I would like to hear about the historical and technical reasons for why Zermelo-Fraenkel set theory with the axiom of Choice became the dominant standard for the foundations of mathematics.

The system has certainly gained a lot of momentum in the academic cirlce since its inception a century ago, but what are the details of the whole story?

I have not been able to find any answer online.


First, "foundations" are not what they once used to be. The idea of "one true logic" and "one true mathematics" justifiable from self-evident truths does not have much currency these days. So interest in real foundations, and belief in their existence or necessity, has been consistently waning, see Azzouni, Is there still a Sense in which Mathematics can have Foundations?

The Frege-Russell's project of transforming mathematical practice into formal enterprise did not come to fruition either. $ZFC$ is getting textbook lip service as a check on paradoxes and a source of mathematical logic curiae, but the actual proofs are still given essentially in the vernacular of naive set theory, see comments in the MO thread for current reactions. It is more prominent in mathematical logic and higher set theory, where the plethora of technical results on first order logic, on independence and consistency strength, and the fact that more complex theories are transparently modeled in it, turned $ZFC$ into a convenient common yardstick, a lingua franca of the field. But that, as with Latin or English, is, in part, a historical contingency.

Now, how it happened, Kanamori's The Mathematical Development of Set Theory from Cantor to Cohen is a detailed source. After the loud controversy over his axiom of choice Zermelo in 1908 set up a system $Z$ of seven axioms (AC not included) that "started from set theory as it is historically given... to exclude all contradictions" and "to retain all that is valuable". The authoritative summary of set theory, Hausdorff's Grundzüge der Mengenlehre (1914), which was to become Bourbaki's inspiration, did not incorporate it, see How Set Theory Evolved From Hausdorff Until Today. Hausdorff considered axiomatizations premature, and used refined naive set theory instead. But in 1910-1913 Russell and Whitehead published their Principia Mathematica, which accomplished a formidable task (with largely unacknowledged help from Schröder's Algebra der Logik (1890-1905)): it convinced the initiated that all of mathematics known to date could, in principle, be fully formalized.

Over the course of 1920-s two major developments took place: von Neumann and Fraenkel added axioms of regularity and foundation to $Z$, and the primality of first order logic started emerging. The latter is often credited to Skolem and Hilbert, see How did first-order logic come to be the dominant formal logic?, but it was solidified by Gödel's theorems that displayed its technical virtues. Ironically, Gödel originally proved incompleteness in $PM$, which was not first order, and Zermelo, who endorsed $Z$ with additional axioms in 1930, the modern $ZF$, advocated its second order reading. How $PM$ was gradually washed out of circulation due to its maze of ramified types and clumsy notation, can be seen, in part, from Who superseded Peano's dot notation in symbolic logic and when? So the early alternatives fell by the wayside. When Bourbaki started putting out their Éléments de mathématique in 1939 their axioms were not quite Zermelo's, but the system was equivalent to $ZFC$ minus foundation, see On Bourbaki's axiomatic system for set theory.

What about later alternatives? Gödel showed that $PM$'s theory of types was equivalent in consistency strength and expressive power to $Z$, which was simpler and closer to the vernacular. Bernays, anticipated by von Neumann, proposed a set theory with classes, $NBG$, adopted by Gödel in 1940, which was proved to be a conservative extension of $ZFC$. Quine, another influential advocate of first order logic, proposed New Foundations in 1937, later $NFU$, which also turned out to be bi-interpretable with $ZFC$. By 1960-s it became clear that genuine alternatives (see SEP survey) concern themselves with matters that ordinary mathematicians need not concern themselves with. And $ZFC$ had the advantages of simplicity and familiarity. For comparisons with later non-set-theoretic "foundational" alternatives, such as category theory or recent univalent foundations, see Dzamonja, Set Theory and its Place in the Foundations of Mathematics.

  • $\begingroup$ This answer is so good I saved it to my PC. $\endgroup$ – Alex Sep 26 '20 at 16:22
  • $\begingroup$ "von Neumann and Fraenkel added axioms of regularity and foundation to 𝑍" Regularity is the same as foundation. Do you mean replacement? $\endgroup$ – Noah Schweber Oct 5 '20 at 2:54
  • $\begingroup$ NFU is definitely not bi-interpretable with ZFC - it has far too weak consistency strength. (And meanwhile NF is not known to be consistent even relative to large cardinals - there's a claimed relative consistency proof by Holmes, according to which NF would have much weaker consistency strength than ZFC, but as far as I'm aware it hasn't been accepted yet despite being around for a few years now.) $\endgroup$ – Noah Schweber Oct 5 '20 at 2:56
  • $\begingroup$ @NoahSchweber SEP's long version:"The NFU world can be understood to be a nonstandard initial segment of the world of ZFC (which could be arranged to include its entire standard part!) with an automorphism and the ZFC world (or an initial segment of it) can be interpreted in NFU as the theory of isomorphism classes of well-founded extensional relations with top (often restricted to its strongly cantorian part); these two theories are mutually interpretable, so the corresponding views of the world admit mutual translation." $\endgroup$ – Conifold Oct 5 '20 at 8:22
  • $\begingroup$ @Conifold That's bizarre. Unless I'm having a terribly stupid moment it's definitely not true since ZFC proves the consistency of NFU (to put it mildly). I think that passage is flat-out wrong. $\endgroup$ – Noah Schweber Oct 6 '20 at 3:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.