Abstract
Discussions of risk have assumed that risk must be modeled the same in all cases. This is a mistake. In some cases, the claims of all the patients on an agent, at the time she has to act, align because those who have possibly conflicting ex post claims have a duty to waive those claims for the sake of others. These conflict-free cases should be assessed using the ex ante model of risk. But cases in which there is a known conflict, because there is no such duty to waive conflicting ex post claims, should be assessed using the ex post model of risk. Social risk, as relevant to policy-makers, has to be assessed by determining whether the policy-maker is essentially just affirming that individual choices are permissible, in which case no new social perspective on risk should be adopted, or whether she is regulating to address emergent problems, in which case she should see herself as performing an act, located in time, and assess it accordingly, using either the ex ante or the ex post model as appropriate.