I am grateful for the opportunity to respond to Neil C. Renic’s thoughtful review of my book The Drone Age: How Drone Technology Will Change War and Peace. Since there are many points on which we agree, I focus this reply on the issues of risk redirection with drone strikes and the responsibility that drone users like the United States bear for it.

My book argues that drones redistribute risk away from pilots and toward a number of other groups, such as informers on the ground, local security forces, and civilians caught in the crossfire. Renic goes further, arguing that “those exposed to drone violence are rational actors who will modify their behavior” and encourage technological innovation (such as the adaptation of drones) to inflict more civilian casualties in response to drone use by their adversaries. His argument hinges on an analogy with the law of conservation and suggests that risk can never be reduced in absolute terms. It can be reduced in one sphere but only at the cost of being necessarily redirected elsewhere.

My objection to this stronger claim is both conceptual and empirical. On the conceptual side, I am not convinced that risk can never be reduced in absolute terms for both sides of a conflict. Some technologies—for example, precision-guided missiles—obviously reduce risk for the perpetrator but may also reduce risks faced on the opposite side if deployed in ways to minimize civilian casualties, carefully monitored to ensure accountability, and used no more recklessly than the alternative options. When precision-guided missiles are compared to a reasonable alternative, such as dropping an unguided bomb from 35,000 feet, the former may actually reduce the overall risk of civilian casualties for both sides. The problem is that the conditions for reducing the overall risk to both sides with new technology are stringent and compliance is difficult to monitor. Over time, states may succumb to moral hazard and use technology like drones more recklessly than they did before. They may also fight more often as a consequence of having new technologies that minimize their risks. This is why I have argued it is crucial to be attuned to how new technology like drones shape our strategic choices, specifically the calculations of risk and goals.

My empirical objection concerns what we know and how we understand the decision-making calculations of terrorist actors. As Renic acknowledges, given the limits of the current evidence, “we cannot know” how much violence against civilians emerges as a compensatory strategy to deal with the inaccessibility of combatant targets. Terrorist actors like the Islamic State are naturally risk-taking and have a range of reasons—strategic, organizational, ideological, and so on—to prey on civilians. Their decision-making processes are opaque, and the precise combination of reasons that leads them to brutalize civilians may be more complex than a reaction to technological asymmetry. We also cannot assume that this decision always stems from a reaction to a Western adversary like the United States, given that such indiscriminate violence is common and has a range of local and regional audiences. For this reason, we cannot automatically assume the moral responsibility of the United States for these attacks on civilians absent reliable evidence of a link between these attacks and the inaccessibility of combatants. That said, the mere possibility of such redirection suggests a precautionary obligation for the United States to do what it can to ensure that civilians are not bearing extra risks in response to its use of drones. Even if we do not know whether these second-order effects are in play, it is wise to act as if they are and to deploy drones in ways that maximize both restraint and accountability.