Probability and Uncertainty
Summary: Good decisions require separating outcome quality from decision quality, making peace with genuine uncertainty, and using base rates to counteract the planning fallacy and overconfidence — skills that run against the brain’s default toward narrative certainty.
Sources: Clippings/Duke-Thinking-in-Bets.md · Clippings/Kahneman-Quotes.md · Clippings/Rosling-Factfulness.md
Source pages: Thinking in Bets — Quotes, Daniel Kahneman — Quotes, Factfulness — Quotes
Last updated: 2026-05-04
Decision quality vs. outcome quality
Duke’s founding distinction: a good decision can produce a bad outcome, and a bad decision can produce a good outcome. “Resulting” — judging the decision by the result — prevents learning:
“What makes a decision great is not that it has a great outcome. A great decision is the result of a good process, and that process must include an attempt to accurately represent our own state of knowledge.” (source: Duke-Thinking-in-Bets.md)
In a world with hidden information and luck (which is most of the world), outcome feedback is noisy. The only way to improve is to evaluate process separately from result. See duke-thinking-in-bets.
Life is poker, not chess
Chess has perfect information. Real decisions don’t:
“Chess, for all its strategic complexity, isn’t a great model for decision-making in life, where most of our decisions involve hidden information and a much greater influence of luck.” (source: Duke-Thinking-in-Bets.md)
Treating decisions as though the correct answer could always be worked out analytically — the chess error — produces overconfidence and misattributes outcomes to skill or blame. Business, health, relationships, and finance all involve poker-like uncertainty. The right response is probabilistic thinking, not more analysis of a fixed board.
Uncertainty is the correct epistemic state
“The secret is to make peace with walking around in a world where we recognize that we are not sure and that’s okay.” (source: Duke-Thinking-in-Bets.md)
“I’m not sure” is not a failure of knowledge — it is an accurate representation of it. Pretending to certainty blocks learning: you can’t update a belief you’ve already marked as settled. See narrative-bias for the mechanism by which the mind produces false certainty through WYSIATI.
The planning fallacy
Kahneman’s term for systematic optimism about future projects:
“We focus on our goal, anchor on our plan, and neglect relevant base rates, exposing ourselves to the planning fallacy.” (source: Kahneman-Quotes.md)
The inside view (“given everything specific to this project”) consistently overestimates speed and underestimates cost. The outside view (“how long do projects like this typically take?”) is almost always more accurate. The fix is not more detail about the current project — it is deliberately importing base rates from comparable past projects. See kahneman-thinking-fast-and-slow and thaler-misbehaving.
Overconfidence is institutionally rewarded
“Experts who acknowledge the full extent of their ignorance may expect to be replaced by more confident competitors, who are better able to gain the trust of clients. An unbiased appreciation of uncertainty is a cornerstone of rationality — but it is not what people and organizations want.” (source: Kahneman-Quotes.md)
False confidence is selected for by clients and markets. This creates structural incentives against the calibrated uncertainty that good probabilistic thinking requires. The forecaster who says “60% chance” loses clients to the one who says “this will happen.” See duke-thinking-in-bets.
Numbers need comparison
Rosling’s rule for interpreting statistics:
“Never leave a number all by itself. Never believe that one number on its own can be meaningful. If you are offered one number, always ask for at least one more.” (source: Rosling-Factfulness.md)
4.2 million infant deaths sounds catastrophic in isolation. Compared to 14.4 million in 1950, it is the lowest in recorded history. Context is not optional for understanding — it is understanding. The base rate is always the second number. Compare the planning fallacy: ignoring base rates is the same error as ignoring the comparison number.
The gap instinct and binary thinking
“human beings have a strong dramatic instinct toward binary thinking, a basic urge to divide things into two distinct groups, with nothing but an empty gap in between.” (source: Rosling-Factfulness.md)
Binary categories (rich/poor, safe/dangerous, good/bad) are cognitively cheaper than distributions. Real-world data almost always clusters in the middle. Treating a continuous distribution as two poles systematically distorts probability estimates. See dual-process-cognition: binary categorization is a System 1 preference.
Possibilism
Rosling distinguishes between optimism (a disposition) and possibilism (a conclusion from evidence):
“I’m not an optimist… I’m a very serious ‘possibilist’… someone who neither hopes without reason, nor fears without reason.” (source: Rosling-Factfulness.md)
The possibilist sees past progress and concludes that further progress is possible — not certain, but probable enough to act on. This is probability applied to worldview: calibrated rather than defaulting to either hope or despair. Compare duke-thinking-in-bets: bets against future selves require honest probability estimation about outcomes, neither wishfully high nor defensively low.
Motivated reasoning blocks updating
Intelligence doesn’t protect against motivated reasoning — it makes it worse:
“IQ is positively correlated with the number of reasons people find to support their own side in an argument.” (source: Duke-Thinking-in-Bets.md)
More cognitive firepower produces better defenses of existing beliefs, not better beliefs. Updating probabilities on new evidence requires noticing that the evidence is against you, which System 1 is designed to avoid. See narrative-bias and dual-process-cognition.