Parsimony

Parsimony is a methodological principle that favors explanations or models that use the fewest assumptions, variables, or entities necessary to account for the evidence. In its commonly cited form, it is associated with Occam's razor: among competing hypotheses, the simplest one that explains the data is preferable. Parsimony is a heuristic, not an inviolable law.

Historically, the idea is linked to William of Ockham, a 14th-century English philosopher and theologian. The maxim entia non sunt multiplicanda praeter necessitatem—entities should not be multiplied beyond necessity—captures the essential intuition: unnecessary complexity should be avoided.

In science and statistics, parsimony guides model selection by balancing fit and complexity. Penalties for additional parameters appear in information criteria, such as AIC and BIC, and in methods that prefer the tree or model with fewer changes in phylogenetics. Across disciplines, parsimonious explanations are valued for their interpretability and falsifiability.

Limitations include the risk of underfitting when a model is too simple to capture important structure, or context dependence in what counts as a "simple" model. Thus, parsimony is a practical heuristic and a guiding principle rather than an absolute criterion.