Regular readers may recall that a while ago I posted a fallacious proof of the continuum hypothesis. They may also recall that a bit more recently I tried unsuccessfully to understand forcing. Forcing is a technique for building models of set theories, which Paul Cohen used to show that the continuum hypothesis doesn’t follow from the ZFC axioms. (The ZFC axioms are the Zermelo-Fraenkel axiomatization of set theory plus the axiom of choice, and I’m told that most normal maths that can be proved at all can in principle be cast in terms of set theory and then its set-theoretic version can be proved from these axioms, if you feel the need to do that.) Anyway, in spite of the evidence that I’m not very good at it, I’ve been thinking about the continuum hypothesis again.
The continuum hypothesis, for those of you who don’t know but are still reading, is the proposition that the number of real numbers is the second smallest infinite number. It’s been proved that the number of integers is the smallest infinite number, that the number of real numbers is bigger, and that there is a second smallest infinite number. Georg Cantor conjectured that the number of real numbers was the second smallest, Kurt Gödel proved that the continuum hypothesis was consistent with the ZFC axioms, and Paul Cohen proved that it didn’t follow from them. So we’ve established that the ZFC axioms, even if true, don’t settle the question. And in the time since Cohen finished the independence proof, set theorists have learned a great deal more about which axiom systems do and don’t settle the continuum hypothesis, which models it is and isn’t true in, and the relationships between them. (I don’t understand this work myself: in order to understand it I’d have to understand how forcing works, and I’m sorry to report that I still don’t.)
Now, some people respond to this situation by saying that the continuum hypothesis (CH from now on) is indeterminate. This might be because they think that the notion of a set corresponding to ZFC doesn’t pin down a particular model, and CH is true in some and not in others. It might be because they think there are multiple set-theoretic universes out there, and CH is true in (or of) some of them but not others. It might be because they think mathematical truth basically amounts to provability, and we know that neither CH nor its negation is provable. Or perhaps they think the indeterminacy comes in somewhere else, for example as fundamental metaphysical indeterminacy.
Some people are set-theoretic Platonists, who think there’s a real, determinate universe of sets out there, and that even though we don’t know enough about it yet to prove that CH is true or false in it, CH is nonetheless true or false in it. To settle the question we would have to learn things about the set-theoretic universe that don’t follow logically from what we already know about it. How exactly we’d go about learning something like that is a vexed question, but (so the story goes) we have already managed to learn some things about it that don’t follow from nothing, so perhaps we could pull off the trick again. I’ve heard something like this view attributed to Gödel, though I’m not sure how fair a reflection of his views what I heard was, or how badly I’ve garbled what I heard.
I’ve got a certain amount of sympathy with this kind of view in principle, but I don’t agree with it. I think sets are made up, and that makes me a fictionalist about sets. I don’t think that integers and real numbers are made up, and my attitude towards them probably makes me a Platonist about those. I sympathize with the Platonist view about CH in principle because I’d be happy to take this attitude towards mathematical entities I didn’t think were made up. But it’s not my view because I do think sets are made up. In fact, the more I hear about sets, the more made up they sound.
Sometimes we talk about things being true or false in fictions, and I think that’s a reasonable way to talk. It’s true in the relevant fictions that Miss Marple solves crimes, for example. I think that there are a few fictions relating to set theory, and that we have pretty tight rules for establishing what is and isn’t true in several of these fictions. (Much tighter rules than we have for Miss Marple.) As is the way with fictions, sometimes the rules don’t settle what’s true in the fictions, and maybe sometimes nothing settles it at all. We tend to be more easygoing about indeterminacy in fiction than about indeterminacy in reality, and that’s probably fair. If you give a set-theoretic statement of CH, that will tend to be true in some of these fictions, false in others, and indeterminate in others. So you might think that a fictionalist about CH should think this is all there is to the truth or otherwise of CH.
Well, that’s not what I think. Take another look at what I said CH was, earlier in the post:
"The proposition that the number of real numbers is the second smallest infinite number."
The eagle-eyed and literal-minded among you will notice that this doesn’t mention sets. It mentions numbers. Since in this instance we don’t need to worry about uninstantiated numbers, we can cast CH without mentioning sets, or even mentioning infinite numbers:
"There aren’t any things such that there are more of them than there are integers but fewer of them than there are real numbers."
We’ll call this CHWS, for Continuum Hypothesis Without Sets. I think that when we ask whether the continuum hypothesis is true, this is the question we’re ultimately interested in, at least under the assumption that the integers and real numbers exist. But I expect that some set theorists will like to think of the set-theoretic formulation of CH as the continuum hypothesis, and that’s fine. I’ll try to sidestep the terminological issue by giving it a slightly different name: CHWS. In deference to these imagined set theorists, from now on I’ll use CH for their set-theoretic version.
I said earlier that I was a realist, and probably a Platonist, about integers and real numbers. That’s more or less true, in that it’s my working hypothesis even if I’m not fully committed to it. As I see it, if you’ve got some things, then it makes sense to talk about how many of them there are, and whether the number of them is infinite, and whether there are more of them than there are of some other things. And none of this presupposes the existence of sets. When you say there are more even numbers than prime numbers between 1 and 100 you’re saying something about numbers, not about sets. We’re able to give a precise explication of these notions in set-theoretic terms, and this explication serves for most purposes, but you might think that one lesson of the independence of CH from ZFC is that ZFC set theory can’t help us answer the question about how many real numbers there are. But if I’m a Platonist about real numbers, which I probably am, and I don’t think that there’s indeterminacy in how many of them there are, which I don’t, then CHWS can still have an answer. It’s just that set theory alone can’t help us find it. Perhaps sometimes when you ask whether there are more Xs than Ys this doesn’t always have a determinate answer, and so CHWS could be indeterminate for that reason. But I don’t see that anything we’ve learned about set theory compels us to think that. I think it’s a very strange idea, although the subject matter seems sufficiently strange that I should be open to the possibility, and so I am. But being open to the possibility that CHWS is indeterminate is very different from thinking it actually is indeterminate. And if I had to guess, I’d say it probably isn’t.
So in summary, here are the things I said I think:
- Sets are made up, so I’m a fictionalist about sets.
- Integers and real numbers aren’t made up: I’m a Platonist about those.
- We can make sense of questions about whether or not there are more Xs than Ys without understanding them in terms of sets.
- Sometimes those questions have determinate answers, and maybe they always do.
- The continuum hypothesis can be cast as a question about how many real numbers there are, without reference to sets.
- This question may have a determinate answer, and nothing about set theory gives us much reason to think it doesn’t.
As I understand it, this isn’t a popular combination of views, and I suppose it might even be provably incoherent. Perhaps there are people who could sensibly be confident that they could prove that it’s incoherent in an afternoon. But I’ve held these views as working hypotheses for a while now, and although I don’t think I’ve converted anyone, I also haven’t noticed them leading to any problems in the general pursuit of truth, so to speak. For somebody who has the information and interests I have, they make a livable position, and livability is a source of evidence in philosophy. But I do worry that I don’t understand the issues well enough to have earned the right to hold this combination of views. Maybe my taking a position on this stuff at all is pure hubris, and my foolish, incoherent position is a result of that hubris. Or maybe I’m more or less right, and my reasons for arriving at the position are sensible enough that being right is some kind of epistemic achievement. (Or maybe I’m entitled to a view, but am nonetheless wrong.) On the one hand the Enlightenment ideal is supposed to involve having confidence in your own rational capacities and making sense of things for yourself, instead of taking things on trust from people you’re told are authorities. On the other hand, that’s exactly how people end up posting flat-earth videos on Youtube.
So, am I being like a flat-earther? The unpopular part of my position, as I understand it, is that I’m a fictionalist about sets but I think CHWS may still have a determinate truth value. It seems to me that a couple of things differentiate me from the flat-earthers. One is that I’ve studied a lot of philosophy, including some philosophy of maths, and this should give me some protection against making really silly mistakes when thinking about this kind of thing. Another is that members of the academic establishment put quite a lot of effort into debunking flat-earthism, whereas nobody’s putting much effort into debunking the combination of set-theoretic fictionalism and realism about a version of the continuum hypothesis. Although like I say, I wouldn’t be all that surprised if a philosophically informed set-theorist with a free afternoon could do it. If there is a quick debunking to be made, the main pressure point seems to be the notion that we can make comparisons of infinite sizes without dealing with sets. More specifically, there’s the notion that we can make comparisons between uncountably infinite sizes without dealing with sets. Can we? It’s not obvious to me whether we can or not. Maybe that’s the question I need to answer before I’m allowed to have an opinion on this stuff.