The Accuracy / Usefulness distinction

Distinctions are useful. They probably would top my mindlist of meta-cognitive tools. The first such distinction I can recall is that between Map and Territory. The first explanation I can personally remember is Neil Gaiman’s, in American Gods;

…The more accurate the map, the more it resembles the territory. The most accurate map possible would be the territory, and thus would be perfectly accurate and perfectly useless.

Alfred Korzybski is usually credited with the idea, but it dates to an earlier story by a mathematician who dabbled in fiction, Charles Lutwidge Dodgson — better known as Lewis Carroll. His work, published when Korzybski was 14, was a fantastical story about a land where they build maps of greater and greater detail, ending with a proposed map on a scale of 1:1 — which was abandoned after farmers complained would block the sun and kill the crops.

Gaiman was wrong, though. The problem, as Carroll makes clear, is that some maps need to be lower scale to be useful— not that they need to be less accurate. Accuracy is a byproduct of scale and resolution; high resolution isn’t bad. And so, climbing up a meta-level, I think we need to revise our map of the map/territory distinction.

The goal of a map is to navigate. That means elucidation and representation are secondary to usable predictive power; a brilliantly lucid, parsimonious, and compactly detailed model is beaten by a kludgy hack of a model that predicts reality better when we need to pick a course of action.

Of course, predictive power is only useful if we can use the map. Newtonian physics sometimes beats general relativity because the cost of the more complex model isn’t worth the added accuracy. The kludgiest trigonometric lookup tables based on trial and error are a better map for innumerate artillerymen than Newtonian physics.

This leads me to my point; maps must be judged by utility for the tasks at hand, not any other metric. Accuracy, i.e. predictive power, is great for judging scientific theories, but reductionism has its limits in practice. As Eliezer said, make sure your map can fit in your glove box.