Definition
The St. Petersburg paradox is Daniel Bernoulli's 1738 problem: a hypothetical gamble with infinite expected value that no reasonable person would pay much to play, used historically as a foundational case for distinguishing expected value from optimal decision-making.
Why it matters
The St. Petersburg paradox is the founding intellectual problem of expected utility theory — Bernoulli proposed log utility as the resolution, beginning a 250-year tradition of using utility curvature to reconcile expected-value calculations with observed and intuitive behavior. Ergodicity economics offers a different resolution: the paradox arises because expected value is an ensemble average, while the question of how much a single individual should pay is a long-run-experience question, and the two diverge sharply for this game. Naming the paradox and its modern resolution is what connects the historical problem to the contemporary ergodicity-economics framework.
How it works
In the St. Petersburg game, a fair coin is flipped until tails appears, and the payoff doubles with each successive head before the tails: one dollar if tails comes up on the first flip, two dollars if it takes two flips, four if three, eight if four, sixteen if five, and so on without limit. The expected payoff, computed the usual way by weighting each possible payoff by its probability and adding the weighted payoffs together, is the infinite sum of unit contributions — and so the expected value of the gamble is infinite. The conventional resolution since Bernoulli has been that a reasonable agent uses log utility (or some other concave utility function), which makes the expected utility of the gamble finite even though the expected payoff is infinite. The ergodicity-economics resolution observes that the expected payoff is an ensemble average over the population of possible game outcomes; the long-run average payoff to a single individual playing the game repeatedly is finite, because the rare extreme payoffs that drive the ensemble average to infinity do not occur in any finite sequence of plays. The two resolutions are mathematically related: under multiplicative dynamics, log-utility maximization coincides with long-run growth maximization.
In practice
For an individual evaluating real-world decisions that ostensibly have very high expected values — venture investments with rare extreme outcomes, lotteries with positive expected payoff, asymmetric bets with fat-tailed payoffs — the St. Petersburg structure is the warning that expected value alone is not a sufficient decision criterion. The practical move is to ask whether the expected-value figure is dominated by rare extreme outcomes that no individual realizes in a finite horizon, and whether the long-run outcome to a single individual differs meaningfully from the cross-sectional expected value. The same logic applies to investment strategies with low-probability, high-magnitude payoffs: the ensemble expected value can be favorable while the long-run outcome to a single investor is unfavorable.
In the Longevity Standard Framework
St. Petersburg paradox is the authoritative historical case in ergodicity economics for the structural distinction between ensemble-average and time-average reasoning, and is the foundational problem from which the modern framework's critique of expected-value reasoning under non-ergodic dynamics descends. The Longevity Standard framework's individual-path orientation — operating on the participant's specific arrangement and planning horizon rather than on ensemble-average expectations — is in the intellectual tradition that the ergodicity-economics resolution of the paradox formalizes. Realized value is the per-individual metric that operates in the same register as the time-average resolution of the paradox: what an individual experiences rather than what the population of possible outcomes averages to.
Related terms
- Expected value paradox
- Expected utility theory
- Log utility
- Ergodicity
- Non-ergodic system
- Kelly criterion
- Time average
- Ensemble average