8.8 KiB
Scout vs Soldier Mindset
Julia Galef's Framework
The Two Mindsets
Soldier Mindset:
- Goal: Defend your position
- Metaphor: Protecting territory
- Reasoning: Motivated by desired conclusion
- Evidence: Cherry-pick what supports you
- Beliefs: Fortifications to defend
- Identity: Tied to being right
Scout Mindset:
- Goal: Map the territory accurately
- Metaphor: Exploring terrain
- Reasoning: Motivated by accuracy
- Evidence: Seek all relevant data
- Beliefs: Working hypotheses
- Identity: Tied to good process
Why Soldier Mindset Exists
Evolutionary Benefits
In ancestral environment:
- Defending tribe increased survival
- Loyalty signaled trustworthiness
- Confidence attracted mates/followers
- Changing mind showed weakness
Result: Humans evolved to defend positions, not seek truth.
Social Benefits (Today)
Soldier mindset helps with:
- Tribal signaling (showing loyalty)
- Self-esteem (feeling right)
- Persuasion (confidence convinces)
- Belonging (fitting in)
The trap: These benefits come at the cost of accuracy.
Why Scout Mindset Wins (For Forecasting)
Accuracy Beats Persuasion
In forecasting:
- No one cares if you "sound confident"
- The territory doesn't care about your loyalty
- Reality punishes inaccuracy
Scouts get promoted. Soldiers die in wrong positions.
Long-Term Reputation
Scout mindset builds:
- Credibility (you admit mistakes)
- Trust (not ideologically motivated)
- Respect (intellectually honest)
Soldier mindset destroys:
- Credibility (predict confidently, wrong often)
- Trust (clearly biased)
- Respect (never update beliefs)
Recognizing Soldier Mindset in Yourself
Telltale Signs
1. Directional Motivated Reasoning
- You want a specific conclusion
- Evidence is ammunition, not information
- You feel defensive when challenged
2. Asymmetric Evidence Standards
- Friendly evidence: Low bar
- Unfriendly evidence: High bar
- "I need extraordinary proof for claims I dislike"
3. Identity Protective Cognition
- Changing mind feels like betrayal
- Beliefs are tribal markers
- Being wrong = being a bad person
4. Affect Heuristic
- If I like it → It must be true
- If I dislike it → It must be false
- Emotions drive conclusions
Cultivating Scout Mindset
1. Separate Identity from Beliefs
Soldier: "I am a [belief-holder]"
- "I am a Democrat"
- "I am an atheist"
- "I am a Bitcoin believer"
Scout: "I currently believe [X] with [Y]% confidence"
- "I lean left on policy with 70% confidence"
- "I assign 90% probability God doesn't exist"
- "I think Bitcoin has 40% chance of long-term success"
Why this helps:
- Beliefs become probabilities, not identities
- Updating doesn't threaten who you are
- Confidence is explicit, not assumed
2. Reframe "Being Wrong"
Soldier framing:
- Being wrong = Failure
- Changing mind = Weakness
- Admitting error = Losing
Scout framing:
- Being wrong = Learning
- Changing mind = Updating
- Admitting error = Intellectual honesty
Practice:
- Celebrate discovering you were wrong (you learned something!)
- Track your updates publicly (builds reputation for honesty)
- Say "I changed my mind because..." not "I was wrong, sorry"
3. Make Accuracy Your Goal
Soldier goal: Win the argument
Scout goal: Get the right answer
Practical shift:
- Don't ask "How can I defend this?"
- Ask "What would I need to see to change my mind?"
4. Value Process Over Outcome
Outcome-focused (Soldier):
- Judge decisions by results
- Outcome bias (good outcome = good decision)
- Rewarded for being lucky
Process-focused (Scout):
- Judge decisions by reasoning
- Good process despite bad outcome = Success
- Rewarded for good epistemics
Motivated Reasoning
What It Is
Definition: Unconsciously selecting evidence and arguments that support a desired conclusion.
Key word: Unconscious. You genuinely believe you're being objective.
How It Works
Normal reasoning:
- Gather evidence
- Weigh evidence
- Form conclusion
Motivated reasoning:
- Have desired conclusion
- Search for supporting evidence
- Stop when you have "enough"
- Ignore or discount contradictory evidence
- Conclude what you wanted from the start
Feels like: Objective analysis Actually is: Rationalization
Detecting Motivated Reasoning
Ask yourself:
1. Do I want this conclusion to be true?
- If yes → High risk of motivated reasoning
2. Would I be bothered if evidence contradicted this?
- If yes → Motivated reasoning likely
3. Am I trying to prove X or trying to find the truth?
- If "prove X" → Soldier mindset active
4. If I was wrong, how would I know?
- If "I wouldn't be wrong" → Motivated reasoning
Intellectual Honesty
What It Looks Like
Honest forecaster:
- States assumptions clearly
- Admits uncertainties
- Shows work (not just conclusions)
- Updates when wrong
- Says "I don't know" when appropriate
- Gives credit to alternative views
Dishonest (or self-deceiving) forecaster:
- Hides assumptions
- Expresses false certainty
- Only shares conclusions
- Never updates
- Always has an answer
- Strawmans opposing views
The "Outsourcing Test"
Question: "If I hired someone else to make this forecast, and I paid them for accuracy, would they reach the same conclusion?"
If NO: You're probably letting motivated reasoning distort your judgment.
Identity and Beliefs
The Trap
When beliefs become identity:
- Challenging belief = Challenging identity
- Changing belief = Losing self
- Defense becomes automatic
Example:
- "I am a Democrat" → Can't update on any right-wing policy
- "I am an entrepreneur" → Can't admit my startup will fail
- "I am a skeptic" → Can't update toward belief in X
The Fix: Hold Beliefs Lightly
Strong opinions, loosely held
Distinguish:
- Values: Core identity, rarely change (e.g., "I value honesty")
- Beliefs: Maps of the world, update frequently (e.g., "I think X is true")
Forecasting-compatible identity:
- "I am someone who seeks truth"
- "I am calibrated"
- "I update my beliefs with evidence"
NOT:
- "I am bullish on crypto"
- "I am a pessimist about AI"
- "I am a contrarian"
Practical Exercises
1. The "And I Could Be Wrong" Suffix
Practice: Every time you state a belief, add: "...and I could be wrong."
Example:
- "I think this startup will succeed... and I could be wrong."
Why it helps:
- Reminds you beliefs are probabilistic
- Reduces emotional attachment
- Signals intellectual humility
2. Steelman the Opposition
Don't strawman (weak version of opposing view)
Do steelman (strongest version of opposing view)
Method:
- State opposing view
- Make it as strong as possible
- Articulate why someone smart would believe it
- Then and only then evaluate it
Example:
Strawman: "Skeptics of Bitcoin just don't understand technology."
Steelman: "Bitcoin skeptics note that high volatility makes it poor as currency, energy costs are unsustainable, regulatory risk is high, and most use cases can be solved with traditional databases. These are serious, well-founded concerns."
3. Ideological Turing Test
Test: Can you explain the opposing view so well that someone from that side can't tell you're not one of them?
If YES: You understand the position If NO: You're strawmanning (Soldier mindset)
4. Pre-Commitment to Update
Before looking at evidence:
State:
- "I currently believe X with Y% confidence"
- "If I see evidence Z, I will update to W%"
Why this helps:
- Locks in update rule before emotions engage
- Prevents post-hoc rationalization
- Makes updating mechanical, not emotional
Integration with Forecasting
Scout Mindset Improves Every Step
1. Reference Class Selection
- Scout: Choose class objectively
- Soldier: Choose class that supports conclusion
2. Evidence Gathering
- Scout: Seek disconfirming evidence
- Soldier: Seek confirming evidence
3. Probability Estimation
- Scout: Calibrate based on accuracy
- Soldier: Express confidence to persuade
4. Updating
- Scout: Update incrementally with evidence
- Soldier: Stick to position regardless
5. Confidence Intervals
- Scout: Wide CIs reflecting uncertainty
- Soldier: Narrow CIs to sound confident
Summary
Scout Mindset:
- Goal: Accuracy
- Process: Seek truth
- Beliefs: Probabilities
- Identity: Good process
- Updates: Frequently
- Confidence: Calibrated
Soldier Mindset:
- Goal: Win argument
- Process: Defend position
- Beliefs: Certainties
- Identity: Being right
- Updates: Rarely
- Confidence: Overconfident
For forecasting: Be a scout. Soldiers get killed by reality.
Return to: Main Skill