Traditional definitions of determinism in terms of causation seem nowadays to have been largely superseded by accounts in terms of predictability. If it were true that all and only caused events were predictable then doctrines of universal causation and universal predictability would be equivalent and it would only remain to ask what advantages if any an indirect epistemological account had over a direct ontological one—none, one might have thought, more especially if the former presupposed the latter. In fact, however, the (...) two are by no means so simply and directly related: being caused is neither alone sufficient nor yet again is it necessary for predictability, or so at least I shall be endeavouring here to show. (shrink)