Epistemic uncertainty

Epistemic uncertainty is uncertainty about facts within a known model — the kind of uncertainty that can, in principle, be reduced by more data, better measurements, or larger samples. It is the uncertainty of the weather forecast, the clinical trial, the election poll: we know what the possibilities are, we have a model for how they arise, and our ignorance is a function of finite information.

Epistemic uncertainty is distinct from ontological uncertainty, which concerns unknown possibilities, and from Knightian uncertainty, which concerns unmeasurable probabilities. In complex systems, the boundary between epistemic and ontological uncertainty is not fixed: a phenomenon that appears to be epistemically uncertain may reveal itself as ontologically uncertain when the model fails. The ludic fallacy is the error of treating all uncertainty as epistemic — of assuming that the world is like a casino where the rules are known and only the outcomes are hidden.