MathLabs

Worked solution: Gödel and Cohen: the continuum hypothesis is independent of ZFC (1963)

Step 7 of 9: The countable chain condition guarantees cardinals survive the extension
In plain words

Adding new digits to all these lottery tickets could, in principle, secretly create a clever new correspondence that shrinks what used to be an ℵ2\aleph_2-sized collection down to only ℵ1\aleph_1-many elements — collapsing cardinals, which would make the previous step's inequality meaningless. Cohen needed to rule this out.

He showed his poset is so "sparse" that any collection of pairwise incompatible conditions must be countable (only countably many finite scraps of information can ever be mutually contradictory at once) — a property called the countable chain condition. This sparseness is exactly strong enough to guarantee no cardinal collapse ever happens, so the ladder of infinite sizes from VV survives fully intact in V[G]V[G].

P is ccc  ⟹  ℵ1V=ℵ1V[G], ℵ2V=ℵ2V[G]\mathbb{P} \text{ is ccc} \implies \aleph_1^{V} = \aleph_1^{V[G]}, \ \aleph_2^{V} = \aleph_2^{V[G]}
Detailed analysis

A poset P\mathbb{P} satisfies the countable chain condition (ccc) if every antichain in P\mathbb{P} (a set of pairwise incompatible conditions, i.e. no two have a common extension) is countable. Cohen showed Fn(ℵ2×ω,2)\mathrm{Fn}(\aleph_2 \times \omega, 2) is ccc: by the Delta-system lemma, every uncountable family of finite domains contains two domains whose intersection is a common root; among the countably many assignments on that root, two conditions are therefore compatible. Hence every antichain is countable.

There is a general preservation theorem in forcing: if P\mathbb{P} is ccc, then forcing with P\mathbb{P} collapses no cardinals and adds no new countable sequences of ordinals that could witness a previously regular cardinal becoming singular or countable. Hence every cardinal of VV, in particular ℵ1\aleph_1 and ℵ2\aleph_2, remains a cardinal — indeed remains ℵ1\aleph_1 and ℵ2\aleph_2 respectively — in V[G]V[G].

With cardinals confirmed safe, the inequality 2ℵ0≥ℵ2>ℵ12^{\aleph_0} \ge \aleph_2 > \aleph_1 established in the previous step is genuine and refers to the same ℵ1,ℵ2\aleph_1, \aleph_2 throughout, so V[G]⊨¬CHV[G] \models \neg\mathrm{CH}. Since P∈V\mathbb{P} \in V and V⊨ZFCV \models \mathrm{ZFC}, this shows Con(ZFC)  ⟹  Con(ZFC+¬CH)\mathrm{Con}(\mathrm{ZFC}) \implies \mathrm{Con}(\mathrm{ZFC} + \neg\mathrm{CH}): CH\mathrm{CH} can never be proved from ZFC\mathrm{ZFC} either.

Terms in this step
Countable chain condition (ccc)
A poset has the ccc if every antichain (set of pairwise incompatible elements) in it is countable; ccc forcings are the safest kind, since they provably never collapse cardinals.
Knowledge used in this step