The questions of whether and how to target individuals for killing, and the consequences of doing so, raise difficult issues. Allen Buchanan and Robert O. Keohane, two acute observers of world politics and international institutions, have offered a template for international regulation and accountability that seeks to address some of the dilemmas raised by targeted drone strikes. No short article can consider every major issue in such a proposal; all work, especially initial attempts to pioneer an area of inquiry and offer the rationale and basic structure for a new international regime, must make assumptions and set limits in order to focus on essential issues. I suggest, however, that the assumptions and the analytical and empirical limits of Buchanan and Keohane's proposal are so problematic that they vitiate the value of the proposed international regime.

Buchanan and Keohane suggest that because of “the large scale of major terrorist attacks,” terrorism is an act of war, not criminal violence. Thus, the “war paradigm”—not the “law enforcement paradigm”—best fits the problem of preventing and responding to large-scale terrorist threats. While the choice to act from within the war paradigm is obviously consequential, I do not have space to parse it here.Footnote ¹ I suggest, however, that the problem of terrorism can and probably ought to be approached from both war and law enforcement paradigms. The question is deciding which paradigm takes precedence and explicating the kinds of circumstances that determine when one or the other should be applied.

In the dozen years since the United States began to use drone strikes for targeted killing there have been relatively few public articulations of the strategy by administration officials. In May 2013, when President Barack Obama made the criteria for U.S. strikes public, he implied that the United States was using both law enforcement and war paradigms:

Buchanan and Keohane put terrorism in the war paradigm because of the “large scale of major terrorist attacks.” However, while the overall number of deaths from terrorist attacks continues to increase, most attacks result in relatively small-scale loss of life.Footnote ³ Specifically, the attacks of September 11, 2001, on the United States resulted in approximately 3,000 killed; by comparison the three worst terrorist attacks in 2013 killed, respectively, 142, 123, and 107 people. Most attacks resulted in many fewer deaths.Footnote ⁴

Regardless of the scale of the attacks planned or actually conducted by terrorists, international criminal law is already available to prosecute those who deliberately target and kill noncombatants in both large and small numbers. Buchanan and Keohane would likely respond that it is impractical to arrest many terrorists, and that the risk to one's own forces or to bystanders of attempting to arrest terrorists, who are not likely to surrender peacefully and who shelter among civilians, may be high. However, we could say the same about the people, including heads of state, who have been indicted, arrested, and tried under the statutes of international criminal courts. Indeed, those who have been indicted by international criminal courts have often already caused much greater harm than the targeted leaders of terrorist organizations, some or perhaps most of whom have not yet planned or executed large-scale attacks.

The belief that drone strikes minimize the risk to soldiers is important. If they indeed reduce risks to combatants, police, and civilians, their use is arguably warranted. If they reduce risk to combatants but hurt civilians directly, or indirectly as a consequence of increased militancy in reaction to drone strikes, then the utilitarian case for drone strikes is weakened. We must consider whether and when the desire to lower risks to law enforcement or military personnel ought to outweigh the importance of lowering risks to noncombatants.

Advocates of the war paradigm may have overestimated the risks of the law enforcement paradigm, specifically the risk to police and civilians of attempting to arrest suspected terrorists. For instance, the journalist Mark Bowden suggests that “under law-enforcement rules, a lot more people, both soldiers and civilians, are likely to be killed.” Bowden uses as an example the famous 1993 “Black Hawk Down” raid in Mogadishu, the aim of which was to arrest Omar Salad and Mohamed Hassan Awale. “We were not officially at war with Somalia, but the ensuing firefight left 18 Americans dead and killed an estimated 500 to 1,000 Somalis.”Footnote ⁵ But this is a mistaken comparison. Yes, the aim of that particular raid was arrest, but the U.S. military was using the tools of war for police purposes in a crowded location in the larger context of a “humanitarian intervention.” It is unlikely that a police operation, using a law enforcement paradigm, would have attempted an arrest of suspects under such conditions. A policing paradigm could have avoided the incident and its attendant carnage altogether, as police would not have had the firepower to cause so many civilian deaths. More importantly, the attempted arrests would not have occurred in the first place, but would have been postponed until the situation was less risky to police and bystanders.

The more apt comparison is between drone strikes and raids by special forces, such as the 2011 attempt to arrest Osama bin Laden in Abbottabad, Pakistan. True, the risks to the Navy Seal team were higher than if the United States had used a missile strike, but much greater discrimination was possible in a special operations raid, and thus there were relatively few noncombatant casualties.

On the other hand, the strongest reason to put drone strikes in the category of the war paradigm is that the strikes—which aim to kill combatants in a foreign country—are themselves an international use of force and thus arguably amount to an act of war. If we do accept, for the sake of argument, that the war paradigm is the appropriate lens through which to examine terrorism and counterterrorism, and drone strikes in particular, it would be helpful to think through the problem of the legitimacy, accountability, and regulation of drone strikes using the ethical frameworks that are available to evaluate uses of force, including but not limited to the just war tradition.Footnote ⁶ This immediately raises jus ad bellum questions of the permissibility of drone strikes as a mode of self-defense, whether the strikes are necessary, and whether they are preventive, preemptive, or punitive. Because drone strikes often target leaders of organizations who are thought to be planning an attack, this puts many drone strikes in the category of preemptive and preventive uses of force. Preemption may be legal and permissible in some circumstances, but preventive strikes are generally not considered legitimate. The categorization of terrorism and drone strikes as war also raises the jus in bello consideration of discrimination and proportionality, discussed at greater length below.

My second area of concern is that Buchanan and Keohane limit the scope of their proposal to the use of remotely piloted vehicle drone strikes. That is, they do not want to consider other modes of “targeted killing,” at this point, as acts to be covered by their proposed accountability regime. Yet the larger strategy, of which drone strikes are a part, is the attempt to kill the leaders of organizations who pose a terrorist threat because they may have planned or conducted terrorist acts. There is no compelling logical reason to accept that drone strikes are distinct from other methods of targeted killing, such as those used by the United States against Viet Cong leaders, or those that were employed by apartheid South Africa against leaders of the African National Congress residing in southern Africa and Europe, or those currently practiced by Israel against leaders of Palestinian organizations. All aim to kill individuals that the targeting state perceives as a threat.

Further, the current U.S. targeted killing program uses various means, not simply drone strikes. For instance, cruise missile strikes, which victims sometimes mistake for drone attacks, have been used for targeted killing in Yemen and Somalia. And on more than one occasion the United States has also sent Navy SEALs to attack suspected and known al-Qaeda militants in Pakistan.

On the other hand, as Buchanan and Keohane suggest, one might argue that drones should be treated first on the reasoning that it may be easier to regulate a technology that is not yet widely available or widely used. That may be the case. But the political prudence argument can cut the other way; if one consequence of a drone accountability regime is to legitimize this method of targeted killing, it may unintentionally legitimate targeted killing by other means that remain unregulated and for which states remain legally unaccountable.

Third, Buchanan and Keohane limit their focus to “the regulation of lethal drone use as it now predominantly occurs [emphasis added] and can be expected to occur in the near to medium future.” They also say that, “if usage changed, then the difference between drones and boots-on-the-ground occupation might diminish or even disappear. This would be the case if ‘clouds’ of drones permanently patrolled airspace within a country and enforced embargos at its borders.” They imply that drone strikes are currently undertaken in small number for limited ends, but should such strikes become frequent and ubiquitous, the authors would reconsider their approach. Yet Buchanan and Keohane fail to appreciate the current pattern of U.S. drone strikes and the already changing nature of the U.S. drone strike program. It is arguably the case that drone strikes have already become part of a larger policy project that violates the sovereignty of peoples.

The U.S. drone strike program has already expanded in several ways. Specifically, the geographic zones where strikes occur have been extended beyond the initial strike zone in Yemen (2002) to Pakistan (in 2004) and Somalia (in 2007). Further, U.S. officials have increased the categories of persons who can be targeted, such that the “kill list” now includes those identified as high-level and immediate threats to the United States as well as lower-level militants. True, we have been assured that the list is vetted at the highest levels under strict criteria; but the use of “signature” strikes—whereby people, whose identities are not known, are targeted based on a pattern of activity—suggests a great deal of room for less “discriminating” results.Footnote ⁷ Furthermore, while the number of drone strikes fluctuates, overall they have become more frequent—and the U.S. capacity to make such strikes is growing. Indeed, the United States plans more drone aircraft development, production, procurement, and deployment. In addition to increasing the number of Reaper drones and the support structure for those aircraft, the United States plans to procure the new armed Gray Eagle drone. The total drone strike force will thus grow beyond what is necessary to conduct a limited program of targeted killing.Footnote ⁸

Moreover, the weapons payload of drones has become increasingly deadly. Predator drones, which carry two “Hellfire” antitank weapons, have been supplemented by the Reaper drone, which carries up to four Hellfire missiles and two laser-guided 500-pound bombs. As the weapons payload of drone aircraft grows, and as the number of targets and strikes increases, we can expect these precise weapons to have less discriminating consequences. A highly accurate laser-guided 500-pound bomb is still a 500-pound bomb, with a radius of blast destruction (depending on where it detonates) of 100 or 200 feet. Drone strikes intended to disrupt and destroy terrorist networks also disrupt and destroy elements of civilian life. Specifically, people who live under drone flight zones in Pakistan and Yemen describe a sense of constant surveillance, in addition to the fear and pain associated with civilian injury and death due to drone strikes.Footnote ⁹

Fourth, while Buchanan and Keohane frame the rationale for drone strikes in consequentialist terms—suggesting the strikes are effective and, indeed, more effective than alternative methods—they do not, for the most part, critically examine the empirical and policy questions of the tactical and strategic role of drone strikes. Nor do they critically evaluate claims about the effectiveness of the strikes and their consequences for civilians. Rather, they assume the advantages of drones (that the drone strike program is inexpensive, accurate, flexible, capable of gathering intelligence, and reduces risks to U.S. combatants) over alternative tools and appear to take the effectiveness of drone strikes for granted.Footnote ¹⁰ For example, echoing arguments made by President Obama and other advocates of drone strikes, Buchanan and Keohane claim that “lethal drones are capable of greater precision in targeting, with less collateral damage to persons or property than conventional weaponry.” While they also state that “the number of civilian deaths from drone strikes is not negligible,” they believe that drone strikes have “violated the discrimination principle much less than other forms of violence directed against terrorism.”

The comparison of drone strikes to conventional weapons is misleading in two senses. Drones are conventional weapons platforms that carry conventional weapons: bombs and missiles with significant payloads. What is perhaps intended is a comparison between small-scale uses of force for limited missions and large conventional military assaults that entail capturing territory, defending it, and taking prisoners. But conventional war is not the relevant comparison; assuming for the moment that targeted killing of militants is legitimate, the relevant comparison is between methods of targeted killing. The most discriminate weapon for targeted killings is arguably a knife or another method of close-in killing. Although the “collateral damage” consequences of drone strikes is disputed, drones using Hellfire missiles or laser-guided bombs would almost certainly have higher levels of collateral damage than other methods of targeted killing, such as strangling, sharpshooting, or raids by special operations forces. Harm to civilians in drone strikes using bombs or missiles is more likely to occur than with these alternative methods simply because militants often shelter with their families and can be indistinguishable from civilians.

The assumption that drone strikes are discriminating and effective must be evaluated against the record. Although the numbers of civilians killed and injured by U.S. drone strikes is disputed, there is no disputing, as Buchanan and Keohane acknowledge, that drone strikes for targeted killing in Pakistan, Yemen, and Somalia have caused civilian deaths and injuries. The following figure, using data compiled by the Long War Journal (LWJ)—a conservative source with ties to the U.S. intelligence community—illustrates the trend in drone strike numbers and deaths in Yemen from 2002 to November 2014. In my view, LWJ undercounts civilian deaths in drone strikes. But, even using their conservative assessment, in the 107 strikes LWJ tracked in Yemen from 2002 to late 2014, it attributes a total of 105 civilian deaths to drones, in comparison to 538 militant deaths. Although LWJ data suggest that civilians account for 16 percent of those killed, they also suggest that the average number of civilian deaths caused by each drone strike has diminished over time.

Figure 1:

Long War Journal Count of Drone Strikes and Deaths in Yemen, 2002–November 2014Footnote ¹¹

The numbers of drone strikes and people killed in Pakistan are higher than in Yemen, and sources often disagree about the number of civilians killed and injured in such strikes. Long War Journal and The Bureau of Instigative Journalism (TBIJ) count approximately the same number of strikes between 2004 and late 2014—373 and 402, respectively—but differ markedly on the number of civilians killed and the proportion of civilian to militant deaths. LWJ counts 156 civilians killed (5 percent of the total), while TBIJ counts between 416 and 953—which would account for between 16 and 25 percent of the total deaths.Footnote ¹²

The requirement announced by President Obama in May 2013 that “before any strike is taken there must be near-certainty that no civilians will be killed or injured” is admirable, but unlikely to be met.Footnote ¹³ Signature strikes against suspected terrorists are almost by definition indiscriminate. Further, even if signature strikes are ended, it is foreseeable—if not foreseen in every case—that drone strikes will cause some civilian casualties for the reasons described above.

If it were clear, or at least plausible, that the military advantage of drone strikes outweighed the cost to civilians, then a logic of proportionality and military necessity might diminish concerns about civilian casualties. But this is not the case. Indeed, it may be that drone strikes are at best ineffective and at worst counterproductive to their stated purpose, namely, diminishing the capacity of militants to attack the United States or its allies. Patrick Johnston and Anoop Sarbahi find a short-term (one week) reduction in terrorist violence in Pakistan after drone strikes.Footnote ¹⁴ Others find that the consequences of the strikes may be negligible or counterproductive, increasing militancy.Footnote ¹⁵ In recent cases, Pakistani militants have claimed that their terrorist attacks on civilians were in retaliation for drone strikes.Footnote ¹⁶

Figure 2 compares data from LWJ with data from the Global Terrorism Database for bombings and explosions by militants in Yemen that target private citizens and their property, NGOs, tourists, and educational institutions.

Figure 2:

Comparing Drone Attacks with Terrorist Attacks in Yemen, 2002–2013Footnote ¹⁷

As this figure suggests, it is not at all clear that drone strikes are diminishing the overall threat of terrorist strikes—including against civilian targets. More analysis is needed before their effectiveness, and assertions of their proportionality and necessity, can be accepted.

Further, whether or not it is the intention of the United States, drone strikes are perceived by the people affected by them as war, and as part of a larger policy project that violates their sovereignty. This is in part because drone strikes are generally not conducted in isolation from other tactics. The strikes in Pakistan and Yemen have been associated with larger-scale conventional attacks by U.S. allies, leading to additional thousands of civilian deaths and injuries. The Pakistani counterterrorist campaign in the areas where militants operate and U.S. drone strikes have occurred have killed, wounded, and displaced thousands of Pakistani civilians since 2004.Footnote ¹⁸ The U.S. State Department and others have documented the abuse of civilians by the Pakistani government. Data by the Pakistan Institute for Peace Studies suggest that more than 150 civilians were killed in Pakistani military “operational attacks” against, or clashes with, militants in 2013.Footnote ¹⁹ Overall, as the Institute for Economics and Peace argues, “the presence of state-sponsored violence, such as extrajudicial killings, political terror, and gross human rights abuses,” is correlated with increased terrorism.Footnote ²⁰

Fifth, and finally, Buchanan and Keohane focus on international regulation and accountability, and largely defer the question of domestic regulation and accountability. They claim that it would be easier to make an international accountability regime before making a domestic system of accountability. This point is arguable; in any case, the ease of establishing such a regime is only one of several important considerations.

With only one power using drones for targeted killing, the place to regulate drone strikes is not first at the international level. Rather, domestic accountability for drone strikes is what is first required. The current level of accountability and oversight in the U.S. drone targeted-killing program arguably undermines democracy and accountability.Footnote ²¹ In his May 2013 speech, President Obama said he asked his “administration to review proposals to extend oversight of lethal actions outside of war zones that go beyond our reporting to Congress.”Footnote ²² The president mentioned options, but more than a year later efforts to increase oversight and accountability have largely failed. While domestic regulation and public accountability is still minimal, we ought not to underestimate the potential for regulating drone strikes through the domestic political process and for demanding accountability within it. While I entirely agree with Buchanan and Keohane that some measure of regulation and accountability of targeted killing via drone strikes is necessary, I am reluctant to call for an international regime when the system of domestic accountability and regulation is currently opaque and anemic.Footnote ²³ Rather, a robust and transparent system of domestic accountability for drone strikes could be the model for an international regime.