Model Validity and Semantics of Information

Do predictions obtained from models constitute information on which reliable decisions can be made? Is it necessary, that to be of interest, predictions and other information generated by models must be true? This paper investigates the relation between the model and reality, information and truth. It will argue that meaningful data need not necessarily be true in order to constitute information. Partially true information or even completely false information can lead to a desirable outcome such as a technological innovation or a scientific breakthrough. Sometimes sheer serendipity gives rise to an invention. A combination of true and false information may result in an epoch-making event such as Columbus’ discovery of America, on his intended voyage to India. An even more basic problem prevents scientists from thinking exclusively in terms of “true” information in the research process. In beginning from an existing theory (say Aristotelian physics), and developing a new theory (say Galilean physics) one can talk about the truth within each model, but during the transition between the two, there is a mixture of old and new concepts in which truth is not well defined. Instead of the veridicity of a model, the two basic concepts that are commonly used in empirical sciences are models correctness (validity) and its appropriateness within a context. The conclusion is that despite the empirical models being in general not true but only truthlike, they may nevertheless produce results from which adequate conclusions can be drawn, and therefore can serve as the grounds for decision-making. In that sense they can yield information vital for improving our knowledge about the actual empirical world that is the precondition for technological innovation and scientific discovery.