The nature of data — From measurements to systems

In spite of the fact that the concept of data is at the foundation of the sciences, we do not present it in our computer education in the context of a scientific theory, based on laws or immutable principles. Rather, by drilling computer programming, we teach it as dependent upon current technology, destined to be obsolete in 10–15 years. The aim of this paper is to reverse this trend. In a historical perspective and across a broad spectrum of disciplines, we trace the development of data from the basic notion of measurement to the description of theoretical systems and models. From this survey we deduce a common framework of ideas which, developed by the great scientists of the past, suggests how we may establish a modern theory of data, conceived as a distinct geometry characterized by data laws or invariances.