Why 30% of Something Is Better Than 100% of Nothing