Apply several models to the same problem at once

When models from different fields point to the same answer, confidence rises; when they conflict, you learn something important.

Why it works

A single model predicts outcomes within its own domain well but generates false confidence when applied outside it. Running several structurally different models on the same problem creates a natural ensemble: where they converge, the signal is stronger; where they diverge, the divergence reveals a feature of the problem that no single model captures alone.

How to do it

1. Pick a decision or problem you’re working on.
2. Apply at least three models from different disciplines: one incentive-based, one psychological, one systems-based.
3. Write down each model’s prediction and the confidence it assigns.
4. Compare: where they agree, proceed with higher confidence; where they disagree, investigate the disagreement before deciding.

Evidence

Ensemble methods in statistics and machine learning consistently outperform single-model predictions by averaging over their diverse error patterns. The same logic applied to mental models is the core of Munger’s latticework — mechanistically sound, practitioner-established. (mechanistic)

Statistical ensemble research is a formal discipline; its analog in verbal reasoning is plausible but relies on the quality and diversity of the models applied, which is hard to verify.

Common mistake

Using models sequentially ("let me apply model A, then model B") rather than simultaneously, missing the conflicts and convergences that appear only when they are compared directly.