We Need More Goat Sacrifices

Major social policies and philanthropic programs should begin with a ritual goat sacrifice - not to appease gods, but to force decision-makers to confront how little they know. Well-intentioned interventions often fail: school grants produced no discernible gains, diversity officers did not increase representation, a major Afghan jobs program could not establish whether women got jobs, and a long-running laptop initiative did not improve students’ academic performance. Other examples include ineffective vaccine messages, Medicaid work requirements that reduced insurance coverage without increasing employment, and relationship counseling that failed to meet its goals. These failures echo sociologist Peter Rossi’s satirical “Iron Law”: large social programs tend to have zero net impact, and rigorous evaluations are especially likely to find null results. Promising findings often weaken on replication or when programs are tried in new settings.

A deeper cause is an illusion of control: people mistake good intentions for evidence that their actions will work. The classic study said to show that people value lottery tickets more when choosing their own numbers actually used football cards, and participants who chose cards simply selected players they liked. Real lottery replications found no picking advantage. Yet people still feel that their thoughts and actions can shape chance, and this confidence can lead them to intervene without testing results. Many federally funded STEM projects are never experimentally evaluated, and some government programs lack even defined goals or progress data. The proposed sacrifice is a comic safeguard against this overconfidence: make evaluation and humility prerequisites for trying to change the world.