or capture operations against terrorist targets around the world.110 Although its operations are shrouded in secrecy, JSOC is widely acknowledged to have expanded dramatically over the last decade.111 By 2015, it had a budget of nearly $1 billion and a staff of approximately 4,000 soldiers and civilians, supplemented by its own drone fleet and intelligence capacity.112 Private contractors also assist with JSOC operations. Contractors from the private security firm Blackwater, which had played a controversial role in Iraq, were responsible for assisting drone operations, even to the point of loading missiles onto Predator aircraft, although they were not permitted to be in the so-called “kill chain,” which authorized drone strikes.113 With this elaborate infrastructure, JSOC has been alternatively described as a “self-sustaining secret army” and “an almost industrial scale counter-terrorism killing machine.”114
The JSOC drone program operates in multiple theaters and across recognized conflicts, such as Iraq and Afghanistan, and undeclared battlefields in Yemen and Somalia. In the latter cases, JSOC conducted targeted killing operations with drones, sometimes in cooperation with the CIA. The relationship between the CIA and JSOC is complex. Although their drone programs operate simultaneously in these theaters, JSOC and the CIA sometimes clash on proposed targets and on information sharing. They also have different surveillance equipment and communications systems that are not fully compatible for sharing information about a target.115 Reportedly, they have different standards for acceptable civilian casualties when identifying targets. Yet despite these differences, JSOC and the CIA have cooperated in implementing targeted killings. For example, JSOC ran secret drone overflights in Pakistan to allow the CIA to identify targets for targeted killings.116 JSOC has “borrowed” CIA drones for targeted killings and sometimes closely coordinated their strikes in real time with CIA station chiefs.117 This arrangement—where JSOC forces operate under the cover of the CIA—was employed by President Obama for the killing of Osama bin Laden in May 2011. It remains unclear how many drone strikes fell under the same blended legal authority, but some experts have argued that the distinction between JSOC and CIA is even more blurry than it appears at first glance.118 The most aggressive counterterrorism operations outside declared warzones—such as manhunts or “snatch and grab” operations in places like Yemen and Somalia and targeted killings in Pakistan—have involved CIA, JSOC, other US military personnel from different branches, and private contractors working together, although their degree of their cooperation was kept secret.
These joint operations are especially controversial because JSOC forces would occasionally operate under the CIA’s legal authority, effectively as seconded units to the CIA itself. This allowed JSOC to operate with looser rules of engagement and less legal scrutiny than would traditional military operations.119 As a result, JSOC-run strikes or joint JSOC-CIA operations may operate with a higher tolerance for civilian casualties. For example, it remains unclear in joint JSOC-CIA operations whose lawyers—and whose standards for acceptable civilian casualties—apply. These operations may be effectively shielded from effective Congressional scrutiny, especially if it is unclear who is actually responsible for firing the missile.120 An absence of public transparency about drone-based targeted killing compounds the problems surrounding responsibility and accountability: it remains as unclear today as it was a decade ago who is really pulling the trigger for some drone strikes.
One additional problem arising from the bureaucratic complexity of the targeted killing program is that different government agencies have developed different kill lists: at one point, the National Security Council, CIA, and JSOC each had their own.121 In some cases, there was substantial overlap between these kill lists, but in others there were sharp differences of opinion about the degree to which an individual posed a threat or should be targeted. To resolve these differences, the US government has built a bureaucratic infrastructure to decide on targets and to administer its targeted killing program. Each of these actors could nominate a potential target for consideration at the so-called “Terror Tuesday” meetings. Biographical and threat-related information on the targets would be collected and condensed onto a short document dubbed the “baseball card” (BBC), which would be distributed to different stakeholders for discussion. In consultation with the National Counterterrorism Center (NCTC) and his senior deputies, President Obama had the final authority over who would be placed on the kill list and would personally approve the targets. The consolidated kill list even earned a different, more anodyne name: the Joint Prioritized Effects list. By 2013, the machinery of killing had become so complex that the US government constructed a master database—called, with Orwellian flourish, a “disposition matrix”—which included names, biographical details, suspected locations, and operational details.122
As some of these details slipped out to the public, the outcry over drones began to grow, and the Obama administration was forced to defend the legal basis of the program publicly.123 In April 2012, Brennan offered a defense of targeted killings, arguing that they were justified on the grounds of self-defense against al Qaeda and they were conducted in ways consistent with the laws of armed conflict.124 In a speech at the National Defense University in May 2013, President Obama described the US drone campaign in Pakistan, Yemen, and elsewhere as part of “a just war—a war waged proportionally, in last resort, and in self-defense.”125 Although he expressed some ambivalence about the expansive terms of the AUMF, President Obama still argued that targeted killings were authorized on the grounds of self-defense against al Qaeda and were legally and morally superior alternatives to using special forces or ground troops to capture or kill terrorist suspects in places where the United States was not formally at war. Yet the controversy continued to grow when legal rationales for targeting US civilians such as Anwar Awlaki became public. These secret memoranda produced by the Department of Justice suggested that killing US citizens was permitted if capture was infeasible and a threat was imminent. These memoranda allowed the US government to kill a citizen without presenting evidence or conducting a fair trial—clear requirements under the Bill of Rights—provided that the citizen joined the ranks of a declared enemy of the United States. These documents also left the standards for feasibility undefined and noted that imminence “did not require the United States to have clear evidence that a specific attack on US persons and interests will take place in the immediate future.”126 The Obama administration’s response had brought some sunlight to the drones program, but had also illustrated how far the United States had drifted from its historical position of limiting the use of force to fighting foreign combatants on active battlefields.
In 2013, President Obama formalized these standards by signing off on new presidential policy guidance (PPG) for drone strikes with requirements for authorizing a strike much tighter than those his administration had used so far. This guidance required that: (1) the United States conclude that it could not feasibly capture a suspect; (2) that there be near certainty that a target is a lawful one and in the location where the strike is contemplated; and (3) that there is a near certainty that non-combatants will not be harmed by the strike.127 It also required: that the US government certify that the target was a “continuing, imminent threat to US persons”;128 that the US government conclude, with legal review, that the government where the terrorist suspect was located could not or would not address the threat; and that the United States itself concluded that capture was not feasible. Once all of these standards had been met, a drone strike could—but would not necessarily be—authorized. In light of these new, stricter requirements, the pace of targeted killings in Obama’s second term slowed considerably, with fewer strikes taken off the “hot” battlefields but more devoted to destroying ISIS forces in the declared war in Syria and Afghanistan.
As part of his May 2013 speech announcing the changes in the PPG, President Obama proposed that the CIA’s drone operations be transferred to JSOC in order to subject it to traditional military oversight. Some prominent CIA officials, including President Obama’s chief counterterrorism advisor and later CIA director John Brennan, were supportive of the plan on the grounds that drone strikes had diverted the CIA from its traditional mission of collecting intelligence.129 The rationale also appeared to subject the CIA drones program to the military chain of command and its associated Congressional reporting requirements. Almost immediately the proposal ran into Congressional opposition from both parties, and Senator Diane Feinstein (D-CA) inserted a clause in a 2014 military spending bill that blocked the proposed transfer of authority. Part of the reason for this opposition was that many in Congress, from both parties, were convinced that the CIA was more careful and discriminate in launching targeted killing than JSOC. They also believed that the CIA, operated with higher evidentiary standards for selecting potential targeted killing and for avoiding collateral damage than JSOC. 130 In July 2016, the Obama administration released official data on the drones program claiming that it had killed only between 64 and 116 civilians between 2009 and 2015.131 Almost immediately, outside groups called these estimates wildly implausible and demanded to know the methodology behind counting the deaths. By the end of the Obama administration, the CIA and JSOC’s programs remained parallel operations and the true number of civilian casualties—as well as who bore legal and moral responsibility for them—remained unclear.
When President Donald Trump came into office in 2017, few people could anticipate what he would do with the targeted killing program that he inherited. On the campaign trail, he said little about it. His campaign rhetoric tended to be extreme—for example, he boasted that he would kill terrorists and their families—but it was unclear whether he would follow through, and if so, whether he would do so with drones.132 In practice, Trump accepted the basic outlines of President Obama’s polices, though he delegated more of the decision-making to the Pentagon and to lower levels. Although his administration has not released a revised PPG, it quietly revoked some of the new standards for avoiding civilian casualties that Obama had put in place in May 2013.133 Trump made substantial changes to targeting standards, throwing out the standard that a target must pose a continuing or imminent threat to US persons.134 He also lowered the standard for targeting individuals from “near certainty” to “reasonable certainty,” effectively lowering the burden of proof for authorizing a drone strike and enabling drone strikes where the risks of civilian casualties were higher.135
In practice, this meant that the pace of drone strikes under President Trump ramped up considerably, although the geographic focus shifted. While the Trump administration continued to strike at targets in Pakistan, it has shifted more toward doing so in support of the counterinsurgency mission in Afghanistan.136 In a tacit recognition of how these battlefields are intertwined, the Trump administration targeted high-level Pakistani militants in Afghanistan, including Mullah Fazullah, the leader of the Pakistani Taliban and the successor of Nek Muhammad.137 Elsewhere, it accelerated the pace of drone strikes against ISIS in Iraq and Syria, although hard data on the number of drone strikes there is hard to come by. Outside the “hot” battlefields, the increase in targeted killings was more striking. In 2017, the Trump administration launched 131 drone strikes in Yemen, more than three times the number of strikes in the previous year.138 These were allegedly against AQAP targets, though in practice it was hard to distinguish the actual purpose of these specific strikes amid the tumult of Yemen’s brutal civil war. In that war, the United States found itself allied with, but not necessarily always attacking alongside, Saudi Arabia and United Arab Emirates in their campaign against Houthi rebels seeking to overthrow the central government in Yemen. The shadow war also spread deeper into Africa: in Somalia, the Trump administration conducted an aggressive drone strike campaign against al-Shabaab militants with thirty-five strikes launched in 2017.139 By 2018, this had increased to forty-five so-called “precision strikes,” although actual details on who was launching the strikes, and what targeting standards were used to allow them, remained shrouded in secrecy.140
President Trump has also followed Obama’s lead in geographically expanding the shadow wars in order to allow for more drone strikes. Although hardly noticed at the time, President Obama ramped up the drone war in Libya, using nearly 300 drone strikes among hundreds of other attacks to force the Islamic State to evacuate Sirte in eastern Libya.141 In December 2016, President Obama quietly designated eastern Libya as a zone of “active hostilities,” thus exempting it from the tighter rules of his 2013 PPG guidance. This decision—effectively designating part, but not all, of a country a war zone—allowed more permissive standards to apply for drone strikes. This was revoked before President Trump took office, but he took his cue from Obama’s practice.142 He re-designated eastern Libya as an active combat zone, but also declared parts of Yemen and Somalia areas of “active hostilities” in order to enable more drone strikes.143 These decisions went alongside the rapid creation and expansion of new drone bases in Niger and lower Somalia to allow the expanded air war to take place.144
The final step that Trump took to amend the drones program pushed it further into the shadows. In 2017, the Trump administration simply ignored the legal requirement imposed by the Obama administration in May 2013 that it should produce a report detailing the civilian casualties from drone strikes.145 By 2018, his administration had taken steps to formally revoke this requirement, arguing that it led to unnecessary burdens on the government.146 As a result, there is no official government data for the growing number of shadow wars that drones enable in Pakistan, Yemen, Somalia, Libya, and elsewhere. This makes it harder for anyone outside the closely guarded government bureaucracy running drone strikes—including those in Congress—to be informed enough to make judgments about its use.
As Trump’s first term in office approached its end, it was clear that the finely calibrated killing machine built by the Obama administration had been handed over to a president who had placed it into overdrive and shielded even more of its operations from public view. In retrospect, the evolution of drone strikes is extraordinary. In less than twenty years, the US government had gone from scrambling to work out how to attach a missile to a Predator drone to developing a legal foundation and bureaucratic infrastructure for the practice of targeted killing in seven or more countries. This unwieldly bureaucratic infrastructure was remarkably different from the clear lines of authority that had traditionally governed military operations. It was internally divided between the customers (CIA and JSOC) and between the customers and those who actually fly the drones (the US Air Force). The entire process as to how someone in a foreign country is selected for killing is shrouded in secrecy and subject to relatively weak Congressional scrutiny and reporting requirements. For the most part, it remains entirely within the executive branch, effectively putting the US president in charge of deciding on the deaths of citizens in foreign countries.147 Most of the key elements of the program—including the standards for selecting targets—remain unclear. Drone technology might have brought the United States the capability of continuously monitoring and striking targets remotely, but it also led the United States to lose sight of its goals and drift into a growing number of conflicts worldwide. The policy governing its use has also been allowed to travel some distance from what was once considered acceptable state practice.
The process for targeted killing is suffused with technology, procedures, and language that places some distance between those flying the missions and their intended targets. If the president approves a person for a targeted killing, a sixty-day window is given to “operators” for a potential strike.148 Inside the government, the process for targeted killings became known in shorthand as “find, fix, and finish.”149 The organization that proposed the strike, typically CIA or JSOC, would generally be assigned principal responsibility for implementing that strike within that window. The first step—finding the target—involved a number of government agencies, including the CIA, NSA, and others, combing through multiple “streams” of information to locate the target. Among these streams were tips from sources on the ground, media reports, intelligence from foreign governments, and signals intelligence such as NSA intercepts of cell phones, emails, and other electronic records. These would be compiled into a coherent picture of the target and its environment. As chapter 4 will discuss, this process reflects the drive toward total battlefield awareness seen in the wars in Afghanistan and Iraq, where drones are helping to provide a level of detail about enemies and their movements that has never been achieved before in warfare.
The most difficult element of finding the target concerns human intelligence. The CIA had long traditionally struggled to develop human intelligence—dubbed HUMINT—from local sources in places like Pakistan and Yemen where tribal loyalty runs deep, and it is considered a violation of local custom to betray “guests” protected by one’s family or tribe. Against these limitations, the CIA has offered cash payments to induce locals to provide information about the location of potential targets and, in some cases, to plant transmitters in their cars, homes, or even on their person so that they could be traced and later killed.150 In other cases, the United States has engaged in social network analysis, tracking those who know or speak to the target in order to identify potential targets for surveillance. Information of the key personnel of a social network would then be passed on to ground sources or given to drone pilots to guide their missions. Many Predator or Reaper drones involved in targeted killings never fired a missile, but rather watched targets and their social networks continuously to understand their regular movements and pinpoint their locations.
Once the target was found, relevant operational details about a potential strike are collected into a “strike package,” which would move up the military chain of command and ultimately arrive at the White House. This would “fix” the target at a location. Among these details were estimates about the feasibility of a potential capture operation and the degree of civilian casualties associated with a potential strike, with corresponding estimates of the degree of confidence in these casualty estimates. The location of the drone strike was conceptualized in three-dimensional space as a “killbox.”151 Inside this space, the target is centrally located, but the planners can also construct probabilistic estimates of how many others will be killed. The strike package would be approved by lawyers working for different government agencies and intelligence analysts who would estimate the degree of certainty of the intelligence behind the strike. One internal JSOC document suggested that targeted killings would not be approved without two pieces of confirming intelligence pinpointing a target and no contrary intelligence that the target was elsewhere.152
The degree of policy and legal scrutiny of both CIA and JSOC operations has varied across the different administrations. Under President Obama, the legal scrutiny was conducted through multiple rounds of review, beginning with the National Security Staff (NSS) and ending with the Deputies and Principals of the National Security Council. In some high profile cases, especially those involving US citizens, President Obama would personally sign off on the “disposition” of the target. The Obama administration required that lawyers be involved in the process of approving targets, but the Trump administration scaled back this requirement because it was too onerous. Trump instead opted for a more delegated approach, sacrificing much of the high-level vetting and allowing the decision to be taken at a lower level with considerably less legal scrutiny.153 Only the decision to engage in direct action (either covert operations or drones) in a new country was taken to the highest level for approval.154 This reflected President Trump’s conviction that Obama had erred on the side of too much legal wrangling and that the process needed to be streamlined in order for the military to “take their gloves off” against enemies like the Islamic State.
Once the decision has been made to the target the individual and the legal review concluded, the final step is to finish a target. Although the precise details remain secret, it is clear that authorization for a targeted killing would come from the president, especially for risky strikes, or from senior deputies for more routine strikes. This authorization would be conveyed by a “customer,” such as JSOC or the CIA, to the drone pilots. Once the authorization to launch a time-sensitive missile strike had been given, a team of government and intelligence officials from different agencies would view the drone’s video feed to make assessments about the feasibility of a strike and the risks of civilian casualties. Communication between the participants in a strike would be done by chat messages, and the entire operation would be supervised by military lawyers who approved strikes.155 Analysts would be able to watch the drone’s video feed in real time, although senior Pentagon officials tried to limit this access to only those strictly needed for the strike. Once the missile strikes were approved and launched, the video feed from the Predator or Reaper drone did not stop, so the pilot and all others viewing that feed would be able to provide a damage assessment and conclude whether they had killed the target. Only at this point would the killbox be closed.
Drone technology is designed to allow states to escape what might be called “the tyranny of distance”—the fact that geography imposes real constraints on military operations and costs militaries valuable time that could be deployed to offensive activities.156 It also does so without posing immediate physical risk to the pilots, who are located in cockpits on a military base thousands of miles from the strike. The process of targeted killings by drones is conducted with antiseptic language designed to create moral distance between the pilots and those killed. But it is clear that drones do not completely eliminate distance just as they do not banish risk. Drone warfare is a surprisingly intimate form of violence, as it allows the perpetrator of violence to observe the prospective target over time, thus breeding more familiarity than is often found with artillery, nuclear weapons, or other forms of violence conducted at a remove.
The issue of distance is crucial to the debate over the ethics of targeted killing. Critics have argued that drone warfare cheapens violence by making it feel and look more like a video game, rather than real combat. The UN Special Rapporteur on Extrajudicial Killings Philip Alston has remarked that drones produce a “PlayStation” mentality, through which the combatants are desensitized to the deaths that drones produce and treat it as a game.157 To some extent, the technology itself reinforces this: the US military designed some of its controllers for drones to reflect those used by modern video game consoles, in part because so many young men were already adept with them. There is also some evidence from leaked documents that some drone pilots have described their victims in dehumanizing ways—for example, “bug splats”—and even called those running from their drones “squirters.” Others described their job in routine terms as “mowing the lawn” by removing insurgents from the battlefield.158
Some pilots have described their operations as “like a video game” and seemed to enjoy the experience and power of delivering death from above.159 Michael Haas, a former drone pilot involved in targeted killing, described how drone technology can desensitize pilots to the reality of targeted killings. “Ever step on ants and never give it another thought? That’s what you are made to think of the targets—as just black blobs on a screen. You start to do these psychological gymnastics to make it easier to do what you have to do—they deserved it, they chose their side. You had to kill part of your conscience to keep doing your job every day—and ignore those voices telling you this wasn’t right.”160 Although the violence was conducted without immediate risk to the pilots, many pilots expressed a willingness to shoot and to be seen as real warriors like the rest of the air force.161 The warrior ethos that underlies much of modern militaries affects drone pilots, many of whom are derided by their peers as not fighting a “real war,” and paradoxically produces coarse language and aggressive behavior designed to insulate them against this charge.
If war becomes “push button”—that is, like a video game and conducted by de-sensitized pilots—there is a danger that it may become too frequent. As chapter 1 suggested, drone warfare, especially targeted killings, is so different from traditional combat that there is a risk that governments will engage in it more often.162 Drones are a seductive type of technology in that they appear bloodless from the vantage point of the person pulling the trigger. For the United States, targeted killings offer the possibility of controlling and limiting risk because its pilots cannot be shot down or killed. The safety of pilots thousands of miles away in an air-conditioned bunker might lead policymakers to conclude that they should use drones more often, perhaps engaging in riskier activities than they would if manned aircraft were employed.
Another danger of remote warfare is that the criteria for using drones for targeting killing will be “steadily relaxed” over time, so that it occurs more often and may not always be directed to those who are actual combatants.163 The Obama administration came under criticism for this: for expanding drone strikes to targets in militant groups only indirectly related to al Qaeda and to lower-ranking operatives, rather than leaders. The Obama administration was accused of counting all “military aged males”—or MAMs, in official parlance—as terrorist operatives, even when there is no evidence for this designation. Some critics have argued that the absence of domestic political risk in targeted killing—specifically, that there is no prospect of downed or captured pilots—has made the Obama and Trump administrations careless in their selection of targets.164 Some have gone even further, arguing that the practice of targeted killing itself shatters the assumption of mutual risk that underlies most combat, rendering it closer to a technologically sophisticated form of hunting, or even execution, rather than warfare.165
At the same time, many drone pilots report that the process of targeted killing is not as capricious or lawless as it is often depicted in the media. T. Mark McCurley, a former drone pilot, has written that his experience of flying drone missions did not show that drone pilots were careless; on the contrary, their fear of making a mistake made them exceptionally careful.166 The fact that many people across the US government can pull up the video feed of a Predator drone ensures mistakes will be caught and properly punished. Drone pilots do not pull the trigger following a snap decision, nor do they make discretionary choices about who is targeted. Reckless killing via drones is possible, defenders argue, but difficult due to the number of people who have some supervisory or input role in the process. Others have noted that drone pilots are embedded in a decision-making process populated by intelligence analysts and lawyers watching their every move, so pilots themselves cannot be careless or indifferent to civilian casualties without consequences. They also argue that the process of targeted killing is more deliberative than the decision-making processes within war because participants have the time to reflect upon their actions before killing.167
Moreover, drone pilots often argue that they are hardly removed from the experience of killing in combat or targeted killings, but rather are intimately connected to their victims because they have been monitoring them for days. In contrast to combat, where violence emerges suddenly between strangers, drone pilots often are able to build up substantial amounts of time monitoring their target and developing some empathy with them.168 Drone pilots have reported seeing their targets drinking tea, attending weddings, and spending time with family members. For this reason, many drone pilots reject the argument that drone warfare is desensitizing. They report having a vivid understanding of the death and suffering resulting from a drone strike.169 With targeted killings, although the pilot was physically removed from the carnage that flows from a strike, he or she would see the human consequences of a drone strike, including the grieving families and sometimes even the burials of the victims. In the words of one unnamed pilot, “we’re not disconnected from what’s happening. We’re not playing video games. With RPAs, you grasp your enormous level of responsibility. You witness it all.”170
Figure 3.2 US Air Force ground control station at Balad Air Base, Iraq, 2007.
Some drone pilots report that the surprising intimacy of targeted killings has given them post-traumatic stress disorder (PTSD) because they witness the visceral effects of death and destruction up close.171 Former US Air Force drone pilot Brandon Bryant reports that the experience of flying Predators over places like Afghanistan and Iraq was both alternately boring and horrifying, especially when he saw targets and those surrounding them in agony after a Hellfire missile strike.172 Some pilots are also haunted by their mistakes, especially when their missiles hit women, children, and even animals. A 2011 air force study found that one third of all Predator and Reaper pilots suffered from burnout, while 17% suffered from “clinical distress.”173 Similar results were found for Global Hawk pilots. A later Department of Defense study found that drone pilots experienced the same levels of anxiety and depression as pilots of manned aircraft who were deployed to Afghanistan or Iraq.174 These findings remain controversial, and at least one skeptic argued that the trauma experienced by drone pilots is due to boredom, rather than the voyeurism of virtual combat, and that claims of PTSD are designed to deflect pressure on the United States for engaging in war at a remote level.175
At a minimum, it is clear that the experience of drone warfare is leading some former pilots to become outspoken critics of the targeted killings program. In 2015, four former drone pilots involved in the program wrote a letter to President Obama calling it “one of the most devastating driving forces of terrorism and destabilization around the world.”176 Calling the program “morally outrageous,” they argued that targeted killings are producing a deep wellspring of hatred of the United States around the world and that excessive civilian casualties are fueling recruitment to terrorist groups like ISIS.177 While these dissenters remain a minority of the total number of drone pilots who flew Predator and Reaper drones, their criticisms cut to the core of the issue: does the targeted killing program work?
The question of the effectiveness of targeted killings using drones is perhaps the most controversial question facing US policymakers.178 There is considerable anecdotal evidence that drone strikes have decimated organizations like al Qaeda by killing their most dangerous operatives. There is no doubt that drones have taken very senior figures in groups like al Qaeda, ISIS, and others off the battlefield. By one estimate in 2013, more than fifty senior al Qaeda and Taliban leaders have been killed by drone strikes.179 In 2011, one CIA official remarked that “we are killing these sons of bitches faster than they can grow them now.”180 By 2015–2016, US and allied drones were also killing senior Islamic State members in Iraq and Syria, forcing the group to adapt to preserve its leadership structure. A secondary benefit of the drone program is that it places so much pressure on terrorist organizations that it becomes hard for them to operate, recruit, or launch strikes.181 Amid the constant fear that their operatives will be taken out in a blinding strike, terrorist organizations can do little more than think of their own survival. For this reason, some government officials argued in 2011 that years of drone strikes had pushed al Qaeda to the point of “strategic collapse.”182
In Washington, there is a consensus among both political parties that drones are effective in disrupting the operations of terrorist organizations. Former Bush counterterrorism advisor Juan Zarate argued that drone strikes had knocked al Qaeda “on its heels” as a result of the death of so many operatives.183 In 2012, former Obama administration official Jeh Johnson publicly raised the question about what happens to the United States’ war on terror once al Qaeda has been decisively defeated, in part due to the use of drones.184 Even as criticism of the civilian casualties from drone strikes increased, the Obama administration insisted that their effectiveness was undoubtable. As President Obama remarked in his speech at the National Defense University in May 2013, “today, the core of al Qaeda in Afghanistan and Pakistan is on the path to defeat. Their remaining operatives spend more time thinking about their own safety than plotting against us. They did not direct the attacks in Benghazi or Boston. They’ve not carried out a successful attack on our homeland since 9/11.”185 In 2016, former CIA Director Michael V. Hayden described the drones program as “the most precise and effective application of firepower in the history of armed conflict.”186 President Donald Trump has said little about targeted killings, but clearly believes that they are an essential part of getting tough with terrorists.
There is anecdotal evidence that terrorist organizations are terrified of drones and that their presence imposes real costs on their ability to operate freely. The journalist David Rohde was held in captivity by the Taliban and reported that his guards were terrified of drones and took steps not to draw the attention of the Predators flying overhead.187 In its publications, al Qaeda lamented the effects of drones, noting that its commanders had been snatched away by planes that are “unheard, unseen and unknown.”188 Elsewhere al Qaeda has discussed the carnage and destruction—as well as the pervasive fear—produced by the drones above. Osama bin Laden was very concerned about drone strikes and advised his followers not to gather in large numbers for fear that they would attract drones.189 He even advised some of his followers to flee Waziristan to avoid their gaze.190 He also recommended a range of operational security measures, such as traveling by road infrequently, carefully monitoring movements to not attract attention, and moving on overcast days to avoid being spotted. He recognized that drones had hollowed out the top leadership of al Qaeda and left the ranks of the leadership populated with younger, less experienced, operatives.191 At the end of his life, he was also obsessed with operational security, particularly with the danger that the CIA would place a tracking device on a courier and reveal his location.192 Perhaps the best evidence of the fear of drones has been the extent to which al Qaeda, the Taliban, ISIS, and other targeted groups have changed tactics and ruthlessly killed those that they suspect turned over information about their whereabouts.193
Because of the difficulties of getting reliable data on violence in places like Pakistan and Yemen and the secrecy surrounding the drones program, there have been relatively few empirical studies of the effectiveness of drones. One such study found a modest reduction in violence and reprisal attacks on tribal elders following US drone strikes in Pakistan.194 Another found that drone strikes weakened the operational effectiveness of terrorist groups by causing such pressure that they experienced desertions and internal political struggles.195 Yet a general problem with analyzing the effectiveness of drones is that there is no consensus on how the effectiveness of drones should be measured.196 Reductions in violence following strikes are an intuitive way to do so, but in these environments the causes of fluctuations in violence are complex. Moreover, some policy advocates of targeted killing have conflated tactical success—that is, removing potentially dangerous actors from the battlefield—with strategic success. In fact, these are quite different: the United States could be eliminating dangerous terrorists in places like Pakistan and Yemen but simultaneously be weakening the government and putting longer-term goals, like regional stability, farther out of reach. Other arguments have confused efficiency—that is, the cost of drone strikes relative to manned aircraft strikes—with effectiveness, arguing that drones are a better, less expensive way to fight a war than a more expensive manned air platform. The problem is that the financial cost of a particular weapon is no guide to its effectiveness. A cheaper weapon is not, by definition, a more effective one. Many of the efficiency arguments are also problematic because they assume that the goals of a targeted killing program are fixed and clear. But for years the Obama and Trump administrations have left the targeted killing program so cloaked in secrecy that it is hard to know what ultimate goals it serves. Does the United States have an end state to a campaign of targeted killing? Or is it simply one of maintenance of an unstable situation by removing “bad guys” from the battlefield? One danger is that drone technology may lead to goal displacement in which the objectives of a targeted killing program become more ambitious because the technology enables new operations once seen as too difficult or risky.
At a minimum, if the goal of the targeted killings programs in Pakistan and Yemen is to defeat al Qaeda and affiliated forces, it cannot be considered a clear success. Instead of pushing al Qaeda to the edge of strategic collapse, drones have accelerated its fragmentation into a series of local affiliates, most of which are weaker than al Qaeda’s central organization was at its height before September 11. Here the effectiveness argument is murkier than it looks: drones have degraded the central organization of al Qaeda, but they have also turned one enemy into a series of loosely connected smaller foes. It is not yet clear whether this fragmentation will render al Qaeda strategically weaker than it was beforehand, or whether in fact this dynamic will give al Qaeda a second life as a franchise outside the original theaters of operation. The proliferation of al Qaeda franchises—and plots flowing from these branches in Yemen and elsewhere—suggest that the claim of effectiveness is not yet proven. Similarly, the rise of the Islamic State in Syria and Iraq out of the ashes of an al Qaeda affiliate decimated by drone strikes suggests that the strikes themselves are not the kind of mortal blow that destroys the organization itself.
If the goal of the targeted killing program is to improve the stability of governments in places like Pakistan and Yemen. While the cause and effect between drone strikes and stability is difficult to determine because so many factors are at play, the countries most targeted outside of conventional wars—Pakistan, Yemen, and Somalia—are no more stable after years of drone strikes.197 Despite the pressure of drone strikes, Pakistan remains a base for dangerous militant groups attacking US forces in Afghanistan, foreign fighters have continued to flow into the Afghanistan-Pakistan region,198 and Yemen has collapsed into a vicious civil war. Perhaps the biggest difficulties involved in assessing these outcomes are that they involve a counterfactual: if the United States had not engaged in a targeted killing program, would these countries have been more stable? In cases like Yemen, where the causes of militancy are complex and not reducible to drone strikes, this is a particularly difficult question to answer. It is impossible to know whether the Houthi rebellion and ensuing civil war would have happened if the United States had not conducted drone strikes in the preceding years. The decision to launch a targeted killing is ultimately based on an ex ante judgment that the country was already unstable enough to justify a targeted killing. It is possible that the United States only engages in targeted killings in countries already sliding into chaos, thus making the adverse consequences of drone strikes hard to measure and to trace back to US “kinetic action.” The relationship between drone strikes and stability remains extraordinarily difficult to disentangle, especially when it remains unclear whether drone strikes have disrupted potential insurgent attacks against the government or merely redirected them elsewhere.
The redirection problem with drones is a serious one because most insurgent and terrorist groups work across national borders. A sustained campaign of targeted killing may scatter an organization rather than cripple it. This contagion effect is recognized by senior government officials. Former CIA Director Leon Panetta argued that one measure of the success of the drones program is the fact that al Qaeda forces in Pakistan had sought refuge elsewhere to avoid the pressure from drone strikes.199 But this may reflect a redirection of the violence rather than a reduction. The CIA found that the Tehrik-i-Taliban Pakistan (TTP) has fled out of the FATA and relocated to cities like Karachi, where they targeted Pakistani civilians with terrorist attacks.200 In other cases, militants flee abroad but bring with them the skills, experience, and weapons needed to turn these local wars into even more fierce and long-lasting conflicts. As the aftermath of the civil war in Afghanistan (1979–1988) showed, the diffusion of trained foreign fighters to Algeria, Bosnia, and elsewhere worsened the civil wars in those countries. A similar dynamic is present in 2019 as foreign fighters that once joined the Islamic State in Iraq and Syria are fleeing to join conflicts around the Middle East and North Africa. If targeted killings push hardened terrorists out of Pakistan but relocate them to Afghanistan, or out of Iraq but into Libya, it is harder to consider targeted killings a net gain.
Most assessments of the strategic effectiveness of drone strikes consider the tactical benefits of drones—removing individual “bad guys” from the battlefield—but fail to consider the other side of the ledger: whether drones are also fueling terrorist organizations by stimulating recruitment of militant networks. Here the evidence is also murky. In Pakistan, there are accounts that drone strikes have convinced some locals to join the ranks of the TTP or other militant groups to fight the United States, or the Pakistani government for its complicity in their deaths.201 One scholar has argued that US drone strikes in Pakistan motivate revenge attacks by Islamist groups against civilians, which can perversely lead to a hostile, uncertain environment which makes recruitment by those same groups easier.202 Yet a comprehensive analysis of court records and of recruitment motives found no evidence that drones or targeted killings were fueling militant Islamism.203
In Yemen, local activists have claimed that drone strikes against AQAP have fostered anti-US sentiment in the tribal regions of the country and encouraged friends and family of killed civilians to join AQAP or other militant networks.204 The drone strikes have allegedly bred “psychological acceptance” of AQAP among Yemenis, in part because they appear to confirm its narrative of a bloodthirsty United States dropping bombs from afar with no concern for who is killed.205 As one local human rights leader put it, “the drones are killing al Qaeda leaders, but they are also turning them into heroes.”206 Because of inadequate data, it is impossible to know for certain whether drone strikes are creating more terrorists than they are eliminating.207 At a minimum the steady drumbeat of popular criticism about drones from Pakistan and Yemen provides a narrative about the brutality of the United States and the fecklessness of their governments that is consistent with the recruitment pitches of dangerous militant groups.
Those recruitment pitches typically focus on one of the most controversial issues surrounding drones: civilian casualties. Advocates of drones compare the relative precision of drones favorably to other methods of warfare and note that drones cause fewer civilian or accidental casualties than attacks such as air strikes or ground assaults. Although the Trump administration has been largely silent on civilian casualties, the Obama administration offered a number of defenses when facing criticism about the human costs of its targeted killings campaign. It argued that the stringent guidelines for selecting drone targets ensure that relatively few, if any, civilians are killed in drone strikes. Former CIA Director John Brennan said that civilian casualties from drone strikes have “typically been in the single digits.”208 In January 2012, President Obama publicly argued that the drone program is “a targeted, focused effort at people who are on the list of active terrorists.”209 The Obama administration also insisted that drone strikes have a “surgical” character and are conducted with a “laser-like focus” on the targets alone.210 In his May 2013 speech, President Obama acknowledged the reality of civilian casualties from drone strikes but insisted that “before any strike is taken, there must be near-certainty that no civilians will be killed or injured—the highest standard we can set.”211
The problem is that casualties from drone strikes are notoriously difficult to count and verify. Most of the strikes are conducted in distant, sometimes ungoverned, territories of Pakistan and Yemen, where few have the ability to interview survivors or even count the dead. Moreover, it is well known that the Taliban, al Qaeda, and other local Islamist groups inflate the casualties for propaganda purposes. Many of the underlying newspaper accounts of drone strikes toss around words like “militants” and “civilians” casually, often without evidence. The underlying difficulties of reporting strikes in these countries has been compounded by the US decision to adopt a classification scheme which counts any male between the ages of eighteen and seventy killed in a drone strike as a “militant” unless posthumous evidence is presented to clear their names.212 With the facts hard to come by, the perception of civilian casualties becomes almost as important as the strikes themselves. As Peter Bergen noted, the perception may be more important than reality, as more than half of respondents to a 2010 survey in North Waziristan believed that the strikes killed mostly civilians.213
A number of critics from inside the US national security establishment have come forward to voice concern that this perception of indiscriminate killing will generate backlash and anti-US sentiment. Admiral Dennis Blair, former Director of National Intelligence, has noted that while drone attacks did reduce the leadership of al Qaeda, it also “increased hatred of America” and harmed “our ability to work with Pakistan [in] eliminating Taliban sanctuaries, encouraging Indian-Pakistani dialogue, and making Pakistan’s nuclear arsenal more secure.”214 Similarly, General James Cartwright, former Vice Chair of the Joint Chiefs of staff, said that with the drones policy in Pakistan, “We’re seeing that blowback. If you’re trying to kill your way to a solution, no matter how precise you are, you’re going to upset people even if they’re not targeted.”215 General Stanley McChrystal, who expanded the use of drones in Afghanistan, remarked that “what scares me about drone strikes is how they are perceived around the world. The resentment created by American use of unmanned strikes . . . is much greater than the average American appreciates. They are hated on a visceral level, even by people who’ve never seen one or seen the effects of one.”216 He also remarked that drones reinforce “perception of American arrogance that says, ‘Well we can fly where we want, we can shoot where we want, because we can.’ ” While that blowback has not yet reached US shores, it is possible that it may in the near future. For example, the Times Square bomber Faisal Shahzad was allegedly trained and deployed by one faction of the TTP to attack New York in response to US drone strikes.217
The risk of backlash against the United States from drone strikes could be worse if drones produce a general climate of fear in the general population rather than just in militant groups. A controversial Stanford/NYU study of the impact of drones in rural Pakistan found that targeted killings had an invidious effect on the social fabric of the societies where they occur.218 As Brian Glyn Williams has noted, drones are often described by local Pakistani villagers as machays (wasps) for their stings or bangana (thunder) for their ability to strike without warning.219 While drones terrify their intended targets, innocent villagers are equally terrified of being in the wrong place at the wrong time when an attack occurs. Drones produce a “wave of terror” among the civilian population which has been described by some mental health professionals as “anticipatory anxiety.”220 In the words of David Rohde, the US journalist who was captured by the Taliban, drones strike fear because they are a “potent, unnerving symbol of unchecked American power.”221 This fear has led some ordinary civilians to refrain from helping those wounded in drone strikes for fear that they will be targeted in a second strike or to attend for funerals for fear that they will wind up in what are known as “double tap” strikes, which target mourners or those gathering at the scene of a strike.222 Critics have also argued that drones have inhibited normal economic and social activity and even made parents reluctant to send their children to schools that might be accidentally targeted.223 The problem, as critics have pointed out, is that these reports of terror from drones are based on select interviews with people in dangerous environments, and it is hard to know how representative their experience has been.224
At a minimum, there is more evidence that targeted killings have turned neighbors on neighbors and fueled communal mistrust in a society where overlapping family, tribal, and social ties are crucial. The targets of drone strikes are often pinpointed by paid informants who allegedly place small electronic targeting devices in the homes or vehicles of suspected terrorists.225 Yet there is no way to tell whether these chips are left with real terrorist operatives or with those with whom the informant has a personal grudge. Rumors of these chips have produced mistrust in the community as “neighbors suspect neighbors of spying for the US, Pakistani or Taliban intelligence or using drone strikes to settle feuds.”226 The journalist Steven Coll reported that this fueled vicious internal power struggles within the societies under their watch as people turn on each other for being spies for the CIA.227 While the drones circling overhead spread fear throughout the population and disrupt normal life, the suspicion produced by these chips and other means of nominating targets further erodes the social trust that underlies much of religious and political life in rural areas. At the same time, there are always some in every country who see drones as a net positive. Some Pakistanis and Yemenis support drones because they eliminate dangerous militants that kill more of their fellow citizens with terrorist attacks. The division between those who support drones and those who do not may fall along a number of lines, including region, class, and the degree of information that people have about drone strikes. For example, some scholars found that highly informed urban elites, and those deeply opposed to the agendas of Islamist networks, are more likely than less-informed and rural people to support US drones strikes even if they violate the sovereignty of the country.228
The depth of the degree of political protests over drones may still matter because it influences the stability of their governments and may make it harder to say “yes” to Washington. In countries like Pakistan and Yemen, drone strikes can corrode the governments’ stability and legitimacy and deepen anti-US sentiment. Even when employed against local enemies of these governments, US drone strikes are a powerful signal to citizens of the helplessness or complicity of their governments and undermine the notion that these governments are a credible competitor for the loyalties of the population.229 From a counterinsurgency vantage point, US drone strikes pose a critical problem, as governments facing ungoverned, unruly regions are usually engaged in a delicate dance with local tribes or clans for the loyalties of the population.230 In September 2012, Pakistani Foreign Minister Hina Rabbani Khar captured this dilemma well, saying that “this has to be our war. We are the ones who have to fight against them. As a drone flies over the territory of Pakistan, it becomes an American war again. And this whole logic of this being our fight, in our own interest is immediately put aside and again it is war which is imposed on us.”231 The extent to which the United States has assumed the role of direct combatant and sidelined the Pakistani government through drone strikes undermines the authority and legitimacy of the government. This makes the establishment of a stable set of partnerships for counterterrorism cooperation difficult, if not impossible, to achieve over the longer term because such plans depend upon strong and legitimate governments as US partners. These political costs need to be weighed against the advantages accrued by removing dangerous militants in an analysis of the effectiveness of drone-based targeted killing.
The practice of targeted killings was not initiated by the rise of drones, but drones did begin to change the political and moral calculus of such practices. Following Israel’s example, the United States has expanded the practice of targeted killing to a global scale, asserting for itself the right to strike al Qaeda and its affiliated forces anywhere in the world. Some of the unique features of drones—their ability to monitor targets over time, their precision, and the absence of physical risk to pilots—make them an attractive option. These technological features are not neutral, but rather affect strategic choice: if it becomes easier to engage in a targeted killing, there is a risk that states will resort to it more often. The danger is that the frequent use of targeted killings will corrode the traditional legal and ethical constraints on the use of force and create precedents that will be exploited by others.232 The Obama administration was aware of this danger and its efforts to discipline the use of drones and improve transparency reflected an awareness that, as Brennan put, it the United States is “establishing a precedent that other nations may follow.”233
There is some evidence that this has already begun to happen. Until recently, states with drones that could launch targeted killings have generally refrained from doing so, although China came close to using drones to kill a drug lord in Myanmar in 2013.234 But now more states are joining the targeted killing club. In 2015, the United Kingdom, which had long supplied intelligence to and cooperated with the United States, authorized a targeted killing of a British national in Syria on the grounds that he was planning imminent terrorist attacks on his homeland.235 In September 2015, Pakistan attached a missile to one of its Burraq drones, possibly with the help of Chinese technology or designs, and struck targets in the Shawal Valley in North Waziristan.236 Less than a decade after secretly greenlighting a CIA drone to kill Nek Mohammed, Pakistan conducted its own targeted killings in the same tumultuous region from which he had emerged. Soon afterward other countries joined the club. Iraq used a Chinese-made drone in an assault against an Islamic State stronghold in Ramadi in December 2015.237 In February 2016, the Nigerian military followed suit by conducting a targeted killing with a Chinese-made Rainbow drone against Boko Haram, a radical Islamist group affiliated with Islamic State.238 More recently, both sides in Yemen’s civil war have used drones for attacks and have even begun to target each other’s leadership as the United States has long done. In April 2018, the United Arab Emirates, part of a coalition led by Saudi Arabia to support Yemen’s government in the civil war, used a Chinese-made drone to kill a senior Houthi leader who had expressed interest in a negotiating an end to the fighting.239
From its origins as a secret practice, hidden from public view in the remote regions of Pakistan and Yemen, targeted killing has now gone global, facilitated by the diffusion of drone technology and the corresponding change in what states can and are willing to do. Yet it remains unclear whether the spread of targeted killing—and the distance that it places between what drones can do today and the traditional respect for the sovereignty of other states—will produce a more peaceful world. The ability to inflict death from above in an almost surgical way reflects a desire among governments to target individuals, rather than groups, and to conduct war in a way that preserves the lives of civilians. It also reflects a desire to control risks to their own pilots, even if the risks to others on the ground increase. But it has also led to goal displacement: the United States may not have intended to become militarily involved in a growing number of countries around the world, but its pursuit of al Qaeda and now the Islamic State through targeted killing has led it do so.240 In the end, the rise of targeted killings by drones may represent a significant technological accomplishment but it is hard to conclude that it is also a moral victory.