Predators and Peace: Explaining the Failure of the Pakistani Conflict Settlement Process in 2013-4

ABSTRACT Did US drone strikes cause the unravelling of the Pakistani conflict settlement process between the government and the TTP in 2013-14? In answering this question, we present strong, fieldwork-based evidence that the effects of leadership decapitation, civilian casualties, and loss of legitimacy and credibility as a negotiation partner by both the government and the TTP interacted in the context of specific social, political and cultural characteristics of a tribal society. We find that drone strikes ‘produced’ some of these factors, but not all, which allows us to conclude with four concrete policy recommendations for rethinking the use of drones.


Introduction
Between 2004 and 2014, successive US Administrations launched more than four hundred strikes by Predator drones in the Waziristan agencies of Pakistan's Federally Administered Tribal Areas (FATA). 1 The frequency of the strikes increased after 2008, rising sharply between 2010 and 2014. 2 By the end of 2014, up to 3,096 fatalities were reported, of whom up to 960 were civilians. 3 Consequently, an anti-drone narrative gained importance in Pakistan, sustained by print and electronic media, politicians, and a crosssection of civil society, which was also aimed at those within the Pakistani military and civilian establishment suspected of colluding with the US. One of the claims made in this narrative is that US drone strikes have been a significant barrier to achieving a peaceful settlement of the conflict between the Pakistani government and the Tehrik-e-Taliban Pakistan (TTP). 4 The main reasons in support of this argument are that the strikes help to create more terrorists than they eliminate (Hudson et al. 2011, Cavallaro et al. 2012; that they foster jihadist sentiment and militant recruitment (Hali 2012, Awan 2013 and that drone strikes 'consistently kill Pakistani civilians, which anger the population, and prompt revenge attacks from the militants' (Bergen and Tiedemann 2010). In addition, the killing of the TTP commander Wali-ur-Rehman by a drone strike in May 2013 and of his successor Hakimullah Mehsud in a strike the following October, are considered as having dealt serious blows to the possibility of peace negotiations between the Pakistani government and the TTP (Anon 2012a(Anon , 2013a(Anon , 2013b(Anon , 2013c(Anon , 2013d. This literature, however, leaves a significant gap in our understanding of the Pakistani conflict settlement process in 2013-14. First, much of it predates the period from late-2013 to mid-2014 when serious efforts to begin formal negotiations between the Pakistani government and the TTP were under way. 5 Second, effects on conflict settlement are, more often than not, a peripheral concern at best, with most of the literature considering drone strikes primarily in the context of US counter-terrorism (Bergen and Tiedemann 2011, Boyle 2013, Hazelton 2013, Johnston and Sarbahi 2016, Jordan 2014, Lehrke and Schomaker 2016 and its domestic and international legal foundations or lack thereof (Reinold 2011, McCrisken 2013, Ahmad 2014, Crawford et al. 2017, Trenta 2018. 6 Third, there is almost no literature specifically on the relationship between drone strikes and peace negotiations in Pakistan. 7 Fourth, there is also a gap in the existing literature from a methodological perspective. We do not suggest that US drone strikes generally have had no negative consequences or specifically have not affected the conflict settlement process in Pakistan; however, we contend that the analysis underpinning the anti-drone narrative is too narrow and even misleading in its construction of the US strikes as the primary reason for the absence not only of a negotiated settlement but of negotiations. The existing literature asks about the effect (success or failure of counterterrorism, counter-insurgency, etc.) of one particular cause (drone strikes) and then concludes that the assumed cause has comprehensive explanatory power for the observed effect. We pursue a methodologically different approach by asking about the causes of effects, starting from the observation of the effect (futile attempts to initiate a process of formal peace negotiations in 2013-14 8 ) and then seeking to establish its causes. This leaves open the possibility of a sole cause, such as drone strikes, but allows us to proceed with a theoretically-driven inquiry in which we first establish a list of theoretically plausible causes, as comprehensively as possible, and then investigate them with a view to identifying all those which are relevant. Doing so, we not only establish probable cause-effect relationships but also eliminate causes that existing theories might suggest but which are not relevant in the case of Pakistan. This approach allows us further to establish, in a more nuanced way, the role that drone strikes have played to date in the failed attempts to bring about negotiations on a settlement of the conflict between the TTP and successive Pakistani governments and to elucidate more comprehensively the causes of this failure, including possible interactions between them.
Filling this gap in the existing literature is not only worthwhile from the perspective of empirical knowledge and theoretical understanding, but also from a policy perspective. While we must be careful not to overclaim the generalisability of findings based on a single case study, we present strong evidence of 'multiple and conjunctural causation' (Ragin 1987, p. 23ff.) of the fate of the attempted peace talks in Pakistan in 2013-14, in which the effects of leadership decapitation, civilian casualties, and loss of legitimacy and credibility as a negotiation partner by both the government (primarily as a result of its alleged complicity with the drones campaign) and the TTP (primarily because of its fragmentation) interacted in the context of specific social, political and cultural characteristics of a tribal society. Drone strikes 'produced' some of these factors, but not all. This suggests four concrete policy implications for the impact of abandoning and/or rethinking the use of drones, the conditions under which either of these may have positive effects for a future peace process, and the trade-offs involved concerning other objectives of the drone campaign, especially US counter-terrorism.
In the next two sections, we outline the theoretical framework that guides the subsequent empirical analysis and reflect in more detail on our methods of data collection and analysis. Section 4 offers an evaluation of our hypotheses based on an analytical narrative for the period under consideration, drawing on thirty interviews and a two-part survey, all conducted during fieldwork in Pakistan in between May and September 2014. The interviewees included current and former politicians and members of the Pakistani security services, as well as diplomats, tribesmen, journalists and analysts, whereas four hundred tribal respondents were approached for the two-part survey. In Section 5, we focus on an additional factor identified in the course of our analysisthe role of tribes. Often neglected in discussions of the impact of drones on the peace process in Pakistan, we demonstrate that the destruction of tribal structures, in part caused by the US drones campaign, is a critical element in the understanding of the difficulties of the Pakistani peace process. We conclude in Section 6, emphasising that the US drone campaign in FATA is only one of several factors that, in conjunction with others, have influenced the attempted conflict settlement negotiations between the Pakistani government and the TTP in 2013-14.

Theoretical Framework: Conflict Settlement in a Counter-terrorism/ Counter-insurgency Context
To understand the failure of peaceful conflict settlement efforts (as opposed to the inability of either side to win a decisive military victory), our principal points of reference are theories of conflict settlement, and specifically the issue of what accounts for the possibility of initiation, conduct, and conclusion of negotiations on a settlement. The first dimensionnegotiation initiationis particularly pertinent in the case of Pakistan and for the period under consideration and it is the primary puzzle we attempt to solve in this study. Why did attempts to start formal negotiations fail over a 12-month period in 2013-14, and why could contacts not progress beyond exploratory talks? Given the context of drone strikes which killed senior TTP figures in May and October 2013, these questions offer a critical point of intersection: a longstanding, and largely inconclusive, debate on the efficacy of decapitation strategies in the counter-terrorism/counter-insurgency literature (e.g., Carvin 2012) that overlaps with a strand of the broader conflict settlement literature emphasising the importance of leadership and leaders' decisions to engage in, or withdraw from, exploratory settlement negotiations (e.g., King 1997).
Without trying to resolve the debates between and within the different schools of thought on leadership decapitation, we use them to examine plausible ways in which such an approach might affect the prospects of initiating negotiations.
On the one hand, some scholars argue that drone strikes are effective in degrading capability and morale of the organisations whose leadership is targeted (Wilner 2010), limiting planning and organisational effectiveness (Anderson 2013), and significantly weakening terrorist groups (Byman 2013). The reasoning underpinning these arguments is that weaker groups are more likely to be willing to negotiate as their prospects of winning an outright military victory diminish. Powerful counter-arguments contend that (1) the use of drones for targeted killing leads to the ascent of less restrained terrorist leaders who are likely to increase civilian targeting (Abrahms and Mierau 2015) and the number of attacks (Lehrke and Schomaker 2016), and (2) organisational decline is less likely following leadership decapitation, especially in separatist and religious organisations (Jenna Jordan 2009).
This debate on decapitation overlaps with the consideration of unintended consequences. Captured in the notion of 'blowback', the argument here is that drone strikes enhance the ability of terrorist organisations to recruit more members, foster an increase in revenge strikes against military and/or civilian targets, undermine the legitimacy of domestic governments, and spread and intensify anti-American and anti-Western attitudes among populations and governments alike, 9 contributing to the phenomena of 'home-grown terrorism' and 'foreign fighters'. This debate is largely inconclusive and findings are often highly context-dependent, even though supporters of blowback appear more numerous (Hudson et al. 2011, Cronin 2013, Boyle 2013) than those who see some utility in drones for select purposes (Johnston and Sarbahi 2016) and those who find little or no evidence to sustain a case for blowback (Swift 2012, Shah 2018. 10 For our argument, the significance of the debates is how drone strikes have affected the TTPgovernment peace process in Pakistan. Several explanations would appear plausible: (1) Leadership decapitation eliminated credible negotiators on the TTP side, increased difficulties for new/remaining leaders to discuss negotiations in a safe environment, required the movement to save face by withdrawing from a nascent peace process, and hardened existing and new members' and new leaders' resolve to seek a military solution. (2) Civilian casualties and damage to infrastructure increased support for the TTP, swelled its ranks with new, vengeful recruits, and made it more difficult for 'moderates' to (continue to) make a case for a negotiated settlement. (3) Amid the alleged complicity of the Pakistani state in the U.S. drone campaign, the government lost legitimacy and credibility as a partner for the TTP in any potential peace process.
While we may find evidence in our empirical analysis to substantiate one or more of these hypothesised causal mechanisms linking drone strikes to the failure to initiate formal negotiations between those on both sides with the authority to make and implement decisions, this would not necessarily equate to a comprehensive explanation. As we are not seeking the effects of a particular cause, but rather the causes of a specific effect, we also need to identify additional plausible explanations and confirm or disconfirm them. For this, we turn back to the broader conflict settlement literature. A number of different arguments have been put forward to explain the failure to initiate negotiations for a conflict settlement. These are related to the conflict parties, their relationship with each other, and the broader external context of the conflict and its potential settlement process.
Existing literature has established the importance of identity (Rothman andOlson 2001, Aggestam 2002), arguing in particular that incompatible identities, denying each other's right to exist on equal terms, make negotiated settlements difficult to achieve. They prevent the reciprocal recognition of the legitimacy of each party's claims (Heraclides 1989, Svensson 2007, Bell 2014) and make it seem less likely that the other side will make meaningful concessions and compromises to a party acknowledged as an 'equal' in negotiations (Albin and Druckman 2012). This reduces the sense of opportunity associated with negotiations (Kelman 1982) and limits parties' motivation to seek, and optimism to find, a negotiated settlement (Pruitt 1997).
The willingness of conflict parties to enter into negotiations on a settlement is frequently also associated with their relative material capabilities. In particular, the notions of the mutually hurting stalemate and the ripe moment for seeking a settlement have become prominent: parties that are relatively equally balanced in their military capabilities and unable to achieve a decisive military victory become locked in a situation in which they must seek alternatives to a military strategy (Zartman 2001). This then creates a specific window of opportunity (Bapat 2005, Ruhe 2015. A similar focus on capability has led to arguments that insurgent groups which are more evenly matched with government forces in military terms are more likely to force negotiations, and even more likely to do so if they expect to be able to extract desired concessions (Cunningham et al. 2009) and are less fearful of a government's future defection from a negotiated settlement (Clayton 2013). In contrast, fragmented or fragmenting conflict parties have an adverse impact on the initiation (and successful conclusion and implementation) of settlement negotiations because they produce multiple veto players with potentially diverging interests (Cunningham 2006), who are therefore less flexible in their approach to negotiations (Walch 2016).
The possibility of initiating settlement negotiations has also been set in the broader external context of the conflict in question. One school of thought emphasises the importance of third-party mediators that help (or nudge) parties to the negotiation table (Bercovitch and Kadayifci 2002). Equally important are guarantors who can offer credible protection against defection (Walter 1997, Pearson et al. 2006, increasing the probability of parties engaging in negotiations as they see a greater likelihood in the long term of retaining concessions obtained in a settlement.
This brief overview of the conflict settlement literature on factors relevant to the initiation of negotiations suggests additional plausible explanations for our case. The reasons why the stage of formal negotiations between the top TTP and the Pakistani decision-makers was never reached in 2013-14, despite offers and acceptances from both sides on different occasions, may also be found in one or more of the following: (4) There is no mutually hurting stalemate between the parties, with either the government expecting to be able to defeat the Taliban militarily, or the Taliban considering itself strong enough to hold out and gradually improve its bargaining position over time. (5) Both sides may be too fragmented internally to make credible, and thus acceptable, offers to initiate formal negotiations. The opposing side sees little likelihood in achieving and retaining desirable concessions in a negotiation process, in part because some factions may deny the other party's legitimacy. (6) The settlement process lacks external mediators and guarantors that could bring the parties to the table and contribute to sustaining a negotiated settlement.
To summarise, our aim in this study is to understand the reasons for the inability of the Pakistani government and TTP to progress to sustained formal conflict settlement negotiations in 2013-14 despite publicly stated commitments to do so. Based on broader engagement with existing literature on why attempts to initiate settlement negotiations succeed or fail, beyond the widely held belief that drone strikes are the primary cause of failure in the Pakistani case, we have established several plausible explanations that can be empirically examined.

Approach
In light of this theoretical framework and our overall question about the failure to move beyond exploratory talks to initiate formal peace negotiations in Pakistan in 2013-14, several methodological considerations follow. In determining our approach, we had to consider our data requirements to decide our methods of data collection and analysis. With six distinct plausible explanations for the failure of conflict settlement efforts in 2013/14, we had distinct data requirements for each. We formulate these as observations that we would expect to make if a potential explanation is valid on its own or in conjunction with other mechanisms. By way of illustration how we operationalise this approach, 11 let us briefly consider our first potential explanation. We formulated this as follows: 'Leadership decapitation eliminated credible negotiators on the TTP side, increased difficulties for new/remaining leaders to discuss negotiations in a safe environment, required the movement to save face by withdrawing from a nascent peace process, and hardened members' and new leaders' resolve to seek a military solution.' If this were true, we would expect to make a number of observation, including that assassinations of insurgent leaders is followed by a breakdown in negotiations, Taliban withdrawal from negotiations, or break-off of pre-negotiation engagements. In addition, we might see that no Taliban negotiation team is nominated and that no joint negotiation platform among Taliban factions emerges. Moreover, attacks on government forces and alleged collaborators would likely increase. In terms of communicating their response, we would expect that Taliban rhetoric becomes more aggressive and that their stated opposition to a negotiated settlement would increase. Evidence of these observations could be gathered from two principle data sources. The first would be relevant interlocutor statements in interviews and the surveys we have carried out, for example pointing out difficulty of remaining senior leaders to meet and discuss negotiations. Second, we would expect media coverage and secondary/grey literature reporting a lack of meetings among senior Taliban, more fractious internal debates and lack of consensus among Taliban factions.
Given the specific circumstances in our particular case, for this and all other potential explanations, we have to consider the feasibility of various data collection methods, assessing and balancing what data is available and accessible and then evaluating its credibility. Working in a fragile and conflictaffected environment like Pakistan and on a politically highly sensitive and emotionally charged topic has potentially significant methodological implications. 12 Data is relatively limited, its credibility not always beyond doubt, and accessibility presents another frequently encountered problem. This affects more or less all types of sources, from surveys and interviews to print and electronic media, from official documents and statements to socalled grey literature and existing scholarship. Even where interlocutors are willing to share information, they may be exposed to retribution. Interviewers are also potentially at risk, and mitigating the risk for them might increase that for interlocutors and/or make it impossible for them to provide useful information, 13 while interviews with others are simply not possible. 14 To mitigate the resulting difficulties, we adopted several strategies. Initially, we built as comprehensive as possible a deep and shared understanding of the case among the three researchers, drawing on the knowledge and understanding of a local and an external expert and of a comparativist with no prior case knowledge, 15 a comprehensive survey of secondary literature, and a detailed examination of media sources and the grey literature, 16 especially as they pertained to the time period under consideration.
Following the development of our theoretical framework, we reached agreement on the range of issues to pursue in semi-structured interviews and in a survey. 17 We then split tasks among the three researchers involved in the project: the local expert conducted the interviews and the survey (allowing us to make the best use of local networks), while the comparativist analysed and subsequently coded the interviews 18 and developed a first draft of the main argument based on the available empirical evidence, including findings from the survey. In parallel, the local expert and the external case expert developed an analytical narrative covering the period under observation, based on the local expert's identification of themes and patterns in the interviews as well as the survey results and other primary material and secondary sources. This way, the development of our final argument based on two separate analyses enabled us to avoid possible confirmation bias, while simultaneously making use of the different types of expertise we all brought to the table and to triangulate data obtained from our interviews, survey and other primary and secondary sources. This triangulation, in turn, helped us to weigh different sources of data in terms of how much significance we would assign to the evidence they provided, i.e., the larger a cross-section of sources supported the same claim the more weight we gave it. 19 This also helped us to avoid being drawn into particular perspectives from just one or two subsets of our interlocutors. For example, we were careful to make sure not to over-rely on the perspectives of political and military insiders in our own analysis, using the insights they offered as data points in our own analytical narrative rather than adopting their perspectives. 20

Assessing the Causes for the Failure to Initiate Sustained Formal Talks
Based on the analysis of our primary and secondary data, we can now offer a more detailed discussion of the plausibility of each of these explanations on their own and in various combinations with each other. Our evidence suggests a nuanced argument: the escalation of the US drone campaign between 2009 and 2013 was not the primary cause of the failure to pursue a political settlement. Neither was it the primary cause for the entry of the Pakistani Government and the TTP into talks in autumn 2013. Strikes did not block or encourage negotiations. 21 The assassinations of TTP leaders in 2013 did not derail progress towards the eventual opening of exploratory talks in early 2014. Conversely, the suspension of drone attacks did not prevent the collapse of the process later in the year. While the strikes and the narratives constructed around them had an effect, they did so as factors within a broader and more complex political, social and cultural context.
In line with Hypothesis 1, on decapitation strikes as an obstacle to the initiation of negotiations, most of our interlocutors saw a clear connections. Almost one-third, across the spectrum of people with whom we talked, pointed to a negative impact of decapitation including the limits put on TTP leaders' freedom of movement. As one of our interviewees, a former Pakistani Ambassador, put it: The Taliban couldn't arrange meetings or talk to each other because of [their] fear [of the drones]. At the very start, when the drone strikes began, they used to be careful during the day time and wandered in the nights. Now, when the technology was developed, they thought these rascals can see us even in the night, so do not get closer. So they couldn't get together to reach a consensus on whether to hold a dialogue with the Pakistani government or not. 22 This view was corroborated by tribesmen noting that 'drone attacks restricted their movement' 23 and that the constant operation of drones 'affects the process of negotiations because the TTP leaders cannot congregate in one place without fear of being killed.' 24 These fears were further heightened by the killing of TTP leader Hakimullah Mehsud on 31 October 2013, days before formal negotiations between the sides were to begin. There was widespread agreement among interlocutors that this could have derailed the political process even before it had started. 'Before the PML-N was even in parliament they killed the protalks Wali-ur-Rehman and later Hakimullah who had also shown willingness to talk,' pointed out a former diplomat hailing from Waziristan, arguing that the killings impacted the possibility of conflict settlement. 25 A FATA journalist summarised the public perception: 'The drone strike that killed Hakimullah was fundamentally detrimental for the peace talks. Had that drone strike not happened, there could have been peace in the area.' 26 But this interpretation fails to take account of the actual chronology of events, in which other developments had forestalled the opening of negotiations. A Pakistani intelligence official and expert on FATA reminded us: The initial peace initiative had broken down almost two weeks before Hakimullah Mehsud's killing because the government had rejected the TTP's demand to release three Taliban commanders. The militants wanted to nominate them as their representatives in the dialogue. These commanders were in military custody and had confessed to killing security personnel in Swat in 2008 and to their links with foreign intelligence agencies. There was no way the military was going to release them. 27 Even before Hakimullah Mehsud's death, the in-fighting threatened to fracture the insurgency, with disagreements between the Punjabi Taliban and the TTP Shura over proposed dialogue. 'It's a tribally structured society,' argued a former diplomat from FATA, 'so people always suspect each other of being a government beneficiary or spy.' 28 Further tensions were fed by claims, circulated by the US State Department, that the May 2013 strike on Wali-ur-Rehman was enabled by intelligence from a source in the Hakimullah faction of the TTP (Murshed 2013).
It might be argued that the drone strikes were catalytic through fragmentation preventing the credible offering and accepting of negotiations. But, if so, they had a paradoxical effect when they were used by the Pakistani Government and military. Far from refraining from any political engagement, Islamabad sought to exploit the divisions withinin the words of Interior Minister Chaudhry Nisar Ali Khan (2013) -'a loose conglomerate of at least 37 different jihadist groups based in the tribal areas' to explore talks with factions willing to engage in the search for a political settlement.
The tactic achieved other goals. While the TTP council was forced to devote its energies to mediating a ceasefire amongst its own ranks (Yusufzai 2014), Pakistani intelligence agencies could exploit the conflict with an eye on outside actors, as an official explained: This divide-and-rule policy of the state exploits tribal and clan-loyalties and familial rivalries . . . this is aimed at keeping militants involved in intra-group wrangling until other interested regional actors [India, Iran, and Afghanistan] show their cards once the US draws down. 29 The US pursuit of a counter-terrorism policy heavily reliant on drones might contribute to the opening of negotiations, but not because Washington envisaged this as a fundamental strategy. Instead, Pakistani actors had to seize upon the strikes, within their counter-insurgency strategy to create suitable conditions through pressure on the Taliban leadership.
Interlocutors with direct knowledge of the process rejected the idea that the fate of the exploratory talks primarily rested upon the status of the drone strikes. As one of them noted with respect to freedom of movement: TTP commanders and their committee kept changing the venue of internal meetings in North Waziristan amid continued flights of US drones . . . With 12 to 15 drones constantly conducting surveillance, it was difficult for them to congregate.
He added about primary responsibility: The suspension of drone strikes [between December 2013 and June 2014] did have some positive effect and helped our efforts to talk to some extent . . . but the main obstacle was that the military did not agree to any exchange of prisoners . . . and so the talks floundered in the very first phase. 30 Instead, just as drone strikes played a role in the process towards political discussions once they were used by local actors, they helped create the conditions dooming the discussions to failuredependent on whether those local actors seized upon them. The process of fragmentation that, with Islamabad's exploitation, brought the TTPwith its '37 different jihadist groups'to the table also inhibited any progress beyond initial explorations. No single faction was strong enough to sustain talks, let alone deliver any future agreement. During the negotiations, the Sajna and Shahryar groupswhose rivalry dates 'to the time of their great grandfathers' 31began fighting over control of the TTP (Khan 2014), with an eventual split in the group when the non-Mehsud Mullah Fazlullah was named TTP chief in May 2014 (Hashim 2014). By then, the government was pushing for exploratory talks about peace deals with individual groups as members of its peace committee concluded that reaching an 'all-inclusive agreement' with the TTP might not be possible (Yousaf 2014a). The Interior Minister argued, 'each [group is] fully independent in decision making . . . holding peace talks with a single group would be useless. ' Significantly, the government dismissed claims that drones had any decisive influence one way or another, arguing that the TTP was never serious about the discussions and only sought a pause to regroup. Moreover, in the words of a member of the government peace committee, the government view was that drone strikes alone do not force the militants to give up . . . There is the fatigue syndrome . . . diminishing number of volunteers . . . and the military was in readiness for the operation. 32 Thus, contrary to Hypothesis 4, any assumption that there was a mutually hurting stalemate in place by early 2014 appears unfounded. Instead, a Pakistani intelligence officer indicated that the military took advantage by blaming the insurgents for the breakdown of exploratory talks while preparing for the North Waziristan offensive in June 2014. According to a senior security official (quoted in Awan 2014), the political track gave way to anticipation of future gains on the battlefield: I think the security services have achieved half of their target with the assistance of government [peace committee] emissaries. The TTP is now visibly divided into pro-peace and anti-peace groups, no matter what their so-called leaders claim.
In a mirror image of the Government's manoeuvre, the TTP also set aside the American operations and focused on Islamabad's position as they refused to extend a month-long ceasefire in spring 2014. They argued that 'the government has failed to respond positively to Taliban's realistic demands of establishing a peace zone, release of non-combatants and suspension of security forces operation in tribal areas' (quoted in Sherazi 2014). A member of the Mehsud tribe argued that 'the TTP leadership belongs to a particular Pashtun [Mehsud] group that does not take kindly to intimidation . . . Thus, any efforts to pressure them to participate in peace talks through drone attacks tend to have the opposite effect.' 33 The Government and TTP positions were compounded by a perceived split in Islamabad, with the army and intelligence services perceived as significantly less committed to achieving a negotiated settlement than the civilian leadership. Thus, in line with Hypothesis 5, fragmentation was not only a problem on the TTP side, and the Taliban were acutely aware of the mismatch between civilian and military objectives. As a former Ambassador pointed out, 'Even after the suspension of drone attacks . . . [the TTP] were afraid of the Pakistani military forces . . . .That's why they demanded a peace zone.' 34 Another interlocutor, close to the TTP, echoed this, noting that 'the militants know that the Army is the strongest player and it has not been sold on the peace process', 35 while another observed with regard to negotiations that 'the government and military do not see eye to eye on this.' 36 The TTPlinked interlocutor bore out these assessments by talking of the consequences 'if the army is not on board' with the political approach We know, whenever there is a peace process, the first demand of the militants would be the release of the prisoners from the custody of the security personnel. This we should have had anticipated before initiating the peace process that they would ask for this and we would have to give them something as the peace process entails give and take policy. So, when we were aware of that and just started the peace process for the political gains, the process had to fail regardless of the drone factor. 37 Other interlocutors also attributed the collapse of the negotiation process in May 2014, at least in part, to the army. Having lost soldiers in clashes in FATA, as well as in high-profile attacks on military installations, 38 commanders were always suspicious of peace deals that had earlier been used by insurgents to regroup and reorganise, for example in South Waziristan (2004 & 2005) and Swat (2009). Moreover, even as the government attempted to sustain exploratory talks, militant violence continued against military and civilian assets. General Musharraf argued that the TTP, as a close affiliate of Al Qa'eda, 'does not recognise the state as legitimate . . . They want to dismantle the system and impose their own brand of religious governance.' He summed up the military's mistrust by calling the peace process 'nothing but a hoax '. 39 This assumption that the TTP denied the state's legitimacy again points to the fundamental incompatibility in approaches to even exploratory talks between the civilian government and the military and serves as further confirmation for Hypothesis 5 and the problem of fragmentation on the Pakistani side. It also highlights problematic disparities in identity and the extent to which some elements in the military considered it inconceivable to accommodate the TTP withintheir own conception of the Pakistani state.
The approach of General Kiyani, who replaced Musharraf, offered a window for negotiations, but with Kiyani's replacement, General Raheel Sharif, it became clear that the Army was focused on an offensive in North Waziristan. Immediate retaliatory aerial bombings in TTP-controlled areas demonstrated a firm response to any attacks by insurgents during the period of exploratory talks. So even as exploratory talks were launched in 2014, for some in the TTP a full-fledged military operation in North Waziristan was always a foregone conclusion. This, rather than drone strikes, was a more immediate source of concern and added to the sense that the Taliban were never serious about peace negotiations. As former ISI chief Ehsan-ul-Haq argued, drone strikes were never a hindrance, although they were used by the TTP as 'a pretext to stall the peace process . . . [Even] when there were no strikes for five months during the peace process, the TTP did not respond positively and continued carrying out lethal attacks.' 40 Though they were not central to the ultimate failure of the exploratory talks, the US drones campaign, in line with Hypothesis 2, was significant in relation to civilian casualties and damage to infrastructure. According to one prominent anti-drone argument, this would have swelled the TTP ranks with new recruits and side-lined moderates who would have made a case for a negotiated settlement. While discussion of the sidelining of moderates was mainly with concern to the fragmentation of the TTP (see above), many of our interlocutors noted how so-called 'collateral damage' and the psychological stress inflicted upon FATA residents generally increased radicalisation and helped the TTP with recruitment. As a senior army and ISI officer put it to us: A tribal Malik linked the drones campaign to the cultural context in which they occurred, 'We live according to Pashtunwali and we can never forget this . . . .It will only increase the younger generation's urge for exacting revenge' 43 , while a sitting Senator noted, 'Our argument has always been that when drone strikes . . . kill an entire family and just one youngster survives, then do you think he cares whether suicide attacks are Islamic or un-Islamic? He will do everything in his power to take revenge . . . ' 44 A journalist observed that 'it is due to these drone attacks that you find suicide bombers in the area in hundreds' 45 . Similar sentiments were expressed by a range of other interlocutors. 46 One interlocutor summarised during a group discussion: 'The concept of badal (the right to avenge an injustice) against an attacker, which is the hallmark of our society, comes into action . . . and because we see the Pakistani state as complicit in drone attacks people do join TTP to take revenge.' 47 Our survey results also offer a perspective on blowback. Of all respondents, 30% noted that the primary impact of drone strikes on the population in the FATAs was a rise in terrorism. The proportion of respondents holding this view increased dramatically from the June survey (23%) to the September survey (37%). Similarly, 29% of all respondents considered drone attacks as most detrimental to daily life and stability in the FATAs among several other choices. Again, there is an interesting shift over time: the figure was 38% in the June survey, dropping to just 21% in the September survey. At the same time, in June 37% of respondents considered the Pakistani army's military campaign as the key factor undermining daily and stability, whereas it was 56% in September. While we cannot infer strong causal links from these survey data, they offer circumstantial evidence in connection with our interview data that some blowback did occur as a direct result of both the US drones campaign and the Pakistani military campaign. 48 This finding is in marked contrast to Shah's study on blowback from the drones campaign in Pakistan. Based on a larger interview sample than ours, he finds no evidence for 'the claim that drones provide a recruitment card for militant organizations' (Shah 2018, p. 82). We have no reason to doubt Shah's findings. 49 The contradiction between his and ours may simply be due to different interview samples and both may adequately reflect genuinely held beliefs by interlocutors, indicating different perceptions and indicating the range of opinion in Pakistani society. But our concern is with the effect that drone strikes, as one plausible factor, may have had in the context of formal peace negotiations. From this perspective, a perception of a blowback effectamong, for example, government and army officials or tribal malikswhether it is correct or not, may be as detrimental as its actuality. This perception changes potential calculations about the prospects of moving beyond exploratory talks. 50 As we suggested in Hypothesis 3, a government seen as condoning the US drone campaign would lose public legitimacy and not be seen as a credible negotiation partner by the Taliban. This collusion was a theme in many interviews. The general consensus among interlocutors was that 'without the blessings of the governments and the military, these drone strikes cannot take place'. 51 What is important for our argument, however, is how interlocutors thought about the effects of collusion. A tribal Malik from North Waziristan added, 'As long as the Pakistani state is seen to be fighting America's war there will be distrust and it is this [distrust] that hinders dialogue.' 52 A senior military official hailing from FATA explained: 'Drones impact conflict resolution by widening mistrust between the state and society . . . the role of intelligence agencies is paramount in this warfare and these operations are by definition shadowy . . . mistrust is rampant (and) hurts any peace process.' 53 Our survey results also support the notion of a loss of government legitimacy as a result of the complex interplay between the US drones campaign and domestic factors. In both surveys, roughly one-quarter of all respondents considered post-9/11 cooperation between Pakistan and the US as the key detrimental factor in causing instability in FATA. 54 Taken together with the fact that 31% and 26% of all respondents, respectively, saw the reason for successive Pakistani governments' cooperation with the US on drones in US financial incentives and Pakistani government corruption, this adds further credence to the notion of a government lacking legitimacy among key actors in society. Moreover, 62% respondents in the June survey considered drone attacks as primarily indicating government duplicity (condemning them in public, while secretly cooperating with the US). However, following the launch of the military operation in North Waziristan, this number drops by 24 percentage points. Instead, the share of respondents who see drone strikes exposing the militants' vulnerability increases to 80 % (from a mere 20% in June). While the magnitude of this shift in opinion may be a function of the larger proportion of more recently displaced people in the September survey, the more general point about the complex interaction of different factors, with drone strikes only one among them, still holds.
However, while collusion clearly was a powerful theme in the projection of a negative impact of the US drone campaign on the peace process, the overall picture was complex, with many more factors at play. A member of the government peace committee noted: The TTP played on the argument that the Pakistani government is too weak and reliant on the US and it can't even ask them to stop drone attacks so how can it talk peace with us . . . So once the government convinced the US to halt attacksbetween December 2013 and June 2014this militant propaganda was challenged. 55 While this might suggest a clear, singular causal connection, our interviews indicate that drone strikes were never the sole, and perhaps not even primary, issue for the TTP. 56 The militants' initial broad demandsstopping the US attacks, severing ties with Washington, and introducing constitutional reforms to make Sharia the supreme law of the landwere narrowed during the exploratory talks to withdrawing the Army from a designated peace zone and allowing free movement to TTP leaders, releasing 'non-combatant prisoners' and ending the crackdown on Taliban members across the country (Mohmand 2014, Yousaf 2014b). These were eventually the (pre-) conditions on which the sides reluctantly agreed after the suspension of the US drones campaign. Yet these talks broke down in May 2014, whilst this suspension was still in force, making it at best a necessary but not sufficient condition for both the initiation and sustenance of negotiations. The US drones campaign and the narratives constructed around it were factors in both the start and collapse of the exploratory talks, 57 but as complementary rather than mutually exclusive explanations. The impact of drones was mediated through different mechanisms, some of which predated both the US drones campaign and the peace process. As a result, there were mixed outcomes. This was perhaps most evident in the complex interplay of decapitation, fragmentation, and the start of exploratory talks in early 2014 and their collapse in May that year.
However, in the course of our analysis, we also identified another critical factor in the fate of the peace process in 2013-14: the role of the tribes in North Waziristan.

The Role of Tribal Actors in the Peace Process
Given only scant attention in existing literature, 58 local tribes were not just a passive observer of the interaction between the government/military, the TTP, and the US. However, more than half of our interlocutors in the interviews emphasised the profound weakening of the tribal system and its traditional cultural values challenged by the rise of an obscurantist version of Islam. The key issue here is that prior to the destruction of tribal fabrics, there were local mechanisms in place to deal with violence between tribes and violence brought to the tribal areas from outside, such as the jirga system. 59 The catalyst for this change was the CIA-ISI use of religion in the 1980s Afghan-Soviet Jihad. It was accelerated by the influx of Taliban fighters from Afghanistan into North Waziristan after 9/11; the inability and unwillingness of the Pakistani government and military to protect the tribes; and the casualties caused by the US drones campaign and Pakistani military actions. The elimination of many elders forced others to emigrate, further weakening the system. As the Taliban and Al-Qaeda fighters used Waziristan as a sanctuary, they thus targeted the already weakened tribal structures: The foreign fighters, who arrived in FATA . . . did not adhere to these customary laws and instead imposed their own extremist ideology on society . . . . [With] their brutality against the weak, kidnapping and murder . . . this has been the main blow to tribal structures. 60 All these factors interacted in complex ways, but overall they contributed, through the weakening of the tribal system, to increasing empowerment of the militants on the one hand, and to growing distrust of the Pakistani state on the other. 61 In our survey, 47% of all respondents ranked Pakistani military operations as the key factor in undermining the stability of tribal structures. This deprived the government of an important ally in the fight against foreign fighters and indigenous insurgents and in any negotiations with the TTP. Unsurprisingly, this number was almost 20 percentage points higher in the September 2014 survey after the launch of the Zarb-i-Azb military operation in North Waziristan (37% in June compared to 56% in September). Almost 30% of all respondents ranked drone strikes as the key factor contributing to the loss of stability, but with a declining share in the September survey (38% in June compared to 21% in September).
By 2013, the tribes saw a Pakistani Army, overstretched from Swat to South Waziristan, which was making convenient agreements with local Taliban. Already in contravention of the pledge of autonomy given to the tribes when they joined the state of Pakistan in 1947, 62 the military was now believed to be acting against insurgents only half-heartedly. There was a growing conviction among tribesmen that the state was either unwilling or unable to protect its citizens from 'terrorists'. Accordingly, argued a tribesman, people saw both the militants and the state as persecutors. 63 Another interviewee similarly observed: The introduction of the army and the assumption by the army of the functions assigned to the civil administration created a mindset in the tribal area that these troops have come to safeguard the American interest in this area and to promote American agenda in this area. This mindset created an insurgency in the tribal area. 64 In addition, a sitting Member of Parliament explained how the military's peace deals further entrenched the TTP in North Waziristan. 'The Taliban came here and disrupted the established jirga system, killed tribal elders and contracted forced marriages . . . [while] the army . . . destroyed the [administrative] authority of the district Political Agent.' Instead of resorting to the traditional methods conflict resolution, he argued further, a senior army commander approached the militants directly. This lack of familiarity with tribal culture was perceived as the army's weakness to the benefit of the terrorists. 65 This growing distrust of the military in particular, was reinforced by a leader of a religious political party representing the TTP in the Sharif initiated peace talks: We do not absolve the Taliban . . . .They have played a big role in destroying the [tribal] system . . . .But one must also remember that when Musharraf deployed military in tribal areas, he violated a long-respected agreement between the tribes and the state of Pakistan . . . This act in itself was detrimental to the tribal system of Jirga and Masharan [elders] because now the army commanders began to hold direct talks with the militants . . . .They were not from the tribal areas and they had no knowledge or experience of the tribal way of resolving disputes or of the psyche of the tribesmen. 66 To strengthen their position, insurgents particularly targeted the tribal jirga (council of elders) to further erode traditional methods of conflict resolution. The military's deployment into FATA, their deals with the Taliban, and their inability to protect the tribal system created conditions in which the TTP's parallel justice system became entrenched, further increasing the gulf between the tribes and the Pakistani state. According to a tribesman from North Waziristan: They give you the option of trial by Rawaj, tribal custom or Sharia, Islamic law. Should one decide the trial by Rawaj, then jirga is held and a compromise is reached; however, in case of Sharia, local Taliban scholars decide the matter.
Taliban have the power to make the guilty party obey the decisions of their jirga. That is why people normally choose Taliban over the state. 67 The US drones campaign further exacerbated these negative effects. Before 9/11, there were thousands of elders across the FATA who were elected from their villages to the jirga, in what the Pashtun see as a representative system of governance. 68 Between 2007 and 2014 the TTP killed hundreds of these eldersdifferent estimates give different numbers 69thereby 'wiping out the traditional leadership', according to refugees from North Waziristan (S. Shah 2014, A. Shah 2018. Targeting the elders to strengthen their control of the region, militants used drone attacks as a pretext for reprisals, eliminating who could have served as mediators. One tribesman summarised this counter-productivity of US drone strikes even when they hit intended victims: 'When US drones target [TTP] leadership, the Taliban also kill our leaders and elders. This is simple cause and effect.' 70 Accordingly, 31% of respondents saw a rise in violence as the primary impact of drone strikes in Waziristan. 71 Using the drone attacks as a reference point, the TTP exploited the tribal value of badal (revenge) not only to create terror but also to sow confusion, pre-empting tribal outrage against atrocities. The murder of elders and state functionaries could be committed and justified in the name of exacting revenge against the military operations of a government that was an ally of the US. As the American drone strikes increased in 2009, Al Qaeda created a group of fighters who tracked down and eliminated suspects (Farooq and Kakakhel 2014). A tribal student said, 'The TTP kills anyone they think is a spy who may have given intelligence for drone strikes.' 72 Even if they wanted to cooperate with state institutions such as the military, tribes now found it far more difficult, as a tribal interviewee explained: Tribal elders [either] no more exist [or] a great number has left the area. The rest who are still living there, are playing both ways as they keep the Taliban and the military happy so that they can survive . . . . If I dare to go and meet any military commander then it is sure that on my way back I will be killed or interrogated by the Taliban. People who work for the army must have links with Talibanthere is always this double game. 73 The erosion of the tribal system and loss of many respected elders in the crossfire between the Taliban, the Pakistani military, and the US drones meant that there was no traditional powerful intermediary between the state and the TTP leadershipan intermediary who could have pressured or incentivised the insurgents to enter into, and remain in, peace negotiations with the government.
The insurgents' hand was further strengthened by the perceived high level of civilian casualties from drone attacks, with one tribal interviewee estimating that '95% of those killed' were not fighters. 74 More than half of respondents -54%said that drone strikes kill more civilians than terrorists. 75 A tribesman cited a March 2011 drone attack on a jirga in North Waziristan, which killed about 40 tribal elders who had gathered to resolve an internal dispute over resources. 76 Other incidents occurred because of the flaws of human intelligence: a tribesman from North Waziristan said information was not always correct, and sometimes faulty intelligence was provided to the US for financial gains or to settle scores leading to the loss of innocent lives. 77 In the 2014 Survey, 65% of tribal respondents did not see drones as effective at disrupting terrorist activities, and 54% did not believe drones were useful in intelligence gathering and reconnaissance. 78 The erosion of the traditional tribal system destroyed local capacity to deal with violence and destabilisation. One of our interviewees put the blame for this squarely on the Pakistani state, 'We who are from FATA were demanding from the beginning don't deploy over there, no shelling . . . no drones . . . There is a tribal way to tackle them (militants).' 79 A tribal elder emphasised that the opportunity to resolve the conflict through the established tribal system was missed because of US and Pakistani interventions, and particularly the failure of the Pakistanis state to protect the tribal system: The tribesmen have their own way of resolving their problems within the parameters of Pashtunwali. The US and Pakistan have made matters worse by directly intervening . . . The people of Waziristan should have been allowed to tackle the problem in accordance with their traditions and the state should have provided the backing and shown the will to protect us . . . this would have given us the incentive to resolve or problems with the militants directly and the militants too would have listened to us knowing that the state is right behind us. 80 The ill-judged actions of the Pakistani state in the tribal areas after 9/11 thus created the conditions for, and sustained, an insurgency while simultaneously undermining local capacity to contribute to any peace process.
The social, political, and cultural dynamics of the FATA, playing out in the wake of the destruction of core elements of the tribal system, were a major factor that amplified the negative effects of the US drones campaign on the peace process. They accentuated the problems of the decapitation strategy (Hypothesis 1) and increased fragmentation (Hypothesis 5), causing significant levels of civilian casualties and damage to infrastructure (Hypothesis 2). In the context of a severely weakened customary system of conflict resolution and in the absence of a mutually hurting stalemate (Hypothesis 4), hardliners were strengthened and had significant opportunities to further undermine the legitimacy of the Pakistani state (Hypothesis 3). It is this understanding of the interactions between the different factors that allows us to offer a meaningful explanation of the initiation and collapse of the short-lived peace process in Pakistan in 2013-14 and of the role of the US drone campaign within it.

Conclusion
Our analysis of the government-initiated peace process between September 2013 and June 2014 highlighted that drone attacks were neither the driving force behind the initiation of the government-TTP talks nor the primary reason for the failure of this exploratory dialogue about formal peace negotiations.
This fills a significant gap in the existing literature on the effects of drone strikes in three dimensions. First, to our knowledge, this is the first comprehensive scholarly analysis of the relationship between drone strikes and the inconclusive peace process in Pakistan in 2013-14. Second, and based on this empirical analysis, we contribute a more nuanced theoretical understanding of the impact of drone strikes on a peace process as part of multiple and conjunctural causes, rather than as a singular cause with comprehensive explanatory power.
Third, by reconnecting our analysis to our theoretical discussion about conflict settlement in a counter-terrorism/counter-insurgency context, and conscious of the limited generalisability of our findings, we can point to several potentially significant policy implications.
Our analysis found that the effects of leadership decapitation, civilian casualties, and both the government's and TTP's fragmentation and loss of legitimacy and credibility as a negotiation partner interacted in the context of specific social, political and cultural characteristics of a tribal society to ultimately doom the prospects of a successful peace process in 2013-14. From a policy perspective, it is important to reiterate that drone strikes 'produced' some of these factors, and exacerbated others, but not all. Thus, the first policy-relevant implication of our analysis is that simply stopping drone strikes is unlikely to be a step that would be sufficient for a different outcome, and that was is required instead is a more comprehensive and sustained conflict settlement strategy that considers these factors and their interaction in the specific context of Pakistan.
Second, our analysis thus also suggests that a more carefully calibrated drone policy, as part of an equally refined settlement strategy, would have better prospects of success in Pakistan, and conceivably beyond. There is some evidence of this already: US drone strikes continue, but are much reduced in number and have caused fewer civilian casualties, 81 the Pakistani government has restored at least some of its authority in FATA (Anon 2015), paving the way for the resettlement of 94% of the displaced in their homes (Zia 2017). Additionally, the differential effect of the drone campaign over time suggests that it is not necessarily incompatible with conflict settlement, provided that careful consideration is given to the timing and sequencing of pressure (e.g., strikes) and incentives (e.g., strikes being suspended). This is an area in which further research might elicit more concrete policy recommendations, including about pressure can remain credible without undermining the prospects of negotiations (because of the consequences of leadership decapitation and blowback) and how incentives can sustain positive momentum towards and during talks rather than being abused by either side to improve its position ahead of renewed violent confrontation.
Third, and directly related to this second implication, our findings indicate that any positive impact on the peace process of changes in drones policy critically depends on factors related to the local conflict parties. The TTP has reduced capability, with their infrastructure in North Waziristan destroyed, the loss of more than 2000 fighters according to military sources (Bennett-Jones 2017), and another military operation, Radd-ul-Fasad, inside Pakistan's urban centres affecting recruitment. Intra-group rivalry has brought more fragmentation, exemplified by the split between TTP and its affiliate Jamaatul-Ahrar in November 2017 (Ahmad 2017). However, the TTP and affiliated groups continue to launch attacks from the border areas of Afghanistan, albeit diminished in frequency and target value (Stancati and Totakhil 2015). The trajectory implied in this is not one of an imminent resumption of negotiations. The Pakistani state appears to consider a military defeat of the TTP insurgency still to be possible, while the TTP has retained sufficient capacity to resist, yet remains too fragmented internally to become a credible negotiation partner should the government decide to change course.
Regardless of the military balance between the sides, the fundamental disruption of the social, political and cultural fabric of tribal structures in FATA has yet to be undone. Deployment of the military in the area for the first time since 1947 and the Army's assumption of administrative functions had already begun the process of undermining the traditional authority of the Jirga, but the CIA-operated drone campaign, driven by US rather than local security interests, furthered the erosion. The covert nature of the campaign also undermined the credibility of the Pakistani state, which found it difficult to acknowledge to its citizens the level of co-operation with the US in the face of anti-US public sentiment. This exacerbated state-society distrust to the benefit of the militants in FATA and beyond. Merely restoring the 'writ of the state' in FATA will not be sufficient to re-establish the trust necessary for a sustainable conflict settlement process. Rather the Pakistani state needs to rebuild its legitimate authority through concrete action so that the residents of the now province of Khyber Pakhthunkhwa do not perceive themselves as second-class citizens less worthy of the state's protection and service delivery.
Fourth, any conceivable changes to drones policy must be calculated against the probability of these local factors and balanced against unavoidable trade-offs with regard to other policy objectives currently pursued through drones (especially counter-terrorism). As long as local conditions are such that with or without drone strikes no settlement process is conceivable, there are few if any incentives for the U.S. to abandon what is arguably its currently best policy option in a protracted counter-terrorism campaign in and beyond Pakistan, motivated by homeland security more than by conflict settlement. Put differently, only once local actors have created a positive dynamic for conflict settlement that demonstrably assuages US concerns about terrorism is it likely that joint US-Pakistani efforts towards a carefully re-calibrated drones policy can be meaningfully aligned with a conflict settlement strategy.
Finally, even if the counter-terrorism driven US drone campaign could be reconciled with a Pakistani strategy to advance peace negotiations, both sides have yet to achieve a level of strategic coherence to enable a reciprocal recalibration of their respective strategies. Here, too, future research is necessary to identify more clearly the conditions under which what kind of strategic coherence could be established and what it would need to entail.
Until then, the likely continuation of pursuing parallel and, at least partially incompatible, strategies by the two presumptive allies will prolong the catalytic effects of drone strikes that we have highlighted in our analysis. Any possibility of conflict settlement between the Pakistani state and the TTP, thus, is likely to remain elusive for some time to come. Notes 1. FATA has been incorporated into the province of Khyber Pakhthunkhwa in 2018. As our focus is mainly on the period 2013-14, we will refer to the areas as FATA throughout. . This view is widely held in journalistic accounts (e.g., Ahmad 2013, Masood and Mehsud 2013, Peralta 2013 and within parts of the scholarly community (e.g., Shaw 2013, Boyle 2013, Fair and Hamza 2016. For a more nuanced perspective, see Williams (2010), International Crisis Group (2013) and Fair (2014). 5. In addition to already cited sources, see, for example, also Tiedemann (2011), Matulich (2012), Plaw and Fricker (2012), and Boyle (2013a). In addition to already cited sources, see, for example, also Tiedemann (2011), Matulich (2012), Plaw and Fricker (2012), and Boyle (2013).
6. There is also a related literature on the new spatial dimensions of drone warfare (e.g., Adamson 2016), on drones in the context of just war theory (e.g., O'Driscoll 2018), on the determinants of public support for drone strikes (e.g., Kreps and Wallace 2016), and on the decision calculus of states to employ airpower (Allen and Machain 2018), all of which are beyond the scope of our argument. 7. The few scholarly discussions of this relationship also predate the period of interest here. See, for example, Zaidi (2010) and Ahmed (2013). For a sophisticated critique of the latter, see Fair (2014), pp. 226-229. 8. Talks between the TTP and the government remained limited to one preliminary meeting between nominated intermediaries on 6 February 2014 and a single round of direct talks between representatives of both sides on 26 March 2014. 9. Note, too, in this context work by Silverman which demonstrates that preexisting bias (e.g., pro-vs. anti-Taliban) shapes perceptions of events (such as drone strikes) and these, in turn, reinforce attitudes (such as anti-Americanism). See Silverman (2018). 10. To our knowledge, the most comprehensive empirical assessment of blowback to date is Shah (2018), who finds no convincing evidence of blowback at the local level. 11. We detail our operationalisation procedure for all potential explanations, corresponding expected observations, and suitable data and data sources in Appendix A. 12. For more on these specific methodological implications of such fieldwork are discussed, among others, by Fujii (2010), Malthaner (2014), Desrosiers and Vucetic (2018), and Knott (2015). 13. One of our interlocutors, for example, cut short the interview, saying, 'Please don't ask me to elaborate . . . We are sitting in this (Intelligence) office . . . they know I am talking to you . . . Please only quote me on the drones not the army or the TTP. This is all I have to say.' Another started the interview insisting, 'Please don't name me in your study . . . my family is in [XXX] and the TTP can label us as spies. It is up to them whether they shoot you or behead you.' 14. Given the security situation and the high levels of mutual distrust, interviewing TTP officials was simply not feasible. We circumvented this problem by interviewing people with knowledge of the Taliban positions on negotiations and triangulated this with official Taliban statements and insights provided by a range of different local experts and analysts. 15. The local expert is a Pakistani academic with a PhD from the UK, the external case expert and the comparativist are both UK-trained and based academics. 16. There are many definitions of grey literature, we rely on the so-called Prague definition as elaborated by Schöpfel (2011): 'Grey literature stands for manifold document types produced on all levels of Government, academics, business and industry in print and electronic formats that are protected by intellectual property rights, of sufficient quality to be collected and preserved by library holdings or institutional repositories, but not controlled by commercial publishers i.e., where publishing is not the primary activity of the producing body.' 17. See Appendix B for interview questions and interviewees, as well as for details on the survey. 18. We used NVivo 11 for Windows for coding and running all data-related queries that form the basis of the descriptive statistics we provide. Coding followed procedures recommended by Bazeley and Jackson (2013). The data was coded in three rounds: the first round looked for evidence to support any of the six hypotheses individually; the second round, following a structuring of the knowledge and understanding gained, looked for connections between the different explanations; and the third round then employed specific text searches to further illustrate and evidence these connections. By way of illustration, the first round, for example, identified mistrust of the army and the Government, collusion between the Pakistani and US Governments, and the destruction of the tribal system as frequently mentioned factors negatively affecting the possibility of a conflict settlement. In round two, a common connection found between these three themes was anti-Americanism, and therefore the third round employed a text search in transcript for 'America' and 'US' to probe, and code, this connection in more detail. 19. For example, 19 of our 30 interlocutors from across the whole range of our sources mentioned psychological stress among residents of the tribal areas as one of the negative consequences of drone use (for strikes as well as surveillance). This effect has also been widely reported in the media and the grey literature. 20. For example, we are careful in our analysis of the impact of drone strikes on fragmentation and the difficulties that this created for negotiations (see below), showing that there was agreement on the fact of fragmentation, but different perceptions of its consequences within the political and military establishment and offering our analysis of this by emphasising that fragmentation created a limited opening for exploratory talks but inhibited any progress beyond such an initial stage. 21. As we demonstrate further below, this was not the purpose of the US drone campaign, which had counter-terrorism as its main objective.  Fair (2014, pp. 224-26), International Crisis Group (2013), and Williams (Williams 2008(Williams , 2010. See also Collombier and Roy ('Tribes and Global Jihadism -Google Books' n.d.). 59. There is a rich literature on the history of the tribal system and its governance in Pakistan, which generally emphasises the importance of tribal elders and the jirga in the administration of justice and maintenance of stability in the tribal area (e.g., Idris 2010, Lyon 2002, Yousaf 2019). See, for example, Ali (1999, p. 188) who argues that a jirga's main 'function is to settle peacefully an existing situation more than to judge right or wrong, determine guilt, or pass sentence.' In addition, the restoration of the jirga system is also considered as an essential contribution to 'uniting people and ensuring peace in the region' (Community Appraisal and Motivation Programme and Saferworld 2012; Z. S. Ahmed andYousaf 2018, Khayyam 2016 • Assassinations of insurgent leaders is followed by a breakdown in negotiations, Taliban withdrawal from negotiations, or break-off of pre-negotiation engagements.
• No Taliban negotiation team is nominated and no joint negotiation platform emerges.
• Attacks on government forces and alleged collaborators increases.
• Taliban rhetoric becomes more aggressive, stated opposition to negotiated settlement increases.
• Relevant interlocutor statements in interviews and survey (e.g., pointing out difficulty of remaining senior leaders to meet and discuss negotiations).
• Media coverage and secondary/grey literature reporting lack of meetings and internal debate/consensus among Taliban factions.
2. Civilian casualties and damage to infrastructure increased support for the Taliban, swelled its ranks with new recruits, and made it more difficult for "moderates" to make a case for a negotiated settlement.
• Surge in expressions of support for Taliban, increase in statements of intent to join their ranks.
• Increase in hard-line statements on Taliban strategy in absolute terms and/or relative to advocacy for negotiated settlement.
• Relevant interlocutor statements in interviews and survey (e.g., discussing their own knowledge of increased support as a response to civilian casualties).
• Media coverage and secondary/grey literature reporting shifting public support and internal debates among Taliban factions. 3. Amid the alleged complicity of the Pakistani state (especially government and army) in the U.S. drone campaign, the government lost legitimacy and credibility as a partner for the Taliban in any potential peace process.
• Denouncing of collusion bet ween government/army and US.
• Expressions of distrust in negotiation process and state representatives.
• Break-off of any formal or informal contacts, likely to be accompanied by statements that do not recognise legitimacy of state and/or its representatives.
• Media coverage and grey literature reporting disengagement by Taliban because of government/ army collusion. 4. There is no mutually hurting stalemate between the parties, with either the government expecting to be able to defeat the Taliban militarily, or the Taliban considering itself strong enough to hold out and gradually improve its bargaining position over time.
• Sides do not take heavy casualties, see no or no significant drop in domestic/external support, and overall retain previous levels of territorial control.
• Relevant interlocutor statements in interviews and survey (e.g., confirming enduring military capabilities and political will to fight).
• Media coverage and secondary/grey literature reporting no significant casualty levels and/or ability to compensate with new recruits or reserves, and/or continuing ability to mobilise resources. 5. Both sides may be too fragmented internally to make credible, and thus acceptable, offers to initiate formal negotiations. The opposing side sees little likelihood in achieving and retaining desirable concessions in a negotiation process, in part because some factions may deny the other party's legitimacy.
• Internal debates prevent the emergence of a joint negotiation platform.
• Where a platform has been publicised, factions disown it or refuse to endorse it.
• One or more factions publicly reject the legitimacy of the negotiation process and/or that of the legitimacy of their own or the other side's representatives.
• One or more factions refuse to negotiate with one or more factions on the other side, rule out concessions, and/or threaten to torpedo ratification and/ or implementation of any agreements.
• Relevant interlocutor statements in interviews and survey (e.g., noting internal power struggles and/or emergence/persistence of veto players).
• Media coverage and secondary/grey literature reporting inability by one or both sides to forge consensus across different factions over strategy and/or legitimacy of other side's negotiators.
6. The settlement process lacks external mediators and guarantors that could bring the parties to the table and contribute to sustaining a negotiated settlement.
• Mediation is not part of the pre/negotiation and/or mediators are ineffective in bringing relevant parties to the table, keeping them there, and/or enabling them to reach a settlement.
• Relevant interlocutor statements in interviews and survey (e.g., pointing to lack of available or acceptable mediators).
• Media coverage and secondary/grey literature reporting noted lack of mediation as obstacle to constructive negotiations.

SURVEY DATA
Based on random sampling, two surveys, using the same set of eleven questions, were carried out before and after the launch of the Zarb-i-Azb military operation in North Waziristan in June 2014. The two-stage nature of the survey was not originally planned. Prioritising safe access to respondents, we had initially planned to carry out one survey only in the D.I. Khan IDP camp bordering South Waziristan, where the majority of displaced people are from South Waziristan, as well as among those who had privately moved to larger cities. The launch of the Zarb-i-Azb military operation in North Waziristan in June 2014, however, created an additional opportunity to extend the survey to IDPs from North Waziristan, many of whom fled to the Bannu IDP camp. Consequently, we carried out a second round of the survey there in September 2014.
In total, we approached 400 participants across the two survey rounds in areas where IDPs from the Waziristan agency were living in government-built, army-run camps (Bannu, D. I. Khan) or in the cities to which they had moved privately (e.g., Peshawar, and Islamabad). The multiple-choice questionnaire used in the surveys was in Urdu, the national language of Pakistan. Each respondent was required to select no more than one answer in response to each question, and they were told that nonresponses were allowed, giving them the option to choose not to answer questions they were not comfortable with. We did not require a justification of non-answers. On this basis, we received between 359 and 389 eligible answers for each of the questions.
In insecure areas access to participants was facilitated through personal contacts in independent think-tanks and the Pakistani Army. In later stages of the survey, these local contacts also acted as enumerators collecting data from our respondents. Enumerators were fully trained and briefed on the research project.
In view of cultural sensitivities, only male respondents were contacted. Survey participants were aged between 19 and 55 years, with the majority between 25 to 40 years. The participants belonged to the two major Mahsud and Wazir tribes as well as the minority Dawar tribe of Waziristan. The sample included both fully literate respondents and those without competent reading ability.
In cases where respondents had limited reading ability, the investigator or enumerator ensured that the questions were read out clearly. It was ensured that the person concerned was competent to make a decision and that the consent was voluntary. In the case of those not well-versed in Urdu, the questions were explained in the local language, Pashto, by translators.
Respondents' consent was obtained and signified, in most cases, by their signature or thumb-impression on the back of the questionnaire or on an attached consent form. In a minority of cases, respondents were not comfortable to give their names or signatures. In these cases, oral consent was obtained.