All Models Are Wrong … But Some Are Useful
Farnam Street (Shane Parrish), after George Box · 2017
"Every mental model is a simplification and therefore, strictly speaking, wrong — the only meaningful question is not 'is this model true?' but 'is this model useful, and for what?' — the necessary caveat on the entire mental-models project."
A model is a simplification that helps understanding — like an architectural model of a building, or a map of a city — and simplification necessarily means leaving out details; that's not a flaw to be fixed but the entire point, since a model with zero simplification (a 1:1 map) would be exactly as unwieldy and useless as the raw, unprocessed reality it was meant to help you navigate.
The practical test for any model is not 'does this perfectly represent reality' (no model does) but questions like: how long has it been around and scrutinized (durability suggests usefulness), does it apply across multiple different areas (versatility suggests real signal, not overfitting to one case), and is it built on first principles rather than requiring infinite regress of further justification — a model that passes these tests earns a place in your 'latticework' even though it remains, technically, wrong.
Who coined the phrase 'all models are wrong, but some are useful,' cited as the origin of this piece's title?
Read more about the topic
The explanation above is written with AI assistance. These are the originals — go to them to check it.
- The Map is Not the TerritoryFarnam Street
- The Great Mental ModelsFarnam Street
How to Use Occam's Razor Without Getting Cut
"Among competing explanations that fit the evidence equally well, the one that requires the fewest assumptions should be preferred — but Occam's Razor is a tool for narrowing initial hypotheses, not a substitute for evidence, and misapplying it (assuming 'simple' always beats 'accurate') can mislead badly."