Volume 62, 2018
Philosophy of Science
Thomas Bonk
Pages 13-18
Measures of Simplicity
There is a broad consensus that the proper measure of simplicity of phenomenological laws (models) is the number of its free parameters. I argue that the “measure” is specious without a prior understanding of what simplicity is. To this end I propose an empirical interpretation of simplicity. Next, I sketch a general method for assigning degrees of simplicity to the elements of a given function space that complements the empirical characterization. It is shown that a “function space” approach can help overcome difficulties of Bayesian accounts of the curve fitting problem.