Knight's distinction between risk, where the probabilities are known, and uncertainty, where they are not, has been eroded in practice by a technical convenience. Subjective probability allows a number to be assigned to any proposition on which someone is willing to bet, which dissolves the distinction formally and preserves it entirely in substance: a probability derived from a long run of similar cases and one derived from an expert's willingness to take a wager are different objects that receive the same notation. Decision procedures treat them identically, and the resulting analysis is exact about quantities that were never measured. There is nothing wrong with the formalism. What has gone wrong is that the provenance of a number stops travelling with it, and by the time it reaches a decision nobody remembers which kind it was.