“All Models are Wrong”: What Does it Mean to “Agree with Experiment” in Computational Chemistry?

Professor Jeremy Harvey

Department of Chemistry, KU Leuven, Leuven, Belgium

In my talk, which comes from the standpoint of a long-time practitioner of computational methods in chemistry (especially those based on quantum chemistry), who also has an interest in philosophy of science, I will try to express some provocative opinions relating to the goals and practice of computational chemistry, and thereby to encourage a renewal of some common beliefs in the field. I will start from the uncontroversial statement made in his famous quote by George Box: “All models are wrong”. I will then argue that given that all models are wrong, the current practice whereby the success of computational studies is often evaluated by whether their results “agree with experiment” is not necessarily desirable, and leads to significant distortions in the field. The second part of the famous quote is that “but some of them are useful”, and I will attempt to explain why I believe it is more important to perform computations as a means to construct useful models than to necessarily agree with experiment.

Recording: