Despite the lead article’s title “Validity Concerns and Usefulness of Student Ratings of Instruction” (Greenwald 1997) in the American Psychologist’’s special section on teacher ratings, the papers did not provide direct evidence on “usefulness.” There is no evidence that the use of teacher ratings improves learning in the long run. The papers do not show that the effects would improve the allocation of effort between teaching and research, or that the quality of the educational experience will be better, or that students and faculty will be happier. Given the evidence to date, the case for student ratings is weak. I raise some questions about usefulness, with a particular emphasis on the ratings’ effects on learning. Are teacher ratings related to learning? The correspondence of ratings to learning should be very close, otherwise, it is not clear what will change. When I was involved with worker incentive standards, the prevailing wisdom was that even modest departures from the desired measure created dysfunctional outcomes. Perhaps the most important finding from the special section is that teacher ratings and learning are not closely related. Furthermore, the ratings/learning relationship seems to be based heavily on studies involving rote learning. The studies provide no breakdown of evidence for studies where skill