Miraculous consilience? Constraints on formulations of the no miracles argument

Document Type


Publication Date



History and Philosophy


The intuition that the success of science is a mark of sciences power to track the nature of our world, to permit us confidence that the descriptions of our best sciences are true (or approximately true), is not new. Puntam (1975) popularized an argument based on this intuition; this argument is known as the No Miracles Argument. Providing a convincing formulation of this argument has proven difficult, and recently Magnus and Callender (2004) have suggested that any probabilistic formulation (given form by Bayes theorem) is fallacious. Earlier formulations suffer from problems just as deep; best explanation formulations are question-begging, for example. In this article, I propose two constraints for those attempting to formulate the argument: first, we ought to narrow the scope of the argument, focusing on only a special set of successes, and, second, the epistemic principle that connects the intuition to the realist conclusion ought to be one that is defeasible (rather than probabilistic or categorical). These constraints serve to avoid problems already set out in the literature, and make the No Miracles Argument a sort of local inference – one that is licensed by the case at hand, rather than a general rule. That is, you may only sometimes infer to the truth or approximate truth from success.