The talk surveys the various asymptotic and small-sample optimality criteria that have been proposed for testing and estimation.  This is followed by a discussion of objections that have been raised against the study of optimality, and of the appropriate role of optimality considerations.