Scientific Judgment (Polanyi)

Scientific judgment is Polanyi's term for the evaluative capacity that distinguishes genuine scientific expertise from mechanical competence in following procedures. It is the ability to recognize which problems matter, which data are significant, which hypotheses are worth testing, which results are surprising, which explanations are elegant. This judgment operates tacitly: the experienced researcher feels when an experiment is on the right track, when a result is too good to be true, when a theory has been stretched beyond its domain of validity. The judgment cannot be reduced to explicit rules—two equally competent scientists examining the same evidence may reach different evaluations, both grounded in genuine expertise, because their tacit grounds differ. What makes scientific judgment reliable is not its conformity to rules but its grounding in the researcher's years of embodied engagement with the domain—the accumulated sensitivity to patterns, anomalies, and the boundaries between the understood and the mysterious. AI systems lack this judgment entirely: they produce outputs consistent with training-data patterns but cannot assess whether those outputs represent genuine advances in understanding or merely probable