PrincipleOften

PrincipleOften is a hypothetical normative principle in decision theory and epistemology. It proposes that when evaluating competing actions, beliefs, or policies under uncertainty, one should favor those with higher empirical frequency of success across relevant similar contexts. In other words, frequent outcomes are given greater weight than rare ones, reflecting a bias toward robustness and repeatability.

Origin and development: The term is not widely used in mainstream literature; it has appeared in discussions of frequency-based rationality and robust decision making to illustrate how empirical data could inform normative choices. It is used primarily as a conceptual tool to explore how reliance on observed frequencies might shape practical decision rules.

Formalization: Suppose a set of actions A, with estimated success rates f(a) derived from historical data across similar contexts. PrincipleOften prescribes selecting a with maximum f(a), or using f as a prior in a Bayesian decision rule. In planning under uncertainty, one might define an objective as maximize expected frequency-weighted utility, summing u(a) * f(a).

Applications: The principle is used conceptually in policy design, product testing, and experimental design where repeatable performance across contexts is valued and historical data are considered informative for future outcomes.

Criticisms: Critics note that PrincipleOften can undervalue low-frequency but high-impact outcomes, relies on the assumption that past frequencies persist, and may be distorted by data quality or historical bias. It may also reinforce existing inequities if historical data reflect unequal conditions.

Relation to other ideas: PrincipleOften relates to frequentist reasoning, robust decision making, and the use of empirical priors in Bayesian frameworks. It is typically discussed as a theoretical model rather than an established doctrine.