It's all in your head -- fine-tuning arguments do not require aleatoric uncertainty

TL;DR

This paper clarifies misconceptions about naturalness and Occam's razor in Bayesian statistics, showing that Bayesian formalism naturally favors simpler, less fine-tuned models across various disciplines.

Contribution

It provides a comprehensive review of naturalness arguments and demonstrates that Bayesian methods inherently discourage fine-tuned, unnatural models.

Findings

Bayesian formalism automatically enforces Occam's razor.

Unnatural models requiring fine-tuning are disfavored by Bayesian reasoning.

The paper bridges perspectives from statistics, physics, and machine learning.

Abstract

Prompted by misconceptions in the recent literature, we review the justifications for naturalness arguments and Occam's razor found in Bayesian statistics. We discuss the automatic Occam's razor that emerges in Bayesian formalism, bringing together points of view from diverse fields, including statistics, social sciences, physics and machine learning. In pedagogical calculations, we demonstrate that this automatic razor disfavors unnatural models in which predictions must be fine-tuned to agree with observation.