For years, the U.S. government has been waging counterterrorism campaigns against al-Qaida and other armed groups in safe havens and weak states. What explains the effectiveness of such campaigns? The variation in effectiveness may result from differences in select tactical, organizational, and technological capabilities of the counterterrorism state and its local partner, captured by the concept of the Legibility and Speed-of-Exploitation System (L&S). Empirical studies, including novel fieldwork data, on the U.S. drone war in Pakistan's Waziristan region from 2004 to 2014 reveal the influence of the L&S on targeted groups. From 2004 to 2007, a lack of U.S. counterterrorism capabilities aligning with the L&S allowed both al-Qaida and the Pakistan Taliban to build their operational infrastructure, expand their bases, engage in extensive recruitment drives, and broker important local alliances. In contrast, as the United States made substantial improvements in the L&S from 2008 to 2014, the campaigns against both groups became increasingly effective. Both al-Qaida and the Pakistan Taliban experienced sustained reductions in operational capabilities, losses of bases, and high desertion rates; they also faced growing political challenges, including from within their own organizations. These findings contrast with the view that counterterrorism offers short-term gains at best and is counterproductive at worst.


The U.S. government is waging counterterrorism campaigns against al-Qaida, the Islamic State, and their affiliates in Africa, the Middle East, and South Asia.1 Many analysts, practitioners, and scholars have long been skeptical of the battlefield effectiveness of this effort, however, arguing that it offers short-term tactical gains at best and is counterproductive at worst.2 One camp, which I call the “civilian-centric camp,” argues that the alienation of civilians produced during counterterrorism campaigns limits their effectiveness. Some members of this camp assert that counterterrorism engenders anger among civilians, turning them into potential recruits for targeted armed groups.3 Others claim that the disaffection of civilians or their concerns over a lack of security, given the limited on-the-ground presence of the counterterrorism state, makes them unlikely to provide the intelligence necessary for targeting terrorists.4 A second camp, which I call the “group-type camp,” suggests that the effectiveness of counterterrorism campaigns depends on the strength of the target. Members of this camp maintain that campaigns directed against weak groups are effective, whereas those against strong groups struggle.5

The empirical record of U.S. counterterrorism campaigns does not align with either the civilian-centric or group-type account. The campaigns in Pakistan, Syria, and Yemen, for example, have generated extensive information on their targets, despite the limited on-the-ground presence of U.S. forces.6 In Yemen, some evidence suggests that drone strikes increased local recruitment among targeted groups.7 In Pakistan, however, evidence of a similar increase in recruitment is more limited.8 Meanwhile, U.S. counterterrorism campaigns have substantially damaged formidable terrorist groups such as al-Shabab in Somalia, al-Qaida in Pakistan, and, most recently, the Islamic State in Iraq and Syria while causing extensive civilian casualties.9 A new explanation is needed to understand these varied dynamics.

This article argues that counterterrorism capabilities vary systematically and that this variation explains the effectiveness of counterterrorism campaigns. To capture this variation, I use the concept of the Legibility and Speed-of-Exploitation System, or the L&S. The L&S, composed of select tactical, organizational, and technological capabilities of the counterterrorism state and its local partner, varies along two dimensions: (1) the “legibility” of the civilian population within which an armed group is based, and (2) the “speed” of exploitation of legibility gains. Legibility involves the counterterrorism state's mapping of the social, economic, and political terrain of the territory being targeted; logging of civilian identities, assets, and behavior; and identification of civilian networks.10 Speed captures the pace at which the counterterrorism state pursues legibility gains to gather information or to strike terrorist targets. The combination of legibility and speed creates a spectrum of counterterrorism capabilities, ranging from low L&S to high L&S, which in turn determines the level of effectiveness of the campaign.

To demonstrate the significance of the L&S, I examine the 2004–14 U.S. counterterrorism campaign against al-Qaida and the Pakistan Taliban in Pakistan's North Waziristan region. I employ a before-after design integrated with process-tracing methods.11 I draw on novel evidence collected during fieldwork in Pakistan and the United States, including interviews with former Pakistani and U.S. officials from the period, members of al-Qaida and the Pakistan Taliban, and civilians from regions affected by the U.S. campaign. The evidence shows that from 2004 to 2007, a period of low L&S, the campaign was ineffective: both al-Qaida and the Pakistan Taliban enhanced their operational capabilities, expanded their number of bases, retained and added new recruits, and increased their levels of political consolidation. In contrast, from 2008 to 2014, a period of high L&S, the campaign's effectiveness improved: both groups suffered sustained reductions in their operational capabilities, a loss of bases, growing numbers of desertions, and political fracturing. These patterns contrast with expectations of both civilian-centric and group-type theories.

This article makes three contributions to the study of counterterrorism, counterinsurgency, and leadership decapitation. To begin, much of the literature on these topics focuses on whether the amount of local violence increases or decreases over a specific period following strikes against targets, infliction of civilian casualties, or provision of development aid.12 I broaden the focus to include metrics that capture the trajectories of armed groups over an entire conflict. Additionally, studies of leadership decapitation and air strikes do not conceptually specify the factors that determine their incidence: Are they a matter of choice for states or a function of an endogenous process? If the latter, what kind of endogenous process? My work specifies that process by conceptualizing the L&S.

Second, the article shows that state-centric factors captured by the L&S exert greater influence on the dynamics of internal conflicts than existing scholarship recognizes. Civilian-centric theories, for example, suggest that civilian collaboration is necessary for the collection of intelligence. In contrast, I show that states can collect information by aligning their own capabilities with the L&S. Similarly, group-type theories suggest that armed groups with institutionalized organizations, embedded social bases, quality recruits, and ties to transnational networks can survive and recover from attacks by the counterterrorism state. My findings show that campaigns with high L&S can severely damage resilient armed groups and constrain their recovery.

Third, the article contributes to the policy debate on counterterrorism by providing novel evidence regarding the impact of U.S. drone strikes on al-Qaida and the Pakistan Taliban. In addition, it suggests that states have limited battlefield-specific incentives to protect civilians. Accordingly, I argue that policymakers should commit to protecting civilians caught up in counterterrorism campaigns for normative rather than strategic reasons.

The article proceeds as follows. First, I conceptualize variation in the effectiveness of counterterrorism campaigns. Second, I introduce my theory of the Legibility and Speed-of-Exploitation System and describe its application to counterterrorism campaigns. Third, I discuss my research design and the U.S. counterterrorism campaign against al-Qaida and the Pakistan Taliban in Pakistan from 2004 to 2014. Fourth, I present evidence on the campaign, tracing the trajectories of both groups in the Waziristan region. I conclude by discussing the potential and limits of counterterrorism strategies.

Assessing the Effectiveness of Counterterrorism

To conceptualize counterterrorism effectiveness, I use variation in the trajectories of targeted armed groups’ organizational capabilities. The level of variation identifies a key objective of the counterterrorism state: degradation of the target's capability.13

Variation in an armed group's trajectory can be identified by examining its levels of operational capability, organizational basing, and collective-action activities, as well as the quality of its political relationships. Operational capability comprises the resources that armed groups use to achieve their political goals—including cash, in-preparation attack plans, training centers, weapons and ammunition dumps, bomb-making facilities, and specialized personnel. Organizational basing refers to the places where armed groups choose to live, operate from, and establish their base of operations. Collective-action activities include the recruitment and retention of members. Political relationships capture a group's ties with nearby groups, from collaboration to active hostilities.14

I disaggregate counterterrorism effectiveness into three categories based on a preponderance of indicators regarding armed group trajectories. A counterterrorism campaign is deemed ineffective if the targeted group increases its operational capability; if it not only holds existing bases, but also establishes new ones; if it maintains robust collective-action efforts; and if its political relationships with other armed groups reflect a level of consolidation through, for example, the creation of alliances, pacts, or mergers. A campaign is considered a stalemate if the targeted group neither increases nor loses operational capability; if it maintains its bases or systematically relocates them; if it preserves collective-action drives without incurring major losses; and if its political relationships with other armed groups neither improve nor worsen. A campaign is categorized as being effective if the targeted group loses huge amounts of operational capability; if it abandons bases (as opposed to relocating them); if its collective-action drives break down; and if its political relationships with other armed groups start to fracture, as evidenced in, for example, fratricide, splintering, or open hostilities.

Theory of the Legibility and Speed-of-Exploitation System

I propose a theory to explain variation in the effectiveness of counterterrorism campaigns. The theory has three actors: a powerful state that does not assume direct political control of the region of operations; a local partner in the area of operations, which may be a state or an armed group; and the targeted group(s).15 I describe (1) the capabilities of the first two actors to apply selective violence against an armed group—that is, the L&S; (2) the mechanisms through which the L&S can disrupt an armed group; and (3) the sources of variation in the L&S. To narrow the potential sources of variation in the L&S, I limit my claims to counterterrorism campaigns.16 Within this scope condition, the theory can be deployed broadly and comparatively.


Many scholars argue that discriminate violence, referred to as “selective violence,” can disrupt armed groups, but that civilian collaboration is necessary for applying such violence.17 As Eli Berman and Aila Matanock note, “A defining aspect of [the existing] literature is that civilians have a consequential role: they can share information (tips) with government forces [which] makes government attacks on rebels much more effective.”18 In my framework, I also posit that the state needs information to engage in selective violence; however, I propose that such information derives from the state's tactical, organizational, and technological capabilities, which are captured by the L&S. As mentioned earlier, the L&S has two components: (1) the legibility of the population in which the armed group is based, and (2) the speed of exploitation of the legibility gains. The interaction of legibility and speed produces a distinct capability, which is associated with observable levels of counterterrorism effectiveness (see figure 1).

Figure 1.

The Legibility and Speed-of-Exploitation (L&S) System and Counterterrorism Effectiveness

Figure 1.

The Legibility and Speed-of-Exploitation (L&S) System and Counterterrorism Effectiveness

Why does the L&S constitute capability in a counterterrorism context? To begin, when states wage counterterrorism campaigns against armed groups, the latter generally have an information advantage. Armed groups develop this advantage because, relative to state forces, they are small and less resourceful materially, lacking the means essential for sustained conventional military warfare. To achieve their objectives, they create clandestine structures for purposes of recruitment, training, planning, decisionmaking, and fundraising, which invariably involves hiding in small groups among the civilian population by leveraging stealth and flexible deployment.19

This technology of warfare challenges the counterterrorism state in two ways. First, the state must identify and locate the members of an armed group concealed within the civilian population. Civilians might be reluctant to identify them for reasons of safety or ascriptive association. I call this the “information-acquisition problem.” Second, the counterterrorism state must respond to the dynamic nature of the armed group, whose members are consistently trying to escape detection by changing locations, hiding their identities, and adopting high levels of operational secrecy. Thus, any information that becomes available to the counterterrorism state is valuable for only a limited time. I call this the “information-execution problem.”

The L&S captures the counterterrorism state's ability to respond to the challenges described above. The legibility dimension addresses the information-acquisition problem. It reflects the depth of a state's knowledge of the population,20 and includes metadata on the social, economic, and political conditions in a particular area; logs of civilian identities, assets, and select behaviors (e.g., movement); and identification of personal, social, and professional networks of civilians in the area. For a counterterrorism state, such legibility is invaluable; it can leverage legibility to generate information—cues, leads, and other details—on an armed group hiding within the population.

Legibility alone, however, is insufficient. The information generated by legibility is useful only if it is processed and acted on in a timely manner. Thus, the speed at which the counterterrorism state exploits legibility gains is important to meeting the information-execution challenge.21 Existing scholarship either ignores information-exploitation processes or assumes that states can always act on any information they generate.22 This is a major problem, because poor bureaucratic processes, a weak capacity to mobilize, and political constraints on intelligence sharing and targeting frequently hinder the ability of states to process and act on information.

The counterterrorism state can improve the L&S by implementing certain tactics, organizational practices, and technologies. From a legibility perspective, because the counterterrorism state has a limited on-the-ground presence, it cannot rely on legibility-enhancing tactics and practices that a domestic counterinsurgent might—for example, censuses, registration of population assets, entry-exit tracking of civilians, and identity cards. Instead, the counterterrorism state must rely on sources that do not require an on-the-ground presence. Thus, the key legibility sources are technology and local partner cooperation.23

The counterterrorism state can deploy interception technologies to monitor the communications of the local population.24 It can use advanced cameras on aerial surveillance platforms to tag and track population behavior.25 It can use aerial platforms with listening devices to detect and identify civilian voices.26 In addition, on-the-ground partners can share metadata on local social, political, and economic conditions, amplifying the usefulness of these and similar technologies.27 Local partners can also share administrative data on the population, access to local communication and financial flows records/infrastructure, as well as local intelligence networks. In tandem with technology, local partners can enhance the counterterrorism state's understanding of civilian behavior and networks.

The counterterrorism state can increase its speed by prioritizing select tactics, practices, and technologies.28 It can use its bureaucratic capabilities for real-time analysis of information. It can draw on database technologies that swiftly search population interception and aerial surveillance logs and other records.29 It can devise flat command-control hierarchies for immediate dissemination of legibility gains to those responsible for targeting.30 It can employ mobilization practices and technologies for rapid targeting, such as positioning ground forces proximate to regions of armed group activity, special forces with airborne assets, and loitering armed drones. It can delegate authority to battlefield commanders while simplifying the bureaucratic process to determine the legality and political appropriateness of its targeting.31

Legibility and speed operate independently of each other. Many of the tactics, practices, and technologies relevant for legibility (e.g., communication interception, aerial surveillance, and population records) do not contribute to speed. Similarly, some capabilities that improve speed (e.g., rapidly deployable forces, swiftly searchable databases, analytical capability, and decentralized decisionmaking) have no value for legibility. In some campaigns, the counterterrorism state might have more of one and less of the other, which amounts to the intermediate L&S capability shown in figure 1.


Two mechanisms associated with variation in the L&S—one kinetic and the other anticipatory—help to determine the effectiveness of counterterrorism campaigns. The kinetic mechanism includes the killing or arrest of armed group leaders and cadres, as well as damage to a group's infrastructure. These outcomes can lead to the severing of internal group ties by removing key organizational nodes, which in turn can degrade operational capability, the ability to maintain bases, recruitment processes, and political ties with allied groups. The anticipatory mechanism includes damage to targets resulting from their anticipation of being targeted, based on previous successful targeting. Such anticipation may constrain the movement of group leaders and cadres, limit their ability to communicate, and increase in-group mistrust. In turn, leaders and cadres may feel compelled to go into hiding, preventing them from pursuing operations, base expansion, and recruitment, as well as managing allied relationships.

Depending on the level of the L&S, the intensity of these mechanisms is likely to vary. When a counterterrorism state has low L&S, it acquires only limited information on which it can act, leading to low targeting activity. Such targeting minimally disrupts the internal ties of the armed group through kinetic and anticipatory mechanisms. When a counterterrorism state has high L&S, it acquires vast amounts of information on which it can act, leading to high targeting activity. Such targeting can cause extensive disruption to the internal ties of the armed group through kinetic and anticipatory mechanisms.

Under high L&S, the kinetic and anticipatory mechanisms can complement each other to prevent recovery of the group. Some armed groups may attempt to leverage their strong institutionalized internal processes, social bases, transnational networks, experience in warfare, or quotidian ties to recover from and adapt to targeting.32 Depending on the level of the L&S, the combination of kinetic and anticipatory effects can nullify such attributes and constrain meaningful recovery.


Two challenges can influence the ability of a counterterrorism state to achieve high L&S. The first challenge involves domestic politics. The L&S requires expensive technology and significant bureaucratic capability, which can vary according to the geographic reach of the campaign and the distance of the counterterrorism state's nearest base.33 The L&S also requires wide political latitude for implementing intrusive surveillance capabilities and speedy oversight processes, as well as a high willingness to accept civilian casualties resulting from quick processing of available leads and lower targeting thresholds. Legislative pressures, interest group politics, and the normative beliefs of policymakers may not permit the investment and political latitude necessary to achieve high L&S.

The second challenge is rooted in the counterterrorism state's relations with its local partner.34 The partner must be capable of and willing to execute tasks such as providing access to local communication infrastructure, assuring unfettered airspace for drones, setting up listening posts, and sharing intelligence in a timely manner. It also must agree to swift targeting protocols. In many situations, the local partner is likely to have constraints on its capacity.35 Even if the counterterrorism state builds up the local partner's capability, operational efficacy may remain elusive.36 Moreover, the local partner may be unwilling to cooperate, perhaps viewing the counterterrorism campaign as undermining its domestic standing.37 Thus, the counterterrorism state's ability to obtain the local partner's cooperation may require brokering complex political pacts.38


The framework described above yields predictions for counterterrorism outcomes and the processes behind them.39 In campaigns/campaign periods with low L&S, the counterterrorism effort will be ineffective. The campaign will struggle to acquire and act on information, leading to limited kinetic and anticipatory effects. Armed groups will internalize the inability of the counterterrorism state to locate them and will maintain robust organizational ties to expand their operational capability, base numbers, levels of recruitment, and political ties with allied groups. Additionally, in the face of losses, groups will be able to recover and adapt.

In campaigns/campaign periods with high L&S, the counterterrorism effort will be effective. The campaign will generate and act swiftly on information, resulting in successful targeting. In turn, the campaign will exhibit considerable kinetic and anticipatory effects. As the level of kinetic damage increases, anticipatory effects will have a constraining effect on the targeted group, which will be evident in breakdowns in in-group trust, communication, and mobility. These, in turn, will lead to a deterioration in the group's internal organizational ties, reducing its operational capability, base numbers, and recruitment and retention levels; the group will also see its relations with allied groups begin to fracture.

Research Design

I probe my predictions on the effectiveness of counterterrorism campaigns with an in-depth examination of the 2004–14 U.S. counterterrorism campaign against al-Qaida and the Pakistan Taliban in Pakistan's North Waziristan Agency. I use a before-after research design integrated with process-tracing methods. There are obvious limits to a single-conflict research design for the purposes of making generalizable claims; but for understanding the effectiveness of counterterrorism campaigns, this type of research design offers unusual advantages. For example, one of the main challenges in studying counterterrorism campaigns is that they are difficult to observe. My design overcomes this difficulty by drawing on granular qualitative data collected through fieldwork to probe the internal validity of my theory and the alternatives.

This approach offers inferential leverage in three ways. First, it provides traction on the sequence of events and causal mechanisms while holding constant major structural factors, such as the nature of the war and geography, and reducing the effects of independent Pakistani state activity, an important political confounder.40 Second, the approach offers within-case variation on the variable of interest for the theory, the L&S, predicting shifts in effectiveness over time.41 I identify variation in the L&S in two temporal categories: 2004 to 2007 (low L&S) and 2008 to 2014 (high L&S). Third, group-type and civilian-centric theories offer predictions different from those of the theory described here in terms of the mechanisms involved and their influence on the outcome. The U.S. counterterrorism campaign was carried out against two different groups: al-Qaida and the Pakistan Taliban. If group-type theories have more explanatory power than the L&S, then group-specific variation in the campaign's effectiveness should be evident. The civilian-centric view predicts variation based on the counterterrorism state's treatment of the civilian population. In the absence of incentives for civilian cooperation with the counterterrorism state, such as the provision of security, campaign ineffectiveness is more likely. Also, in periods of civilian harm, there should be evidence of targeting-information deficits and a surge in armed group recruitment.


The research design used in this study requires granular accounts of what the targeted groups looked like at the start of the counterterrorism campaign, how these groups changed over time, and the role of the campaign in bringing about those changes. The main evidence I use comes from sixty-six semi-structured interviews with individuals in Pakistan and the United States conducted in 2016 and 2017.42 The interviews were conducted using the purposive sampling strategy, a nonprobability sampling strategy “to obtain information about highly specific events and processes” for process tracing.43 I supplement these interviews with primary source materials, including a published history of the Mehsud faction of the Pakistan Taliban, letters to and by Osama bin Laden released by the U.S. government, local conflict historiographies, select al-Qaida documents obtained during fieldwork, U.S. government documents, and U.S. and British press reports.44


Bordering Afghanistan, Pakistan's Federally Administered Tribal Areas (FATA) have been a semi-autonomous region since the country's formation in 1947. The terrorist attacks of September 11, 2001, put the tribal areas at the center of international politics. In late 2001, al-Qaida and other targets of the U.S. military campaigns in Afghanistan traveled from Afghanistan's Shahikot mountains to FATA. There, al-Qaida rallied local leaders opposed to Pakistan's support for the U.S. invasion of Afghanistan. By 2004, the resistance had grown to become the Pakistan Taliban.

In 2004 the U.S. government, in cooperation with the Pakistani government, launched a counterterrorism campaign to target al-Qaida and its local affiliates. The campaign was governed by a secret pact reached in June 2004 between the U.S. Central Intelligence Agency (CIA) and Pakistan's Inter-Services Intelligence (ISI) agency. The agreement demarcated “flight boxes” over two regions of FATA above which U.S. drones could fly (i.e., the Waziristan and Bajaur Agencies); it also established CIA forward bases in the tribal areas and provided for an intelligence-sharing arrangement between the two agencies regarding the use of armed drones.45 The negotiations leading to the pact were complex. Pakistani leaders worried about the potential fallout of the campaign on the country's domestic politics and the possibility that the United States might use the campaign as an excuse to spy on nuclear and other facilities in Pakistan, as well as key ISI detachments.46 The final pact took these concerns into account.

In 2008, the two countries renegotiated the pact to expand the program. In January 2008, CIA Director Michael Hayden and U.S. Director of National Intelligence Mike McConnell met with Pakistani President Pervez Musharraf and brokered a wide-reaching intelligence-sharing and targeting arrangement featuring looser rules of engagement than contained in the 2004 pact.47 An ISI official with knowledge of the 2008 pact explained the context at the time: “General Musharraf was under tremendous pressure … He was worried about his own survival in office and wanted American support.”48 Pakistan agreed to the expansion of the campaign, but terminated the flight box over the Bajaur Agency.

The U.S. Counterterrorism Campaign's L&S in Waziristan

In this section, I discuss the broad patterns of the L&S in the North Waziristan Agency across two time periods: 2004 to 2007 and 2008 to 2014.49


From 2004 to 2007, multiple indicators suggest that the L&S in North Waziristan was low. First, in terms of legibility, the U.S. counterterrorism force had only limited availability of drone “orbits” to monitor much of Waziristan.50 (“Orbit” here refers to a metric used to determine the number of surveillance platforms employed by the U.S. government.) Interviews with officials from Pakistani intelligence and the U.S. government suggest that surveillance drones were often unavailable during this period of the campaign.51 Second, the ISI did not provide the CIA access to its domestic communication-interception infrastructure.52 Third, the campaign had limited human intelligence resources (including linguists and analysts). The CIA and ISI joint human intelligence program to identify drone targets was small and lacked funding.53

In terms of speed, the campaign also experienced significant limitations. First, the targeting protocol was severely constrained. The U.S. government process for determining the legality and utility of targeting was cumbersome and created significant time lags.54 From 2004 to 2006, all strikes had to be cleared first by the director-general of the ISI and then by the Pakistani president.55 In 2006, the U.S. government further constrained the process with the imposition of a “triple trigger” system, requiring the agreement of three officials before a subject could be targeted: the CIA's station chief in Islamabad, a U.S. military commander, and the U.S. ambassador to Pakistan.56 Second, the campaign suffered from a paucity of U.S. armed-strike platforms, especially as the war in Iraq intensified.57 Consequently, in a number of instances where the CIA had some intelligence leads, it would pass them on to its Pakistani interlocutors, who would mobilize regular infantry units to carry out raids.

RAMPED-UP L&S, 2008–14

From 2008 to 2014, the U.S. counterterrorism campaign attained high L&S in the North Waziristan Agency. Four factors contributed to this change. First, starting in 2007, the U.S. government significantly augmented its technical capability to undertake intensive aerial surveillance by bringing online a stream of Predator and Reaper drones.58 By 2008, these drones were operating over much of the North Waziristan Agency. The U.S. Air Force Central Command reported 3,764 Predator and Reaper sorties until 2007 in regions within the Central Command, which includes Pakistan; this number increased to 7,185 in 2008 and 9,415 in 2009.59 In interviews, Pakistani officials also noted a surge in “ISR [intelligence, surveillance, and reconnaissance platform] activities” and continuous surveillance by multiple drones over North Waziristan.60

Second, the U.S. government increased its surveillance and analysis resources designated for the campaign. According to a former U.S. official involved with the campaign, the level of surveillance resources starting in 2008 was substantially higher than it was for operations in Somalia and Yemen until 2014.61 A U.S. intelligence official noted that the National Security Agency (NSA) “threw the kitchen sink at the FATA”; to achieve its “huge coverage of the FATA, NSA had 10 times the manpower, 20 times the budget, and 100 times the brainpower” compared to the CIA's Information Operations Center.62

Third, the U.S. and Pakistani governments deepened their level of cooperation under the 2008 pact. One interviewee remarked on their wide-ranging communication interception and population data-sharing arrangement, part of which focused on Waziristan. Additionally, ISI increased U.S. access to the region's communication infrastructure to facilitate interception.63 The ISI cooperated with a number of U.S. NSA surveillance programs, including the Boundless Informant Program, to identify telephone numbers and record phone calls.64 The ISI's Bannu detachment (renamed Counterterrorism Center Bannu in 2008) shared population records available through North Waziristan's political administration.65 The ISI also provided the CIA's forward bases with analysts, linguists, and mapping information until 2009.66

Fourth, the CIA upgraded its human intelligence infrastructure for North Waziristan, with significant support from the ISI. By 2008, there were two human-intelligence networks on the ground. One was set up by the ISI's counterterrorism wing, which coordinated its operations with the CIA's Islamabad station.67 The second was managed directly by the CIA's Islamabad station. In addition to Afghanistan-based units in regions bordering FATA, this network relied on retired Pakistani intelligence officials and operated without the knowledge of the ISI.

The influx of Predator and Reaper drones concentrated in a relatively small region ensured that the CIA had ample rapid-strike capability without having to compromise continuous aerial surveillance operations, as had been common in the past. This capability translated into high speed, for two reasons. First, the U.S. government gave the CIA considerable leeway in undertaking strikes against unknown targets.68 Given the “Agency's understanding with the Pakistanis in 2008,” noted one former U.S. government official, the CIA began to ignore the triple-trigger requirement in early 2008 and officially shelved it in early 2009.69 The ISI agreed to let the U.S. government execute strikes at times and locations of its choice.70 Second, the U.S. policy expanded beyond targeting “high-value” al-Qaida leaders to include fighters whose “patterns of suspicious behavior [could be] detected from drone surveillance cameras.”71

Although the overall capacity and rules of the campaign remained similar throughout this period, there was one exception. Interviewees noted that tenuous relations between the U.S. and Pakistani governments from early 2011 to mid-2012 curtailed both the level of cooperation and the political latitude available to the U.S. government for targeting. Three incidents contributed to the near breakdown in ties: the January 2011 shootings in Lahore by U.S. spy Raymond Davis, the May 2011 raid that killed Osama bin Laden, and the November 2011 U.S. attack on Pakistani military outposts.72

The Trajectories of al-Qaida and the Pakistan Taliban, 2004–14

In this section, I process trace the trajectories of al-Qaida and the Pakistan Taliban from 2004 to 2014 along four dimensions of effectiveness: operational capability, organizational basing, collective action, and political relationships with other armed factions. Across three periods, I first compare the trajectories of the two groups along each of these dimensions. I then analytically trace the mechanisms on targeting patterns and internal group dynamics that can be expected under different levels of the L&S.73


From 2004 to 2007, the U.S. counterterrorism campaign against al-Qaida and the Pakistan Taliban was ineffective, and the trajectories of both groups were on the rise. At the start of the campaign, al-Qaida had rebuilt itself in the South and North Waziristan Agencies of Pakistan's tribal areas, receiving support from young local leaders. The organization had two major factions: the central faction reporting to Osama bin Laden, but led by Hadi al Iraqi and seconded by Abu Laith al-Libi; and a Pakistan-based faction, led by Khalid Khubaib.74 Until late 2004, al-Qaida's main base of operations was located in the South Waziristan Agency. By 2005, al-Qaida, under Libi's supervision, had ordered the group's core members to move to the North Waziristan Agency.75 From 2006 to 2007, al-Qaida established elaborate bases in areas of Data Khel, Dosali, Mir Ali, and Miramshah—all in the North Waziristan Agency. By 2006, its leadership had begun referring to the Agency as the Islamic State of Waziristan.76

In this period, al-Qaida had ample cash flow, which it used to set up a vast operational infrastructure. A Karachi-based shura (consultative body) assembled by the mastermind of the terrorist attacks of September 11, 2001, Khalid Sheikh Mohammed, funneled a steady stream of funds to Waziristan.77 Al Iraqi used these funds to “establish training centers, suicide bomb training, IED [improvised explosive device] production, weapons and explosive handling, material printing, and lodging facilities.”78 He also set up specialized units such as Jundullah and Jaish-ul-Qiba: Jundullah would target the U.S. and other Western interests in Pakistan, with the goal of neutralizing Pakistan's support for the United States, while Jaish-ul-Qiba would conduct operations in Afghanistan.79

From 2004 to 2007, al-Qaida aggressively sought out new members, especially “emotionally stable, well-educated, and experienced cadres.”80 It recruited fighters from across the broader Middle East and expanded its cadres of Pakistani fighters—in particular, trained cadres of “jihadi organizations [who] had fought in [Indian] Kashmir.”81 Journalist Saleem Shahzad notes that “slowly and gradually this strategy began to work, and brought thousands of new recruits into the [al-Qaida] fold.”82 Al-Qaida attracted fighters from Harkat-ul-Mujahideen, Harkat-ul-Jihad al-Islami, Lashkar-e-Jhangvi, and Jamaat-e-Islami—all of which had been involved in the so-called Jihad-e-Kashmir, or fight for Kashmir.83

Starting in 2006, al-Qaida began to consolidate its Pakistan-centered operations under the commander of the 313 Brigade, Ilyas Kashmiri. Kashmiri started training “fidayee [suicide bombers who would carry out assaults before exploding their suicide vests], [other] suicide bombers, and IED saboteurs.”84 In parallel, al-Qaida focused on training cadres for operations in border regions of Afghanistan, given that “opportunities to launch operations from North Waziristan Agency were better.”85 In addition to its ever-growing military capability, al-Qaida expanded its public outreach through outlets such as Al-Sahab Media and the Urdu magazine Ummat.86

From 2004 to 2007, al-Qaida consolidated its political ties with other armed groups in the region and solidified control over quasi-independent factions. In the South Waziristan Agency, it formed an alliance with Nek Muhammed's Yargulkhel Wazir tribe and co-opted the fighting units of two other local armed group leaders, Abdullah and Baitullah Mehsud. In the North Waziristan Agency, it created alliances with influential armed groups such as the factions led by Maulvi Sadiq Nur and Abdul Khaliq Haqqani.87 Al-Qaida also backed the fighting force of the Islamic Movement of Uzbekistan.88 From 2006 to 2007, it took control of important independent factions in the region, including the Libyan Islamic Fighters Group and the Egyptian Takfiri group.89 It also made pacts with influential local groups such as the Hafiz Gul Bahadur group and the Haqqani network.90

The Pakistan Taliban's trajectory was similar but less linear than al-Qaida's. Nek Muhammed, based in South Waziristan but with some influence in the tribal areas of North Waziristan, emerged as a key leader of the Taliban movement. He started out with “a small force without much operational capability; he had some weapons and only a few hundred men.”91 By early 2004, his operational strength had grown, and he had gained recognition as a supporter of al-Qaida. In the same time period, Abdullah Mehsud had created a group with influence in North Waziristan.92 Baitullah Mehsud similarly had organized a small militia and engaged in sabotage in South Waziristan.

By 2005, Baitullah's militia had eclipsed both the Nek and Abdullah factions. (Nek died in a U.S. drone strike in the South Waziristan Agency in June 2004.) Baitullah expanded his group to the North Waziristan Agency, building bases in Dand-e-Darpa Khel, Data Khel, Dosali, Mir Ali, and Miramshah.93 By 2006, he had set up “dozens of marakaz [centers] for training suicide bombers and IED factories.”94 By late 2007, his group's operational capability was comparable to that of “a division-sized conventional military force.”95

Factions of the Pakistan Taliban recruited local tribal youth as both midlevel commanders and full-time fighters. In 2004, Baitullah set up offices to recruit in select tehsils (subdistricts) of the South Waziristan Agency; by 2006, he had expanded his recruitment drive to the North Waziristan Agency. Baitullah organized his factions into halqas, the equivalent of “brigades organized down to the level of tehsils.96 By late 2006, he had created nearly a dozen halqas.97 In response to the Pakistani military's 2007 assault on Islamabad's Red Mosque, he rapidly expanded all of his halqas, making “appeals to avenge the blood of the martyrs of Red Mosque.”98

From 2004 to 2007, the Pakistan Taliban's political trajectory in the North Waziristan Agency was messier than al-Qaida's, though one of consolidation. Abdullah Mehsud's and Nek Muhammed's factions cooperated loosely in this period. After Nek's death in June 2004, his faction maintained an alliance with both Abdullah and Baitullah Mehsud. Abdullah and Baitullah were estranged, however, having accused the other of being on the ISI's payroll. In February 2005, Baitullah struck a deal with the ISI, substantiating Abdullah's accusation. That deal, however, collapsed within a few months, and Baitullah expanded his group for jihad against Pakistan. By late 2005, Baitullah's faction had grown more powerful than Abdullah's. Baitullah was able to convince major fighting units in Abdullah's and Nek's factions to merge with his group.99 He also maintained a strong alliance with al-Qaida, which supplied him not only with funds, but also trainers and commanders to provide supervision to his organization. In late 2007, Baitullah emerged as the amir of the Pakistan Taliban, which he renamed Tehrik-i-Taliban Pakistan (TTP). According to a tribal elder from North Waziristan, “Baitullah's growing force helped him gain al-Qaida's endorsement and cash for buying out other factions from across the tribal belt.”100 Besides merging factions from across FATA and the Khyber Pakhtunkhwa Province, Baitullah co-opted groups in North Waziristan by giving them a seat on the TTP's shura.101

During the 2004–07 period, the U.S. government carried out only a limited number of drone strikes against al-Qaida and the Pakistan Taliban. Mostly, it provided information to Pakistani forces to check the growth of both groups. The U.S. drone strike that killed Nek Muhammed was coordinated by Pakistani intelligence.102 In late 2005, three drone strikes were reported.103 In one strike, senior al-Qaida leader Hamza Rabia was killed in the village of Asoori, near the town of Mir Ali. In 2007, the United States launched three strikes in the North Waziristan Agency.104 Using information provided by the CIA, the Pakistani military conducted raids with limited success on villages near Mir Ali, Miramshah, Razmak, and Shawal.105

The low number of strikes during the 2004–07 period can be attributed to difficulties associated with locating targets. A former U.S. official noted how the lack of leads was a source of continuing frustration for the CIA's counterterrorism leadership.106 Drone surveillance generated some leads, and according to an interviewee, “the ISI's Bannu detachment and its offices in Miramshah and Mir Ali cooperated with the CIA.”107 Such cooperation was sporadic, however, and the CIA did not depend entirely on the ISI for obtaining leads. As another interviewee stated, “The CIA's Islamabad station generated leads from intelligence gathered in Afghanistan and Iraq on potential high-value targets in the tribal area,” which in turn led to drone strikes.108 In internal meetings, CIA Director Hayden fussed over the failure to remove “a single al-Qaida leader from the battlefield in 2007,” using the phrase “0 for 07” to signify the dearth of intelligence and successful strikes.109

As noted earlier, to the extent that the United States was able to acquire information, stringent rules made the execution of strikes difficult. According to a U.S. government document, “In 2007, the CIA sought ISI approval for 15 strikes, received prompt approval for three and a single ‘forced approval.'”110 One former ISI official observed, “The process of maturing the target was long, which led to targets disappearing, and the final bureaucratic approval was complicated on the American and Pakistani sides; it went as far as the DG [director-general] ISI and then president [of Pakistan], all of which contributed to many near misses.”111 Given the difficulty experienced by the United States in calling in drones, “the CIA's Miramshah forward base provided leads to the [Pakistani military's] 117th Brigade [on the ground in Miramshah].”112 Often, the brigade “acted on the leads but was invariably late.”113

In sum, the minimal pressure applied by the United States against al-Qaida and the Pakistan Taliban from 2004 to 2007 allowed both groups to recover from their losses relatively quickly and to avoid future losses while maintaining their growth trajectories. During this period, al-Qaida became more deeply institutionalized and attracted large numbers of specialized cadres. And by late 2007, the Baitullah Mehsud-led Taliban had become a formidable and sizable armed group.


As noted earlier, under the renegotiated CIA-ISI counterterrorism pact and with an influx of technology resources, the U.S. counterterrorism campaign from 2008 to 2011 attained high L&S. In line with the theory's expectation, the United States’ increased surveillance of North Waziristan generated many more targeting leads from 2008 to 2011 than it did from 2004 to 2007. Pakistani officials were surprised at the richness of the video feeds available to U.S. ground operatives in Bannu, Mir Ali, Miramshah, and Islamabad.114 One official involved with the program stated, “Drones started identifying al-Qaida and the Pakistan Taliban factions under training, notable midlevel commanders, and important marakaz used to host local jihadis and foreign fighters from Afghanistan and Iraq.”115

By early 2010, the operational capabilities of both al-Qaida and the Pakistan Taliban had fractured. Until 2007, al-Qaida had been freely training and plotting operations against targets inside and outside Pakistan: “During Sheikh Libi's time, al-Qaida had global ambitions,” noted one TTP operative.116 By 2009, however, it was struggling to maintain its global and local operational activities, scrapping a “dozen plans” for attacks in this period.117 According to another interviewee, starting in early 2008, al-Qaida began moving personnel from “villages in the Nauraq tribe area and those along the Mir Ali-Miramsah road.”118 Another noted that “marakaz in Spulga, Degan, Yusuf Khel, [and] Tappi” villages were closed.119 By late 2009, al-Qaida had pulled back most of its master trainers and midlevel commanders who had been helping Taliban forces.120 Al-Qaida's leadership started preparations to relocate bases out of Waziristan, such as those created by the eastern Afghanistan leader Farook al-Qatari, and to safe houses in mainland Pakistan. In a letter to bin Laden in June 2010, North Waziristan–based al-Qaida leader Sheikh Atiya Abd Rehman advocated pulling a “large portion of the group out of Waziristan.”121

Until late 2009, the Pakistan Taliban also showed signs of disruption, though not to the same degree as al-Qaida. Baitullah Mehsud preserved some of the Taliban's operational infrastructure in the South Waziristan Agency at “a safe distance from al-Qaida [Central].” Baitullah saw al-Qaida as the main target of U.S. drone strikes.122 He ordered the factions under the command of “Wali-ur-Rehman, Mufti Noor Wali, Khan Saeed, and Maulvi Shamim [all TTP commanders] to remain in the Mehsud area [of the South Waziristan Agency]” and to focus “on training rather than active sabotage.”123 He also asked Qari Hussain Mehsud, a master trainer of suicide bomb squads, to “prepare a stream of suicide bombers to bleed the Pakistani state.”124 The Pakistan Taliban staged suicide bomb attacks in Pakistani urban centers. One Taliban interviewee claimed that the group “planned ninety-four attacks.” He added, “I know it because I myself played a role in delivering messages and shifting eleven to twelve suicide bombers and suicide jackets … They carried out successful suicide attacks.”125

By late 2010, however, U.S. drone strikes had significantly damaged the Pakistan Taliban's operational capability. The onset of a Pakistani counterinsurgency campaign in neighboring South Waziristan Agency in 2009 and 2010, known as Operation Rah-e-Nijat, precluded the group from shifting its operational capability away from the North Waziristan Agency. After the discovery of Taliban leader Qari Hussain Mehsud's involvement in the December 2009 sabotage operation against the CIA forward base in Khost, Afghanistan, the United States intensified its surveillance and targeting of the Pakistan Taliban's infrastructure in the North Waziristan Agency. In response, the group systematically closed a number of its facilities, including training and IED manufacturing centers. Hakimullah Mehsud, the successor to Baitullah Mehsud (who was killed in a drone strike in August 2009) as the amir of the Pakistan Taliban, asked his aide “Latifullah [Mehsud] to minimize the signatures on his marakaz and create mobile training units.”126 By late 2010, Hakimullah had relocated some of his North Waziristan–based cadres closer to Pakistani urban areas, including Karachi.

By 2009, al-Qaida was beginning to suffer a growing number of desertions. A Pakistani official who monitored Waziristan during this time stated, “I read intercepts of al-Qaida and Taliban for around three years [2008 to 2011]. There was chaos in these groups by 2010 due to drone strikes. I heard al-Qaida's leadership, known for its cool and strong discipline, regularly lose its composure. So much of the chatter was about ‘my men have left me!'.”127 A member of al-Qaida who was in North Waziristan from 2009 to 2010 observed, “Drone strikes became a major concern for all jihadi organizations, including al-Qaida … the majority of the members were not as committed ideologically. I noticed that many left … I was troubled by this and spoke with [fellow jihadis]. We concluded most want to be ghazis [holy fighters who return from the battlefield alive]. Only a few want to be shaheed [martyrs].”128 In November 2010, al-Qaida leader Sheikh Atiya wrote to bin Laden, “We are facing difficulties due to the grave shortages in personnel in some cadres.”129 A Karachi-based journalist tracking jihadi activity noted that, by 2010, many jihadis from Waziristan were returning “because of drone strikes.”130

From 2010 to 2011, the Pakistan Taliban also experienced an increasing number of defections. Until 2008, ISI officials based in the towns of Wana, Dera Ismail Khan, Miramshah, and Bannu had expressed increasing concern about “the growing halqas … Boys were flocking to the TTP offices in different tehsils of North Waziristan Agency.”131 By early 2010, however, the ISI had begun noticing a gradual thinning of the Taliban's ranks.132 In separate interviews, two Taliban operatives initially declared desertions to be “a propaganda of ISI.” Later in one of the interviews, however, one of the operatives confirmed that there had been some desertions, describing those leaving the groups as “non-ideological.”133 An al-Qaida operative declined to discuss al-Qaida's manpower challenges, but noted a spate of desertions from the Pakistan Taliban in 2010 and 2011.134

By 2010, al-Qaida's and the Pakistan Taliban's political strategies of consolidation and co-optation had begun to unwind. Both organizations experienced within-group fratricide and breakdowns in relations with other groups in the region. Al-Qaida grew wary of the Haqqani network.135 Its top leadership in North Waziristan considered the network too close to the ISI; senior and midlevel leaders even thought it might be involved in the kind of intelligence gathering that led to drone strikes. Al-Qaida also fell out with the Libyan Islamic Fighting Group over differences concerning al-Qaida's strategic direction.136 The Pakistan Taliban experienced similar political challenges. Under Baitullah, the Mehsud faction fell out with the Hafiz Gul Bahadur group in early 2008. After Baitullah's death, senior commanders under Mehsud Taliban, Hakimullah Mehsud, and Wali-ur-Rehman Mehsud began a fight over succession.137 Hakimullah Mehsud's relations with important local commanders also began to suffer.138

Al-Qaida's and the Pakistan Taliban's dislocation from bases, degradation of operational capability, manpower challenges, and breakdowns in their political strategies of consolidation resulted, in part, from an increase in the number of drone strikes, or what I call “kinetic effects.” The strikes killed leaders and specialized personnel from both groups.139 According to data compiled by the New America Foundation, from 2008 to 2011 al-Qaida lost twenty-five prominent leaders and the Pakistan Taliban lost ten. Some of my interviewees directly linked these kinetic losses to ongoing degradation. In January 2008, for instance, senior al-Qaida leader Abu Laith al Libi was killed in a drone strike in Mir Ali after the timely exploitation of a lead generated by the CIA and triangulated by the ISI.140 As one Taliban operative noted, “Libi … was one of the most important persons of al-Qaida … His death was a big loss for amir sahib [Baitullah Mehsud].”141 According to another interviewee, “[In 2008] an al-Qaida leader who was responsible for jihadi activities in Pakistan was killed in a drone strike. After his death, many plans came to a halt.”142 Another noted how the 2009 killings of Qari Tahir Jan of the Islamic Movement of Uzbekistan and his field commanders in an air strike set back both al-Qaida and the Pakistan Taliban. The interviewee added, “[It was] a great loss, as they were commanding many Uzbek, Turkmen, and other foreign fighters who would fight alongside both al-Qaida and the Pakistan Taliban.”143

The targeting of al-Qaida's Sheikh Abu Khabab and his explosives project reveals how losses of specialized personnel undermined al-Qaida operationally.144 Khabab ran a robust operation in North Waziristan, with a number of marakaz dedicated to training explosives experts for local and global operations. He produced instruction materials, led lectures, and mentored Pakistan Taliban leaders. His death in a 2008 drone strike was a major loss for al-Qaida and had consequences for the Pakistan Taliban as well. Compounding this loss were the deaths from drone strikes from 2009 to 2011 of personnel he had trained to produce dirty bombs, land mines, and shaped charges, as well as instructors and examiners affiliated with his explosives training course “Dor-al-Mutafarja.”145 In the process, the explosives project was transferred initially to a Pakistani faction and later to Khabab's teenage son, Luqman. The Pakistani faction tried to revive the project by translating Khabab's manuals from Arabic to Urdu for mass circulation. Without reliable trainers, however, the written materials could not compensate for gaps created by the loss of the elder Khabab and his team. By late 2011, the quantity and quality of al-Qaida's and the Pakistan Taliban's explosives expertise had dropped dramatically.

Multiple interviewees noted that, as with al-Qaida, the growing exposure of the Pakistan Taliban to drone strikes starting in 2009 led to substantial operational disruption. One interviewee noted, “We had to close so many important centers after drone strikes. So many plans fell through. I will give you example of one center which we had to close. There were eighteen drone strikes in a single day on [one] center in which many Arab sheikhs were killed. I don't remember exactly whether it was 2009 or 2010, but I know it was almost a month after the [October 2009] attack on GHQ [Pakistan's military headquarters], because the ones who attacked at GHQ were trained in that center. Another attack, this one on Sri Lanka's cricket team … was also planned there.”146

Some interviewees linked the disruption of the Pakistan Taliban's operational infrastructure to the targeted killing of Qari Hussain Mehsud. In the words of one interviewee, “Qari Hussain was considered the ‘nuclear power’ of TTP. Many great plans remained incomplete due to his killing.”147 Another stated, “Under [Qari Hussain's] leadership, a number of tasks were completed successfully until drones came in our way. I remember one drone attack in Mir Ali, in which seven Germans were killed. They were soon to undertake suicide attacks in Germany and France … A few days later Qari sahib was martyred in the Wazir Kushali area [of North Waziristan]. Due to drone strikes, the plan of sending the German suicide bombers to France and Germany was never realized.”148

Kinetic effects, however, were not the only mechanism pushing al-Qaida and the Pakistan Taliban toward further degradation. Both groups suffered from anticipatory effects as well. For example, the remaining leadership and rank and file had to stay on the move. The intensified surveillance, which frequently led to strikes, made the top leadership and important midlevel commanders of al-Qaida and the Pakistan Taliban “feel extremely vulnerable.”149 A Taliban operative described the challenge of operating under the deluge of drone activity: “When drones started flying [in 2008], we became very careful about our movement. They surely made many of us anxious. It was as if a fan is turned on at the peak of winter. You cover yourself in a blanket but still feel very cold due to the fan being on. And you have to bear the noise of the fan.”150

Interviewees from al-Qaida and the Pakistan Taliban stated that both groups feared being spied on from within their ranks: “Commanders would regularly change their locations and vehicles. Also, they kept their locations secret, as there were reports that people within were providing information to the CIA.”151 According to multiple interviewees, the pervasive fear of spies had an especially detrimental effect on recruitment. One interviewee spoke about Hakimullah Mehsud turning away new cadres for just this reason: “[In late 2009] after arrival of fresh cohorts from Karachi, Punjab, and Mohmand Agency, the number of drone strikes increased. So those in Mir Ali and Miramshah used to see them suspiciously. Amir sahib [Hakimullah] decided that they should go back where they had come from.”152

Interviewees from al-Qaida and the Pakistan Taliban also mentioned how fear arising from the inability to move about freely limited the ability of leaders to make standard organizational decisions and maintain operational capability. Some factions of al-Qaida and the Pakistan Taliban, such as Jundullah and the 313 Brigade, were more watchful than others, believing they were more likely to be targeted. One Pakistani intelligence official noted that, “because of drone strikes on moving vehicles, groups started using motorbikes.”153 In a letter to bin Laden, Sheikh Atiya suggested “stopping many of the operations so we can move around less, and be less exposed to strikes.”154

Constraints imposed on the ability of groups to communicate also damaged within-group ties. Both al-Qaida and the Pakistan Taliban became increasingly wary of using communication devices, though the Pakistan Taliban was initially much less cautious than al-Qaida; many of its cadres used mukhabras, a Pashto word for shortwave radio sets.155 By 2011, however, even the Taliban had become fearful of communication devices, knowing that [Pakistani] “intelligence and its sidekicks were listening in on all radio communication, even PTCL [landline phones].”156 Groups instead relied on carefully chosen human messengers for within-group communication, leading to long gaps in communication and weak command and control. According to one interviewee, “Many were deeply frustrated that senior leaders would not stay in contact. In my second trip, I remember not seeing any senior leader nor getting communication for two months. We were told to concentrate on reading jihadi material.”157

Challenges to mobility, effective communication, and trust operated in tandem, constraining the ability of al-Qaida and the Pakistan Taliban to engage in damage control and recovery.158 In a letter to bin Laden, Sheikh Atiya noted, “We will focus on defensive security (counterespionage) by focusing on striking the spy plane bases using special operations, and on patience, persistence, hiding as well as decreasing our presence at least this year because it is an important year.”159 Such descriptions of sustained degradation were repeated by al-Qaida and the Pakistan Taliban members interviewed for this study: “The challenges faced by al-Qaida and the Pakistan Taliban were the same. Training became very difficult because a number of strikes took place where training was carried out … Movement became especially hard [and] people would maintain distance from those [midlevel commanders and rank and file] they did not trust.”160

Writing in 2011, an al-Qaida leader known by the pseudonym Hikmatullah Lodhi summed up the overall impact of drone strikes on al-Qaida and the Pakistan Taliban from 2008 to 2011: “We need to do our utmost to recover from the losses due to drone strikes … if leaders continue to be killed, the jihadi movement's entire direction and pace can suffer; if field commanders continue to die, training will be poorer and the operational capabilities of mujahideen will suffer, and the the next generation of mujahideen will not be of high quality. In short, drone strikes can overwhelm the strength of the mujahideen.”161


From 2012 to 2014, the CIA maintained high legibility of Waziristan, having made improvements following years of experience surveilling the region and maintaining close cooperation with the ISI.162 During this period, it also successfully exploited numerous leads not just on al-Qaida, but also on the Pakistan Taliban. As a result, both groups suffered—often in similar ways. By 2012, for example, al-Qaida had begun moving its senior leadership and midlevel commanders out of Waziristan. According to one TPP interviewee, “Many of the Turks, Uzbek, and other foreign mujahedeen shifted to some new safe houses.” Others “went to Afghanistan. Some even went back to Chechnya, Egypt, Iran, Iraq, Turkmenistan, and Uzbekistan.”163 The remaining leaders and cadres spent most of their time in temporary hideouts in Mir Ali and Miramshah: “With increasing drone strikes hitting the targets,” observed another TPP interviewee, “many had moved to settled neighborhoods.”164

The Pakistan Taliban also continued to disperse its bases and factions, often within but also away from the North Waziristan Agency. For example, Hakimullah Mehsud asked “Mullah Nazir [a South Waziristan–based commander] and, after his killing in a drone strike, Bahwal Khan, to house some of his fighters.”165 Two civilians held prisoner by the Pakistan Taliban from 2012 to 2014 noted that the Taliban closed multiple marakaz where they had been held captive for fear of being targeted in drone strikes.166 By 2012, increasing numbers of “Pakistan Taliban fighters [had] moved to Karachi.”167 Hakimullah Mehsud also sent some of his fighters to Khost and Paktika in Afghanistan.168

The operational capabilities of al-Qaida and the Pakistan Taliban also continued to diminish during this period. By 2012, al-Qaida was struggling to train and produce fighters. It closed most of its training centers, explosives production facilities, and lodging marakaz in and around the town of Mir Ali, which had once powered both global and local operations.169 The Pakistan Taliban still operated small training centers in and around Miramshah, which were significantly more in number than what al-Qaida had; much of its capability, however, was “a shadow of the past.”170 Hakimullah Mehsud sought to build underground bases in Miramshah and Mir Ali to revive his group's training and explosives production activities. Although construction on some of these bases was completed, most projects remained unfinished.171

Defections continued to plague both al-Qaida and the Pakistan Taliban. The number of foreign fighters as well as fighters from Pakistan's jihadi organizations available to al-Qaida dried up. One interviewee from the Pakistan Taliban, who claimed to have interacted with al-Qaida, noted: “Compared to local jihadi organizations, al-Qaida was in dire need of manpower.”172 Another interviewee stated that by late 2013 an unspecified number of Pakistani jihadis who had originally come from Karachi had left al-Qaida.173 Two important commanders of the Pakistan Taliban, Khalid Sajna and Shehryar Mehsud, “worried about increasing desertions” of full-time fighters.174 Throughout 2013 until the start of Pakistan's military operation against the TTP, known as Zarb-e-Azb, in June 2014, a number of Mehsud Taliban cells surrendered to the ISI.175

From 2012 to 2014, al-Qaida's political relationships with armed groups in Waziristan worsened significantly. Its “cold war” with the Haqqani network came out into the open, with many leaders criticizing Siraj Haqqani, the leader of the network, for the harm inflicted on al-Qaida by drone strikes.176 Within al-Qaida, relationships between the senior leadership and various subgroups also deteriorated. Al-Qaida's Pakistani cadres argued incessantly with the Arab leadership over operational strategy, a lack of direction, and the leadership's extreme operational security measures. According to one interviewee, al-Qaida leader Aiman-al-Zawahiri's 2013 pledge to create an al-Qaida branch on the Indian subcontinent, with expansive operational goals for the region, was in part a response to the growing sense of frustration expressed by Pakistani cadres.177

Finally, the Pakistan Taliban's political ties continued to fray. In the words of one interviewee, “Hakimullah felt he was under siege; he did not trust the Mehsud halqa; he also grew wary of the Punjabi factions.”178 Such distrust led to the splintering of the Pakistan Taliban into three distinct groups: one led by Shahryar Mehsud, another by Khalid Sajna, and a third by Asmat Muawiya. After Hakimullah's killing in November 2013, some of his group's surviving cadres joined forces with Shahryar Mehsud in an organization called Karwan-e-Hakimullah (the Caravan of Hakimullah). Others aligned themselves with Maulvi Fazlullah and Khalid Sajna.179

As they had from 2008 to 2011, the anticipatory effects of U.S. drone strikes caused enormous harm to both al-Qaida and the Pakistan Taliban from 2012 to 2014; multiple interviewees highlighted challenges to conducting operations, maintaining bases, and finding new recruits.180 One Pakistan Taliban interviewee noted, “The senior leaders struggled to contact their subordinate commanders due to frequent flying of drones over the area … When [Hakimullah Mehsud] came to Miramshah, he would often quit meetings, avoid contact until drones flew overhead. Like him, other leaders avoided interacting and travelling with drones flying.”181 When asked to clarify why drones caused such shifts in behavior, the interviewee added: “Once a man from London told amir sahib that data on his eyes, hair, dress, cap, etc. had been recorded and that he could be traced anywhere with this information. He even advised him not to look upward to the sky, as he could be traced by the saved data of his eyes.”182

Al-Qaida's and the Pakistan Taliban's political relationships in the region suffered from both kinetic and anticipatory effects. From the perspective of al-Qaida's leadership, the killing of senior leaders empowered junior commanders, who were harder to manage. Assertive junior commanders made operational calls, managerial changes, and basing decisions, which engendered friction with more senior members.183 Such agency problems metastasized into in-group trust issues, with many leaders claiming that the newly empowered cadres had “sold out” to the CIA and the ISI.184 Similar dynamics also contributed to the Pakistan Taliban's political fragmentation. Senior leaders, including Hakimullah Mehsud, felt constrained in making political decisions. One of the two captive civilian interviewees mentioned earlier recounted conversations with senior commander Khalid Sajna about the political challenges stemming from the increasing use of drones: “Sajna would often come eat with me. He once said there are many days now when I have no cash, and I have nothing to pay my people.”185

Kinetic and anticipatory effects directly fueled in-group discord. The sidelining and subsequent breakaway of the Punjabi Taliban faction in 2013 resulted from severe in-group trust issues. Hakimullah Mehsud was aware of the ISI's overtures to the Punjabi Taliban, but disarray in his own group prevented him from sanctioning the faction's leader, Asmat Muawiya. As one TTP interviewee remarked, “Amir sahib [Hakimullah Mahsud] was told by a number of other leaders that Asmat Muawiya is, in fact, an ISI agent and that you should not assign him the important position of the commander of TTP in Punjab. Amir sahib could not give much attention to this.”186

Two mechanisms contributed to the degradation of al-Qaida and the Pakistan Taliban that my theory does not predict. First, as both groups in North Waziristan struggled to find new safe havens, they increasingly became targets. Until 2010, both remained wary of relocating to other districts of FATA or Afghanistan because of a lack of reliable local hosts. A 2009 Pakistani military operation known as Rah-e-Nijat prevented the Pakistan Taliban from moving training centers, lodging facilities, and IED factories to South Waziristan. Al-Qaida seriously contemplated moving to eastern Afghanistan, but the U.S. surge of forces from 2010 to 2012 ruled out a wholesale move.187 Second, the rise of the Islamic State in Iraq and Syria added to al-Qaida's challenges in Waziristan. Its emergence captured the attention of many cadres in Waziristan, particularly those who differed with bin Laden and Zawahiri on the revival of the caliphate; al-Qaida's soft approach toward Iran, and Shia populations, more generally; and tacit acceptance of the political authority of nation-states.188 Multiple interviewees implied that some members defected to the Islamic State, but none provided details.

Alternative Explanations for U.S. Counterterrorism Effectiveness

In this section, I evaluate two alternative explanations that scholars have proposed for understanding counterterrorism effectiveness: one group-type, the other civilian-centric. Based on the available evidence, I process trace select necessary implications for each explanation to be valid.189

Group-type theories, which predict that group attributes determine a group's survival strategy, its ability to recover in the face of targeting, or both, cannot account for the trajectory of effectiveness for the 2008 to 2014 period. These attributes include embedded social bases, the quality of the group's recruits, the degree of group institutionalization, links to transnational resistance networks (including jihadi networks), and the experience of prior conflict.

Neither al-Qaida nor the Pakistan Taliban possessed all of these attributes. Al-Qaida employed better methods for attracting a steady stream of foreign fighters and high-quality recruits, using ideological appeals to attract motivated jihadis from Pakistan. Pakistani factions of al-Qaida such as Badar Mansoor, Jundullah, and the 313 Brigade drew significant support from Waziristan's Mehsud, Dawar, and Wazir tribes. Al-Qaida also was a highly institutionalized organization.190 Some of its factions had experience fighting strong states in Egypt, Iraq, Indian Kashmir, Libya, and Jordan.191 Al-Qaida also had deep ties to global Salafi-jihadi networks. On the other hand, the Pakistan Taliban employed poorer recruitment methods, using cash and tribal linkages as opposed to strong ideological appeals. Its level of institutionalization paled in comparison to al-Qaida's. Its main factions did not have significant experience dealing with state repression. The Pakistan Taliban's only advantage relative to al-Qaida was its social base: the deep tribal network in Waziristan.

Despite their differences, both al-Qaida and the Pakistan Taliban struggled in the face of U.S. targeting. I did not find evidence of al-Qaida's superior organization helping the group survive the increasing rate of U.S. drone strikes or recover through adaptation. The Pakistan Taliban, too, fractured like al-Qaida, although it exhibited less dysfunction than al-Qaida during certain periods. For example, it maintained a reasonable operational trajectory and a sizable number of bases in 2008–09. From 2010 to 2014, however, the Pakistan Taliban's trajectory continued downward, as the group failed to make meaningful progress toward restoring its pre-2008 growth trajectory. There is support for the group-type view to the extent that al-Qaida tried to develop sophisticated defensive measures to manage its vulnerability to drone strikes, but that appears not to have stopped its degradation.192

Is the Pakistan Taliban's superior trajectory, compared to that of al-Qaida from 2008 to 2009, compatible with a group-type explanation? Not necessarily. That trajectory was less a function of its type and more a function of the variation in the targeting preferences of the counterterrorism operation. U.S. and Pakistani officials noted that the main focus of the targeting in that period was al-Qaida, not the Pakistan Taliban.

Civilian-centric theories also cannot account for the trajectories of al-Qaida and the Pakistan Taliban from 2008 to 2014. A set of civilian-centric theories emphasizes a hearts-and-minds strategy as the key to effectiveness.193 One test of these theories is to determine whether the United States implemented such a strategy and the level of civilian security during the campaign. All available evidence suggests that the United States made no effort either directly or through its Pakistani partners to improve civilian security in the targeted regions. Instead, with al-Qaida and the Pakistan Taliban searching for spies who were purportedly tipping off the CIA/ISI, civilian security seemed to worsen from 2008 to 2011.194 Al-Qaida and the Pakistan Taliban created two units, Lashkar-e-Khorasan and Saif-ul-Furqan, for the purpose of counterespionage. Civilian interviewees noted that the units engaged in excesses and were widely feared.195

Another way to assess a civilian-centric explanation is to examine the main sources of intelligence available to the counterterrorism campaign. For this civilian-centric account to hold, tips from civilians should be the main source of intelligence.196 I found only limited support of this. As I have argued, intelligence collection was a function of the L&S. Two ISI officials who worked with the CIA until 2011 on intelligence gathering described the role of civilian informants as marginal, claiming that the CIA depended primarily on technology and cash incentives to members of armed groups.197 There are limits to my sources on this point, but their assessment aligns with an internal assessment by the U.S. government's intelligence, surveillance, and reconnaissance task force in two other counterterrorism theaters: Somalia and Yemen.198 That assessment emphasizes tactics such as document exploitation and a variety of technological tools as the key sources of intelligence.

Another civilian-centric theory suggests that counterterrorism creates re-sentaient, especially when it involves the killing of civilians, and thus more local recruits for the targeted groups.199 The recruitment trajectories of al-Qaida and the Pakistan Taliban do not align with this view, however. There is substantial evidence of civilian harm caused by U.S. drone strikes in Waziristan, although its exact level is unclear.200 Yet, as discussed earlier, multiple interviewees from al-Qaida and the Pakistan Taliban denied a surge in recruitment in response to drone strikes; more dominant explanations centered on manpower shortages and desertions starting in 2008. One al-Qaida operative tellingly noted, “When I see the news that drone strikes have helped the militants to recruit more people, I consider it false analysis … [both] al-Qaida and Pakistan Taliban had serious manpower shortages.”201


This article has argued that the effectiveness of counterterrorism campaigns is best explained by variation in specific tactical, organizational, and technological capabilities of the counterterrorism state, captured by the Legibility and Speed-of-Exploitation System. The case of U.S. counterterrorism in Pakistan illustrates the utility of the concept of L&S in understanding counterterrorism effectiveness. It shows that U.S. counterterrorism, backed by a surge in capabilities subsumed under the L&S following 2008, carried out an increasing number of drone strikes and pushed both al-Qaida and the Pakistan Taliban toward similar paths of collapse. As one Taliban leader stated, “Drone strikes had a major impact … it was not Zarb-e-Azb [Pakistan's military operation to clear North Waziristan starting June 2014] that shattered Talibanization but drone strikes. They are the real cause of our downfall. Although jihad can never be eliminated as it will continue till the day of judgment, but the drone strikes greatly weakened the Taliban [and al-Qaida].”202

These findings have two significant implications for U.S. policymakers. First, counterterrorism is a powerful option for degrading armed groups in safe havens and weak states. Such degradation appears to be a strategic priority for U.S. policymakers in many conflicts. For the Barack Obama administration, for instance, scaling back al-Qaida's capability to execute plots against the U.S. homeland and reducing the threat of al-Qaida and the Pakistan Taliban to a nuclear-armed Pakistani state were major strategic priorities. The Obama administration largely achieved these goals.

Second, counterterrorism can play a greater role in U.S. grand strategy than many scholars and some policymakers recognize. U.S. policymakers are confronting the challenge of responding to threatening armed groups in the face of growing domestic opposition to manpower-intensive military deployments and resurgent great power competition. My findings support the view that U.S. policymakers have a viable military alternative to manpower-intensive deployments. With the right mix of technology and local partner cooperation, the U.S. government can keep its armed group foes in check.

The promise of counterterrorism should not be overstated, however. My work suggests three critical challenges facing policymaking. First, counterterrorism can entail serious political costs. My theory is agnostic on the national and transnational effects of such campaigns, but such effects are plausible. The U.S. campaign in Pakistan, for example, was deeply unpopular in Pakistan nationally and transnationally.203 It complicated the broader U.S. approach to Pakistan.204 Other major campaigns, such as that against the Islamic State, have not been as divisive, at least transnationally.205 Future research needs to examine when such adverse political effects are likely and the circumstances in which policymakers prefer the gains of counterterrorism over their negative political effects.

Second, policymakers need to recognize that not all counterterrorism deployments are alike and that poorly resourced campaigns are likely to be ineffective. Much of the debate about counterterrorism has focused on whether it is more successful than counterinsurgency, not on what kind of counterterrorism is optimal.206 My work suggests that while counterterrorism strategies can be effective, realizing requisite capabilities for a large-scale campaign remains challenging. It requires substantial domestic political support for investments in technology and tremendous latitude for selecting targets. It also requires extensive cooperation from local partners. For example, a robust counterterrorism strategy in Afghanistan in 2009–10 would have been enormously resource intensive. The geographic scale of the insurgent challenge in Afghanistan was much greater than in Pakistan, and the capability of the Afghan government to support the effort was very weak. The U.S. government would have struggled to implement it the way it did in Pakistan.

Finally, counterterrorism can be effective at the battlefield level even when the counterterrorism state does not provide protection to civilians. Thus, policymakers need to reconsider approaches to minimize civilian harm. Current approaches suggest that civilian protection is essential for effectiveness.207 These approaches rest on the belief that the strategic logic of civilian protection will compel states to minimize civilian harm. My theory and findings offer a contrasting perspective: if states master many of the capabilities identified under the L&S, they may achieve substantial battlefield efficacy despite collateral damage. Given this deeply troubling implication of my study, civilian security should be decoupled from efficacy in doctrine, policy, and human rights discourse. Policymakers must protect civilians for purely normative rather than utilitarian reasons.


For comments on earlier versions of this article, the author thanks Nicholas Anderson, Kara Camarena, Austin Carson, Irina Chindea, Dara Cohen, Martha Crenshaw, Lesley-Ann Daniels, James Fearon, Dotan Haim, Sana Jaffrey, Stathis Kalyvas, John Mearsheimer, Dylan Moore, Scott Sagan, Aqil Shah, Paul Staniland, Rachel Tecott, Harold Trinkunas, Jeremy Weinstein, Kevin Weng, and Amy Zegart. He also thanks participants at the 2018 CISAC (Center for International Security and Cooperation) Social Science Seminar at Stanford University; the 2017 Harvard International Security Conference; the 2017 University of Chicago Program on International Politics, Economics, and Security; the Institut Barcelona Estudis Internacionals’ 2018 Symposium on Micro-Dynamics of Armed Conflicts; conference participants at the 2017 annual meeting of the American Political Science Association, the 2018 annual convention of the International Studies Association; and the anonymous reviewers. He is especially grateful to the many interlocutors who participated in or assisted with the study. The names of many of interviewees have been withheld to protect their safety. Finally, the author thanks the Committee on Southern Asian Studies and the Division of Social Sciences at the University of Chicago for financial support that helped fund the fieldwork for this project.



I define “counterterrorism” as the use of force by a powerful state, including the use of airpower, surveillance technologies, special operators, local allies, or a combination thereof, to target armed groups in safe havens and weak states without taking direct political control of the region of operations.


Views on the limits of counterterrorism were central to the policy debate in 2009 and 2010 on the Afghan troop surge. See Fred Kaplan, “CT or COIN?” Slate, March 24, 2009, http://www.slate.com/articles/news_and_politics/war_stories/2009/03/ct_or_coin.html. Andrew Exum notes that “an orthodox [counterterrorism] campaign is almost certainly destined to fail spectacularly in Afghanistan and Pakistan.” Exum, “On CT vs. COIN” Small Wars Journal, 2009, http://smallwarsjournal.com/jrnl/art/on-ct-vs-coin. For major accounts that are skeptical of counterterrorism's battlefield efficacy, see Michael O'Hanlon and Bruce Riedel, “Plan A-Minus for Afghanistan,” Washington Quarterly, Vol. 34, No. 1 (Winter 2011), pp. 123–132, at p. 127, doi:10.1080/0163660X.2011.538308; Audrey Kurth Cronin, “U.S. Grand Strategy and Counterterrorism,” Orbis, Vol. 56, No. 2 (Spring 2012), pp. 192–214, doi:10.1016/j.orbis.2012.01.004; Michael J. Boyle, “The Costs and Consequences of Drone Warfare,” International Affairs, Vol. 89, No. 2 (January 2013), pp. 1–29, doi:10.1111/1468-2346.12002; Mark Mazetti and Scott Shane, “As New Drone Policy Is Weighed, Few Practical Effects Are Seen,” New York Times, March 21, 2013; Gen. John P. Abizaid (U.S. Army, ret.) and Rosa Brooks, “Recommendations and Report of the Task Force on U.S. Drone Policy,” 2nd ed. (Washington, D.C.: Stimson, April 2015); David Kilcullen, Blood Year: The Unravelling of Western Counterterrorism (New York: Oxford University Press, 2016), pp. 53–66; Robert A. Pape, “Don't Overestimate Drone Strikes’ Power to Kill High Value Targets” (Washington, D.C.: Cipher Brief, September 10, 2017); and Jacqueline L. Hazelton, “Drone Strikes and Grand Strategy: Toward a Political Understanding of the Uses of Unmanned Aerial Vehicle Attacks in U.S. Security Policy,” Journal of Strategic Studies, Vol. 40, Nos. 1–2 (February 2017), pp. 68–91, doi:10.1080/01402390.2016.1196589. Important exceptions include Austin Long, “Small Is Beautiful: The Counterterrorism Option in Afghanistan,” Orbis, Vol. 54, No. 2 (Spring 2010), pp. 199–214, doi:10.1016/j.orbis.2010.01.010; and Daniel Byman, “Why Drones Work: The Case for Washington's Weapon of Choice,” Foreign Affairs, Vol. 92, No. 4 (July/August 2013), pp. 32–43.


See, for example, Boyle, “The Costs and Consequences of Drone Warfare,” p. 18; David Kilcullen and Andrew Exum, “Death from Above, Outrage Down Below,” New York Times, May 16, 2009; James Cavallaro, Stephan Sonnenberg, and Sarah Knuckey, “Living under Drones: Death, Injury, and Trauma to Civilians from U.S. Drone Practices in Pakistan” (Stanford, Calif.: International Human Rights and Conflict Resolution Clinic, Stanford Law School; New York: Global Justice Clinic, New York University School of Law, September 25, 2012), pp. 133–134; and Gregory D. Johnson, testimony before the U.S. Senate Foreign Relations Committee, January 20, 2010, 111th Cong., 2nd sess., https://www.foreign.senate.gov/imo/media/doc/johnsentestimony100120a1.pdf. For a theory of how resentment, anger, and a desire for agency in the face of repression may encourage civilians to join an armed group, see Elisabeth Jean Wood, Insurgent Collective Action and Civil War in El Salvador (New York: Cambridge University Press, 2003).


O'Hanlon and Riedel, “Plan A-Minus for Afghanistan,” p. 127. The notion that civilians need to be provided protection, governance, aid, or some combination thereof for intelligence gathering is seminal to counterinsurgency theory. See, for example, Eli Berman, Joseph Felter, and Jacob Shapiro, Small Wars, Big Data: The Information Revolution in Modern Conflict (Princeton, N.J.: Princeton University Press, 2018). See also the literature review in Eli Berman and Aila M. Matanock, “The Empiricists’ Insurgency,” Annual Review of Political Science, Vol. 18 (May 2015), pp. 443-464, doi:10.1146/annurev-polisci-082312-124553.


This literature identifies group attributes as causal variables of insurgent trajectories in face of counterinsurgency, counterterrorism, or both. See, for example, Jeremy M. Weinstein, Inside Rebellion: The Politics of Insurgent Violence (New York: Cambridge University Press); Jenna Jordan, “Attacking the Leader, Missing the Mark: Why Terrorist Groups Survive Decapitation Strikes,” International Security, Vol. 38, No. 4 (Spring 2014), pp. 7–38, doi:10.1162/ISEC_a_00157; Austin Long, “Whack-a-Mole or Coup de Grace? Institutionalization and Leadership Targeting in Iraq and Afghanistan,” Security Studies, Vol. 23, No. 3 (August 2014), pp. 471–512, doi:10.1080/09636412.2014.935229; Paul Staniland, Networks of Rebellion: Explaining Insurgent Cohesion and Collapse (Ithaca, N.Y.: Cornell University Press, 2014); Monica Duffy Toft and Yuri M. Zhukov, “Islamists and Nationalists: Rebel Motivation and Counterinsurgency in Russia's North Caucasus,” American Political Science Review, Vol. 109, No. 2 (May 2015), pp. 222–238, doi:10.1017/S000305541500012X; Evgeny Finkel, “The Phoenix Effect of State Repression: Jewish Resistance during the Holocaust,” American Political Science Review, Vol. 109, No. 2 (May 2015), pp. 339–353, doi:10.1017/S000305541500009X; and Pape, “Don't Overestimate Drone Strikes’ Power to Kill High Value Targets.”


For an assessment on U.S. intelligence gathering, see Long, “Small Is Beautiful,” pp. 210–211.


Johnson, testimony before the U.S. Senate Foreign Relations Committee.


Aqil Shah, “Do U.S. Drone Strikes Cause Blowback? Evidence from Pakistan and Beyond,” International Security, Vol. 42, No. 4 (Spring 2018), pp. 47–84, doi:10.1162/isec_a_00312.


Both the Bureau of Investigative Journalism and Airwars provide estimates of civilian harm by the U.S. government. See Bureau of Investigative Journalism, “Drone Strikes in Pakistan” (London: Bureau of Investigative Journalism, n.d.), accessed September 22, 2018, https://www.thebureauinvestigates.com/projects/drone-war/pakistan; and “Iraq and Syria: Civilian and ‘Friendly Fire’ Casualties from International Military Actions,” n.d.), accessed September 22, 2018, https://airwars.org/civilian-casualty-claims. See also Azmat Khan and Anand Gopal, “The Uncounted,” New York Times Magazine, November 16, 2017, https://www.nytimes.com/interactive/2017/11/16/magazine/uncounted-civilian-casualties-iraq-airstrikes.html.


The concept of legibility is drawn from James Scott, Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed (New Haven, Conn.: Yale University Press, 1998). For recent work on legibility, see Melissa M. Lee and Nan Zhang, “Legibility and the Informational Foundations of State Capacity,” Journal of Politics, Vol. 79, No. 1 (January 2017), pp. 118–132, doi:10.1086/688053.


Alexander L. George and Andrew Bennett, Case Studies and Theory Development in the Social Sciences (Cambridge, Mass.: MIT Press, 2005), p. 166.


For a review, see Berman and Matanock, “The Empiricists’ Insurgency.” Studies on drones that take similar approaches include Patrick B. Johnston and Anoop K. Sarbahi, “The Impact of U.S. Drone Strikes on Terrorism in Pakistan,” International Studies Quarterly, Vol. 60, No. 2 (June 2016), pp. 203–219, doi:10.1093/isq/sqv004; and Asfandyar Mir and Dylan Moore, “Drones, Surveillance, and Violence: Evidence and Theory from a U.S. Drone Program,” University of Chicago, 2018.


This approach sidesteps political effects such as host-nation stability or transnational effects. I discuss implications for such effects in the conclusion.


Brian J. Phillips, “Terrorist Group Cooperation and Longevity,” International Studies Quarterly, Vol. 58, No. 2 (June 2014), pp. 336–347, doi:10.1111/isqu.12073; and Tricia Bacon, “Is the Enemy of My Enemy My Friend? How Terrorist Groups Select Partners,” Security Studies, Vol. 27, No. 3 (2018), pp. 345–378, doi:10.1080/09636412.2017.1416813.


This framework is agnostic on the relationship of the targeted armed group to the local population.


The logic of disruption that results from variation in the L&S applies to domestic and external counterinsurgency contexts.


Berman and Matanock, “The Empiricists’ Insurgency,” p. 446; Berman, Felter, and Shapiro, Small Wars, Big Data, p. 55. For conception of selective violence, see Stathis N. Kalyvas, The Logic of Violence in Civil War (New York: Cambridge University Press, 2006), pp. 173–209.


Berman and Matanock, “The Empiricists’ Insurgency,” p. 446.


Charles Edward Callwell, Small Wars: Their Principles and Practice (London: HM Stationery Office, 1903); and Kalyvas, The Logic of Violence in Civil War, pp. 87–110.


Scott, Seeing Like a State, pp. 11–52. Legibility is analytically distant from intelligence. As the repository of knowledge on the population, legibility allows states to generate intelligence.


U.S. Gen. Stanley McChrystal described the importance of rapid exploitation in McChrystal, “It Takes a Network: The New Frontline of Modern Warfare,” Foreign Policy, February 21, 2011, https://foreignpolicy.com/2011/02/21/it-takes-a-network.


Eli Berman, Jacob N. Shapiro, and Joseph H. Felter, “Can Hearts and Minds Be Bought? The Economics of Counterinsurgency in Iraq,” Journal of Political Economy, Vol. 119, No. 4 (August 2011), p. 773, doi:10.1086/661983.


On the importance of technology, see U.S. special forces accounts including Michael T. Flynn, Rich Juergens, and Thomas L. Cantrell, “Employing ISR: SOF Best Practices,” Joint Forces Quarterly, Vol. 50, No. 3 (July 2008), pp. 56–61; Stanley McChrystal, My Share of the Task: A Memoir (New York: Penguin, 2013); and Richard H. Shultz, Military Innovation in War: It Takes a Learning Organization, a Case Study of Task Force 714 in Iraq (MacDill Air Force Base, Fla.: Joint Special Operations University Press, 2016).


Daniel Byman, “The Intelligence War on Terrorism,” Intelligence and National Security, Vol. 29, No. 6 (2014), pp. 846, doi:10.1080/02684527.2013.851876; and William M. Arkin, Unmanned: Drones, Data, and the Illusion of Perfect Warfare (New York: Little, Brown and Company, 2015).


Doug Richardson, “Continuous Clandestine Tagging, Tracking, and Locating (CTTL)” (Tampa, Fla.: U.S. Special Operations Command, 2007), https://www.wired.com/images_blogs/dangerroom/files/Richardson_Continuous.pdf.


On voice-matching technology as discussed in leaked National Security Agency (NSA) documents, see Adolf Cusmario, “Al-Zawahiri: At It Again!” (Washington, D.C.: NSA, March 17, 2006), https://theintercept.com/snowden-sidtoday/4755587-al-zawahiri-at-it-again.


Joseph L. Votel et al., “Unconventional Warfare in the Gray Zone,” Joint Forces Quarterly, Vol. 80, No. 1 (January 2016), pp. 101–109.


Timothy D. Haugh and Douglas W. Leonard, “Improving Outcomes: Intelligence, Surveillance, and Reconnaissance Assessment,” Air & Space Power Journal, Vol. 31, No. 4 (2017), pp. 4–16.


Shultz, Military Innovation in War, p.51.


On dissemination of information, see Byman, “The Intelligence War on Terrorism,” p. 855.


McChrystal, My Share of the Task, p. 269.


Sarah Elizabeth Parkinson, “Organizing Rebellion: Rethinking High-Risk Mobilization and Social Networks in War,” American Political Science Review, Vol. 107, No. 3 (August 2013), pp. 418–432, doi:10.1017/S0003055413000208; Jordan, “Attacking the Leader, Missing the Mark”; Long, “Whack-a-Mole or Coup de Grace?”; Staniland, Networks of Rebellion; Toft and Zhukov, “Islamists and Nationalists”; and Finkel, “The Phoenix Effect of State Repression.”


Intelligence Surveillance Reconnaissance Task Force, “ISR Support to Small Footprint Operations—Somalia/Yemen,” February 2013, p. 6, https://theintercept.com/document/2015/10/14/small-footprint-operations-2-13/#page-1.


Byman, “The Intelligence War on Terrorism,” p. 857.


Daniel Byman, “U.S. Counterterrorism Intelligence Cooperation with the Developing World and Its Limits,” Intelligence and National Security, Vol. 32, No. 2 (2017), p. 152, doi:10.1080/02684527.2016.1235379.


Stephen Biddle, “Building Security Forces and Stabilizing Nations: The Problem of Agency,” Daedalus, Vol. 146, No. 4 (Fall 2017), pp. 126–138, doi:10.1162/DAED_a_00464.


Byman, “U.S. Counterterrorism Intelligence Cooperation with the Developing World and Its Limits,” pp. 152–153.


On determinants of compliance by local partners, see Walter C. Ladwig III, The Forgotten Front: Patron-Client Relationships in Counterinsurgency (Cambridge: Cambridge University Press, 2017); and Barbara Elias, “The Big Problem of Small Allies: New Data and Theory on Defiant Local Counterinsurgency Partners in Afghanistan and Iraq,” Security Studies, Vol. 27, No. 2 (2018), pp. 233–262, doi:10.1080/09636412.2017.1386935.


I offer predictions for high and low L&S, because the empirics fall into these categories.


Pakistan did not undertake unilateral military action in North Waziristan from 2008 to 2014. See Paul Staniland, Asfandyar Mir, and Sameer Lalwani, “Politics and Threat Perception: Explaining Pakistani Military Strategy on the North West Frontier,” Security Studies (forthcoming), doi:10.1080/09636412.2018.1483160.


Alexander George and Andrew Bennett write, “One of the difficult requirements of a before-after research design is that only one variable can change at the moment that divides the longitudinal case neatly into two.” This case plausibly meets that requirement. See George and Bennett, Case Studies and Theory Development in the Social Sciences, p. 166.


The interviewees included members of al-Qaida and the Pakistan Taliban (n = 12), civilian captives of al-Qaida and the Pakistan Taliban (n = 3), civilians from areas proximate to sites of drone strikes (n = 22), journalists in contact with al-Qaida and the Pakistan Taliban (n = 8), Pakistani officials involved in coordinating the U.S. counterterrorism campaign (n = 13), Pakistani police officials involved in counterterrorism operations (n = 2), and former U.S. officials involved in counterterrorism activities in Pakistan (n = 6).


Qisín Tansey, “Process Tracing and Elite Interviewing: A Case for Non-Probability Sampling,” PS: Political Science & Politics, Vol. 40, No. 4 (October 2007), p. 768, doi:10.1017/S1049096507071211. The interviews were conducted under University of Chicago Internal Review Board Protocol 16–0234. The online appendix, at doi:10.7910/DVN/QUKFFH, contains discussion of the interview process, health of data, and fieldwork ethics. I was challenged by issues of interviewee safety, enumerators, and fixers. On the importance of ethics in fieldwork, I took cues from Jesse Driscoll and Caroline Schuster, “Spies Like Us,” Ethnography, Vol. 19, No. 3 (September 2017), pp. 411–4130, doi:10.1177/1466138117711717, especially on the concern of security services seeing research team members as “spies.” I recognize the limits of my data and am open to considering further evidence.


Key sources include the bin Laden letters in Criminal Docket No. 10–19 (S-4) (RJD), United States v. Abid Naseer (Eastern District of New York, February 15, 2015); the history of the Mehsud Taliban and role of al-Qaida in Wazirstan in Mufti Abu Mansoor Asim and Mufti Noor Wali, Inqilab-e-Mehsud (Mehsud revolution) (Birmal, Afghanistan: Maseed Computer Center, 2017); and the book by al-Qaida leader Hikmatullah Lodhi, Drone hamley aur un ka tour (Drone attacks and their solution) (Waziristan, Pakistan: n.p., 2011). I also filed Freedom of Information Act requests with the U.S. government. I am still waiting for responses for many of them, including those sent to the U.S. Director of National Intelligence (DF-2016-00300), Central Intelligence Agency (F-2017–01836), and U.S. Central Command (USCENTCOM 17–0486).


For a map of this area, see the online appendix. The demarcation was initially done on “million-maps” in Pakistan's Military Operations Directorate and ISI headquarters. Author interview with former Pakistani military official, 2016.


Adrian Levy and Catherine Scott-Clark, The Exile: The Flight of Osama bin Laden (New York: Bloomsbury, 2017), p. 222.


Author interview with Pakistani military official, 2016. Hayden's meeting with Musharraf has been referenced in Chris Woods, Sudden Justice: America's Secret Drone Wars (New York: Oxford University Press, 2015), p. 111.


Author interview with former ISI official, 2016.


In the online appendix, I offer a year-wise breakdown to operationalize L&S.


Author interview with former ISI official, 2017; and author interview with U.S. government official, 2016.


Author interviews with former ISI officials; and author interviews with former U.S. government officials in Chicago and Washington, D.C., 2016 and 2017.


Ibid. See, for example, the discussion of constraints on the ISI's signals interception capabilities and cooperation with the U.S. government as observed by an NSA staffer in 2006 in “A Firsthand View of Pakistan and Its SIGINTers” (Washington, D.C.: NSA, May 31, 2006), https://theintercept.com/snowden-sidtoday/4755718-a-firsthand-view-of-pakistan-and-its-siginters.


Author interviews with former ISI and Pakistani military officials, 2016 and 2017.


Robert L. Grenier, 88 Days to Kandahar: A CIA Diary (New York: Simon and Schuster, 2016), pp. 381–382; and author interview with former U.S. government official, 2016.


Author interview with former ISI official, 2017; and Woods, Sudden Justice, p. 102.


Author interview with former U.S. government official, 2017.


Author interview with former ISI official, 2016; and author interview with former U.S. government official, 2016.


Robert M. Gates, Duty: Memoirs of Secretary at War (New York: Vintage, 2013), p. 133; and author interview with former U.S. government official, 2016.


Declassified by U.S. Central Command under a Freedom of Information Act request (2016–04358-F).


Author interview with former Pakistani military officials, 2016 and 2017.


Author interview with former U.S. government official, 2017.


Greg Miller, Julie Tate, and Barton Gellman, “Documents Reveal NSA's Extensive Involvement in Targeted Killing Program,” Washington Post, October 16, 2013.


Author interview with former ISI official, 2016. For a reference to the NSA's one signal interception operations center operated with the support of ISI in Islamabad, called the Combined SIGINT Operations Center, see “The Day the Earth Shook in Pakistan” (Washington, D.C.: NSA, January 1, 2006), https://theintercept.com/snowden-sidtoday/4755482-the-day-the-earth-shook-in-pakistan.


See, for example, discussion of bulk Pakistani cellular call records to the NSA and their use for finding al-Qaida and Taliban members in “Skynet Applying Advanced Cloud Based Behavior” (Washington, D.C.: NSA, 2012), https://assets.documentcloud.org/docurnents/2074625/skynet-applying-advanced-cloud-based-behavior.pdf. The interviewee noted that the records were shared by the ISI's T division with the CIA and NSA. Privacy International mentions deep cooperation between the ISI and NSA in “Tipping the Scales: Security and Surveillance in Pakistan” (London: Privacy International, 2015), p. 6, https://privacyinternational.org/sites/default/files/2018-02/PAKISTAN%20REPORT%20HIGH%20RES%2020150721_0.pdf.


Author interview with former ISI official, 2016.


ISI's support to the forward bases in 2010 and onward remained unclear in interviews.


Author interview with former ISI official, 2016.


Woods, Sudden Justice, p. 114.


Author interview with former U.S. government official, 2017.


Author interview with former Pakistani military official, 2016.


Jonathan S. Landay, “Obama's Drone War Kills ‘Others,’ Not Just al Qaida Leaders” (Washington, D.C.: McClatchy DC, April 9, 2013), https://www.mcclatchydc.com/news/nation-world/world/article24747826.html. Woods notes that this targeting criteria came into effect in January 2008. See Woods, Sudden Justice, p. 111.


For a discussion of these tensions, see Steve Coll, Directorate S: The C.I.A. and America's Secret Wars in Afghanistan and Pakistan (New York: Penguin, 2018), pp. 513–630.


This approach constitutes “analytical process tracing” of the new theory. See, for example, George and Bennett, Case Studies and Theory Development in the Social Sciences, pp. 225–228.


Author interview with former ISI official, 2016; and author interview with al-Qaida operative, 2017.


Author interview with former ISI official, 2016.


Author interview with former TTP operative, 2017.


Author interview with former ISI official, 2016.




Author interview with TTP operative, 2016; and Saleem Shahzad, Inside Al-Qaeda and the Taliban (London: Pluto, 2011), p. 7.


Author interview with police official, 2016.


Author interview with former ISI official, 2016.


Shahzad, Inside Al-Qaeda and the Taliban, p. 9.


Author interview with TTP operative, 2016.


Author interview with journalist, 2016.


Shahzad, Inside Al-Qaeda and the Taliban, p. 18.


Author interview with journalist, 2016. Both outlets were producing material in 2004.


Author interview with Dawar tribal elder, 2016.


Author interview with al-Qaida operative, 2016.


Author interview with former ISI official, 2016.


Author interview with former ISI official, 2016.


Author interview with former ISI official, 2016.


Author interview with Mehsud tribal elder, 2016.






Author interview with former ISI official, 2016; and Asim and Wali, Inqilab-e-Mehsud, pp. 165–167.


Author interview with TTP operative, 2016. Halqa in Pashto signifies a sizable unit.


Author interview with TTP operative, 2016.


Author interview with TTP operative, 2016; author interview with Mehsud tribal elder, 2016; and Asim and Wali, Inqilab-e-Mehsud, pp. 161–166.


Asim and Wali, Inqilab-e-Mehsud, p. 444.


Author interview with Dawar tribal elder, 2016.


Author interview with TTP operative, 2017.


Author interview with former Pakistan military officials, 2016.


Bureau of Investigative Journalism, “Drone Strikes in Pakistan.”




Author interview with former Pakistani military official, 2016; and author interview with former ISI official, 2017.


Author interview with former U.S. government official, 2017.


Author interview with former ISI official, 2016.


Author interview with former ISI official, 2016.


Author interview with former U.S. government official, 2017; and Michael Morell, The Great War of Our Time: The CIA's Fight against Terrorism—from al-Qa'ida to ISIS (New York: Twelve, 2015), p. 130.


Jonathan S. Landay, “U.S. Secret: CIA Collaborated with Pakistan Spy Agency in Drone War” (Washington, D.C: McClatchy DC, April 9, 2013), https://www.mcclatchydc.com/news/nation-world/world/article24747829.html.


Author interview with former ISI official, 2017.


Author interview with former Pakistan military official, 2016.


Author interview with former Pakistan military official, 2016.


Author interview with former ISI official, 2017. On ISI's access to live feed from U.S. drones, see Azaz Syed, The Secrets of Pakistan's War on Al-Qaeda (Islamabad: Narratives, 2015), p. 73.


Author interview with former Pakistan military official, 2016.


Author interview with former TTP operative, 2017.


Author interview with al-Qaida operative, 2016. In addition, see Government Exhibit 429 10-CR-019(S-4)(RJD), in United States of America v. Abid Naseer.


Author interview with former ISI official, 2017.


Author interview with former TTP operative, 2017.


Author interview with TTP operative, 2017.


See Government Exhibit 421 10-CR-019(S-4)(RJD), in United States of America v. Abid Naseer.


Enumerator interview with TTP operative, 2016.


Author interview with journalist, 2016.


Author interview with TTP operative, 2016.


Author interview with TTP operative, 2016.


Author interview with former ISI official, 2016.


Author interview with former Pakistani military official, 2016.


Author interview with al-Qaida operative, 2017.


See Government Exhibit 429 10-CR-019(S-4)(RJD), in United States of America v. Abid Naseer.


Author interview with journalist, 2016.


Author interview with former ISI official, 2016.


Author interview with former ISI official, 2016.


Author interview with TTP operative, 2017; and author interview with TTP operative, 2017.


Author interview with al-Qaida operative, 2016.


Enumerator interview with TTP operative, 2016; and author interview with al-Qaida operative, 2017.


Author interview with former ISI official, 2016.


Author interview with journalist, 2016; and enumerator interview with TTP operative, 2017.


The leaders mentioned included Maulvi Sadiq Noor, Khaliq Haqqani, and Hafiz Saeed. Author interview with journalist, 2016.


Asim and Wali, Inqilab-e-Mehsud, pp. 445–446.


Author interview with former ISI official, 2017.


Author interview with TTP operative, 2017.


Author interview with TTP operative, 2016. In an interview while in Nangarhar prison, al-Qaida operative Sheikh Jafar made a similar assessment. See Faizullah Khan, Durand line ka kaidi—qaid se rehai tak (Prisoner of Durand Line—from imprisonment to release) (Karachi, Pakistan: ZAK, 2016), p. 140.


Author interview with TTP operative, 2016.


Author interview with TTP operative, 2016; another interviewee confirmed the same chain of events but noted that Khabab survived a 2008 strike and died much later, probably in 2010.


I obtained materials on “Dor-ul-Mutafarja” during fieldwork.


Author interview with TTP operative, 2016.


Author interview with TTP operative, 2016.


Author interview with TTP operative, 2016.


Author interview with former TTP operative, 2017.


Author interview with former TTP operative, 2017.


Author interview with TTP operative, 2016; and Lodhi, Drone hamley aur un ka tour, p. 35.


Author interview with TTP operative, 2016.


Author interview with former Pakistani military official, 2016.


See Government Exhibit 421 10-CR-019(S-4)(RJD), in United States of America v. Abid Naseer.


Author interview with TTP operative, 2016.


Author interview with TTP operative, 2016.


Author interview with al-Qaida operative, 2017.


Lodhi, Drone hamley aur un ka tour, pp. 35–37.


See Government Exhibit 421 10-CR-019(S-4)(RJD), in United States of America v. Abid Naseer.


Author interview with al-Qaida operative, 2016.


Lodhi, Drone hamley aur un ka tour, pp. 60–61.


See, for example, details on the availability of bulk cellphone records of major Pakistani telecommunication firms and their use for target acquisition by the NSA as late as 2012, in “Skynet Applying Advanced Cloud-Based Behavior.”


Author interview with TTP operative, 2017.


Author interview with TTP operative, 2017.


Author interview with former ISI official, 2016.


Author interviews with former civilian captives, 2016 and 2017.


Author interview with journalist, 2016.


Author interview with journalist, 2016.


Author interview with former ISI official, 2016; and Lodhi, Drone hamley aur un ka tour, p. 34.


Author interview with former ISI official, 2016.


Author interview with Pakistani military official, 2017.


Author interview with TTP operative, 2016.


Author interview with al-Qaida operative, 2017.


Author interview with al-Qaida operative, 2016.


Multiple North Waziristan–based civilians and journalist respondents.


Author interview with al-Qaida operative, 2017.


Author interview with al-Qaida operative, 2016.


Author interview with Mehsud tribal elder, 2016. He was in contact with Hakimullah Mehsud from 2009 to 2013.


Author interview with TTP operative, 2016.


Author and enumerator interviews with TTP operatives, 2016–17.


Author interview with TTP operative, 2017.


Author interview with TTP operative, 2017.


Author interview with Dawar tribal elder, 2017.


Author interview with al-Qaida operative, 2016. Interviewees highlighted such problems in Jundullah, the Egyptian Takfiri Group of Issa Al Masri, the Turkish Jamaat of Abu Ahmed, and the Roshan Aka faction.


Author interview with former civilian captive, 2016.


Author interview with TTP operative, 2017.


Select leaders moved to Karachi and some al-Qaida commanders relocated to Iraq and Syria. These findings should not be seen as downplaying those effects.


Levy and Scott-Clark, The Exile, pp. 240–241; Abu Jarar Shamsi, “Al-Qaeda of Waziristan: A Testimony from Within,” Dabiq, No. 6 (2014), pp. 40–55; and author interview with al-Qaida operative, 2017.


This constitutes the hoop test of process tracing in David Collier, “Understanding Process Tracing,” PS: Political Science & Politics, Vol. 44, No. 4 (October 2011), p. 825, doi:10.1017/S1049096511001429. If the evidence bears out the implications, it affirms relevance of the theory but does not confirm it.


Jordan, “Attacking the Leader, Missing the Mark,” pp. 22–25.


Author interview with al-Qaida operative, 2017; and author interview with former ISI official, 2017.


Lodhi provides sophisticated guidelines to jihadi leaders on defending their organizations against drone strikes. See Lodhi, Drone hamley aur un ka tour.


O'Hanlon and Riedel, “Plan A-Minus for Afghanistan,” p. 127; and Berman, Felter, and Shapiro, Small Wars, Big Data, p. 55.


Author interview with TTP operative, 2017; and enumerator interview with TTP operative, 2016.


Author interviews with civilians in Dera Ismail Khan, Mir Ali, and Peshawar, 2016 and 2017.


Berman and Matanock, “The Empiricists’ Insurgency.”


Author interview with former ISI official, 2016; and author interview with former ISI official, 2017.


Intelligence Surveillance Reconnaissance Task Force, “ISR Support to Small Footprint Operations—Somalia/Yemen.”


This is a narrow slice of the broader drone blowback thesis. For a review of that thesis, see Shah, “Do U.S. Drone Strikes Cause Blowback?”


The Bureau of Investigative Journalism estimates that 421 to 960 civilians were killed in Pakistan from 2004 to 2014. See Bureau of Investigative Journalism, “Drone Strikes in Pakistan.” Reports by the U.S. Director National Intelligence claim there were 117 civilian casualties outside Afghanistan and Iraq from 2009 to 2016. See, for example, “U.S. Counterterrorism Strikes outside Areas of Active Hostilities” (Washington, D.C.: Office of the Director of National Intelligence, 2016), https://www.dni.gov/files/documents/Newsroom/Press%20Releases/DNI+Release+on+CT+Strikes+Outside+Areas+of+Active+Hostilities.PDF. Steve Coll notes that civilian casualties from select U.S. drone strikes fueled an “imagery of callous atrocity.” Coll, The Directorate S, p. 535.


Author interview with al-Qaida operative, 2016.


Author interview with Taliban operative, 2016. For a similar assessment and language on overall effects of drone strikes on jihadi organizations, see Asim and Wali, Inqilab-e-Mehsud, p. 444.


Pew Research Center, “Widespread Opposition to Drones” (Washington, D.C.: Pew Research Center, July 11, 2014), http://www.pewglobal.org/2014/07/14/global-opposition-to-u-s-surveillanceand-drones-but-limited-harm-to-americas-image/pg-2014-07-14-balance-of-power-0-01/. On transnational effects, see Boyle, “Costs and Consequences of Drone Warfare”; and Abizaid and Brooks, “Recommendations and Report of the Task Force on U.S. Drone Policy,” pp. 10–11.


Hillary Clinton, Hard Choices: A Memoir (New York: Simon and Schuster, 2014), pp. 159–188.


Richard Wike, Bruce Stokes, and Jacob Poushter, “Global Publics Back U.S. on Fighting ISIS, but Are Critical of Post-9/11 Torture” (Washington, D.C.: Pew Research Center, June 23, 2015), http://www.pewglobal.org/2015/06/23/global-publics-back-u-s-on-fighting-isis-but-are-critical-of-post-911-torture/.


See, for example, Michael J. Boyle, “Do Counterterrorism and Counterinsurgency Go Together?” International Affairs, Vol. 86, No. 2 (March 2010), pp. 333-353, doi:10.1111/j.1468-2346.2010.00885.x.


See Cavallaro, Sonnenberg, and Knuckey, “Living under Drones”; Abizaid and Brooks, “Recommendations and Report of the Task Force on U.S. Drone Policy”; and Joseph H. Felter and Jacob N. Shapiro, “Limiting Civilian Casualties as Part of a Winning Strategy: The Case of Courageous Restraint,” Daedalus, Vol. 146, No. 1 (Winter 2017), pp. 44–58, doi:10.1162/DAED_a_00421.