State Minimization Re-visited

A well-known circumscription policy in situation calculus theories of actions is to minimize the Abnormality predicate by varying the Holds predicate. Unfortunately this admitted counter-intuitive models. A different policy of varying the Result function eliminated these models. Explanations of how it did this are not entirely satisfactory, but seem to appeal to informal notions of state minimization. We reexamine this policy and show that there are simple justifications for it that are based on classical automata theory. It incidentally turns out that the description "state minimization" for the varying Result policy is more accurate than the original nomenclature had intended.