Chance, Probability, & Uncertainty at the Edge of Human Reasoning
What is Knightian Uncertainty?
Key Finding
Knightian uncertainty is not merely extreme risk or high ambiguity — it is a categorically distinct epistemic condition where probability itself has no meaning.
Overview
This paper provides a rigorous conceptual clarification of what Knightian uncertainty actually is — and what it is not. Despite a century of use in entrepreneurship theory, the concept remains frequently conflated with risk, ambiguity, and other forms of incomplete knowledge. The paper systematically distinguishes these concepts and maps the precise boundaries where human reasoning reaches its limits.
Contribution to the Research Program
This is foundational work for the Knowledge Problems & Entrepreneurial Reasoning stream. By clarifying the concept that underlies the entire research program, this paper ensures that subsequent work — on AI and computability, on epistemic risks of generative AI, on entrepreneurial judgment — rests on solid conceptual ground. It is the definitional anchor for the lab's treatment of uncertainty.
Key Insights
- Knightian uncertainty is categorically distinct from risk (quantifiable probability) and ambiguity (unknown probability)
- Under genuine uncertainty, the sample space itself is unknown — not just the probabilities
- This distinction has practical implications for how entrepreneurs reason and how AI systems can (and cannot) assist
- The paper maps the boundary conditions where probabilistic reasoning breaks down entirely