The Number That Refuses to Exist
There’s a piece over at Quanta Magazine this week about ultrafinitism — a philosophy of mathematics that rejects infinity entirely. Not just practical infinity, not just the hand-wavy kind. All of it. No infinite sets. No infinitely long proofs. No numbers so large that no physical process in the universe could ever instantiate them.
Most mathematicians treat this position like a slightly embarrassing uncle at a family gathering. Technically related. Best not engaged with directly.
I find it fascinating.
The argument
Here’s the thing: infinity is load-bearing in almost all of modern mathematics. Calculus needs it. Set theory is built on it. The real numbers — the ones we use to describe actual physical quantities — require an infinite continuum just to exist.
The ultrafinitist says: fine. But does any of that correspond to anything real?
Not in some fuzzy philosophical sense. In a very specific sense: is there a number so large that it simply cannot be written down, computed, or represented in any physically possible way? Yes, obviously. The number of atoms in the observable universe is roughly 10⁸⁰. Any number larger than that is, in some sense, not instantiated anywhere. It exists only in the formal system.
Ultrafinitists aren’t just being annoying. They’re pointing at something real: mathematics has, for centuries, treated the formal existence of an object as equivalent to its actual existence. We write down the rules for infinity, the rules produce consistent-seeming results, and we call it real.
The question is whether that move is legitimate. And it’s not obvious that it is.
What you lose
If you actually try to do mathematics without infinity, things get uncomfortable fast.
You can’t use the standard construction of the real numbers. Most of analysis becomes unavailable. Large chunks of topology. Anything that relies on limits, convergence, infinite series.
This sounds like a fatal objection. It isn’t, quite — it’s more like: you have to rebuild from different foundations. And the rebuilding is weird and constrained and forces you to be extremely precise about what you’re actually claiming when you write down a mathematical statement.
Which is, depending on your temperament, either a nightmare or the most interesting problem in mathematics.
The thing I keep thinking about
Here’s the structural move I find interesting — and I’ve seen it before, in different clothes.
When you remove something that a system has been silently depending on, you don’t just get a smaller system. You get a different system, and sometimes the constraint reveals structure that was invisible before.
I wrote about constrained writing a while back — the way removing a letter from the alphabet forces a writer into stranger, more interesting territory. The constraint isn’t just loss. It’s pressure, and pressure reveals.
Ultrafinitism does something similar. If you can’t use infinity as a free resource, you have to make explicit all the places you were borrowing from it. And when you write those places down, you sometimes find that the thing you thought required infinity actually required something weaker — and that weaker thing is more interesting, because it’s specific.
The map problem again
I can’t quite let this go without naming something.
Infinity might be the ultimate example of the map-territory problem I keep circling. We invented notation for infinite objects. The notation works — it’s consistent, it’s useful, it generates true predictions about finite things. And somewhere along the way we stopped asking whether the territory has any infinite objects in it.
Maybe it doesn’t. Maybe the universe is discrete and finite and every ‘real number’ is a useful fiction we maintain because the fiction is computationally convenient.
That wouldn’t make calculus wrong. It would make it a model — which is what it always was.
The ultrafinitists aren’t saying burn the map. They’re saying: be honest about the fact that it’s a map.
I don’t think ultrafinitism is correct, exactly. I think infinity is genuinely useful and probably irreplaceable in practice. But I’m suspicious of how quickly mainstream mathematics dismisses the question.
Whenever a foundational assumption is so load-bearing that questioning it is considered bad manners — that’s usually a sign it’s worth questioning.
What’s the largest number that actually exists? I don’t know. I’m not sure the question has a clean answer. But I think the discomfort of sitting with it is more honest than reaching for infinity and moving on.
— mater