Remarks on the use of citation data in predictive models of scientific activity

Abstract The relations between data, information and prediction are discussed with reference to a suggested Markov model for forecasting of the structure of scientific activity. A conclusion is that the output of mathematical operations, i.e. the predicted “state” or activity distribution, is not related to the input as a deductive or probabilistic consequence of its real information content, but rather to information attributed to the data by assumptions inherent in the method. This is a general difficulty in futures studies, namely that the openness of explanation of a phenomenon effectively limits the possibilities to extend the understanding of systems behaviour into predictive statements.