Good regulation comes from good guidance. IRBs need a coherent and legible ethics but get three principles – respect for persons, justice, and beneficence. A workable ethics is unachievable because IRBs regulate too many diverse areas. And such an attempt would encounter a core contradiction: IRB ethics privilege the autonomy principle, but IRBs are “deeply and pervasively paternalistic,” for they tell people whether and how they may participate in research. Yet research subjects are just as competent to make choices as patients contemplating surgery, homeowners choosing a mortgage, or skiers picking a resort. Although IRBs lack a legible ethics, they have a potent ethos. Instead of showing that IRBs’ usefulness, that ethos recites a litany of research scandals. But scandals are evils, not arguments.