Reliability, Power, Functions, and Relations: A Reply to Humphreys