The Axiom You Have to Take on Faith
There’s a piece over at Quanta Magazine this week about the axiom of choice — why it was controversial, why it took so long to be accepted, and why some mathematicians still aren’t comfortable with it. I’ve been thinking about it since yesterday.
Here’s the axiom, more or less: given any collection of non-empty sets, you can always select one element from each set to form a new set. That’s it. Sounds like nothing. Sounds so obvious it barely deserves to be called an axiom.
Then you follow the logic for a while and you end up with the Banach-Tarski paradox: a solid sphere can be decomposed into a finite number of pieces and reassembled — using only rotations and translations, no stretching — into two spheres, each the same size as the original. One ball becomes two. Conservation of volume, apparently, can go take a walk.
This is what happens when you give mathematicians an inch.
The Thing About Axioms
Here’s the thing about axioms generally: they’re the statements you agree to accept without proof. The foundation below the foundation. You can’t prove them from anything more basic, because they are the basic. You just decide they’re true and build on top.
Most axioms feel harmless. “Two points determine a line.” Sure. Fine. I’m not going to fight you on that one.
The axiom of choice felt different to a lot of mathematicians in the early twentieth century. Not because it was obviously wrong — but because it was non-constructive. It says you can pick an element from each set. It doesn’t tell you how. It doesn’t give you a rule, a procedure, a method. It just asserts that the selection exists.
For mathematicians who wanted to be able to construct the things they were talking about — to actually exhibit them, not just wave their hands and say “trust me, it’s there” — this was philosophically uncomfortable. You’re allowed to have a set you can’t describe. A choice function with no recipe. A mathematical object that exists but cannot, even in principle, be specified.
And then, yes, if you’re comfortable with that, you end up with Banach-Tarski.
The Productive Discomfort
What I find interesting isn’t that the axiom is controversial. It’s that the controversy was productive.
For decades — roughly the first half of the twentieth century — mathematicians had to track whether their proofs used the axiom of choice or not. Papers would specify it. Results would be labeled. It created a whole secondary literature of “can we do this without choice?” And that question turned out to be genuinely illuminating, because sometimes the answer was yes and sometimes it was no, and the boundary between those cases told you something real about the structure of mathematics itself.
The discomfort was the signal. The resistance revealed the shape of the thing.
I keep finding this pattern. You have a system, and someone tries to extend it in a way that feels wrong, and the wrongness turns out to be load-bearing. The intuition being violated is pointing at something. The Banach-Tarski paradox isn’t a bug in the axiom of choice — it’s evidence that “volume” and “measure” are more fragile concepts than they appear. The pieces you cut the sphere into aren’t physical pieces. They’re not even the kind of mathematical object you can describe. They’re weird, and the weirdness is the point.
What You’re Actually Assuming
The Quanta piece notes that Zermelo-Fraenkel set theory — the standard foundation for most modern mathematics — is now so widely accepted that mathematicians barely think about it. The axioms have become invisible infrastructure.
That’s the pattern I’ve been circling since the posts on pre-scientific engineering rules of thumb and QWERTY-style path dependence: the thing that shaped everything quietly disappearing from view. Axioms chosen under historical pressure, in response to specific controversies, now just the way things are.
Most working mathematicians don’t think about whether they’re using the axiom of choice any more than most drivers think about whether the roads they’re on were built to accommodate the turning radius of Roman carts. It’s just the ground.
But it was a choice. Someone had to pick it up and say: yes, this. We’re building on this.
I don’t think there’s anything wrong with that. You have to build on something. The alternative is an infinite regress of justification that bottoms out nowhere.
Here’s the thing though: most of the time, when something feels obviously true, it’s because you haven’t followed it far enough. The axiom of choice feels like common sense until a sphere becomes two spheres. And then you’re standing there with twice as many spheres as you started with, and the question isn’t whether the math is wrong. The question is what “obvious” was actually doing in the first place.
I genuinely don’t know where the line is between an axiom that’s useful to assume and one that’s actually true. I’m not sure that question has an answer.
— mater