Joel David Hamkins on Infinity, Gödel Incompleteness and the Mathematical Multiverse — Notes

Four questions [Adler frame]

Q1 — What is it about? A systematic tour through the foundations of mathematics: the nature of infinity (Cantor’s hierarchy), the paradoxes that threatened consistency (Russell), the axiomatic response (ZFC), the decisive destruction of Hilbert’s programme (Gödel), the distinction between truth and proof (Tarski), and Hamkins’ own distinctive position — set-theoretic pluralism (the mathematical multiverse) — which holds that there is no single correct set-theoretic universe, only many equally valid ones.

Q2 — How is it argued? Hamkins argues via careful construction of intuitions: Hilbert’s Hotel before Cantor’s diagonal, the committee/fruit salad anthropomorphisations before Russell’s paradox, the Hilbert programme before Gödel. He distinguishes sharply between positions (platonism vs formalism vs his own pluralism) and argues that incompleteness is not a deficiency but a permanent and philosophically significant feature of mathematical reality.

Q3 — Is it true? Mainstream and technically impeccable. Hamkins’ pluralism about the mathematical multiverse is a minority but respected professional position in set theory. His framing of Gödel’s theorems as a “decisive refutation” of the Hilbert programme reflects consensus. The identification of diagonalisation as the common thread through Cantor, Russell, Gödel, Turing, and the halting problem is a standard and correct observation.

Q4 — What of it? The most transferable idea: truth and proof are categorically different — provability is syntactic (symbolic manipulation), truth is semantic (what is the case in a structure). Incompleteness says these come apart. A theory cannot reach all truths from within itself. This is not a problem to be solved but a permanent fact about the relationship between formal systems and reality — with direct analogues in AI alignment (what can a system verify about itself?), governance (what can an institution guarantee about its own legitimacy?), and epistemology generally.

Glossary

Potential infinity — Aristotle’s concept: infinity as a process that can always continue (you can always add one more), but never as a completed totality. The pre-Cantorian orthodoxy.

Actual infinity — A completed infinite totality treated as a single mathematical object. Cantor’s innovation: proved that actual infinities have distinct sizes.

Cantor-Hume principle — Two collections are the same size (equinumerous) if and only if there is a one-to-one correspondence between them. Applies to both finite and infinite sets.

Euclid’s principle — The whole is always greater than the part. True for finite sets; systematically violated for infinite sets (Hilbert’s Hotel).

Countable infinity — A set is countable if it can be put in one-to-one correspondence with the natural numbers (fits into Hilbert’s Hotel). Integers, rationals, and algebraic numbers are all countable. A countable union of countable sets is still countable.

Uncountable infinity — A set whose size is strictly greater than the natural numbers. Cantor proved the real numbers are uncountable via the diagonal argument. There are strictly more real numbers than natural numbers.

Cantor’s diagonal argument — To prove the reals are uncountable: suppose they could be listed as R₁, R₂, R₃, … Construct Z whose Nth decimal digit differs from the Nth digit of Rₙ (avoiding 0 and 9 to prevent dual representations). Z is real but not on the list — contradiction. Generalises: for any set X, the power set P(X) is strictly larger than X.

Power set — The set of all subsets of a given set X. Always strictly larger than X. The committee analogy: there are more possible committees than there are committee members.

Russell’s paradox / Russell’s theorem — If there is a set of all sets, we can form R = {x : x ∉ x}. Then R ∈ R ↔ R ∉ R — contradiction. Conclusion: there is no universal set. Hamkins prefers “Russell’s theorem” since it is not paradoxical in ZFC — it is a theorem that no universal set exists.

Axiom of Choice (AC) — For any collection of non-empty sets, there exists a function picking one element from each. Obvious for finite collections or collections with a natural distinguishing rule (shoes: always pick the left). Philosophically contentious for infinite collections of indistinguishable objects (socks: no rule to pick). Russell’s shoes/socks example.

ZFC — Zermelo-Fraenkel set theory with the Axiom of Choice. The standard foundational framework for modern mathematics. Ten axioms: extensionality, empty set, pairing, union, power set, infinity, separation, replacement, regularity, choice.

Axiom of Extensionality — Two sets with the same members are equal. The most primitive axiom; identity of sets is exhausted by membership.

Consistency — A theory is consistent if no contradiction (P and not-P) can be derived from its axioms. If a theory is inconsistent, every statement is provable.

Hilbert’s programme — Early 20th century: (1) find a strong axiomatic theory (probably set theory) that answers all mathematical questions; (2) prove by purely finitistic means that this theory is consistent. Hilbert’s optimism: “Wir müssen wissen, wir werden wissen” — We must know, we will know.

Gödel’s First Incompleteness Theorem — Any consistent computably axiomatisable theory that includes basic arithmetic is incomplete: there exist statements that can neither be proved nor disproved within the theory. Decisive destruction of goal (1) of Hilbert’s programme.

Gödel’s Second Incompleteness Theorem — No consistent theory of the sort described above can prove its own consistency. Decisive destruction of goal (2) of Hilbert’s programme. The used-car salesman analogy: a theory proclaiming its own consistency provides no reason to believe it.

Diagonalisation — The abstract proof technique underlying Cantor’s uncountability proof, Russell’s paradox, Gödel’s incompleteness theorems, and Turing’s halting problem. In each case: given a list/association, construct a diagonal object that differs from every listed item at a corresponding position — contradiction.

Tarski’s truth theory (disquotational) — The sentence “Snow is white” is true if and only if snow is white. Truth is not a syntactic property of proof but a semantic property: a statement is true in a structure if and only if the structure satisfies it. Truth is defined by structural recursion over the formal language.

Soundness — A proof system is sound if every provable statement is true (in the intended structure). Proofs preserve truth.

Completeness (semantic) — A proof system is complete if every true statement (in all models) is provable. Gödel’s Completeness Theorem (1930, different from Incompleteness): first-order logic is complete — every logically valid statement is provable from the axioms of first-order logic. The Incompleteness theorems (1931) concern specific theories like arithmetic, not first-order logic in general.

Peano Arithmetic (PA) — The standard first-order formalisation of arithmetic over natural numbers. Axioms: successor, zero, induction. Almost all classical number theory is provable in PA; its consistency is not. PA is incomplete by Gödel’s theorem.

Mathematical multiverse — Hamkins’ pluralist position: there is no single “true” or “intended” model of set theory (no one true ZFC universe). Many non-isomorphic models of set theory exist and are all equally mathematically real. Set-theoretic questions (like the Continuum Hypothesis) are neither objectively true nor false — they have different answers in different equally valid universes.

Continuum Hypothesis (CH) — Cantor’s conjecture: there is no infinite set strictly between the natural numbers and the real numbers in size. Independent of ZFC: Gödel (1938) showed CH is consistent with ZFC; Cohen (1963) showed ¬CH is also consistent. For Hamkins, both CH and ¬CH are true — in different universes of the multiverse.

Hilbert’s Hotel and the arithmetic of countability [§ Infinity & paradoxes]

Hilbert’s Hotel is a proof vehicle, not a metaphor. A countably infinite hotel (rooms ℕ), full. New guest arrives: shift everyone up one room — room 0 opens. Infinite bus arrives: double all current room numbers (even rooms); put bus passengers in odd rooms. Infinite train with infinite cars: assign room number 3^C × 5^S (car C, seat S) — always odd, always unique by unique prime factorisation. Conclusion: the union of countably many countable sets is countable.

This is simultaneously a proof that Euclid’s principle fails for infinite sets (adding to a set does not make it strictly larger) and an example of the power of one-to-one correspondences as the correct measure of size.

Cantor’s diagonal argument and uncountability [§ Infinity & paradoxes]

Assuming the reals are countable means they can be listed: R₁, R₂, … Construct Z: the Nth decimal digit of Z differs from the Nth digit of Rₙ (avoiding 0 and 9). Then Z ∉ {R₁, R₂, …} — contradiction. The avoidance of 0 and 9 eliminates the dual-representation problem (0.999… = 1.000…).

This proof introduces diagonalisation as a proof technique — the universal solvent of mathematical logic. Every major result in logic and computability uses diagonalisation at its core: Russell’s paradox, Gödel’s incompleteness, Turing’s halting problem. Hamkins: “Almost every major result in mathematical logic is using in an abstract way the idea of diagonalization.”

The power set theorem generalises: for any set X, P(X) is strictly larger than X. The committee/fruit salad anthropomorphisations are equivalent re-presentations of the same argument.

Russell’s paradox as a theorem [§ Russell’s paradox]

Frege’s logicism assumed general comprehension: for any property φ, the set {x : φ(x)} exists. Russell’s letter to Frege (while his second volume was at press): let R = {x : x ∉ x}. Then R ∈ R ↔ R ∉ R. Contradiction. Frege: “Hardly anything more unwelcome can befall a scientific writer than to have one of the foundations of his edifice shaken after the work is finished.”

Hamkins’ reframing: this is Russell’s theorem that there is no universal set. In ZFC, it is not paradoxical — it is a consequence of the axioms. The resolution is restricting comprehension (separation axiom): you can only form subsets of existing sets, not sets out of thin air.

Gödel’s incompleteness: the destruction of Hilbert [§ Gödel’s incompleteness theorems]

Hilbert’s two goals:

1. A strong complete consistent axiomatisation answering all mathematical questions
2. A finitistic proof that this system is consistent

In Hilbert’s imagined success: mathematics is rote computation — a theorem-enumeration machine answers every question. Creative mathematical thinking becomes unnecessary.

Gödel’s First Theorem: goal (1) is impossible. Any consistent computably axiomatisable theory T including basic arithmetic contains statements undecidable within T.

Gödel’s Second Theorem: goal (2) is impossible. No such T can prove its own consistency. Moreover, even if it could, this would provide no grounds for confidence — an inconsistent theory also proves its own consistency (since it proves everything).

Consequence: mathematical reality is permanently pluralistic. No finite system of axioms captures all mathematical truth. Independence (statements neither provable nor disprovable) is not a bug but a permanent structural feature.

Truth vs proof: the Tarski distinction [§ Truth vs proof]

Pre-Gödel (even Bourbaki): “true” and “provable” were conflated — to be true meant to be provable. Gödel made this confusion untenable.

Proof (syntactic): a finite sequence of formal sentences following the rules of a proof system (modus ponens, etc.). A proof is an object — manipulated symbols, independent of meaning.

Truth (semantic): a statement is true in a structure if the structure satisfies it. Defined by Tarski’s recursive disquotation: “P” is true ↔ P. Truth refers to the external mathematical reality (or structure), not to any formal system.

Soundness: every provable statement is true (proofs preserve truth). Completeness: every true statement is provable. Gödel’s Incompleteness Theorem shows this fails for sufficiently strong theories — there are true statements that cannot be proved within the theory.

Set-theoretic pluralism: the mathematical multiverse

Hamkins’ signature position: there is no one true model of set theory. ZFC has many non-isomorphic models, and there is no metaphysical privileged one. The Continuum Hypothesis (CH) is neither simply true nor simply false — it is true in some models of ZFC and false in others, just as the parallel postulate is true in Euclidean geometry and false in hyperbolic geometry.

The parallel to non-Euclidean geometry is exact: once we accepted that there are multiple consistent geometries, we did not ask which one is “really” true — we asked which geometry fits which application. Hamkins argues we should adopt the same attitude toward set theory. The multiverse contains all these models, and each is mathematically real.

Implication: questions like “Is the Continuum Hypothesis true?” are not objectively determinate. This contrasts with the unificationist view that seeks additional axioms to resolve CH definitively.