Square b8 · Mathematics · Queenside

Incompleteness

Gödel: a consistent theory cannot prove its own consistency.

Meaning

The displayed statement is a compact form of Gödel’s second incompleteness theorem: a consistent, sufficiently strong formal theory TT (for example, Peano arithmetic) cannot prove its own consistency statement Con(T)\mathrm{Con}(T). Here ⊬\nvdash means “does not prove,” and Con(T)\mathrm{Con}(T) is a sentence in the language of TT that expresses, via a Gödel numbering of syntax, that no contradiction is derivable in TT.

Gödel’s first incompleteness theorem is the companion result: any such consistent TT is incomplete—there exist sentences in its language that TT neither proves nor refutes. The second theorem strengthens the philosophical bite: TT cannot certify its own reliability by an internal consistency proof.

Context

Kurt Gödel announced the incompleteness theorems in 1930 and published them in 1931, in the paper “Uber formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I.” The target was Hilbert’s program, which had hoped to secure all of classical mathematics by finitary consistency proofs. Gödel showed that for any consistent recursively axiomatizable theory rich enough to encode basic arithmetic, consistency is not provable inside the theory itself (assuming that consistency is expressed in the standard way).

The proofs proceed by arithmetizing syntax: formulas and derivations are coded as numbers, so that self-reference becomes possible through a carefully constructed fixed-point (diagonal) lemma. The “Gödel sentence” asserts its own unprovability; the consistency sentence is closely related.

Incompleteness does not say that mathematics is riddled with unknowable truths in an absolute sense, nor that every interesting statement is independent. It does say that no single formal system of the usual kind can capture all arithmetical truths or prove its own consistency. Subsequent work (Rosser, Tarski, Turing, Feferman, and many others) refined the scope, connected incompleteness to undecidability and computability, and produced natural independent statements (for example, in combinatorics and set theory).

On the Canon Board, T⊬Con(T)T\nvdash\mathrm{Con}(T) stands for the limitative theorems of twentieth-century logic—the precise sense in which formal axiomatic method cannot be total. It pairs with the continuum square (a8) as a reminder that foundations have theorems, not only definitions—and that some of those theorems are about what cannot be proved.

Sources