Experiment and the Pursuit of Ugly Models

TL;DR

This paper explores how in modern high-energy physics, the experimental context favors pursuing complex, ad hoc models that are either radically novel or easily testable, even if they lack traditional virtues.

Contribution

It analyzes the shift in scientific model pursuit towards 'ugly' models driven by experimental constraints and presents criteria under which such models are considered pursuitworthy.

Findings

Experimental context influences pursuit of complex models.

Four case studies illustrate pursuit of 'ugly' models.

Conditions for pursuitworthiness are outlined.

Abstract

Scientists do not merely choose to accept fully formed theories, they also have to decide which models to work on before they are fully developed and tested. Since decisive empirical evidence in favour of a model will not yet have been gathered, other criteria must play determining roles. I examine the case of modern high-energy physics where the experimental context that once favoured the pursuit of beautiful, simple, and general theories now favours the pursuit of models that are ad hoc, narrow in scope, and complex; in short, ugly models. The lack of new discoveries since the Higgs boson, together with the lack of a new higher energy collider, has left searches for new physics conceptually and empirically wide open. Physicists must make use of the experiment at hand while also creatively exploring alternatives that have not yet been explored. This encourages the pursuit of models that have at least one of two key features: i) they take radically novel approaches, or ii) are easily testable. I present four cases, neutralino dark matter, the relaxion, repulsive gravity, and models of lepton flavour universality violation, and show that even if they do not exhibit traditional epistemic virtues, they are nonetheless pursuitworthy. I argue that experimental context strongly determines pursuitworthiness and I lay out the conditions under which experiment encourages the pursuit of ugly models.