The independence axiom (which forces linearity of preferences over lotteries and underlies expected-utility maximization) is a contingent assumption, not an unavoidable fact. Dropping it yields consistent, well‑studied alternative decision frameworks (e.g., prospect theory, rank‑dependent utility) that change how we should model rational choice under risk and uncertainty.
— If policymakers, economists and AI designers stop treating expected utility as sacrosanct, regulation, risk assessment, and algorithmic decision‑systems may be redesigned around different, possibly more realistic, norms of rationality.
Ihor Kendiukhov
2026.03.15
100% relevant
The article invokes the von Neumann–Morgenstern axioms, singles out independence as analogous to Euclid’s parallel postulate, and points to existing non‑expected‑utility theories and Dutch‑book arguments to illustrate the alternative.
← Back to All Ideas