In some legal contexts, decision-making is guided by algorithms that rely on statistical generalizations about groups to tell us what to believe about individuals. Sometimes, these algorithms incorporate characteristics like race, gender, and socioeconomic status, raising particularly acute equity concerns. The NFL’s “race norming” scandal (in which race-based assumptions about intelligence made it harder for Black players to obtain settlement payouts for cognitive injuries) is a recent prominent example, and the NFL abandoned these algorithms under pressure. But similar practices pervade other important legal settings, including criminal sentencing and calculation of civil damages—not to mention many non-legal settings that the law has something to say about, such as medical diagnosis. So what should we think about such practices? This lecture will explore and critique the deep inconsistencies in the law’s treatment of them, and place this discussion in the context of influential economic thinking about the differences between “statistical discrimination,” “taste-based discrimination,” and “stereotyping.”