Scientific innovation and the limits of social scientific prediction

Abstract

Philosophers and historians of philosophy have come to recognize that at the core of logical positivism was an attachment to prediction as the necessary condition for scientific knowledge.1 The inheritors of their tradition, especially the Bayesians among us, continue to seek a theory of confirmation that reflects this epistemic commitment. The impor tance of prediction in the growth of scientific knowledge is a com mitment I share with the positivists, so I do not blanch at that design ation, much less employ it as a term of abuse. Precisely expressing and conclusively establishing the claims of pre diction as a necessary condition for certifying claims as increments of knowledge is a goal that has so far eluded us post-positivists. Philoso phers know the problems well: defining a positive instance, distin guishing projectible from nonprojectible predicates, deciding whether retrodiction is as epistemically probative as prediction.2 But I can't help thinking that these problems are technicalities important and arresting, but not impediments to embracing the positivist demand that increments in scientific knowledge withstand tests of predictive success.