Worked solution: Gödel and Cohen: the continuum hypothesis is independent of ZFC (1963)
For two thousand years, mathematicians tried to prove Euclid's fifth postulate (about parallel lines) from his other four axioms, and always failed; eventually it turned out the postulate is genuinely optional — geometries where it holds (flat, Euclidean) and geometries where it fails (curved, non-Euclidean) are both perfectly consistent.
(Zermelo–Fraenkel set theory with choice) is the standard rulebook mathematicians use for building sets. Saying is independent of means it behaves exactly like Euclid's fifth postulate: the rulebook itself simply does not decide whether or its opposite holds, and both choices lead to equally consistent worlds of sets.
A statement is independent of a consistent theory if neither nor can be proved from 's axioms; equivalently, both and are themselves consistent theories. Proving independence therefore always takes two separate halves, each a relative consistency result.
Half one: build a model (a universe of sets) in which every axiom holds and also holds, showing . Half two: build a different model in which every axiom holds but holds instead, showing . Neither half alone settles independence — Gödel supplying only half one by 1940 left open the possibility that might still be provable from ; only Cohen's 1963 construction of the second model closed that gap for good.
The next three steps present Gödel's half (Steps 3–4), then the following three present Cohen's half (Steps 5–7), before Step 8 combines them into the full independence theorem.
- Model of a theory
- A structure (here, a universe of sets together with the membership relation ) in which every axiom of a given theory comes out true. Building a model of is how one proves is consistent with .