Scholars in many fields of research not only try to explain real world phenomena but also aim to predict the future. Such fields are, among many others, technology assessment, innovation studies, or – less abstract – climate research. At least these examples also show that frequently policy consultancy is at stake; to shape the future politicians in particular and society in general need, interalia, scientific advice. However, in such cases one important question often is not asked: How reliable are predictions of the near and far future? To the contrary, the use of computer simulations in order to produce predictions of future events and to visualize these predictions seems to suggest that such predictions are particularly reliable.
Therefore, it is imperative for those who must base their decisions amongst other things on scientific advice to know whether this advice is resilient. Philosophy in general and philosophy of science in particular can be applied to find at least some criteria to measure the reliability of predictions based on computer simulations.
In the proposed presentation at first the methodological pitfalls of the use of computer simulations to predict the future shall be identified – these are theoretical issues, problems arising from implementation, and questions concerning the empirical foundation of computer simulations. In order to do so, examples from diverse scientific disciplines will be used. Secondly, it will be demonstrated how these pitfalls can corrupt the outcomes of computer simulations and how they might computer simulations primarily might support our understanding of the respective field of research. However, hope that using computer simulations to predict the future provide for reliability is misguided. The outcomes of computer simulations cannot be more reliable than the theories, the empirical foundation, and the implementation they are basing on are reliable.