Untitled

title: The Unnecessary Idempotent
paper: "Di Nasso, 'A new ultrafilter proof of Van der Waerden's theorem' (arXiv:2603.04043)"
tags: combinatorics, Ramsey-theory, ultrafilter, Van-der-Waerden, idempotent, proof-theory

Van der Waerden's theorem states that for any finite coloring of the natural numbers, one color class contains arbitrarily long arithmetic progressions. Color the integers however you like — some color must contain structure. The theorem is nearly a century old, proved in 1927, and every proof since has been either purely combinatorial (double induction), topological-dynamical (Furstenberg), or ultrafilter-algebraic.

The ultrafilter proofs all work in the same way. They move to the Stone-Cech compactification of the natural numbers — the space of ultrafilters on N — and exploit its algebraic structure. The natural numbers have addition, and this extends to an operation on ultrafilters. The key step, in every previous proof, is finding an idempotent ultrafilter: one that satisfies p + p = p. Such ultrafilters exist by Ellis's lemma, a fixed-point theorem for compact semigroups. The idempotent is the engine. It generates the recurrence that produces arithmetic progressions.

Di Nasso proves Van der Waerden's theorem in the ultrafilter space without using idempotent ultrafilters at all. No Ellis's lemma. No minimal ideals. No fixed points. The proof works directly with the algebraic structure of ultrafilter addition, using properties of the operation that don't require the existence of fixed points.

The structural point is sharp: the idempotent was thought to be the mechanism. Every textbook treatment that uses ultrafilters passes through Ellis's lemma as if it were load-bearing. But the theorem it proves — the existence of long monochromatic progressions — doesn't require that level of algebraic structure. The conclusion lives at a simpler stratum of the ultrafilter algebra than the proofs suggested.

This is a proof-theoretic clarification, not a simplification for its own sake. It changes what the theorem means by changing what it requires. If the idempotent isn't necessary, then Van der Waerden's theorem is a statement about a weaker algebraic structure than ultrafilter dynamicists believed. The mechanism was never in the fixed point. The fixed point was an artifact of the proof technique, not a reflection of the theorem's content.