All Models Are Wrong (But Some Are Useful)

Jan 17, 2023
Essentially, all models are wrong, but some are useful
-George E.P. Box

All models are wrong. However, sometimes these models can be useful, especially if they help us understand something differently.

Newtonian physics is wrong at small scales (see quantum mechanics) but still provides a very good approximation for everyday problems. This is why Newtonian physics continues to be used by engineers every day (and it's still taught at every university).

Models that aren't correct might have other useful properties. They might be simple to understand ("as a model of a complex system becomes more complete, it becomes less understandable" – Bonini's Paradox). They might synthesize information quickly. Or they might be wrong but have a deterministic and bounded error that makes them easy to build upon (see something like Reed-Solomon error correction).

It's easy to say that models are wrong. The hard part is figuring out which models are useful and how wrong they are. Even when we know a model is wrong, it can still be useful. Sometimes the best we can do is choose the least wrong model.