research judgment

Mechanism-first vs Prediction-first Science

A short essay seed on when prediction helps mechanism and when it becomes a substitute for thinking.

Seed

Prediction-first work can be useful when it compresses many observations into a manipulable model. It becomes weaker when accuracy is treated as explanation.

Mechanism-first work can be useful when it identifies causal handles. It becomes weaker when it ignores noisy reality and only protects a favored story.

The interesting zone is where a model proposes a perturbation that changes what experiment should be done next.