Name: 赌注思维
Author: Annie Duke
ISBN: 978-0735216358

One-sentence summary: Better decisions come from treating choices as bets under uncertainty, separating decision quality from outcome luck, and building habits that make beliefs easier to update.

Key Ideas

1. Every Important Choice Is a Bet

Annie Duke argues that decisions are not declarations of certainty; they are bets about how the future might unfold. A bet does not have to involve money. Choosing a job, hiring a candidate, launching a product, trusting a forecast, or deciding how to respond in a relationship all involve staking resources on an uncertain future.

This framing matters because it changes the psychology of decision-making. When people say "I know this will work," they often hide uncertainty from themselves and from others. When they say "I am betting this has a 70% chance of working," they make their assumptions visible. The bet frame turns vague confidence into something that can be discussed, challenged, and improved.

A bet also forces tradeoffs into the open. There is no option with no risk; there are only different distributions of possible outcomes. The better question is not "Can I be sure?" but "What odds am I getting, what do I stand to gain or lose, and what would make me change my mind?"

Practical application: Before a consequential decision, write it as a bet: "I am choosing Option A because I estimate X% chance of benefit, Y% chance of cost, and Z assumptions must hold." This converts intuitive confidence into a decision record that can later be reviewed.

2. Resulting Distorts Learning

One of the book's central concepts is "resulting," the habit of judging a decision by how it turned out. If an outcome is good, we assume the decision was good. If an outcome is bad, we assume the decision was bad. Duke shows why this is an unreliable way to learn.

Poker makes the problem obvious. A player can make the statistically correct move and still lose because the next card is unlucky. Another player can make a reckless move and win because randomness happens to favor them. If both players judge only by the final pot, the skilled player may abandon a sound process while the reckless player becomes overconfident.

Life has the same structure, but the randomness is harder to see. A startup can fail despite a strong decision process because market timing shifts. A risky shortcut can appear harmless because no immediate damage occurs. Resulting teaches the wrong lesson from both kinds of event.

Practical application: Run postmortems that separate process from outcome. Ask two questions independently: "Was the outcome good or bad?" and "Was the decision process good or bad given what we knew at the time?" Improvement comes mostly from the second question.

3. Good Thinking Requires Probabilistic Language

Duke's approach depends on replacing all-or-nothing language with probability estimates. Beliefs are rarely simply true or false in practical life. They are more like degrees of confidence that should move as evidence changes.

Saying "I am 60% confident" is more useful than saying "I think so." The number may be imperfect, but it invites better conversation. Someone can ask why the estimate is not 40% or 80%. They can ask what evidence would move it. They can compare it with their own estimate.

Probabilistic language also reduces ego attachment. If a belief is a percentage, updating it is not a humiliation; it is calibration. The goal becomes accuracy rather than victory in an argument.

Practical application: In meetings and personal decisions, ban unsupported certainty for high-stakes questions. Replace "definitely," "obviously," and "no chance" with confidence levels, ranges, and explicit assumptions.

4. Beliefs Are Bets We Keep Updating

Duke treats beliefs as wagers on reality. Every belief implies predictions about what evidence should appear if the belief is accurate. That means beliefs should be updated when evidence arrives, just as a bettor updates odds when new information changes the game.

The hard part is that humans protect beliefs as if they were possessions. Once a belief becomes part of identity, contrary evidence feels like a personal attack. Motivated reasoning then steps in: people search for confirming evidence, reinterpret disconfirming evidence, and remember events in ways that defend the original view.

Thinking in bets asks for a lighter grip. A belief should be held with enough conviction to act, but not so tightly that it cannot be revised. This is especially important for leaders, investors, scientists, product teams, and anyone operating in environments where feedback is ambiguous.

Practical application: Attach a review trigger to important beliefs. For example: "If churn rises above 8%, I will reduce my confidence in our onboarding hypothesis," or "If three independent candidates reject the offer for similar reasons, I will revisit our compensation assumptions."

5. Decision Quality and Outcome Quality Are Different Variables

The book gives readers a simple but powerful distinction: outcome quality is what happened; decision quality is how sound the choice was when made. Confusing the two is one reason people overlearn from luck and underlearn from process.

A good decision can produce a bad result because uncertainty remains after the decision. A bad decision can produce a good result because luck sometimes pays. Over enough repetitions, process matters. Over a single event, outcome can be noisy.

This distinction makes decision review more honest. It prevents blaming people for every bad result and prevents celebrating every lucky success. It also helps people stay calm when a well-made choice fails, because the failure becomes data rather than proof of incompetence.

Practical application: Use a two-axis review matrix after major decisions: good process / bad process on one axis, good outcome / bad outcome on the other. Celebrate good process even when the outcome was unlucky, and investigate bad process even when the outcome was lucky.

6. A Truth-Seeking Group Beats Solo Rationality

Duke emphasizes that people are poor at detecting their own motivated reasoning. We are much better at spotting weak logic, missing evidence, and convenient stories in other people's thinking. A well-designed group can therefore improve decision quality if it is committed to truth rather than social comfort.

The group must have norms. Members should reward uncertainty, challenge reasoning without attacking identity, and avoid judging only by outcomes. A decision group should be able to say, "That was a lucky win," or "That was a good decision with a bad result," without turning the conversation into blame or flattery.

Duke draws on scientific norms sometimes summarized as CUDOS: shared information, universal standards, disinterested evaluation, and organized skepticism. These norms are hard to maintain informally, so they need explicit agreement.

Practical application: Create a small decision circle for important work or life choices. Before asking for advice, share your confidence level, the evidence you have, what you may be biased toward, and what would change your mind. Ask the group to evaluate the reasoning, not to validate your preference.

7. Mental Time Travel Improves Choices

Many bad decisions come from giving too much weight to the immediate emotional moment. Duke uses the 10-10-10 style of thinking to widen the time horizon: how will this decision feel in 10 minutes, 10 months, and 10 years?

The point is not that the long term always wins. Sometimes immediate needs matter. The value of the exercise is that it separates short-term emotion from medium- and long-term consequences. A painful conversation may feel terrible in 10 minutes but beneficial in 10 months. A tempting shortcut may feel good now but embarrassing later.

Mental time travel also helps reduce regret. When people make a decision with multiple time horizons in view, they are less likely to be surprised by predictable future feelings. They can choose with a fuller picture of the emotional and practical consequences.

Practical application: For emotionally charged decisions, write three short paragraphs: the 10-minute self, the 10-month self, and the 10-year self. Let each perspective vote, then decide with all three in view.

Frameworks and Models

The Decision Quality Matrix

Use this matrix to avoid confusing process with luck.

	Good outcome	Bad outcome
Good process	Earned success or favorable variance	Unlucky loss; preserve the process and review assumptions
Bad process	Lucky win; do not reinforce the behavior	Clear failure; fix the process

The most dangerous quadrant is bad process with good outcome. It feels like success, but it teaches people to repeat fragile behavior. The most emotionally difficult quadrant is good process with bad outcome. It feels like failure, but it may contain the best evidence that the team can think clearly under uncertainty.

Step 1: Record the decision process before the outcome is known.
Step 2: After the outcome, classify process quality and outcome quality separately.
Step 3: Reward process improvements, not just visible wins.

Bet Framing Template

A decision can be written as a structured bet.

I am choosing: [option]
I estimate: [probability distribution]
The upside is: [benefits]
The downside is: [costs]
My key assumptions are: [assumptions]
Evidence that would change my mind: [signals]
Review date: [time]

This template is especially useful when the decision is complex enough that memory will later rewrite the original reasoning. It creates a record of what you believed before knowing the outcome. It also helps other people challenge the decision while the assumptions are still visible.

Belief Updating Loop

Belief -> Confidence level -> Prediction -> Evidence -> Updated confidence

The loop treats beliefs as living estimates. A belief that never changes despite new evidence is probably serving identity, status, or comfort rather than accuracy.

Belief: State what you think is true.
Confidence level: Attach a percentage or range.
Prediction: Define what you expect to observe.
Evidence: Compare reality with the prediction.
Updated confidence: Move the estimate up, down, or sideways.

Truth-Seeking Group Norms

A strong decision group needs operating rules.

Accuracy over agreement: The goal is to be less wrong, not to preserve every opinion.
Process over outcome: Review reasoning even when the result was good.
Uncertainty over certainty theater: Confidence levels are preferred to declarations.
Specific evidence over vibes: Claims should be tied to observations, data, or explicit reasoning.
Change of mind as progress: Updating a belief is treated as a win, not a loss.

These norms prevent advice sessions from becoming social reassurance. The group exists to improve judgment. Without norms, a group can easily become a confirmation machine. With norms, it can become a practical substitute for the rationality people wish they had alone.

Decision Journal Protocol

A decision journal is the operational version of thinking in bets. It preserves the state of mind before hindsight rewrites it. It does not need to be elaborate; it needs to be consistent.

Context: What decision is being made, and why now?
Options: What alternatives are seriously available?
Odds: What probabilities are assigned to the main outcomes?
Assumptions: What must be true for the choice to work?
Emotions: What feelings may be distorting the estimate?
Review: When will the decision be evaluated?

The value of the journal appears over time. Patterns become visible. Some people discover that they are habitually overconfident. Others discover that they avoid high-upside options because they overweight embarrassment or short-term discomfort.

Common Failure Modes

Outcome worship: Treating the scoreboard as the full explanation of decision quality.
Certainty performance: Using confident language to gain status or avoid scrutiny.
Identity defense: Protecting a belief because changing it would feel like losing face.
Selective evidence: Gathering data mainly from sources that support the preferred option.
Single-story review: Creating one clean narrative after the fact and ignoring alternate paths.
Unpriced downside: Focusing on expected benefit while leaving the cost of failure vague.
No review loop: Making decisions without setting a point for calibration.

Each failure mode can be reduced with explicit probability estimates, written assumptions, and a group that is allowed to challenge reasoning. The methods are simple, but the social discipline is demanding. That is why Duke returns repeatedly to process design rather than merely telling readers to be smarter.

Key Quotes

"Life is poker, not chess." — Annie Duke

"What makes a decision great is not that it has a great outcome." — Annie Duke

"Resulting is a routine thinking pattern." — Annie Duke

"All decisions are bets." — Annie Duke

"Being wrong feels bad." — Annie Duke

Connections with Other Books

thinking-fast-and-slow: Kahneman's distinction between fast intuition and slow analysis explains many of the cognitive biases Duke wants readers to manage with probability language and decision review.
the-undoing-project: Michael Lewis tells the story behind the research tradition that made bias, heuristics, and judgment under uncertainty central to modern psychology.
nudge: Thaler and Sunstein focus on choice architecture; Duke focuses on the internal and social architecture required to make better choices under uncertainty.
the-signal-and-the-noise: Nate Silver's work on forecasting complements Duke's emphasis on probability, calibration, and updating beliefs when new evidence arrives.
influence-the-psychology-of-persuasion: Cialdini shows how people are moved by social and cognitive triggers; Duke shows how to defend decision quality against similar psychological pressures.
the-psychology-of-money: Housel's treatment of luck, risk, and behavior in financial life echoes Duke's warning that outcomes can hide the difference between skill and chance.
thinking-in-systems: Meadows helps explain why outcomes are often delayed, nonlinear, and hard to attribute, which is exactly why Duke warns against simplistic learning from results.

When to Use This Knowledge

When a user is judging a past decision only by whether it worked out.
When a team needs to review a launch, hire, investment, or strategic bet without blame-driven hindsight.
When someone is overconfident and needs to express uncertainty more precisely.
When a leader wants to build a culture where changing one's mind is respected.
When a product, business, or investment decision has several plausible futures rather than one obvious answer.
When a person is stuck in motivated reasoning and needs to ask what evidence would change their mind.
When a group needs a practical structure for decision journals, premortems, or postmortems.
When short-term emotion is dominating a decision with long-term consequences.

Agent Usage Notes

Use this book when the user needs help improving judgment, not merely when they ask about poker. The poker examples are a teaching device for uncertainty, variance, and feedback quality. The transferable lesson is that reality often gives noisy feedback, so humans need better processes for learning.

This book is especially useful for strategy discussions. It can help an agent ask for base rates, confidence levels, decision records, and review criteria. It also gives language for challenging a user's conclusion without sounding dismissive: "What odds would you put on that?" is often more productive than "Are you sure?"

The book is less useful when the decision is purely deterministic or rule-bound. It is most useful when the situation contains incomplete information, delayed feedback, emotional stakes, or multiple plausible futures. In those contexts, thinking in bets gives both a mindset and a workflow.

# 赌注思维 > **One-sentence summary:** Better decisions come from treating choices as bets under uncertainty, separating decision quality from outcome luck, and building habits that make beliefs easier to update. ## Key Ideas ### 1. Every Important Choice Is a Bet Annie Duke argues that decisions are not declarations of certainty; they are bets about how the future might unfold. A bet does not have to involve money. Choosing a job, hiring a candidate, launching a product, trusting a forecast, or deciding how to respond in a relationship all involve staking resources on an uncertain future. This framing matters because it changes the psychology of decision-making. When people say "I know this will work," they often hide uncertainty from themselves and from others. When they say "I am betting this has a 70% chance of working," they make their assumptions visible. The bet frame turns vague confidence into something that can be discussed, challenged, and improved. A bet also forces tradeoffs into the open. There is no option with no risk; there are only different distributions of possible outcomes. The better question is not "Can I be sure?" but "What odds am I getting, what do I stand to gain or lose, and what would make me change my mind?" **Practical application:** Before a consequential decision, write it as a bet: "I am choosing Option A because I estimate X% chance of benefit, Y% chance of cost, and Z assumptions must hold." This converts intuitive confidence into a decision record that can later be reviewed. ### 2. Resulting Distorts Learning One of the book's central concepts is "resulting," the habit of judging a decision by how it turned out. If an outcome is good, we assume the decision was good. If an outcome is bad, we assume the decision was bad. Duke shows why this is an unreliable way to learn. Poker makes the problem obvious. A player can make the statistically correct move and still lose because the next card is unlucky. Another player can make a reckless move and win because randomness happens to favor them. If both players judge only by the final pot, the skilled player may abandon a sound process while the reckless player becomes overconfident. Life has the same structure, but the randomness is harder to see. A startup can fail despite a strong decision process because market timing shifts. A risky shortcut can appear harmless because no immediate damage occurs. Resulting teaches the wrong lesson from both kinds of event. **Practical application:** Run postmortems that separate process from outcome. Ask two questions independently: "Was the outcome good or bad?" and "Was the decision process good or bad given what we knew at the time?" Improvement comes mostly from the second question. ### 3. Good Thinking Requires Probabilistic Language Duke's approach depends on replacing all-or-nothing language with probability estimates. Beliefs are rarely simply true or false in practical life. They are more like degrees of confidence that should move as evidence changes. Saying "I am 60% confident" is more useful than saying "I think so." The number may be imperfect, but it invites better conversation. Someone can ask why the estimate is not 40% or 80%. They can ask what evidence would move it. They can compare it with their own estimate. Probabilistic language also reduces ego attachment. If a belief is a percentage, updating it is not a humiliation; it is calibration. The goal becomes accuracy rather than victory in an argument. **Practical application:** In meetings and personal decisions, ban unsupported certainty for high-stakes questions. Replace "definitely," "obviously," and "no chance" with confidence levels, ranges, and explicit assumptions. ### 4. Beliefs Are Bets We Keep Updating Duke treats beliefs as wagers on reality. Every belief implies predictions about what evidence should appear if the belief is accurate. That means beliefs should be updated when evidence arrives, just as a bettor updates odds when new information changes the game. The hard part is that humans protect beliefs as if they were possessions. Once a belief becomes part of identity, contrary evidence feels like a personal attack. Motivated reasoning then steps in: people search for confirming evidence, reinterpret disconfirming evidence, and remember events in ways that defend the original view. Thinking in bets asks for a lighter grip. A belief should be held with enough conviction to act, but not so tightly that it cannot be revised. This is especially important for leaders, investors, scientists, product teams, and anyone operating in environments where feedback is ambiguous. **Practical application:** Attach a review trigger to important beliefs. For example: "If churn rises above 8%, I will reduce my confidence in our onboarding hypothesis," or "If three independent candidates reject the offer for similar reasons, I will revisit our compensation assumptions." ### 5. Decision Quality and Outcome Quality Are Different Variables The book gives readers a simple but powerful distinction: outcome quality is what happened; decision quality is how sound the choice was when made. Confusing the two is one reason people overlearn from luck and underlearn from process. A good decision can produce a bad result because uncertainty remains after the decision. A bad decision can produce a good result because luck sometimes pays. Over enough repetitions, process matters. Over a single event, outcome can be noisy. This distinction makes decision review more honest. It prevents blaming people for every bad result and prevents celebrating every lucky success. It also helps people stay calm when a well-made choice fails, because the failure becomes data rather than proof of incompetence. **Practical application:** Use a two-axis review matrix after major decisions: good process / bad process on one axis, good outcome / bad outcome on the other. Celebrate good process even when the outcome was unlucky, and investigate bad process even when the outcome was lucky. ### 6. A Truth-Seeking Group Beats Solo Rationality Duke emphasizes that people are poor at detecting their own motivated reasoning. We are much better at spotting weak logic, missing evidence, and convenient stories in other people's thinking. A well-designed group can therefore improve decision quality if it is committed to truth rather than social comfort. The group must have norms. Members should reward uncertainty, challenge reasoning without attacking identity, and avoid judging only by outcomes. A decision group should be able to say, "That was a lucky win," or "That was a good decision with a bad result," without turning the conversation into blame or flattery. Duke draws on scientific norms sometimes summarized as CUDOS: shared information, universal standards, disinterested evaluation, and organized skepticism. These norms are hard to maintain informally, so they need explicit agreement. **Practical application:** Create a small decision circle for important work or life choices. Before asking for advice, share your confidence level, the evidence you have, what you may be biased toward, and what would change your mind. Ask the group to evaluate the reasoning, not to validate your preference. ### 7. Mental Time Travel Improves Choices Many bad decisions come from giving too much weight to the immediate emotional moment. Duke uses the 10-10-10 style of thinking to widen the time horizon: how will this decision feel in 10 minutes, 10 months, and 10 years? The point is not that the long term always wins. Sometimes immediate needs matter. The value of the exercise is that it separates short-term emotion from medium- and long-term consequences. A painful conversation may feel terrible in 10 minutes but beneficial in 10 months. A tempting shortcut may feel good now but embarrassing later. Mental time travel also helps reduce regret. When people make a decision with multiple time horizons in view, they are less likely to be surprised by predictable future feelings. They can choose with a fuller picture of the emotional and practical consequences. **Practical application:** For emotionally charged decisions, write three short paragraphs: the 10-minute self, the 10-month self, and the 10-year self. Let each perspective vote, then decide with all three in view. ## Frameworks and Models ### The Decision Quality Matrix Use this matrix to avoid confusing process with luck. | | Good outcome | Bad outcome | |---|---|---| | **Good process** | Earned success or favorable variance | Unlucky loss; preserve the process and review assumptions | | **Bad process** | Lucky win; do not reinforce the behavior | Clear failure; fix the process | The most dangerous quadrant is bad process with good outcome. It feels like success, but it teaches people to repeat fragile behavior. The most emotionally difficult quadrant is good process with bad outcome. It feels like failure, but it may contain the best evidence that the team can think clearly under uncertainty. - **Step 1:** Record the decision process before the outcome is known. - **Step 2:** After the outcome, classify process quality and outcome quality separately. - **Step 3:** Reward process improvements, not just visible wins. ### Bet Framing Template A decision can be written as a structured bet. ```text I am choosing: [option] I estimate: [probability distribution] The upside is: [benefits] The downside is: [costs] My key assumptions are: [assumptions] Evidence that would change my mind: [signals] Review date: [time] ``` This template is especially useful when the decision is complex enough that memory will later rewrite the original reasoning. It creates a record of what you believed before knowing the outcome. It also helps other people challenge the decision while the assumptions are still visible. ### Belief Updating Loop ```text Belief -> Confidence level -> Prediction -> Evidence -> Updated confidence ``` The loop treats beliefs as living estimates. A belief that never changes despite new evidence is probably serving identity, status, or comfort rather than accuracy. - **Belief:** State what you think is true. - **Confidence level:** Attach a percentage or range. - **Prediction:** Define what you expect to observe. - **Evidence:** Compare reality with the prediction. - **Updated confidence:** Move the estimate up, down, or sideways. ### Truth-Seeking Group Norms A strong decision group needs operating rules. - **Accuracy over agreement:** The goal is to be less wrong, not to preserve every opinion. - **Process over outcome:** Review reasoning even when the result was good. - **Uncertainty over certainty theater:** Confidence levels are preferred to declarations. - **Specific evidence over vibes:** Claims should be tied to observations, data, or explicit reasoning. - **Change of mind as progress:** Updating a belief is treated as a win, not a loss. These norms prevent advice sessions from becoming social reassurance. The group exists to improve judgment. Without norms, a group can easily become a confirmation machine. With norms, it can become a practical substitute for the rationality people wish they had alone. ### Decision Journal Protocol A decision journal is the operational version of thinking in bets. It preserves the state of mind before hindsight rewrites it. It does not need to be elaborate; it needs to be consistent. - **Context:** What decision is being made, and why now? - **Options:** What alternatives are seriously available? - **Odds:** What probabilities are assigned to the main outcomes? - **Assumptions:** What must be true for the choice to work? - **Emotions:** What feelings may be distorting the estimate? - **Review:** When will the decision be evaluated? The value of the journal appears over time. Patterns become visible. Some people discover that they are habitually overconfident. Others discover that they avoid high-upside options because they overweight embarrassment or short-term discomfort. ### Common Failure Modes - **Outcome worship:** Treating the scoreboard as the full explanation of decision quality. - **Certainty performance:** Using confident language to gain status or avoid scrutiny. - **Identity defense:** Protecting a belief because changing it would feel like losing face. - **Selective evidence:** Gathering data mainly from sources that support the preferred option. - **Single-story review:** Creating one clean narrative after the fact and ignoring alternate paths. - **Unpriced downside:** Focusing on expected benefit while leaving the cost of failure vague. - **No review loop:** Making decisions without setting a point for calibration. Each failure mode can be reduced with explicit probability estimates, written assumptions, and a group that is allowed to challenge reasoning. The methods are simple, but the social discipline is demanding. That is why Duke returns repeatedly to process design rather than merely telling readers to be smarter. ## Key Quotes > "Life is poker, not chess." — Annie Duke > "What makes a decision great is not that it has a great outcome." — Annie Duke > "Resulting is a routine thinking pattern." — Annie Duke > "All decisions are bets." — Annie Duke > "Being wrong feels bad." — Annie Duke ## Connections with Other Books - [[thinking-fast-and-slow]]: Kahneman's distinction between fast intuition and slow analysis explains many of the cognitive biases Duke wants readers to manage with probability language and decision review. - [[the-undoing-project]]: Michael Lewis tells the story behind the research tradition that made bias, heuristics, and judgment under uncertainty central to modern psychology. - [[nudge]]: Thaler and Sunstein focus on choice architecture; Duke focuses on the internal and social architecture required to make better choices under uncertainty. - [[the-signal-and-the-noise]]: Nate Silver's work on forecasting complements Duke's emphasis on probability, calibration, and updating beliefs when new evidence arrives. - [[influence-the-psychology-of-persuasion]]: Cialdini shows how people are moved by social and cognitive triggers; Duke shows how to defend decision quality against similar psychological pressures. - [[the-psychology-of-money]]: Housel's treatment of luck, risk, and behavior in financial life echoes Duke's warning that outcomes can hide the difference between skill and chance. - [[thinking-in-systems]]: Meadows helps explain why outcomes are often delayed, nonlinear, and hard to attribute, which is exactly why Duke warns against simplistic learning from results. ## When to Use This Knowledge - When a user is judging a past decision only by whether it worked out. - When a team needs to review a launch, hire, investment, or strategic bet without blame-driven hindsight. - When someone is overconfident and needs to express uncertainty more precisely. - When a leader wants to build a culture where changing one's mind is respected. - When a product, business, or investment decision has several plausible futures rather than one obvious answer. - When a person is stuck in motivated reasoning and needs to ask what evidence would change their mind. - When a group needs a practical structure for decision journals, premortems, or postmortems. - When short-term emotion is dominating a decision with long-term consequences. ## Agent Usage Notes Use this book when the user needs help improving judgment, not merely when they ask about poker. The poker examples are a teaching device for uncertainty, variance, and feedback quality. The transferable lesson is that reality often gives noisy feedback, so humans need better processes for learning. This book is especially useful for strategy discussions. It can help an agent ask for base rates, confidence levels, decision records, and review criteria. It also gives language for challenging a user's conclusion without sounding dismissive: "What odds would you put on that?" is often more productive than "Are you sure?" The book is less useful when the decision is purely deterministic or rule-bound. It is most useful when the situation contains incomplete information, delayed feedback, emotional stakes, or multiple plausible futures. In those contexts, thinking in bets gives both a mindset and a workflow.