The Decision Problem
Every day you make decisions. Small ones (what to eat), medium ones (how to spend your weekend), and occasionally large ones (career changes, relationships, major purchases).
Most decisions are made poorly. We go with gut instinct, which is often biased. We avoid thinking about hard choices until we're forced. We let default options choose for us.
Better decision-making isn't about being smarter. It's about using better processes — frameworks that correct for common errors and help you think systematically.
The Anatomy of a Decision
Every decision has components:
Options: What are the possible choices? Outcomes: What might happen with each choice? Probabilities: How likely is each outcome? Values: How much do you care about each outcome?
Many bad decisions come from missing one of these:
- Not seeing all options (missing creative alternatives)
- Not considering all outcomes (ignoring risks or benefits)
- Misjudging probabilities (overconfidence, availability bias)
- Being unclear about values (not knowing what you actually want)
Good decision processes address all four.
Framework 1: Expected Value Thinking
What it is: Weighing options by their expected outcomes.
How it works:
- List possible outcomes for each option
- Estimate probability of each outcome
- Estimate value (positive or negative) of each outcome
- Calculate: Expected Value = Probability × Value, summed across outcomes
Example: Should you accept a job offer?
Option A (Accept):
- 60% chance it works out well: +100 value
- 30% chance it's okay: +30 value
- 10% chance it's terrible: -50 value
- Expected value: (0.6 × 100) + (0.3 × 30) + (0.1 × -50) = 60 + 9 - 5 = 64
Option B (Stay):
- Expected value of current situation: +40
This suggests accepting, but the numbers are illustrative. Real value is in the systematic thinking, not precise math.
Limitations:
- Hard to estimate probabilities accurately
- Values aren't always quantifiable
- Doesn't capture risk tolerance
- Ignores irreversibility
AI prompt:
I'm deciding whether to [decision].
Help me think through expected value:
1. What are my options (including ones I might not have considered)?
2. What are the possible outcomes of each?
3. What probabilities seem reasonable for each outcome?
4. How much would I value each outcome?
5. What does this suggest about my decision?
Framework 2: Regret Minimization
What it is: Choosing the option you'll least regret when looking back.
How it works:
- Imagine yourself at age 80, looking back on this decision
- Which choice would you most regret not taking?
- Choose to minimize that regret
When it's useful:
- Big life decisions (career, relationships)
- Irreversible or hard-to-reverse choices
- When you're tempted to play it safe by default
Jeff Bezos version: "I knew that when I was 80, I would never regret having tried. I would only regret not trying."
Caution: Works better for decisions involving action vs. inaction. Less useful for choosing between multiple active options.
AI prompt:
Decision: [Your decision]
Help me apply regret minimization:
1. If I don't take this opportunity, what might I regret at 80?
2. If I do take it and it fails, what would I regret?
3. Which regret seems worse?
4. What does this suggest?
Framework 3: Reversibility Assessment
What it is: Considering how easily you can reverse the decision.
How it works:
- How reversible is this decision?
- Reversible decisions: Decide quickly, learn, adjust
- Irreversible decisions: Slow down, deliberate carefully
Why it matters: Many people apply careful deliberation to reversible decisions (what to order for dinner) while making irreversible decisions impulsively.
Types of decisions:
- One-way doors: Hard or impossible to reverse. Go slow.
- Two-way doors: Easy to reverse. Go fast, adjust as needed.
AI prompt:
I'm deciding about [decision].
Help me assess reversibility:
1. How easily could I reverse this if it doesn't work?
2. What would reversing cost (time, money, relationships)?
3. Is this a one-way or two-way door?
4. How should that affect my decision process?
Framework 4: Pre-Mortem Analysis
What it is: Imagining the decision failed and figuring out why.
How it works:
- Imagine it's one year from now
- The decision turned out terribly
- Write down why it failed
- Now: Can you prevent those failures?
Why it works: Normal planning is optimistic. We imagine success. Pre-mortems force you to imagine failure, surfacing risks you'd otherwise miss.
AI prompt:
I'm planning to [decision/plan].
Conduct a pre-mortem:
1. Imagine it's one year later and this completely failed
2. What are the most likely reasons it failed?
3. What risks am I underweighting?
4. What could I do now to prevent the most likely failure modes?
Framework 5: Second-Order Thinking
What it is: Thinking beyond immediate consequences to downstream effects.
How it works:
- What's the immediate result of this choice?
- And then what? (Second-order effects)
- And then what? (Third-order effects)
- Continue until you've considered the full chain
Why it matters: Most people stop at first-order effects. But many decisions have second and third-order effects that dominate.
Example: "If we cut prices, we'll sell more." (First order) → "Competitors will cut prices too." (Second order) → "Industry-wide margin compression." (Third order) → "Maybe cutting prices isn't as good as it seemed."
AI prompt:
I'm considering [decision].
Help me think through second-order effects:
1. What are the immediate (first-order) results?
2. What happens next as a result of those? (Second order)
3. And then what? (Third order)
4. Are there effects I'm probably not seeing?
5. Do the later-order effects change how I should think about this?
Framework 6: Outside View
What it is: Using base rates from similar situations rather than relying on your inside view.
How it works:
- What's the reference class? (Similar decisions/situations)
- What's the base rate? (How often do these succeed/fail?)
- What would you predict based only on the base rate?
- Now adjust for your specific circumstances — but not too much
Why it matters: People overweight their specific situation and underweight how similar situations usually turn out. We think we're special; we're usually not that different.
Example: "What's the success rate for startups in this category? What makes me think I'll beat the base rate?"
AI prompt:
I'm [decision/situation].
Help me take the outside view:
1. What's the reference class for this situation?
2. What's the typical outcome (base rate)?
3. What would I predict based only on base rates?
4. What are my specific factors that might justify adjusting?
5. Am I overweighting my specialness?
Framework 7: Value Clarification
What it is: Getting clear on what you actually want before deciding.
How it works:
- What do I actually want from this decision?
- What are my real priorities?
- What values are in conflict?
- Which values matter more?
Why it matters: Many "hard" decisions are hard because of unclear or conflicting values. Clarify values first, and the decision often becomes obvious.
AI prompt:
I'm struggling with this decision: [Decision]
Help me clarify my values:
1. What do I actually want here? (Not what I should want)
2. What values or priorities are in conflict?
3. Which values matter more to me?
4. If I were being honest about my priorities, what would I choose?
Combining Frameworks
Different frameworks suit different decisions:
Big, irreversible life decisions:
- Regret minimization
- Pre-mortem
- Value clarification
Risky business/financial decisions:
- Expected value
- Outside view
- Second-order thinking
Uncertain but reversible:
- Reversibility assessment
- Decide quickly, adjust
Decisions you're procrastinating:
- Often need value clarification
- The indecision itself is a decision
Common Decision Errors
Analysis Paralysis
Problem: Overthinking reversible, low-stakes decisions.
Fix: Set a time limit. For small decisions, 5 minutes is enough. Decide and move on.
Going With Gut on Big Decisions
Problem: Using intuition where systematic thinking is needed.
Fix: For irreversible or high-stakes decisions, force yourself through a framework. Your gut is useful but not sufficient.
Ignoring Opportunity Cost
Problem: Evaluating options in isolation rather than comparing to alternatives.
Fix: Always ask: "What else could I do with this time/money/energy?"
Seeking More Information Instead of Deciding
Problem: Gathering information to avoid the discomfort of deciding.
Fix: Ask: "Would more information actually change my decision?" Often it wouldn't.
Letting Default Win
Problem: Not deciding is deciding — for the status quo.
Fix: Recognize that inaction is a choice with consequences.
Using AI for Decisions
Full Decision Analysis
Help me make this decision: [Decision]
Walk me through:
1. What are all my options? (Including ones I might not have considered)
2. What are the likely outcomes of each?
3. How should I think about probabilities?
4. What are my real priorities here?
5. What does each framework suggest?
- Expected value
- Regret minimization
- Reversibility
- Pre-mortem
- Outside view
6. What would you recommend and why?
Devil's Advocate
I'm leaning toward [choice].
Play devil's advocate:
1. What's wrong with this choice?
2. What am I probably not seeing?
3. What could go wrong?
4. Is there a better option I'm dismissing too quickly?
Decision Review
I made a decision recently: [Decision]
The outcome was: [Outcome]
Help me review:
1. Was it a good decision given what I knew at the time?
2. What could I have done better?
3. What should I learn for similar decisions?
Distinguish decision quality from outcome quality — good decisions can have bad outcomes.
What's Next
Decision-making often involves other people. And people often disagree.
Chapter 7 covers arguing without fighting — how to disagree productively, change minds (including your own), and escape tribal thinking.