LAWS and the Law: Rules as Impediment to Lethal Autonomy 39

Existenz: An International Journal in Philosophy, Religion, Politics, and the Arts

judgments about collateral harm; they could be designed to incorporate better sensors and more real-time information processing than human soldiers; and they would not be subject to the same cognitive biases that sometimes undermine human judgment (GLB 29–30). For such reasons, Arkin is not alone in advocating for the potential moral benefits of LAWS, and similar arguments have been advanced by others, including recent work by Don Howard.5 Whether these differences would indeed lead to greater compliance with the laws of armed conflict is hard to say, however, the dangers in presuming too much of the technology seem obvious. As much as one should care deeply about the ways that soldiers fail morally, it would be a mistake to ignore the many amazing things that they are capable of doing in virtue of their being human. Arkin is right to point out that human soldiers sometimes fail to follow the laws of war, but with rare exceptions, they fail to do so despite having the ability to understand what the law requires of them. Even when they fail to follow the laws of war, soldiers achieve something that machines thus far have been generally unable to accomplish: In understanding the ways that the law obligates them and constrains their actions, soldiers achieve something that is morally important, even if they fail to make good use of that achievement. Machines may eventually be capable of the same kind of understanding, but quoting my colleague, Richard Schoonhoven, at this point machines "aren't even stupid."6

Making Sense of the Law

The most sensible starting point for analyzing the demands that following the law makes on agents is the concept of law itself. The benefit of starting there, I hope, is to make clear that even the concept of law is not simple—the law consists in various types of rules and rule-like contents, each of which imposes different demands on its subjects. In this section, I

briefly describe three categories of legal contents, before turning to the ways in which the heterogeneity of laws makes legal rule-following especially difficult. In his analysis of law, Paul Boghossian identifies three distinct types of content that compose law: rules, imperatives, and normative propositions.7 On most philosophical analyses of law these three contents compose the law in some measure, but there is significant disagreement regarding the relative contributions of each one. John Austin, for example, argued that laws are the commands (or imperatives) of a sovereign ruler backed by force of sanction, although he recognized that some commands take the more general form of rules.8 Alternatively, H. L. A. Hart argued that laws are social rules, drawing a distinction between those rules that guide conduct and those that allow the creation or modification of new laws.9 In stark contrast with both Austin and Hart, Ronald Dworkin argues that in addition to rules and normative propositions, the law consists not only in rules, but in moral principles and policies, meaning that there are aspects of any legal system that cannot be explained strictly in terms of social facts.10 The point is that there is significant disagreement regarding the concept of law and the role that rules, imperatives, and normative propositions play in constituting the law as such. Importantly, each of these perspectives suggests that there are distinct difficulties for the legal subject charged with, knowing, understanding, and applying the law. Hence, it is important to address the similarities and differences between these types of legal contents.

Boghossian argues that imperatival contents, sometimes called commands or instructions, specify both a condition and an action that is to be performed whenever that condition obtains (ER 474). Imperatives tend to be highly specific in their formulation of both the applicable condition and the requisite behavior.