SEMINAR SCIENTIFIC REALISM: ON THE ADVANTAGE OF NOVELTY

It would seem that when a theory accommodates known data, it is only moderately confirmed, but if it predicts new and unexpected phenomena, this cannot happen by chance: the theory must be true. This also provides the "ultimate argument for scientific realism". On the other hand, if confirmation is a logical relation between theory and data, it should not matter at all whether the data were known in advance or not. Moreover, what about the successful predictions of theories later proven false? Besides, what exactly is novelty? Being altogether unknown before? Or unknown to the theorist? Or not explained by any other theory? Or not used in developing the theory? Perhaps there are not only data for theories, but also data about theories, and perhaps the novelty of predictions is one of them. If so, the modalities of discovery matter to justification. Or perhaps there is no sharp dichotomy, but a gradual passage between accommodation and novel prediction, marked by varying degrees of unlikelihood: sometimes accommodation may be highly convincing, and novel prediction may impress little. Ê