Killing Norms Softly: US Targeted Killing, Quasi-secrecy and the Assassination Ban

Abstract This article argues that when actors engage in controversial new security practices, it is misconceived to view secrecy as an opposed, counterproductive alternative to the pursuit of legitimation. Rather, we propose, deployment of “quasi-secrecy”—a combination of official secrecy with leaks, selective disclosure, and de facto public awareness—can be an effective strategy for achieving normalization and legitimation while containing the risks entailed by disclosure. We support this claim via a detailed case study of US targeted killing. First, we establish the existence of an American norm against targeted killing during the period 1976–2001. We then detail the process by which an innovation in practice was secretly approved, implemented, became known, and was gradually, partially officially acknowledged. We argue that even if quasi-secrecy was not in this instance a coherently-conceived and deliberately pursued strategy from start to finish, the case provides proof of concept for its potential to be deployed as such.


Introduction
How does a once-forbidden practice become normal, legitimate, even routine? More specifically, how can those in government who desire this outcome make it so? We ask this question apropos of a particular case: the program of targeted killing conceived and executed by the United States under the presidencies of George W. Bush and Barack Obama. In constructing an answer, we proceed as follows. First, we define targeted killing. Next, we survey general theory regarding how political actors accomplish normative shifts, scrutinizing the intuitively plausible idea that engineering a stable new normative settlement requires direct and persuasive public advocacy for any innovation. In this framework, secrecy is an alternative to legitimating new practices-and likely a counter-productive one. Using US targeted killing as a detailed case in support, we propose a contrary analysis: that official secrecy, deployed in a partial form we term "quasi-secrecy," can play an Andris Banka is assistant professor in International Relations at Ça g University in Turkey. Adam Quinn is senior lecturer in International Politics at the Department of Political Science and International Studies, University of Birmingham, UK.
instrumental role in the process of normalizing potentially controversial shifts. With the central argument thus outlined, section 2 concludes by providing a concise explanation of the concept of quasi-secrecy and the context for its deployment in this case, as well as some methodological reflection regarding the parameters of what we seek to demonstrate.
Turning to the case in detail, we first establish that between 1976 and 2001, US administrations operated as though the executive order banning assassination tightly constrained, and perhaps prohibited altogether, US Central Intelligence Agency (CIA) operations with the direct aim of killing specified individuals. This prohibition was domestic in origin, stemming from the Church Committee's recommended reforms of the intelligence agencies in the 1970s. Second, we establish that after September 2001, this prohibition was reinterpreted to permit a category of killing previously treated as proscribed. Facilitated by technological advance, this practice subsequently became frequent and routinized. While this shift did not constitute a total erasure of the norm against assassination, it did represent a substantial revision of it and constriction of its scope. Third, we detail and analyze the process by which the shift in practice became publicly known and via which the executive sought to legitimate it. Here we demonstrate the operation of quasi-secrecy and illustrate its utility for normalizing potentially controversial innovation. We conclude that this case provides proof of concept for quasi-secrecy's viability as a legitimation strategy. Even if we do not suppose it to have been clearly and comprehensively conceived as such at the outset in this instance, it could indeed be deployed as deliberate strategy in the future.
Defining Targeted Killing (and Assassination) Targeted killing, for our purposes here, means the deliberate killing of an identified individual, specified in advance as the target, approved by an authorized part of a government bureaucracy without independent judicial process. 1 The distinction between permissible targeted killing and assassination-and the sustainability of such a distinction-is a matter of contest. In common parlance, assassination refers to any planned, individually-targeted extrajudicial killing with a political or ideological rationale, especially where the target occupies a leadership role. In international law, the permissibility of such killing hinges on whether the target can be legitimately classified as a combatant in war, their location relative to the lines of battle, and whether treachery or perfidy is involved in the execution. 2 Preemptive use of lethal force against individuals outside a conventional battlefield may be justified by appeal to the principle of selfdefense, but legal scholars have sought to establish criteria limiting this. The threat must be clear and imminent, not "distant, unviable, or merely foreseeable." 3 The killing should not simply be punishment for past action: the individual targeted must be "actively involved in an imminent attack." 4 The extent of collateral damage must be weighed, 5 and the government carrying out the killing should articulate the legal basis for its actions. 6 Much remains unresolved in the interpretation of these criteria, however. For example, "there is considerable debate as to how far in anticipation [of harm to oneself] it is legitimate to act." 7 In the analysis that follows, we remain always cognizant of the contested legal and terminological status of lethal acts targeted at individuals designated as terrorists, militants, or enemy combatants. Taking aim at such targets is indeed distinct from plotting to kill a prominent figure in a foreign government, the category of act to which assassination has historically been applied with least controversy. Some may consider that the possibility of drawing such a distinction disposes a priori of the question of how to reconcile contemporary US practice with the assassination ban: the killings carried out by the United States did not violate the ban because they were not, by definition, assassinations. To assert this, however, is simply to beg the question; in fact, the point is both contestable and contested. 8 It was precisely the goal of the executive, during the period of our focus, to achieve acceptance as unproblematic for a practice that had been treated as prohibited during the previous two and a half decades. We are not concerned with the intellectual or moral correctness of that project, only with the process by which it successfully unfolded.

Norm Establishment and Erosion
The established definition of norms in the literature is "collective expectations for the proper behaviour of actors with a given identity." 9 Broadly speaking, norms serve as "rules of the road." 10 They tell us "what the playing board will look like, and which moves are acceptable. 11 Norms-when established-may be violated, but this carries a price for the violator, and thus they shape behavior. 12 A rich body of scholarship documents instances of norm-building, norm-diffusion, and socialization. 13 Instances of normweakening have also been charted, though such studies are less numerous. 14 The norm life cycle model, introduced by Martha Finnemore and Kathryn Sikkink (1998), provides the archetypal framework for norm establishment: a norm is first proposed by an entrepreneur, then achieves broad acceptance from a critical mass of actors, and finally is widely internalized such that "conformance … is almost automatic." 15 Ryder McKeown (2009) modeled the reverse process in the "norm death series." Here, first revisionists challenge a widely-internalized norm through "quiet changes in policy away from compliance." 16 Struggle then occurs "both in public discourse and within government institutions." If the change achieves widespread acceptance, the previously established norm suffers a crisis of legitimacy and expires. 17 Several scholars note that major events-war, revolution, economic crisis-may trigger or accelerate such shifts. 18 Our analysis takes McKeown's model as its starting point, with a slight adjustment to the level of analysis. The prohibition on assassination is of domestic American rather than international origin, brought into being-as we shall discuss later-by the Church Committee inquiry and resulting reforms in the 1970s. This does not, however, present a conceptual obstacle to our adopting the basic architecture of conventional norm theory; as Finnemore and Sikkink 19 and Wayne Sandholtz 20 note, it is readily transferable between levels. It requires only minor alteration to the identitycontent of the model's component parts. In our case here, the executive is the initiating agent of change, rather than the US government as a unitary whole, while the primary audience for legitimation efforts is not the global public or international institutions, but the American public and domestic institutions such as Congress and the courts.

Strategies of Legitimation
A norm is a bearer of legitimacy. 21 Legitimacy cannot be bestowed unilaterally; its existence presupposes a community "able to pass judgment on appropriateness." 22 Actors sometimes defect from established rules, but even powerful ones are "limited in the costs they can tolerate" and will therefore seek to ground their actions in some claim to legitimacy. 23 Vincent Charles Keating provides a typology of strategies available to an actor engaged in behavior that transgresses established normative boundaries. 24 It posits four alternatives. The first is to act openly in violation of the existing norm with no accompanying effort at justification. If the norm is well-established, however, this can only be a temporary course, since 17 McKeown, "Norm Regress," 12. 18  others will demand explanation for transgressive acts, and by definition such behaviour carries cost. The second strategy is overt justification: to acknowledge the relevant actions but assert that one remains compliant with the established norm nevertheless. This tactic will likely involve some creative redescription of either the requirements of the norm or the characteristics of the acts in question. The third strategy is overt innovation. In this case, an actor actively advocates amending or overturning the established norm to render their actions permissible.
Finally, there is secrecy. In this case, the actor conceals-and denies the existence of-norm-breaking behavior. This strategy does not seek either to justify actions with reference to prevailing norms or to advocate for amending those norms, but rather to evade the need for legitimation altogether. Secrecy is conventionally framed as an undesirable resort, to be employed when an actor does not believe their actions can be successfully publicly defended. 25 Ian Hurd, for example, contends that secrecy is a high-risk strategy that can only be a temporary measure. 26 Daniel Byman is also critical of secrecy, arguing that without transparency and the resulting public debate, a policy can be held hostage to perfection. If policies are not endorsed beforehand by the public and the political opposition, they will provoke intense controversy when abuses and mistakes occur-as they inevitably will." 27

Quasi-secrecy: Meaning, Model and Context
Building on Keating, our proposition here (supported by analysis of the case) is that although secrecy may sometimes be a temporary, counterposed and counterproductive alternative to legitimation, it is misconceived to think of it exclusively, or even primarily, in this way. We contend the case of US targeted killing demonstrates that, on the contrary, when official secrecy and de facto public disclosure are combined-the phenomenon we call quasi-secrecy-this can provide an effective mechanism for normalizing controversial practice. In order that the purpose of the following detailed case analysis be maximally clear, we briefly summarize here how and why quasi-secrecy took shape in this instance and present in outline form the conceptual model for its operation.
The basic fact that the US government was killing individuals on foreign soil using a new, remotely-controlled technology was known and publicly reported from almost the moment it commenced. The first operational 25  installment of the post-9/11 targeted killing program-in Yemen in 2002was accompanied by a single, under-considered instance of public avowal by a US official (see section 4 for details). The practice itself was therefore de facto not a secret for long. Yet, after that moment of initial confusion, it proceeded for years under a blanket of official secrecy. On grounds of national security, government officials refused to confirm even rudimentary facts and declined to publicly articulate any reasoning reconciling the new practice with established law and norms. Indeed, the program's very existence remained highly classified.
This strategy served two functions for the executive. The first was to restrict access to operational data, such as number of strikes, targets, and casualties. Such information would be steadily and painstakingly assembled through the investigative efforts of journalists, researchers, and others. But official non-acknowledgement of these facts lifted any obligation from government officials to engage publicly with potentially uncomfortable questions regarding collateral damage, cost-benefit analysis, and other sensitive issues. The second function was to obviate the need to publicly articulate the legal basis upon which the policy rested. This curtailed the possibility that public officials would be pressed-on the record-regarding the ultimate logical implications of that legal reasoning, most especially the key question of where (if anywhere) was the outer limit of the executive's asserted prerogative to kill.
Over time, the Obama administration gradually loosened the secrecy governing the program with regard to both operational facts and legal reasoning. It did so first through strategic leaks designed to make publicly known that the United States was pursuing a major, proactive counterterrorist program and portray it in the most favorable possible light. This strategy allowed the administration to claim credit for taking effective action while retaining a shield against critical interrogation on the record. The ground having been thus laid, the administration did later open up officially (to a limited extent), allowing it to claim a belated good faith effort at openness and accountability. Indeed, these claims may have reflected the reality of such good faith. Even at this stage, however, the government retained a regime of classification sufficient to allow (a) selective acknowledgement or denial of operational facts at its own discretion, and (b) curtailment, at the point of its own choosing, of official engagement with discussion of the program's ultimate legal implications.
The result was a regime of quasi-secrecy: the coexistence, over an extended period, of official secrecy and de facto public disclosure regarding an ongoing practice. This allowed relevant audiences, including the public, to grow accustomed to the program's existence via regular references in the news media and wider culture. At the same time, it contained the risk of focused controversy and backlash to a level lower than would have been entailed in full disclosure and direct, uninhibited advocacy for a new normative dispensation.
The Bush and Obama administrations, especially the latter, did justify their actions up to a point. But official secrecy was not an alternative to such efforts at legitimation-it was complementary to them. By declining to comprehensively avow operational facts or present with full transparency a claim to new legal authority of specifiable scope and basis, the executive deprived those who would oppose its actions of two things essential to a public contest of ideas through the political system: a mutuallyacknowledged baseline of facts and authoritative official interlocutors. Yet there is no evidence that doing so was ultimately counterproductive from an executive branch perspective. On the contrary, when official figures for drone strikes and civilian casualties were finally released in the last year of the Obama presidency, they generated only muted reaction. A major reason for this outcome, we contend, is that those receiving this new information did not perceive it as such. Having been ambiently aware of targeted killing and drone strikes for years, a non-expert member of the public might reasonably-if incorrectly-have supposed the facts provided for the first time in 2016 to have been long since officially acknowledged. Consequently, they might have supposed it to be far too late-and therefore redundantto initiate discussion about the legal and normative implications of targeted killing; surely such fundamental issues must have been robustly debated and resolved in the proper forums at the appropriate, considerably earlier time? This point will be discussed further in section 6.
We should be clear that the realistic objective of any effort to legitimate controversial new practice is not universal support. In this case, the key domestic audiences were the public, Congress, and the judicial establishment: in short, those among whom majority opposition could materially impede the executive's ability to continue the targeted killing program. Some elements of society were unlikely to ever be persuaded of targeted killing's legitimacy, including human and civil rights lobby groups, liberal lawyers, and longstanding critics of expansive executive war powers on both the political left and right. Resistance from such irreconcilables being a given at the outset, the objective was to prevent this foreseeable opposition from gaining wider purchase among higher-priority audiences. A secrecy-skeptical model of legitimation would warn that failure to be transparent at the outset and engage in overt persuasive advocacy for a new normative settlement would prove counterproductive in this regard. But it is our contention that in this case, while secrecy and quasi-secrecy may have frustrated-and indeed continue to frustrate -such elite groups, they did not demonstrably impede the executive in ultimately securing and maintaining the consent of its priority audiences.
To abstract to the general conceptual level: the model for normalization via quasi-secrecy that we posit here has four stages. First, policy is changed to permit acts previously prohibited, and this is kept secret. Second, credible reports of the new practice come from the field and are published, but the executive adopts a blanket stance of refusing to confirm or deny their veracity. Third, the executive itself becomes a source of public information about the practice via leaks designed to portray it favorably but continues to maintain official secrecy across the board. Fourth, if the previous stages unfold without generating sufficient opposition to force reconsideration, the executive publicly avows the new practice and may also present on the record selected operational facts and justificatory arguments. This acknowledgement should not be confused with total transparency, however: a significant level of official secrecy may be retained and may continue indefinitely.
Finally, a note on context: the success of any legitimation effort is contingent at least to some degree on the propitiousness of the circumstances in which it takes place. Our focus in this article is on formal US policy and the actions and statements of government officials. But the events described should be viewed in the light of a substantial scholarly literature on the wider cultural and historical context in which these events occurred.
Stephanie Carvin and Michael John Williams have insightfully located twenty-first century American targeted killing within a longstanding national project to craft a "legal-scientific way of warfare that … seeks to balance the need to employ overwhelming force with legal and humanitarian concerns through science and technology." 28 Other scholars similarly remind us that contemporary US actions are situated within a longer-term pattern of development in both liberal-democratic and legal norms 29 and the technologies of targeting and violence. 30 Additionally, we posit more immediate factors that made quasi-secret targeted killing by the United States after 2001 possible. These include: the inaccessibility of locations where most strikes took place, making independent reporting difficult; the CIA's standing authority to carry out deniable covert operations at the instruction of the president and therefore its ready suitability to serve as the agent of such a program; and the disproportionate international power of the United States, which afforded it latitude to brush off outside inquiries about its conduct, such as from the United Nations.

Reflections on Methodology, Case Selection, and Sources
Our claims in this article should be viewed within the following parameters. A starting premise of our argument is that the practice of targeted killing has been successfully normalized and legitimated within the American polity. It is therefore not a primary task here to substantiate this premise, but we do provide support for it in section 6, citing public opinion data and acceptance of executive practice by Congress and the courts. Rather, our overriding focus is on the process by which normalization was achieved. In methodological terms, the case evidence and analysis presented below fulfill two functions. First, they serve to falsify two closely related, potential hypotheses: (a) that secrecy is necessarily an oppositional alternative to seeking and obtaining legitimacy for a normative innovation; and (b) that secrecy is always counterproductive to achieving that end. 31 To put it another way, this case is shown to be a noncongruent counterexample with regard to those propositions. 32 The case's second, parallel function is to provide an initial probe of the coherence and plausibility of the proposition that what we term quasi-secrecy can serve as a viable mechanism for normalizing controversial new practice. This is demonstrated via historical process-tracing of how US practice shifted in the case of targeted killing and how that shift was accounted for publicly. Our findings here provide, we contend, grounds for further pursuit of this line of inquiry in future research, such as detailed comparison with cases in which secrecy also played a role, but where legitimation of new practice proved more elusive, such as the United States' use of torture and extra-judicial detention. It is important to specify some claims we do not make. Our core claims here, as specified above, are distinct from arguing that a strategy of quasisecrecy is either necessary to securing legitimation (the only mechanism by which it can be achieved) or sufficient to do so (in other words, assured to succeed in all cases). Testing either of those claims-neither of which we consider prima facie compelling-would require a different research design. We contend only that quasi-secrecy was the mechanism by which legitimation occurred in this instance, and this success is suggestive of its viability and potential utility. Second, we do not maintain that quasi-secrecy was a 31 On the validity of using single cases to contradict hypotheses so long as the case is well-suited as a test of claim, see singular, coherently-conceived and deliberately pursued strategy on the part of the Bush or Obama administrations. As the case detail will demonstrate, more than one factor contributed to bringing about the blend of simultaneous classification and disclosure: the demands of foreign governments to obscure their complicity; prudential concern for restricting sensitive operational detail; the desire to win public credit for effective counterterrorist measures; and (perhaps) sincere commitment to some degree of transparency and accountability. Our claim is that whether conceived as a deliberate whole or simply the product of a synthesis of competing pressures, the ultimate de facto strategy of quasi-secrecy proved a highly effective mechanism for achieving normalization of a controversial innovation. We propose that this should be considered initial proof of concept for quasi-secrecy as a strategy that might be deliberately adopted by future administrations in analogous circumstances. Given the objectives just set out, our focus specifically on the American case-as opposed to the practice of targeted killing more widely-can be justified by reference to two considerations. The first is the unusually explicit status of the United States' prior prohibition of this activity. All countries are governed in principle by the same legal and normative restraints in this area, at least internationally. But the existence of an executive order explicitly banning assassination, interpreted before 2001 as tightly constraining the kind of targeted killing carried out in the years thereafter on large scale, meant that American administrations faced a more acute challenge than others might have when seeking legitimacy for these actions in the eyes of domestic audiences. The second consideration is the scale of the US program and, consequently, the lack of ambiguity as to whether a major shift in practice took place. A small number of other states-including Israel, the United Kingdom and France-engaged in similar lethal actions during the same period. But the United States carried out the largest number of targeted killings and drone strikes by a vast margin. The sheer size of the American program, combined with its systematic and institutionalized character, eliminates scope for any plausible claim that killings in this period merely represented rare anomalous exceptions to a still-standing general prohibition.
The sources available for use in this inquiry vary by period. Our portrayal of the Church Committee era, when the ban on assassination was established, can be grounded fully in archival sources that include declassified materials from the National Security Archive and the Ford Presidential Library. For later periods, during which classification remains in fuller force, the official record is necessarily less complete. Nevertheless, using a combination of public documents, memoirs, and secondary sources based on interviews with participants, it is possible to construct a robust account of how and when practices changed. The account we provide of official acknowledgement of the post-2001 targeted killing program by the executive-what was disclosed publicly (and when) regarding operations and their legal basis-is based on a comprehensive survey of officially available speeches, press conferences, hearings, Congressional debates, legal documents, and published interviews from the period. The account of parallel disclosure-by-leak is likewise grounded in a comprehensive survey of contemporary print media.
For certain kinds of inquiry into a practice such as targeted killing, high levels of government secrecy might present a major obstacle. 33 The gap between what is known about the practice through indirect reportage and what is officially avowed has been a source of frustration even to those with relatively strong claims to rightful access to information. Philip Alston, UN Special Rapporteur on extrajudicial, summary, or arbitrary executions, complained in 2011 of "an almost surreal tendency on the part of the executive and the courts to pretend that information that has been comprehensively leaked, remains unknown or at least uncognizable," 34 Fortunately for our purposes here, it is precisely this disjunction between facts widely known and those officially acknowledged that is of interest. We do not claim to uncover here new data regarding US covert practice. For the underlying facts of the targeted killing program, we draw upon on the deep reservoir of reportage already assembled on the public record. The significance of our contribution lies rather in analysis of how quasi-secrecy served the executive's interest in normalizing it.

Establishment of the Norm Against Assassination
The origins of the prohibition on US government deployment of assassination lie in domestic pressures and processes. Specifically, it resulted from the Church Committee investigation into "illegal, improper, or unethical" activities on the part of intelligence agencies. 35 Despite the Ford administration's public position that it would provide "maximum assistance," 36 the committee faced resistance throughout its investigation, first to its full access to intelligence material, 37 then to the public release of its report. 38 That report condemned US use of assassination as immoral, logistically precarious, and liable to be counterproductive. 39 It also declared that "a system which relies on secrecy" increases the "risk of confusion and rashness in the very areas where clarity and sober judgment were most necessary." 40 The committee favored an assassination ban written into statute, 41 but the ultimate outcome was a compromise short of that: an executive order issued on February 18, 1976. 42 The order did not define "assassination," with the result that the scope of the prohibition has been contested among both scholars and officials ever since. Some argue this ambiguity was deliberate, engineered to preserve the future option of narrow interpretation as to what it forbade. 43 Others contend that it barred the US government "from directing, facilitating, encouraging, or even incidentally causing the killing of any specified individual." 44 During the period 1976-2001, the assassination ban served as a strong constraining influence on US action. It came under strain from two sources. The first-of lesser importance for our purposes-was provocations from dictators such as Libya's Muammar Gaddafi, Iraq's Saddam Hussein, and Panama's Manuel Noriega. These cases have received extended consideration elsewhere, and we do not discuss them here because targeting heads of state entails the most unambiguous possible violation of the assassination ban; therefore, its consideration in these cases raises issues distinct from those posed by the post-9/11 targeted killing program. Our focus is the second source of strain: terrorism and militancy emerging from the Middle East. This factor invited the US government to narrow its interpretation of the ban to permit targeting terroristic non-state actors. Two examples from this period merit detailed attention, and we treat them individually below. Importantly, neither led to the prohibition being set aside; on the contrary, they serve to 37  demonstrate its continuing force. However, they did instigate the first private articulation of legal rationales that would become relevant in the post-9/11 period, making these episodes important precursors to practices later to come.

The Reagan Administration and Lebanese Hit Squads
The Reagan administration came close to carrying out targeted killings in Lebanon in reaction to the threat posed by militant Islamist groups such as Islamic Jihad and Hezbollah. Consideration of the tactic began in the spring of 1984 with a secret proposal from National Security Council (NSC) official Oliver North to covertly contract local teams to eliminate those responsible for bombing US facilities. 45 Stanley Sporkin, General Counsel to the CIA, privately argued that a legal basis for such action could be found in self-defense and a robust legal distinction could be drawn between North's proposal and killings of the kind carried out prior to Church and prohibited after 1976. 46 This line of reasoning received firm pushback from the CIA, whose leadership at that time personally recalled the reputational damage suffered by the agency due to the Church revelations; CIA Director Richard Helms and his deputy John McMahon believed the proposal violated the assassination ban. 47 In the summer of 1985, through a task force on counterterrorism strategy under Vice President George H. W. Bush, officials for the first time formally discussed adopting a narrow interpretation of the ban's scope that excluded terrorists. 48 Despite the CIA's objections, Reagan initially approved the Lebanon proposal. He secretly issued presidential findings-directives, clearing the legal path for the plan-and approximately $1 million was allocated for training and support. 49 The United States aborted the plan, however, when a prospective Lebanese contractor carried out an unapproved car bombing, killing eighty and wounding two hundred. 50 The presidential findings were also rescinded, although even as White House press secretary Marlin Fitzwater acknowledged this, he denied that the findings involved authorizing assassination, a tactic Vice President Bush also stated publicly would be "absolutely criminal." 51 Anonymous officials later told the Washington Post that the president's directives had been deliberately intended to circumvent the existing bar on assassinations. 52 This led White House reporter Helen Thomas to directly ask: "Mr. President, did you sign two orders, directive intelligence orders, which appeared to circumvent the assassination directive-ban on assassinations?" Reagan denied he had ever issued a "permit to assassinate anyone in any of the things that we were doing." 53 After leaving office, Reagan would admit that targeted killing had been considered, but claim that he had ultimately decided it was a "game that America couldn't and didn't play." 54 Robert Oakley, State Department counterterrorism coordinator under Reagan, later recalled there was "a great debate about whether or not one could do this, and a lot of the laws and regulations and executive orders were studied very, very carefully." 55 There were "differences of opinion within the executive branch," but in the final analysis the president decided, "no, we are not going to go that route." 56 In light of the role new technology would play later, it is worth taking note that this decision was heavily influenced by practical concerns regarding the likelihood of successful execution, not just internal resistance on legal and normative grounds.

The Clinton Administration and Osama bin Laden
Until 1998, Osama bin Laden, though monitored by the CIA, was perceived as a "low priority" threat. 57 This assessment changed in 1998 after the bombings of US embassies in Kenya and Tanzania, which killed hundreds. In their wake, President Bill Clinton signed a presidential finding authorizing covert action to capture bin Laden and kill him if he resisted. Clinton would later amend the secret memorandum of understanding that set parameters for the operation a number of times: first expanding the list of permitted targets beyond bin Laden to include a small number of key lieutenants and later authorizing the shooting down of any aircraft or helicopter he might use to escape Afghanistan. 58 Richard Clarke, Clinton's counterterrorism head, notes that the administration was unwilling to approve a straightforward killing and therefore produced a series of unusual documents giving "extremely specific authorities for particular CIA operations aimed at bin Laden." 59 According to Clarke, "there was concern in both the Justice Department and in some elements of the White House and some elements of the CIA that we not create an American hit-list that would become an ongoing institution that we could just keep adding names to and have hit teams go out and assassinate people." 60 Investigative reporter Jeremy Scahill, who has had access to many of the Clinton-era national security files, notes the baroque quality of the order: "The authorization for killing bin Laden was built in a way that there almost was one scenario where he could be killed-when he was in certain kind of a house, with a particular brand of lock on the door and only then you can strike the house." 61 According to CIA Director George Tenet, "almost every authority granted to CIA prior to 9/11 made it clear that just going out and assassinating Bin Laden would not have been permissible or acceptable." 62 Years later, during 9/11 Commission hearings, CIA officials and lawyers uniformly said that they had interpreted authorities signed by Clinton as instructing them to try to capture bin Laden alive and that the only acceptable context for killing him would be during a credible operation aimed at capture. 63 Two senior CIA officers later said they would have been "morally and practically opposed to getting CIA into what might look like an assassination." One went so far as to say he would have "refused an order to directly kill Bin Laden." 64 The Bush Administration After 9/11: A Covert Shift Narrowing the Scope of the Prohibition Before 9/11, only one country-Israel-engaged programmatically in targeted killing, and it was criticized by the United States for doing so. 65 In July 2001, Secretary of State Colin Powell said of Israel's targeted hits: "We … express our distress and opposition to these kinds of targeted killings and we will continue to do so." 66 US Ambassador to Israel Martin Indyk stated: "The United States government is very clearly on record as against targeted assassinations. … They are extrajudicial killings, and we do not support that." 67 In the aftermath of 9/11 however, many criticized the legacy of Church for having neutered the US intelligence community's capabilities, including Paul Bremer, chairman of the bipartisan National Commission on Terrorism, and Henry Kissinger. 68 Former Secretary of State James Baker openly endorsed fully repealing the executive order banning assassination. 69 Former President George H. W. Bush said there was a need to "free up the intelligence system from some of its constraints." 70 Within government, several senior figures made statements about the need to revisit restrictions on covert operations, including Vice President Dick Cheney, 71 Secretary of State Powell, 72 Chairman of the Senate Intelligence Committee Bob Graham, 73 and Vice-Chairman Richard Shelby. 74 A former CIA official observed that in the post-9/11 environment, there was a feeling that "the things we couldn't do before, now we can do them." 75 Georgia Congressman Bob Barr even made a formal proposal to overturn "those portions of executive orders purporting to prohibit the government from directly eliminating terrorist leaders." 76 His initiative secured fourteen cosponsors, though it ultimately went no further. 77 These statements drew some public resistance from figures such as Chair of the House Foreign Affairs Committee Lee Hamilton, 78 former CIA Director and later Secretary of Defense Robert Gates, 79 and Democratic Senator Christopher Dodd. 80 Outside government, nongovernmental organizations (NGOs) and international lawyers argued for preserving pre-9/11 standards. Human Rights Watch sent a letter to President Bush urging that the United States should remain committed to "investigation, arrest, trial and punishment," not "executions or targeting non-combatants." 81 M. Cherif Bassiouni, an eminent scholar of international criminal law, warned against having "the intelligence agencies be judge, jury and executioner all wrapped into one. The potential for abuse is too big and the symbolism is too harmful." 82 In the event, there would be no open declaration that the restrictions governing targeted killing had narrowed. Rather, there was simply a shift in practice, authorized and executed secretly. As Jonathan Ulrich (2014) notes, the president may legally amend or revoke an executive order without "publicly disclosing that he has done so." 83 During NSC meetings after 9/ 11, President Bush reportedly requested a scorecard: a list of top al Qaeda members to be crossed out after their elimination. 84 According to Richard Clarke, Bush told staff on the day of the attacks: "Everything is available for the pursuit of this war. Any barriers in your way, they're gone." 85 CIA headquarters would become the hub for counterterrorism operations. Bob Woodward reported a senior official in 2001 as stating, vaguely but ominously: "The president has given the Agency the green light to do whatever is necessary. The gloves are off. Lethal operations that were unthinkable pre-September 11 are now underway." 86 A landmark moment came on September 15, at a strategy meeting of Bush's national security team at Camp David, when CIA Director George Tenet put forward a plan to hunt and kill al Qaeda members and associated forces without geographical limitation. 87 John Rizzo, a CIA lawyer involved in drafting the proposal, later commented that he had "never seen a presidential authorization as far-reaching and as aggressive in scope." 88 According to Afsheen John Radsan, the assistant general counsel of the CIA, the authorization was "generally worded," amounting to: "Go out and get the bad guys. Disrupt them, kill them, interrogate them." 89 William C. Banks characterizes the authority transferred to the CIA as the "most sweeping and most lethal" since the agency's creation. 90 A decade later, President Bush would acknowledge having granted broad authority to the CIA for covert actions, including "permission to kill Al Qaeda operatives," 91 something he knew to be "a significant departure from America's policies over the past two decades." 92 Following this authorization, instructions cascaded through the national security agencies. Josef "Cofer" Black, director the CIA's Counterterrorism Center, reportedly briefed his team as follows on September 19th: "I want to give you your marching orders, and I want to make them very clear. I have discussed this with the President and he is in full agreement … I don't want bin Laden and his thugs captured, I want them dead. … They must be killed. I want to see photos of their heads on pikes. I want bin Laden's head shipped back in a box filled with dry ice." 93 Gary Schroen, in charge of the subsequent CIA incursion into Afghanistan, later recalled that it was the first time he had received an order to kill rather than seek to capture a target. 94 The CIA prepared a list of about "two dozen terrorist leaders" who could now could be targeted and killed. 95 Tenet requested and received additional authorization for the CIA to add further names to this list without further presidential approval. 96 In terms of both operational substance and oversight, this represented a major rollback of key elements of the Church reforms. 97 The first targeted strike outside a conventional battlefield took place in Yemen on November 3, 2002, killing Qaed Salim Sinan al-Harethi, a highlevel al Qaeda operative, and five others. 98 The president was not asked to authorize the specific strike; it was approved by senior officials in the CIA and military using the authority delegated to them. 99 The event also served, in a backward sort of way, to underscore the imperative of secrecy. In the 89  immediate aftermath of the strike, Deputy Secretary of Defense Paul Wolfowitz took a celebratory public stance, praising the strike on CNN as "a very successful tactical operation. … One hopes each time you get a success like that, not only to have gotten rid of somebody dangerous, but to have imposed changes in their tactics and operations." 100 Unwittingly, Wolfowitz triggered a diplomatic crisis with his remarks: the Yemeni government had agreed with Washington that Yemen would claim responsibility, and it did not receive this surprise reversal well. 101 After this singular initial misstep, however, a wall of official secrecy descended. It would be almost a decade before any high-level government official would next comment on the record regarding a targeted killing.

Technological Advance as Driver and Enabler
Unmanned aerial vehicles (UAVs or drones) were first used in nonarmed form during the Balkan crisis of 1995. 102 The idea of using them as a tool for targeted killing emerged gradually during the Clinton years. A CIA surveillance drone in 1999 spotted bin Laden in Afghanistan, but the five hour preparation time required by the Pentagon to organize a missile strike allowed him to escape. 103 Following this, the project of arming drones (with Hellfire missiles) was put on fast-track. Initially, the CIA was opposed; just a week before 9/11, Director Tenet was quoted saying it would be "a terrible mistake" for the agency to take on this operational role. 104 His assessment immediately changed with 9/11. In his September 15 Camp David presentation, Tenet briefed administration officials for the first time on armed drones. 105 The first armed drone mission took place in Afghanistan on October 7. 106 The program of drone strikes outside of conventional battlefields started slowly during the Bush administration, with just a few strikes against highlevel targets. Following the first one, President Bush was sufficiently impressed to ask: "Why do we fly only one Predator at a time, we ought to have fifty of these things." 107 But the second strike did not take place until the summer of 2004, in Pakistan, targeting Taliban leader Nek Mohammad. Per agreement with Washington, the Pakistani government claimed responsibility, denying American involvement as "absolutely absurd." 108 The campaign had gathered pace by the time Bush's second term was drawing to a close, however. By one account, the administration authorized thirty-six strikes (outside of conventional military combat) in its final year, having carried out just nine in all previous years. 109 Some targets were high-level al Qaeda figures, such as Ayman al-Zawahiri (who evaded efforts to kill him) and Abu Musab al-Zarqawi; others were unknown to the wider public. According to Long War Journal, Bush authorized forty-six such strikes during his presidency. 110 The New America Foundation put the count at forty-eight, with an estimated 399-540 deaths resulting. 111 Official secrecy meant the administration refused to provide its own numbers.
A link between technological innovation and normative change has previously been noted by several scholars. 112 If the United States had not acquired armed drone capabilities, targeted killings would likely still have been part of the "War on Terror"; prior to expansion of the drone program, special forces and private contractors had been directed to prepare for such missions. But as noted earlier, logistical difficulty and associated risk had been a discouraging factor in previous administrations' consideration of plots to kill. Drone technology altered this calculus, increasing the confidence with which policymakers could expect targeted killing operations to achieve their objectives at acceptable risk and cost. By so doing, it incentivized a more permissive interpretation of rules constraining such action.
The development of drone technology was a double-edged sword with regard to maintaining secrecy. On the one hand, by protecting US personnel from physical jeopardy, drones eliminated the need to account for American casualties in the field should things go wrong, a common trigger for forced disclosure of violent overseas operations. On the other hand, although the technology made it easier to keep any single strike secret, it incentivized an increase in their number and frequency that made public reportage-and thus ultimate disclosure of the program-more likely. Once disclosure did take place, however, technology was broadly helpful to the executive's need to keep domestic audiences' favor. Removing loss of American lives from among the possible liabilities of targeted killing operations eliminated a primary reason for opposing them. In addition, it provided some basis in fact for claiming that the United States could strike targets with unprecedentedly high precision and low collateral damage, easing concerns of those worried about effectiveness or civilian casualties.

The Legal Basis
The administration's shift in practice gave rise to important questions regarding the legal basis on which it was proceeding. Gary Solis, then a law professor at the US Military Academy at West Point, believed that the first targeted killings were precedent-setting: "Until just a few months ago, we would all have expressed abhorrence … of targeting individuals off the battlefield. … But now … we are taking a new approach." 113 Amnesty International (2005) pointed out that the United States had historically condemned such actions. 114 To allow its strikes to be carried out, the Bush administration was required to craft legal findings internally to the effect that they were permissible. With the exception of Wolfowitz's CNN gaffe, however, the targeted killing program went entirely unacknowledged officially for the duration of the Bush presidency. The administration did not publicly proffer any legal basis for its actions. It was therefore unclear as a matter of public record whether the executive order banning assassination had been, in any formal way, amended or a new interpretation adopted.
Despite the absence of official acknowledgement, the outline of the legal interpretations supporting the new policy-and reconciling it with existing rules-did begin to take shape in media coverage and subsequent discussion within the legal community. 115 The foundation of the administration's legal case was the sweeping authority of the Authorization for Use of Military Force, 116 passed one week after 9/11 with only a single member of Congress opposed. This legislation was ground-breaking in empowering the president to target non-state actors "even to the individual level," 117 and shifting counterterrorism from a criminal justice framework to a war paradigm, in which terrorists could be designated enemy combatants. This facilitated invoking self-defense as grounds for use of lethal force, 118 buttressed by an expansive interpretation of the concept of imminence as it pertained to terrorist threats. The administration's National Security Strategy, published in 2002, asserted that the United States needed to "adapt the concept of imminent threat to the capabilities and objectives of today's adversaries." 119 One administration official, speaking on condition of anonymity, later explained that this legal framework was crafted precisely because the administration did not wish to simply rescind the assassination ban: "The self-defense exemption was a legal fabrication to save face, to say, 'Yes, it still applies, but just not in these cases.'" 120 Deputy Secretary of State Richard Armitage later recalled that President Bush issued a further order in early 2002, supplementing his earlier authorization of the CIA's targeting program. "I don't recall necessarily the words, 'targeted killings,' but it was clearly that. It was loosening the Executive Order 12333 against assassinations. And the reasoning as I recall was, its wartime, it's not an assassination, it's war." 121 Robert Grenier, the CIA's leading counter-terrorism official during this period, would later acknowledge that a significant shift in legal interpretation had taken place: "Activities that before 9/11 we would have said were assassination-now we are simply exercising our sovereign right of self-defense." 122

Secrecy as Alternative to Justification or Advocacy of Normative Innovation
Excepting the Wolfowitz moment in November 2002, the Bush administration, unlike its successor, committed fully to official secrecy regarding the targeted killing program. Officials refused to publicly state the number of drone strikes authorized, the number of estimated casualties, or by what criteria individuals were selected for the list of targets. When journalists pressed the White House for such details, spokesman Ari Fleischer stonewalled: "There are going to be things that are done that the American people may never know about." 123 It similarly rebuffed international inquiries. Letters from advocacy groups, including Amnesty International, requesting clarification of such strikes' legal basis simply went unanswered. 124 Published strategic documents contained no mention of either targeted killings or drone strikes. Officially, for all externally-facing purposes, the targeted killing program simply did not exist.
As noted earlier, one motive for this secrecy was diplomatic. Permission from the Pakistani and Yemeni governments for US strikes on their territory was predicated on an American commitment to not publicly claim them. As awareness of the strikes widened through media reporting, however, the utility of secrecy as a tool for deflecting awkward questions became apparent. The United States had previously condemned the Israeli policy of targeted killing, but how did this new US program differ? The administration had no credible answer ready for public consumption. Privately, however, administration officials would admit the US policy rested on the same legal rationale as Israel's, 125 the difference being only of "scale and frequency." 126 Anonymous officials occasionally offered verbal assurances that the process for selecting targets and carrying out strikes was careful: "We have more lawyers than Predator pilots," one was quoted saying. 127 But no outsider had access, via any official route, to the information required to evaluate such claims. With such opacity at the official level, Bradley J. Strawser argues, each strike "might as well be considered an assassination or just plain murder." 128

Inheriting and Expanding Targeted Killing
On taking office, President Obama quietly inherited the secret targeted killing program without public statement. On December 9, 2008, CIA Director Michael Hayden briefed the president-elect on all classified missions run by the agency, "the nature of those actions, and the written findings from Bush and other presidents." 129 Four days before his inauguration, Obama met the outgoing president, who advised him that the drone strike program was one of two he would find most valuable. 130 The appeal of targeted strikes for Obama can be readily understood in light of his positions as a candidate. He had been critical of large, costly military deployments such as that in Iraq. He had promised to make the US military "more stealthy, agile, and lethal in its ability to capture or kill terrorists." 131 He had also promised action in Pakistan: "If we have actionable intelligence about high-value terrorist targets and President Musharraf won't act, we will." 132 The new administration almost immediately increased the frequency of drone strikes. Three days after Obama's inauguration, the CIA carried out a targeted strike inside Pakistan. This hit the wrong target, resulting in "a tense back-and-forth over the CIA's vetting procedures for drone attacks" between the President and Hayden. 133 But despite this, "there was no serious disagreement with the decision to continue the program." 134 Less than a month later, the CIA launched another attack, targeting Pakistani Taliban leader Baitullah Mehsud, which killed more than thirty people. 135 President Obama ended his first year having authorized more strikes than Bush had in eight years. 136 Those killed included high-value militants, such as Mehsud, Osama bin Laden's oldest son Saad bin Laden, and Tahir Yuldashev, leader of the al Qaeda-associated Islamic Movement of Uzbekistan. 137 The following year, the number of strikes doubled, to 128. 138 CIA requests for wider zones of permitted targeting in Pakistan and more armed UAVs were approved. 139 This information was not released publicly; the program remained shrouded in official secrecy. Public records were instead compiled by outsiders, such as the New America Foundation (NAF) and the Bureau of Investigative Journalism (BIJ), counting and detailing drone strikes to the best of their ability using information gleaned from other sources. 140 By the end of Obama's two terms in office, the BIJ's numbers suggested he had authorized a total of 563 strikes in Pakistan, Somalia, and Yemen. 141 NAF estimated 353 strikes carried out in Pakistan, killing between 1,934 and 3,094 (estimated), with between 129 and 162 of those thought to be civilians. 142 In addition to being more capable of carrying out targeted killings thanks to drone technology, the Obama administration had a further incentive to favor them. Obama had criticized both extrajudicial detention and enhanced interrogation (considered by many to be torture) as practiced by the Bush administration. Targeted killing could obviate the need for either, while still acting to neutralize the perceived threat. Hayden, who concluded his tenure as CIA Director in February 2009, believed this dynamic partly motivated the incoming administration's enthusiasm for the program. 143 Likewise, Rizzo judged that the administration "never came out and said they would start killing people because they couldn't interrogate them, but the implication was unmistakable." 144 We might also surmise that an increase in the number of drones deployed-and advances in their technological capabilities-boosted US capacity to identify new targets without the need for detention and interrogation to acquire intelligence regarding the identity and location of hostile actors.

From Secrecy to Quasi-secrecy: Selective Disclosure and Strategic Leaks
During its first years, the Obama administration maintained its predecessor's approach to secrecy, keeping the targeted killing program tightly under wraps. Following the first strike of the Obama presidency, White House press secretary Robert Gibbs refused to officially acknowledge the 140 An important side-point should be noted here: NAF and BIJ counted drone strikes. Not all these were targeted killings by our criteria: some were so-called signature strikes, authorized based on observed behavior or circumstances without identity of targeted individuals being known. Though the number of both types of strike increased during these years, separating the two without official confirmation of targets is challenging. Both targeted killings and signature strikes are in turn distinct from killing by US military personnel operating in conventional combat spaces, although issues of secrecy and disclosure may also be relevant to these. spymuseum.org/multimedia/spycast/episode/playing-to-the-edge-an-interview-with-gen-michael-hayden/. 144 Mazzetti, Way of the Knife, 281. event, saying only: "I'm not going to get into these matters." 145 This would become a routine response. During this period, the rule book for targeted strikes was considered so highly classified that it was "hand-carried from office to office rather than sent by e-mail." 146 "We didn't even know if we were allowed to write the word 'drone' in an unclassified e-mail," one State Department official reported. 147 After leaving office, Gibbs revealed that "when I went through the process of becoming press secretary, one of the first things they told me was: "You are not even to acknowledge the drone program. You're not even to discuss that it exists." 148 Behind the wall of official silence, there was disagreement on the merits of such strict secrecy, 149 especially as the increasing frequency of operations made public reportage of strikes inevitable. Secretary of State Hillary Clinton privately complained that blanket denial made it impossible to rebut exaggerated accusations regarding civilian casualties. 150 Leon Panetta, who oversaw the program for years as CIA Director, confirmed later that some officials advocated "full public explanation of each operation," and that he also felt President Obama should be "far more transparent" in explaining the policy. 151 With time, secrecy came into tension with another imperative: securing credit for waging an effective counterterrorism campaign. This led to numerous instances in which officials sought to walk the line of publicly praising the program's effectiveness while simultaneously refusing to directly acknowledge its existence. Two months into the Obama drone campaign, Chairman of the Joint Chiefs of Staff Mike Mullen was asked in an interview about the apparent increase in strikes under the new president. He declined to confirm details, but stated that threats in Pakistan "need to be addressed, have been addressed, and will continue to be addressed." 152 In similar fashion, John Brennan, then deputy national security advisor for counterterrorism, refused to directly address the program but offered the assurance that all counterterrorism operations were "legal," "highly effective," and "very focused." 153 CIA spokesman Paul Gimigliano declined to acknowledge targeted killings, stating only that tools used by the agency were "exceptionally accurate, precise and effective." 154 CIA Director Panetta, questioned at the Pacific Council in 2009, adopted a similar approach: "Obviously because these are covert and secret operations I can't go into particulars. I think it does suffice to say that these operations have been very effective because they have been very precise in terms of the targeting and it involved a minimum of collateral damage." 155 In another interview, he praised the CIA's counterterrorism efforts as the "most aggressive" in history, yet stopped short of stating exactly what methods he was referring to. 156 Most controversially, the administration further engineered favorable coverage by means of leaks to the media on condition of anonymity. In 2011, David Ignatius of the Washington Post reported that the White House was willing to discuss the top-secret drone campaign with him, but only when resulting coverage promoted the precise and effective nature of the program. 157 "These rules about covert activities can be bent when it becomes politically advantageous," he explained. Jonathan Landay of Reuters reported a similar experience, noting that when information worked in the administration's favor, "you get quite a detailed readout." 158 Leaks such as these allowed officials to advance, for public consumption, information carefully selected to portray the efficacy and legality of the program in the most favorable possible light. Alston maintains that leaks "played a powerful role in legitimizing the targeted killings program." 159 Meanwhile, the posture of official secrecy retained in parallel provided a protective barrier behind which officials could step at any moment of their choosing. This option was especially useful when faced with the most challenging questions arising from the practice, such as the outer limit of the legal authority to kill being claimed by the executive or the details of targeted strikes gone awry. In this way, the antagonistic incentives for secrecy and disclosure interacted such that the program became, as Mark Phythian has observed, "neither fully covert nor overt." 160 The Gradual, Partial Official Opening Up In 2010, State Department Legal Adviser Harold Koh presented an assessment to the American Society of International Law, a few paragraphs long, of how targeted killings could fall within the laws of war. 161 Prior to the speech, he had resisted engaging publicly with legal debates on the subject. 162 An administration lawyer later stated that Koh's public defense was the end result of an "unbelievably excruciating process of crafting a public statement that all the agencies can agree on." 163 Alston welcomed the statement as "a good start," but noted that the UN had "been asking for a legal rationale for quite a long time," and Koh's statement failed to answer key legal questions. 164 2011 was an inflection point in intensifying pressure on the administration to account in some on-the-record way for the program and its legal basis, perhaps because it saw the two highest-profile killings to occur during the Obama presidency. First, on May 2, Osama bin Laden (whom the United States had been hunting since before 9/11) was killed by a Navy SEAL helicopter raid on a compound in Abbottabad, Pakistan to which he had finally been traced. This was not a typical installment in the targeted killing program as it is defined here, for two reasons. First, it was carried out under the command of the US military rather than by intelligence operatives or contractors (although the CIA was intimately involved). Second, it was immediately announced by the president as a major success and reported in detail by the press, with administration assistance. While it had been planned secretly, it was not a covert operation in that there was no intention to maintain deniability after it had been carried out. Nevertheless, after the initial wave of celebratory reaction, pointed questions did arise regarding the legal basis for what appeared to be a summary execution. 165 The administration generally avoided addressing the point, though Attorney General Eric Holder, who found himself before the Senate Judiciary Committee on May 4, responded under questioning that "the operation in which Osama bin Laden was killed was lawful. He was the head of al Qaeda, an organization that had conducted the attacks of September 11. He admitted his involvement. … It was justified as an act of national self-defense." 166 The bin Laden killing thus drew unprecedented attention to some of the key underlying legal issues also raised by the CIA drone strike program. At the same time, however, the ad hoc, one-off nature of the Abbottabad raid and the singularly high public profile of the target-not to mention the direct connection to 9/11-gave the bin Laden killing a sui generis quality.
The next substantive public articulation by the administration justifying the broader targeted killing program came at Harvard on September 16, 2011, when Brennan offered a defense for killing beyond "hot" battlefields based on an expansive understanding of imminence of threat. 167 He disclosed no operational facts not already in the public domain, but in combination with Koh's prior statement, Brennan's remarks began in earnest the process of presenting, on the record, an official legal and normative case for what the executive had been doing for the past decade. At the time of these limited steps, however, the very existence of the program to which this legal reasoning pertained remained officially secret, leading to a degree of self-conscious absurdity. Asked directly, "Does the CIA have a drone program?," Brennan refused to plainly acknowledge it, replying jokingly: "If the agency did have such a program, I'm sure it would be done with the utmost care, precision … ." 168 The New York Times report of the event records that the conclusion of Brennan's sentence "was garbled by the laughter of the audience." Brennan's speech came just fourteen days before the second landmark targeted killing that occurred in 2011-that of Anwar al-Awlaki, carried out by CIA drone strike in Yemen on September 30. This incident placed further strain upon secrecy because the president was eager to publicly claim al-Awlaki's elimination as a counterterrorism success 169 and-because al-Awlaki was American-born-it raised the constitutional stakes regarding the executive's asserted right to target and kill.
Tara McKelvey (2013) notes a surge in press scrutiny from mid-2011 onward: "[From] July 2011 to July 2012, the New York Times, the Washington Post, and the Christian Science Monitor published roughly 120 articles, or more than four times the number of articles from a comparable period in the previous twelve months, that looked at legal aspects of the drone program. In addition, these newspapers published 33 articles that looked at moral aspects of the program, more than three times the number of articles during the previous twelve-month-long period." 170 Responding to this pressure, Brennan addressed an audience at the Wilson Center in 2012 and finally unequivocally admitted: "Yes, in full accordance with the law, and in order to prevent terrorist attacks on the United States and to save American lives, the United States Government conducts targeted strikes against specific al-Qaida terrorists." 171 He further noted that he was speaking following an instruction from President Obama "to be more open with the American people about these efforts," 172 and he proceeded to set out elements of its underlying legal basis. This speech was a landmark moment in the administration's public handling of the program. Brennan's remarks were part of a concerted move by the administration: during March and April 2012, several high-level officials engaged in public defense of the administration's use of drones, including Secretary of Homeland Security Jeh Johnson, 173 Attorney General Holder, 174 and CIA General Counsel Stephen Preston. 175 Finally, in 2013, the president himself delivered a speech at the National Defense University seeking to clarify the administration's approach to counterterrorism, especially with regard to targeted drone strikes. 176 In these public statements aimed at legitimating the program, factors cited as relevant included: the state of exception created by a new type of terrorist threat; the unconventional nature of terrorists as non-uniformed armed combatants; the inaccessible locations of those targeted, which put them beyond the reach of capture or law enforcement; the US government's duty to prioritize the lives of the American population; the high risk of imminent attack if the United States did not act; the precision of the technology now available for targeting; and the seriousness and professionalism with which the president and other officials undertook the responsibility of targeting.
With regard to the norm against assassination, the administration fell (to use Keating's terminology) somewhere in between offering justification and advocating innovation. "Assassination," administration officials said, was "repugnant," "unlawful," and not something that the United States practiced or supported. 177 But these killings were not assassinations. "The use of that loaded term is misplaced," said Holder. 178 In his speech, Johnson (2012) directly addressed the relevant history, contending that there was a substantial difference between the US government assassination plots of the 1970s and present practice. "Lethal force against a valid military objective, in an armed conflict is consistent with the law of war," he noted, asserting that the current strikes qualified as such. 179 Key to the administration's justification was the expansive definition of imminence. As noted earlier, this was first sketched officially and publicly by Brennan in his 2011 Harvard speech: "We are finding increasing recognition in the international community that a more flexible understanding of 'imminence' may be appropriate when dealing with terrorist groups, in part because threats posed by non-state actors do not present themselves in the ways that evidenced imminence in more traditional conflicts. … Over time, an increasing number of our international counterterrorism partners have begun to recognize that the traditional conception of what constitutes an 'imminent' attack should be broadened in light of the modern-day capabilities, techniques, and technological innovations of terrorist organizations." 180 Brennan was, to put it mildly, putting a positive spin on the issue of world opinion. Hayden, who had been intimately involved in the drone program during the Bush years, was more frank: "There isn't a government on the planet that agrees with our legal rationale for these operations, except for Afghanistan and maybe Israel." 181 Nevertheless, this position was codified in a Justice Department memo stating that: "The condition that an operational leader present an 'imminent' threat of violent attack against the United States does not require the United States to have clear evidence that a specific attack on U.S. persons and interests will take place in the immediate future. … By its nature … the threat posed by Al Qaeda and its associated forces demands a broader concept of imminence … 182 Resisting Scrutiny and Oversight: The Limits of Openness Throughout the tenure of the Obama administration, the American Civil Liberties Union (ACLU) struggled persistently-using the Freedom of Information Act (FOIA)-for greater transparency regarding targeted killing operations and their legal basis. The White House, in response, fought tenaciously to withhold documents and information. In 2010, the ACLU opened its first lawsuit, asking for "disclosure of the legal basis, scope, and limits on the targeted killing program." 183 In 2012, three additional cases were filed: a lawsuit challenging the killing of al-Awlaki; 184 an information request about 2012 targeted killings in Yemen; and an additional request to disclose information about a drone strike in the al-Majalah region in Yemen. 185 Aside from one minor ACLU victory in 2014, 186 the executive generally succeeded in resisting such demands for disclosure.
The administration also successfully minimized the role of Congress. In 2012, twenty-six Members of Congress signed a request for greater transparency, arguing that targeted killings carried major implications for the United States and the public had the right to know what was being done in their name. 187 When such requests failed to produce results, Senators used confirmation hearings as a forum to press for more information. In 2011, the Senate Select Committee on Intelligence held a hearing on the nomination of General David Petraeus to be CIA Director. Roy Blunt, Republican of Missouri, used the occasion to press the general for details on targeted killings by drones. In response, Petraeus carefully and cannily focused on drone operations in Afghanistan, where such missions operated in an entirely different legal context from those in Pakistan, Yemen, and Somalia. In so doing, he highlighted only the positives: "I would note that the experience of the military with unmanned aerial vehicles is that the precision is quite impressive, that there is a very low incidence of civilian casualties in the course of such operations." 188 In 2013, the same committee pressed Brennan (Obama's nominee to replace Petraeus) harder about the program, since in his previous role he had been one of its chief architects. In his opening statement, he noted that there was a "widespread debate" within the administration about counterterrorism and that policymakers "wrestled with" lethal operations. But he defended the practice by pointing out that the United States was at war with al Qaeda. 189 Senator Ron Wyden of Oregon complained that the committee had never seen a full list of countries in which the CIA carried out lethal operations. 190 Further, it had seen only "two of an estimated 11 legal opinions" on the program. 191 The hearing as a whole served to expose just how little information Congress had, prompting Committee Chair Diane Feinstein to tell reporters afterwards: "I think that this has gone about as far as it can go as a covert activity." 192 In November 2013, the Committee voted 13-2 to require the CIA to reveal how many individuals it believed to have been killed or injured in its strikes. 193 By April 2014 however, this initiative was thwarted when the full Senate removed this specific demand from the relevant bill. Director of National Intelligence James Clapper led administration lobbying against the requirement, arguing that: "To be meaningful to the public, any report including the information described above would require context and to be drafted carefully so as to protect against the disclosure of intelligence sources and methods or other classified information." 194 Official numerical data regarding strikes and casualties would not ultimately be released by the administration until July 2016, and the figures provided at that time were markedly lower than those compiled by outside organizations, leading the New York Times and others to complain that they "answer few questions, and raise many." 195 The administration was similarly resistant to full disclosure of its detailed legal reasoning. It shared with Congress the memos providing legal rationale for the targeted killing of al-Awlaki only when one of the co-authors, David Jeremiah Barron, had his nomination for the 1st US Circuit Court of Appeals blocked until the administration yielded. Even then, it only "allowed lawmakers from a secure room in the Senate, to view copies of two memos written by Barron." 196 The administration declined to share them with the press or public. In 2016, a Stimson Center report accused the administration of "obstructing efforts to develop greater oversight and accountability mechanisms" and reinforcing a "culture of secrecy surrounding the use of armed drones." 197

The Normalization of Targeted Killing
The preceding sections have demonstrated that a substantial shift in government practice took place during the Bush and Obama presidencies. A category of killing that had been treated as forbidden between 1976 and 2001 became routine practice. The change occurred first in secret, then became more widely known under the Obama administration as the scale of the program increased.
A shift of this kind had self-evident potential to generate controversy and opposition. But the Obama administration was ultimately successful in securing and maintaining public support for its actions. A 2012 Washington Post poll found 83% support for Obama's targeted killing policy. 198 A year later, a different survey taken by Gallup showed that almost two-thirds of Americans (65%) approved of the government striking targeted individuals in Pakistan, Yemen, and Somalia. 199 While there were fluctuations over subsequent years, majority support was consistent. 200 This was true among both Republicans and Democrats and held even when respondents were prompted to consider strikes against American citizens living overseas or reminded of the distinction between drone strokes carried out by the CIA and the US military. 201 The reasonable conclusion to be drawn is that public acceptance of the practice was by this time active, not merely premised on ignorance. Organizations dedicated to civil liberties, such as the ACLU, continued to object strongly. But notwithstanding their opposition, the executive was also successful in obtaining consent from the overwhelming bulk of the political elite and governmental institutions. Congress, regardless of the party in the majority-which varied over the course of the period-took no meaningful steps to obstruct the policy. Neither did the courts. Support within the relevant executive agencies was not unanimous. Elliot Ackerman, a CIA officer during the Obama administration, notes the presence of internal dissent, even when official lawyers had carefully articulated for internal purposes a distinction between targeted killing and assassination and blessed the legality of the former. "The discomfort of my colleagues, where it existed, didn't stem from the act itself. … The discomfort existed because it felt like we were doing something, on a large scale, that we'd sworn not to. Most of us felt as though we were violating Executive Order 12333. Everybody knew what was happening-senior intelligence officials, general officers, the administration, even the American people, who ostensibly would not tolerate assassinations carried out in their name." 202 Similarly, Cameron Munter, Obama's Ambassador to Pakistan, resigned from his post, complaining privately that "he didn't realise his main job was to kill people." 203 Yet such objections were marginalized and could not obstruct the program's operation.
A major indicator of how fully the practice was normalized was the extent to which the targeted killing program altered the structures and activities of the CIA itself. Locating and targeting militants on a global basis became the agency's primary task. 204 "We went from a purely espionage organization to more of an offensive weapon, a paramilitary organization where classic spying was less important," a senior officer explained. 205 From having three hundred employees before 9/11, the CIA's Counterterrorism Center grew to two thousand. 206 Mark Mazzetti estimates that more than half of those joining the agency after 9/11 focused exclusively on manhunt and targeted killing operations. 207 While Obama did not initiate this institutional shift, he did accelerate it, propelled by technological advance. Under his authority, the CIA modernized Bush-era targeting practices, turning the kill list into a more sophisticated, constantly-updated database in which "biographies, locations, known associates and affiliated organizations" were catalogued. 208 Updating the target list became a routinized bureaucratic process, with more than one hundred members of the national security apparatus vetting whom should be targeted and where. 209 Targeted killings were no longer exceptional or rare. On the contrary, they were routine and administered systematically. 210 The administration could have sought legitimacy for this shift at the outset through, in Keating's terminology, either overt justification or innovation: that is, through open and active advocacy for the proposition that targeted killing was compatible with established norms, or else that those norms should be updated to render it permissible. Alternatively, it could have done all in its power to keep the killings entirely secret, thus evading the need for legitimation and the risk of failure and backlash entailed in seeking it.
The case detail we have presented here makes it clear that, in practice, the administration declined to make a straight choice between these alternatives, pursuing instead a hybrid path to legitimation via quasi-secrecy. That is: it maintained a tight formal regime of official secrecy, combined with the simultaneous, often unofficial, release of select information designed to portray the efficacy and legality of the program in the most favorable light. Such disclosures allowed the public to become accustomed over time to the existence of targeted killing as US government practice, and this contributed to the goal of normalizing it in the eyes of key domestic audiences. At the same time, official secrecy relieved officials of the need to publicly address inconvenient or unpleasant facts arising from operations or tackle the ultimate logical implications of the program's legal underpinnings, such as locating definitively the outer limits of the executive's asserted prerogative to kill. The administration's success in this case suggests we should consider secrecy not as a binary-and (as some have argued) temporary and counterproductive-alternative to legitimation. Rather, blended with selective disclosure in a strategy of quasi-secrecy, it can be part of an effective mechanism for normalizing potentially controversial innovation. By the time the Obama administration finally officially avowed the targeted killing program, articulated its legal basis, and began to present an overt case in support of its legitimacy, the core operational facts had been de facto publicly known-and painted in a positive light-for years. Consequently, its ultimate, belated official acknowledgement was received by the public not as the shocking announcement of a highly controversial innovation, but merely as confirmation of a long-established government practice, the legal and normative merits of which a reasonable but non-expert observer might suppose had already long since been debated and settled.
The moment at which-in an overt-advocacy-centred model of legitimation-targeted killing should have been pitched to the public as a major but desirable shift in practice simply never occurred. At the time when the shift took place and became de facto publicly known, and for many years thereafter, the executive's position was that official secrecy prevented officials from debating its merits or even frankly acknowledging that it had occurred. By the time the administration was prepared to avow the operational facts of its actions and the legal reasoning underlying them-a sine qua non for meaningful debate-they had ceased to seem sufficiently novel to generate the energetic public engagement they once might have. To the observer's eye, targeted killing appeared to move directly from the category of "outside the bounds of official discussion," to that of "uncontroversial long-established practice," with no way-station in between. For years, the official view was that the time was not yet right to openly litigate the merits of the policy, until all at once it was too late. The utility of quasi-secrecy lies precisely in facilitating this move: it advances the goal of legitimation by cultivating a widespread impression that open debate, resulting in consent, must surely have occurred at an earlier moment, while in fact serving to avert its occurrence at any point. In this way, even a norm of substantial weight may be killed sufficiently softly that the precise moment of its passing fails to register. The ethical and political virtues of such a strategy areclearly-open to question, but this case provides proof of concept for its efficacy. This will no doubt be of interest to future government officials keen to adopt and legitimate potentially controversial new practices. Those who do not identify with the executive branch in this story, however, may consider it a warning regarding maneuvers for which they should remain vigilant.