Make Information in Science Meaningful Again

Abstract

Although the everyday notion of information has clear semantic properties, the all-pervasive technical concept of Shannon information was defended being a non-semantic concept. In this paper I will show how this measure of information was implicitly ‘semantized’ in the early 1950s by many authors, such as Rothstein's or Brillouin's, in order to explain the knowledge dynamics underlying certain scientific practices such as measurement. On the other hand, I will argue that the main attempts in the literature to develop a quantitative measure of semantic information to clarify science and scientific measurements, such as Carnap-Bar-Hillel, or Dretske, will not successfully achieve this philosophical aim for several reasons. Finally, I will defend the use of a qualitative notion of semantic information within the information-theoretical framework MacKay to assess the informational dynamics underlying scientific practices, particularly measurements in statistical mechanics.