Experiment as Sole Judge

The core methodological commitment of physics — and, Feynman insists, of science — is that experiment is the only authority that can confirm or refute a theory. No argument from first principles, no appeal to authority, no degree of internal logical coherence can substitute for checking whether a theory's predictions match what actually happens in the world.

Feynman's formulation is blunt: "It does not make any difference how beautiful your guess is. It does not make any difference how smart you are, who made the guess, or what his name is. If it disagrees with experiment, it is wrong."

What This Rules Out

This principle is not merely a procedural nicety — it has teeth. It rules out:

  • Armchair theorizing without testable predictions. A theory that can explain any outcome in retrospect but predicts nothing in advance has no scientific standing. It must be falsifiable (to use Karl Popper's later vocabulary) to count.
  • Appeal to authority or beauty. A brilliant physicist's intuition is not evidence. A mathematically beautiful theory is not confirmed until its predictions survive experimental test. History is littered with beautiful theories that were wrong and ugly ones that turned out to be right.
  • Consensus as truth. If a measurement shows a universally accepted theory is wrong, the theory is wrong, full stop. Newton's mechanics were accepted for two centuries; they are still wrong in the regime where relativity matters.

The Asymmetry of Confirmation

Experiment can decisively refute a theory — one clear contradictory observation is enough — but it can never fully confirm one. A theory that survives ten thousand experiments might fail on the ten-thousand-and-first in a slightly different regime. What grows with experimental success is confidence, not certainty. All physical laws are approximations: they hold within some domain, with some precision, under some conditions. Every law has boundaries where it breaks down and a more complete theory takes over.

This is uncomfortable but honest. Physics does not deliver absolute truths; it delivers the best available descriptions that survive every experiment we have yet tried.

The Other Side: Theory Guides Experiment

The principle does not mean experiment without theory. You need a theory — a "guess," as Feynman calls it — to know what to measure and what counts as a significant result. The cycle is: guess, deduce consequences, check against experiment, revise. Neither experiment without theory nor theory without experiment produces knowledge. The experiment is the judge; the theory is the defendant.

Connections

Sources