Probleme der theoretischen Deduktion von Prognosen

As futurology gets more and more attention and while science and technology take more and severe interest in the phenomena of future and their prediction a question becomes highly relevant: what are the presuppositions, the methods, and instruments in order to predict future states and events in a qualified scientific manner? With respect to the philosophy of science there seems to predominate the opinion, scientific laws and theories should be necessary for those predictions because only they can guarantee what we may call a methodologically sufficient base. This reasoning will be demonstrated as inadequate. In the following article hence the author discusses the problems which arise when predictions are drawn from laws and theories by logical deduction. He further examines several possible ways as potenþial strategies for solving or avoiding these problems. By that, the role and power of deductive and inductive logic are described referring to the distinction between the problem of discovery and that of justification in the special case of making predictions. For demonstration the author gives first an analysis of the logical structure of deductive-nomological explanations, predictions, and retrodictions drawn from scientific laws or theories. The problem articulated here is then analyzed in the following three directions: (a) concerning the validity of the L-implication between antecedens and consequens, (b) as a problem of empirical (factual) truth resp. (un-) certainty of the universal laws (referred to in the antecedens), and (c) as a problem of practical application of the laws, leading to the insight that it is necessary to predict the conditions stated in the antecedens too, because they are entirely prospective in character as well. For this reason there will arise the problem of an infinite regress because the process of getting to an absolutely nomologically prediction (as opposed to the retrodiction) must take place in an multi-level hierarchy of interlocking predictions, which has no definite end. Four alternative strategies, as potential ways to solve or to avoid that problem are considered: (a) conditional predictions, (b) assumptions or ad hoc hypothesis, (c) purely instrumental predictions on the basis of total closed systems, and (d) an active strategy to realize antecedent conditions and/or to increase their probability. The discussion of these strategies - being of particular pragmatic and methodological importance - shows that they do