This study investigates how uncertainty works in science policy debates by considering an unusual case: one in which uncertainty-based arguments for delay come from the scientific community, rather than industry actors. The case I present is the central use of value-added modeling (VAM) in the evaluation of individual teachers, a controversial trend in education reform. In order to understand how policy actors might counter inconvenient statements of uncertainty from experts, I analyze speeches from Education Secretary Arne Duncan, a committed and influential advocate of VAM. I identify a three-part rhetorical tactic, the "Overcaution Allegation," and describe its persuasive potential to legitimize policies that elicit caution from the scientific community because they are built on uncertain science.