• What is Scout Mindset?
• When to Use This Skill
• Interactive Menu
• Quick Reference
• Resource Files

• Forecasting requires intellectual honesty
• Biases systemically distort probabilities
• Emotional attachment clouds judgment
• Motivated reasoning leads to overconfidence

• Prediction feels emotional - You want a certain outcome
• Stuck at 50/50 - Indecisive, can't commit to probability
• Defending a position - Arguing for your forecast, not questioning it
• After inside view analysis - Used specific details, need bias check
• Disagreement with others - Different people, different probabilities
• Before finalizing - Last sanity check

• Detect motivated reasoning
• Validate evidence standards
• Expose special pleading

• Linear scaling test
• Reference point calibration
• Magnitude assessment

• Entropy principle
• Change vs stability energy
• Default state inversion

• Surprise test
• Historical calibration
• Overconfidence check

• All major cognitive biases
• Systematic checklist
• Prioritized remediation

• Read Scout vs Soldier Mindset
• Read Cognitive Bias Catalog
• Read Debiasing Techniques

• Prediction: [Event]
• Probability: [X]%
• Direction: [High/Low confidence]

1. Polls show A ahead by 5%
2. A has more campaign funding
3. Expert pundits favor A
4. A has better debate ratings

If this reversed evidence existed, would I accept it as valid and change my prediction?

• "This source is biased" (only when it disagrees)
• "Sample size too small" (only for unfavorable polls)
• "Outlier data" (only for data you dislike)
• "Context matters" (invoked selectively)

• Option 1: Reject the evidence entirely (if you wouldn't trust it reversed, don't trust it now)
• Option 2: Accept it in both directions (trust evidence regardless of direction)
• Option 3: Weight it appropriately (maybe it's weak evidence both ways)

• Number of people (100 vs 10,000 vs 1,000,000)
• Dollar amounts ($1K vs $100K vs $10M)
• Time duration (1 month vs 1 year vs 10 years)

• If raised $1M: ___%
• If raised $10M: ___%
• If raised $100M: ___%

• Generate probability from scratch without looking at others
• Then compare and reconcile differences
• Don't just "split the difference" - reason about why estimates differ

• 1 month: P(success) = ___%
• 10 months: P(success) = ___%
• 100 months: P(success) = ___%

• Market share: To maintain requires matching competitor innovation → Energy required: High → Status quo is HARD
• Policy: To maintain requires no proposals for change → Energy required: Low → Status quo is easier

• Normal framing: "Will X change?" (Default = no)
• Inverted framing: "Will X stay the same?" (Default = no)

1. Is this system open or closed?
2. Is energy being input to maintain/improve?
3. Is that energy sufficient?

• Reduce P(status quo) by 10-20%
• Increase P(change) correspondingly

• Point estimate: ___%
• Lower bound: ___%
• Upper bound: ___%
• Width: ___ percentage points
• Confidence level: ___ (usually 80% or 90%)

• 80% CI → Should be shocked 20% of the time
• 90% CI → Should be shocked 10% of the time

• A) "Yes, I'd be very surprised" - ✓ CI appropriately calibrated
• B) "No, not that surprised" - ⚠ CI too narrow (overconfident) → Widen interval
• C) "I'd be amazed if it landed in the range" - ⚠ CI too wide → Narrow interval

1. Collect last 20-50 forecasts with CIs
2. Count how many actual outcomes fell outside your CIs
3. Compare to theoretical expectation

1. Calculate standard deviation of reference class outcomes
2. Your CI should roughly match that variance

• If overconfident: Multiply current width by 1.5× to 2×
• If underconfident: Reduce width by 0.5× to 0.75×

• Did I seek out disconfirming evidence?
• Did I give equal weight to evidence against my position?
• Did I actively try to prove myself wrong?

• Did I rely on recent/memorable examples?
• Did I use systematic data vs "what comes to mind"?
• Did I check if my examples are representative?

• Did I generate my estimate independently first?
• Did I avoid being influenced by others' numbers?
• Did I adjust sufficiently from initial anchor?

• Do I have an emotional preference for the outcome?
• Did I separate "what I want" from "what will happen"?
• Would I make the same forecast if incentives were reversed?

• Did I run a premortem?
• Are my CIs wide enough (surprise test)?
• Did I identify ways I could be wrong?

• Did I attribute success to skill vs luck appropriately?
• Did I consider situational factors, not just personal traits?
• Did I avoid "great man" narratives?

1. Severity: High / Medium / Low
2. Direction: Pushing probability up or down?
3. Magnitude: Estimated percentage point impact

• Julia Galef's framework
• Motivated reasoning
• Intellectual honesty
• Identity and beliefs

• 20+ major biases
• How they affect forecasting
• Detection methods
• Remediation strategies

• Systematic debiasing process
• Pre-commitment strategies
• External accountability
• Algorithmic aids

1. Truth over comfort - Accuracy beats wishful thinking
2. Seek disconfirmation - Try to prove yourself wrong
3. Hold beliefs lightly - Probabilistic, not binary
4. Update incrementally - Change mind with evidence
5. Separate wanting from expecting - Desire ≠ Forecast
6. Check your work - Run bias audits routinely
7. Stay calibrated - Track accuracy over time

Scout mindset is the drive to see things as they are, not as you wish them to be.

• scout-vs-soldier.md - Mindset framework
• cognitive-bias-catalog.md - Comprehensive bias reference
• debiasing-techniques.md - Remediation strategies