Key Insights: Thinking in Bets
Annie Duke spent two decades as a professional poker player before turning to cognitive science. Her argument is narrow: the quality of a decision is not the same as the quality of its result, and most of our habits of judgment assume it is.
5 Pillars of Success
1. Resulting: stop grading decisions by the scoreboard
Duke's term for judging a decision purely by how it turned out is *resulting*. Every outcome blends two things we can never separate - the quality of the decision and the luck surrounding it - and we observe only the blend. A good call that loses gets filed as a mistake; a reckless call that wins gets filed as genius.
Her running example is the 2015 Super Bowl, when Seattle's Pete Carroll called a pass from the one-yard line instead of handing the ball to his running back. The pass was intercepted, and the call was instantly declared one of the worst in football history. Duke argues it was defensible given the down, the clock and the defensive personnel. One result is weak evidence about one decision, even though better decisions win over time.
2. Life is poker, not chess: convert beliefs into odds
Chess has complete information; poker hides cards, and identical reasoning can produce opposite results. Duke argues that business, hiring and medicine are poker. Her mechanism is a question: would you bet on that, and at what odds?
Stating an opinion costs nothing, which is why we state them with false confidence. Pricing a belief forces a number out of the holder, exposes how thin the evidence is, and makes the claim scoreable later. That is why Duke treats "I am 60 percent sure" as accuracy rather than weakness: a belief with no probability attached cannot be audited.
3. Accuracy over being right: build a truth-seeking pod
We do not reason toward truth; we reason toward the conclusion we already prefer. Motivated reasoning picks the destination and recruits the arguments, and contrary evidence is held to a higher standard than evidence that flatters us.
Duke's fix is structural: a small group that agrees in advance to pursue accuracy rather than approval. Challenge ideas and not people, ask for probabilities, treat "I don't know" as a contribution. She describes the poker players she traded hand histories with, a group whose purpose was finding each other's errors, and recommends a "decision swear jar" for hindsight phrases such as "I should have known."
4. Mental time travel: 10-10-10, premortems, backcasting
Emotional heat shortens our time horizon, so Duke moves the question into the future. The 10-10-10 exercise, borrowed from Suzy Welch, asks how a choice will feel in ten minutes, ten months and ten years; the immediate feeling loses authority against the longer horizons.
Applied to a team, the same tool is the premortem: declare that the project has already failed and have everyone write the history of that failure. Prospective hindsight produces more specific reasons than a generic risk list, and it gives people permission to voice doubts they would not otherwise raise. Backcasting is its mirror: assume success and reconstruct how it happened.
5. Precommitment and a written record
Knowing your biases does not remove them, and judgement is worse in the middle of an emotional state than before it. Duke's answer is to decide earlier than the moment of temptation: a Ulysses contract that binds your future self through a penalty, an obstacle to the bad option, or a person who must approve the choice.
The other half is memory. Hindsight bias rewrites what we believed before the result arrived, so a decision journal - options, probabilities and reasoning written down before the outcome is known - is what makes calibration possible. The target is saying 70 percent and being right seven times in ten.
Key Takeaway
"Decision quality is not identical to outcome quality. Even the most brilliant strategy can lose due to pure luck."
Practical Task
1. Pick one consequential decision from the last month. Write down what you knew at the time, the alternatives and your probabilities - before checking the outcome. Grade the process separately from the result. 2. Start a decision journal. For your next five decisions record the options, a percentage for each and one sentence of reasoning, then review an entry only after the result is known. 3. Replace one word this week: when you catch yourself saying "definitely," say a number instead, out loud, until the percentage stops feeling like an apology. 4. Ask "would you bet on that, and at what odds?" in your next meeting. If nobody will price the claim, treat it as enthusiasm rather than evidence. 5. Run a premortem before your next commitment: twenty minutes writing the story of how it already failed, then turn the two most plausible failures into mitigations. Recruit one or two people who will tell you the truth about your decisions.