"The model is always incomplete. The world never is."
All models are wrong. Some are useful. The danger is not using a map — you need one to navigate. The danger is forgetting it is a map, and mistaking the simplified representation for the actual, messy, ever-changing territory beneath it.
— Munger's application of Alfred Korzybski · Science and Sanity, 1933
The core distinction
The Map
Your model of reality
A spreadsheet, a framework, a theory
Selective — it omits what it can't measure
Frozen at the moment it was made
Useful precisely because it simplifies
Becomes dangerous when mistaken for truth
≠
The Territory
Actual reality
Complex, context-dependent, dynamic
Contains what the model left out
Changes continuously — often without notice
Indifferent to your beliefs about it
Always the final referee
How maps diverge from reality over time
Three ways this kills you
FAILURE 01
Precision Mistaken for Accuracy
A model with six decimal places feels more true. It isn't. Wall Street's VAR models were mathematically precise and catastrophically wrong about 2008 — because they couldn't model correlated defaults.
FAILURE 02
The Map Stops Updating
You build a model in 2019, it works. By 2023, the world has shifted — interest rates, competition, customer behaviour — but you're still running on the old map. Kodak trusted a map of a world that no longer existed.
FAILURE 03
Unmeasured = Unreal
Anything the model can't quantify gets ignored. Morale, trust, brand, culture — these are territory, not map. Companies optimise the metrics and destroy the things metrics can't capture.
Munger's antidote — hold your map lightly
STEP 01
Name the assumptions your model depends on. Write them down. Most people skip this.
STEP 02
Ask: if these assumptions are wrong, what happens? Pre-mortem the map before trusting it.
STEP 03
Actively seek disconfirming evidence. When did you last look for proof your model is wrong?
STEP 04
Use a lattice of models. Multiple imperfect maps triangulate closer to territory than any one perfect map.
"
To a man with only a hammer, every problem looks like a nail. The antidote is to have a full toolkit of models — not because any single one is right, but because reality is more complex than any map can capture, and the overlapping imperfections point toward truth.