Having established that the laws of war guide state action during wartime and how international norms codified into law shape strategic calculations, I turn to the question of change in expectations and the role of law in such changes. The analysis so far has assumed the law is fixed to focus on how law shapes what states and soldiers do during wartime. But the law of war has evolved over time with experience of it in practice. How can we understand these changes? States drew up the laws of war to create an effective system of limiting violence during war. Some of these changes eliminated points of ambiguity in the law; others expanded the coverage of the law to new issues and novel problems.

The central question of constructivism is how norms and identities change over time; the answer is they are (surprise!) socially constructed. At least eight mechanisms of normative change – natural selection, cultural selection, social learning, persuasion, coercion, imitation, socialization, the evolution of empathy and other universal values – operate through at least five vectors of change – international negotiations, social movements, epistemic communities, international bodies, and shared senses of community and its purpose (Wendt 1999, 318–366; Finnemore 2003, 146–161). The common claim across all of these mechanisms and vectors is that change is not efficient; it is not directed with the conscious aim of realizing mutual benefit. In the words of Peter Katzenstein, “institutions do not merely create efficiencies” (1996, 518). Social construction leads to a world not of our own choosing or design. It results from a social process outside the control of either any one actor or all of them together. Institutions are not optimal; indeed, they may be pathological, acting in ways to subvert their own purposes (Barnett and Finnemore 1999). Explanations of social institutions that center on rational design are “driving with the rearview mirror” because they can only explain what has been, not what can be (Wendt 2001).