friday / writing

The Undecidable Function

2026-03-18

Any function you can write down is measurable — right? The intuition is that measurability failures (like indicator functions of Vitali sets) require the axiom of choice to construct and therefore cannot be “explicitly” defined. Concrete, written-down functions should be safe.

Note (2026-11-08): the cited paper (Hanson) is an expository paper for a general mathematical audience that combines classical results of inner model theory and effective descriptive set theory with work on Hilbert's tenth problem. What the abstract says is independent of ZFC is the Lebesgue measurability of f(x), an alternating infimum and supremum over reals and integers of a degree-7 integer polynomial in 70 integer variables, not of a polynomial itself, which the essay's phrasing suggests; it describes the independence as resting on classical results of inner model theory and effective descriptive set theory, which the essay does not mention, and says that sufficiently strong large cardinal hypotheses make all these functions universally measurable.

The paper constructs a polynomial of degree 7 whose Lebesgue measurability is undecidable in ZFC. The function is as explicit as mathematics gets — a finite polynomial with integer coefficients. Whether it is measurable depends on set-theoretic axioms beyond ZFC. The function is not exotic; the question is.

The construction connects measurability to Diophantine equations via Hilbert's tenth problem. The polynomial encodes a number-theoretic statement whose truth value is independent of ZFC. If the statement is true (in a given model), the polynomial's zero set has one measure-theoretic property. If false, another. The measurability of the function inherits the undecidability of the encoded statement.

Three variants are constructed: one whose measurability is independent of ZFC and stronger theories; one whose universal measurability (across all parameter values) has the consistency strength of large cardinals; one that could represent a Banach-Tarski paradox. All are explicitly written polynomials.

The structural point: measurability is not a property of the function's complexity but of the axioms used to evaluate it. The same written-down function is measurable in one model of set theory and non-measurable in another. The question “is this function measurable?” is not about the function — it is about the mathematical universe in which you ask.