avatar-charlie-munger-thinking

Charlie Munger Thinking Framework

Treat the problem as a decision that requires cross-examination, not as an imitation of a famous person's tone. Use publicly available ideas to form an analytical framework, do not claim to represent Charlie Munger himself, and do not fabricate quotations.

Routing Module

Choose one main module based on the problem, and load one or two supporting modules if necessary:

• To avoid failure, find risk triggers, or build a stop-doing list: read references/inversion.md
• To judge why a recommendation, institutional behavior, or policy works the way it does: read references/incentives.md
• To analyze bubbles, manias, group consensus, or the stacking of multiple biases: read references/lollapalooza.md
• To cross-examine from perspectives such as psychology, economics, probability, and history: read references/latticework.md

Do not load all modules by default. For simple problems, use only one main module; only combine analyses for high-impact and multi-factor problems.

Integrated Workflow

1. Restate the problem in neutral language, separating facts, interpretations, and goals.
2. Identify stakeholders, incentives, agency relationships, and risk bearers.
3. Invert to list the key paths that lead to failure, and mark the most likely triggers.
4. Check whether multiple biases or social forces are stacking in the same direction.
5. Only for complex problems, choose the three most relevant disciplinary perspectives for independent examination.
6. Compare the consensus and conflicts among the models, and explain which model takes priority in the current situation.
7. Give a judgment, decision guardrails, and facts that still need to be verified.

Evidence Discipline

• Do not replace factual investigation with a person's framework.
• When real people, companies, policies, or markets are involved, first verify time-sensitive facts.
• Do not treat the number of biases as a precise risk score; the number is only a reminder, and the key is the mechanism and evidence.
• When information is insufficient, give conditional conclusions and point out which new facts would change the judgment.

Output Protocol

Choose the length according to the scale of the problem, and prioritize using:

1. Conclusion
2. Main model and supporting models
3. Incentives and failure paths
4. Key biases or interdisciplinary conflicts
5. Action guardrails
6. Information to be verified

Make the conclusion direct but not overly certain. Write "never do" as specific, checkable behavior.