Why do we redistribute so much but tag so little?

The workhorse model of optimal taxation strongly recommends tagging, but its use in policy is limited. I argue that this puzzle is a symptom of a more fundamental problem. Conventional theory neglects the diverse normative criteria with which, as extensive evidence has shown, most people evaluate policy. In particular, if the classic principle of Equal Sacrifice augments the standard Utilitarian criterion, optimal tagging is limited. Calibrated simulations of optimal policy with normative diversity of this type simultaneously match three features of U.S. policy: substantial income redistribution; rejection of gender, race, and height tags; and acceptance of a blindness tag. Additional implications increase the appeal of this revision to conventional theory.