Distinguish risk from ambiguity before reacting

Label whether you’re facing known odds or genuinely unknown odds — the right tool depends on the answer.

Why it works

Most discomfort about uncertainty conflates two different things: risk (calculable odds) and Knightian uncertainty (unknown probability distribution). Labeling which you face changes the appropriate response: for risk, expected-value tools apply; for genuine ambiguity, you need robust strategies that perform acceptably across scenarios rather than strategies optimized for assumed probabilities you don’t actually have. The label itself — “this is ambiguity, not risk” — is psychologically useful because it shifts the frame from “I don’t know enough” to “no one knows enough here, so I need a different approach.”

How to do it

1. Write down the decision you’re facing.
2. Ask: do I have reliable historical data on outcome probabilities? If yes, you have risk, not ambiguity.
3. If no reliable data exists, label it “ambiguity” and shift from expected-value reasoning to scenario-robustness reasoning.
4. Log the decision type in IX Coach and note which reasoning mode you applied.

Evidence

Knight’s (1921) risk/uncertainty distinction is foundational in decision theory. Ellsberg (1961) confirmed experimentally that people treat the two differently. No randomized trials exist for the labeling practice itself, but the conceptual framework is theoretically well-grounded. (mechanistic)

The distinction is conceptually clean but difficult in practice: most real decisions fall in a gray zone between calculable risk and pure Knightian uncertainty.

Sources

• Knight, F.H. (1921). Risk, Uncertainty and Profit. Houghton Mifflin.
• Ellsberg, D. (1961). Risk, ambiguity, and the Savage axioms. Quarterly Journal of Economics, 75(4), 643–669.

Common mistake

Treating all uncertainty as risk and applying expected-value calculations to distributions you “made up” — fabricated probability estimates are worse than acknowledging ambiguity.