Induction

Induction is reasoning from specific observations to general claims: "every raven I've seen is black, so all ravens are black." It is the foundation of classical empiricism and most natural science as usually practised.

CF, following CR, rejects induction. The famous counter-example-vs.-confirmation asymmetry makes induction logically indefensible:

• One white raven refutes "all ravens are black".
• One million black ravens don't confirm "all ravens are black".

The next raven could be white.

Why CF rejects induction

1. Logical asymmetry. Confirmation is much weaker than refutation.
2. Regress. Induction requires a principle of induction, which needs its own justification, and so on.
3. Conjectures and refutations is a better description of how science actually works: we guess, we test, we throw out what fails. We don't infer general laws from finite observations.
4. error-correction is the actual mechanism. We learn by finding and fixing errors, not by accumulating confirming cases.

What this doesn't mean

• CF doesn't say observations are useless. They are essential for tests that refute hypotheses.
• CF doesn't deny that science works. It says science works because of refutation, not because of induction.

"We learn by an evolutionary process focused on error correction, not by induction or justification." — criticalfallibilism.com