It is a commonly expressed opinion (among statisticians) that causal structure cannot be inferred from correlational structure and (among philosophers) that scientific induction cannot be automated. Recently, the philosopher Paul Humphreys and statistician David Freedman attacked Peter Spirtes, Clark Glymour and Richard Scheines for presuming to automate scientific induction by inductively inferring causal from correlational structure in their computer program TETRAD II. Their attacks are not dependent upon the particularities of TETRAD II and, if true, would undermine all current attempts to automate the inductive inference of causal models and Bayesian networks. We assess their arguments and find them wanting.