News for hackers. Stuff that matters.
Re:How many real numbers exist? New proof moves closer to an... by dwohnitmok on Thursday July 15, 2021 @02:38PM (#27848398)
Man there's a lot of juicy stuff in this article (Woodin's Ultimate L program gets briefly alluded to at the end of the article).

I just want to point out, because the HN crowd seems to generally not be mathematical Platonists, that this entire article is implicitly assuming a Platonist philosophical foundation. This may cause confusion for lay readers who are not mathematical Platonists.

In other words the article assumes that mathematical objects have an objective existence: they either exist or they do not. Hence every single logical axiom has a truth value. You do not get to arbitrarily choose what logical axioms you want. If you do, you can end up choosing the "wrong one." Therefore it is an important question to understand whether the Continuum Hypothesis is true or not, even if it's independent of ZFC (and hence requires ultimately philosophical rather than mathematical arguments).

If you aren't a Platonist and instead view logical axioms as having no inherent truth value, but rather foundations that you can pick and choose from as necessary (where in one case you may choose to use an axiom and in another you may choose its negation), then all that might sound very strange to you. In that case, you should mentally substitute every instance of "true" or "false" in the article with "agrees or disagrees with the meta model used to examine the semantics of a logical theory." In particular, whenever we talk about a formal treatment of the semantics (i.e. model) of a theory, whether that be something like ZFC or a programming language, we must always make those statements relative to a meta-model.

For example if we talk about the formal semantics of a language like C, we must first posit a meta-model which already contains notions of things like "integer" and "natural number" which can be used to give meaning to statements such as "performed an operation n times." If you're not a Platonist then you probably believe that there are multiple possible meta-models you could use.

Re:How many real numbers exist? New proof moves closer to an... by IngoBlechschmid on Friday July 16, 2021 @11:52AM (#27857574)
Indeed! And in fact, there are set theorists who argue that the traditional dream solution to the continuum hypothesis, adopting new axioms, is doomed to fail, and that we should instead embrace all models of set theory, all possible universes of mathematics.

I tried to write an introduction to this "set-theoretic multiverse philosophy" which is accessible to the HN crowd here:

https://iblech.gitlab.io/bb/multiverse.html

Re:How many real numbers exist? New proof moves closer to an... by ummonk on Saturday July 17, 2021 @07:11PM (#27869524)
Yeah, I tend to lean towards the side of realism (and constructivism), explicitly for the reasons exemplified by your rock / bottle analogy downthread.

But even from that perspective, I found the article's discussion of "true" / "false" rather weird, since it seemed to be taking certain positions (namely regarding the continuum hypothesis) for granted even though those positions don't seem to be grounded in any material fact.

It's as if people are conjuring up imaginary fictional entities and then asserting certain attributes about those entities as obvious and objective.

Re:How many real numbers exist? New proof moves closer to an... by red_trumpet on Thursday July 15, 2021 @05:06PM (#27850147)
Thanks for that perspective! Searching around I found this article https://plato.stanford.edu/entries/platonism-mathematics/#Tr... which even distinguishes mathematical Platonism and truth-value realism. Interesting stuff!
Re:How many real numbers exist? New proof moves closer to an... by dwohnitmok on Thursday July 15, 2021 @07:58PM (#27851575)
Indeed, that article is correct. More accurately I should be calling this position mathematical realism (although the lines between the two can get a bit blurry).
Re:How many real numbers exist? New proof moves closer to an... by zajio1am on Thursday July 15, 2021 @07:13PM (#27851196)
> this entire article is implicitly assuming a Platonist philosophical foundation

Is mathematical platonism still significant position between mathematicians? I thought it is outdated since Lobachevsky.

Re:How many real numbers exist? New proof moves closer to an... by saithound on Friday July 16, 2021 @04:05AM (#27853971)
Note that a view can be a majority position even if it is extremely outdated. This often happens when the outdated position is much easier to explain than the more nuanced alternatives. E.g. when surveyed, evangelical preachers often endorse Arianism, despite it being considered outdated since 325AD.
Re:How many real numbers exist? New proof moves closer to an... by dwohnitmok on Friday July 16, 2021 @12:04PM (#27857728)
Even though I'm personally not much of a realist, I feel like it's worth explaining why it's so popular when studying foundations of mathematics, because outdated is selling it short and it's actually a fairly nuanced position. The basic question is are mathematicians fundamentally discovering mathematics (Platonism) or creating mathematics (non-Platonism)?

The first step is that it seems like natural numbers are "real" and "tangible" in some sense. So for example, even though we can have modular arithmetic that would say maybe 2 + 5 = 2 under addition mod 5, it clearly feels "artificial" compared to 2 + 5 = 7. More concretely there seems to be some universal idea that seems to underly the fact that if you have three stones and add another stone you get four stones, or if you have three bottles and add another bottle you get four bottles. "three", "one", and "four" might not exist in the same as "bottle" or "stone" exist, but clearly they seem to exist and are "true" in some sense (it is clearly incorrect to say if you have three bottles and add another bottle you still have three bottles).

Indeed, to even talk about mathematics it seems like you need some intuitive grasp of natural numbers that precedes mathematical axioms, or at least be able to understand at a deep level that if you have n things and you add another thing, then you have n + 1 things, and that any axioms that violate this are more "artificial" than axioms which don't violate this. More generally, if you have axioms that create results which violate the rules of "normal" natural numbers, it seems reasonable to say that those axioms are "artificial" and not "real" in some sense. And so one interpretation of mathematicians' jobs, at least when tightly scoped to natural numbers, is to find those axioms of the natural numbers that reflect our natural numbers really work in "our universe" as opposed to a hypothetical other universe. And that approach seems to presuppose that natural numbers exist, at least in our universe, in some objective some that we are "discovering" with our axioms, rather than "creating" ex nihilo.

Indeed being able to say the term "normal natural numbers" and have even lay, non-mathematicians understand what that means (as opposed to "weird natural numbers" such as modular arithmetic) seems to suggest a certain objectiveness to the natural numbers.

Or to put it another way, the abstract notion of "counting" seems to exist objectively and outside of our formal mathematics theories, and our formal theory of natural number is really trying to describe "counting" rather than create the notion of "counting" from scratch.

Now if you accept that "counting" as an abstract idea exists in some sense that makes it possible to say either "yes that is an accurate description of counting" or "no that is not an accurate description of counting," then Godel's incompleteness theorems become quite interesting, namely that every useful theory of the natural numbers will have additional axioms that are independent of the ones already included in that theory. The non-realist looks at the incompleteness theorems and says, "Well yeah those new axioms just create different 'natural numbers' no biggie." The realist says, "That seems totally at odds with the fact that even though hypothetically you could have multiple notions of 'natural number,' there is only one notion of 'natural number' that holds true in our current universe and agrees with our similarly abstract notion of 'counting' and not a hypothetical other universe!"

And if you accept that for the natural numbers... well a lot of things have ramifications for the natural numbers. One classic example, as mentioned elsewhere in this HN discussion, is the busy beaver function, a function whose input is the number of instructions in a given Turing machine and whose output is the maximum number of steps the machine could possibly take for any input before it must be non-terminating on that input (basically an end run around the halting problem).

Any set of axioms (including ZFC) that includes the notion of arithmetic makes a statement on what it think a finite number of busy beaver values are (and only a finite number, again the halting problem prevents us from knowing more). And yet it seems like every value of the busy beaver function has some objective truth value in our universe! I can just run the Turing machine and find out! And that seems like an objective yardstick that we can use to determine whether a given axiom system is "true" or not in our universe (granted it's not one of very much utility because you are waiting to see if a machine runs forever, but still something that seems to have obvious physical ramifications).

Re:How many real numbers exist? New proof moves closer to an... by zajio1am on Friday July 16, 2021 @03:38PM (#27860349)
> The basic question is are mathematicians fundamentally discovering mathematics (Platonism) or creating mathematics (non-Platonism)?

My main issue with mathematical platonics is not whether we are discovering or creating mathematic (which seems to me as semantically empty question), but whether existence of matematical object is an absolute property, or a property relative to a specific model/structure, and whether some models are metaphysically exceptional, or all models are metaphusically equal, only some are more convenient to use, so they are more worth studying.

Consider some poly-platonist, who thinks that both natural-number-structure and non-standard-number structure (and both models of ZFC+CH and models of ZFC+non-CH) exist and are "real" and "tangible", and we discover internal relations in these structures as mathematical knowledge.

> Or to put it another way, the abstract notion of "counting" seems to exist objectively and outside of our formal mathematics theories, and our formal theory of natural number is really trying to describe "counting" rather than create the notion of "counting" from scratch.

I can say that we all have clear informal concept of what are natural numbers, just limitations of logic prevent us to formalize that.

But i cannot see how such argument can be extended to set theory, where are plenty of arbitrary choices how to axiomatize that (e.g. ZFC vs. New Foundations).

Re:How many real numbers exist? New proof moves closer to an... by dwohnitmok on Friday July 16, 2021 @04:18PM (#27860771)
To keep my Platonist hat on... Well if you accept that there is an objective truth to the natural numbers, the set theory axioms you choose will have ramifications for the natural numbers. (Elsewhere in these threads a commentator brings up that Not(Con(ZFC)) will tell you a Turing machine terminates when in our universe it doesn't)

Put another way there will always be a purely arithmetical statement of the natural numbers that is independent of your set theory axioms. What is the truth value of that statement? If you think you have a clear and absolute conception of the natural numbers it seems like you should believe that an arithmetical statement has an objective truth value, and if that's true, then that seems like your objective benchmark to decide on how to accept a new axiom (does it prove or refute this true arithmetical statement?). There's your absoluteness.

Re:How many real numbers exist? New proof moves closer to an... by zenon on Friday July 16, 2021 @01:04AM (#27853186)
In this survey, most "Philosophers of mathematics" endorsed Platonism: https://philpapers.org/surveys/results.pl?affil=Target+facul...
Re:How many real numbers exist? New proof moves closer to an... by saithound on Friday July 16, 2021 @04:00AM (#27853946)
edit: I did not pay enough attention, so what I wrote is wrong, see the comment below

In the survey you linked, it looks like 45.7% of the surveyed philosophers of mathematics endorse Platonism. That's a lot, but not "most", by any measure.

Re:How many real numbers exist? New proof moves closer to an... by oehtXRwMkIs on Friday July 16, 2021 @04:20AM (#27854060)
You're looking at the result for aesthetic value not Platonism.

And by the measure of "most" meaning majority it would still be most.

Re:How many real numbers exist? New proof moves closer to an... by dwohnitmok on Thursday July 15, 2021 @08:01PM (#27851594)
It is a significant position among mathematicans and logicians who study the foundations of mathematics.

It is probably not how most non-foundations mathematicians think of mathematics (especially larger infinities).

Re:How many real numbers exist? New proof moves closer to an... by a1369209993 on Thursday July 15, 2021 @09:47PM (#27852236)
I thought it was outdated since Epimenides. It requires the axiom of the excluded middle, which is about as false as axioms can possibly be, on account of having concrete counterexamples. (eg "This proposition is false.")
Re:How many real numbers exist? New proof moves closer to an... by drdeca on Friday July 16, 2021 @01:54PM (#27859087)
Intuitionistic logic, which is what you get from not assuming lem, still reaches a contradiction if you permit that as a valid proposition. So, doing away with LEM isn’t sufficient. You could use a logic that explicitly has more values though.
Re:How many real numbers exist? New proof moves closer to an... by kachnuv_ocasek on Thursday July 15, 2021 @05:07PM (#27850158)
Thanks for the clarification. I was indeed a bit confused by some of the implicit assumptions as an intuitionist myself.
Re:How many real numbers exist? New proof moves closer to an... by goldenkey on Thursday July 15, 2021 @03:59PM (#27849392)
Individually, or independently, axioms have no truth value. But when you put together a system of multiple axioms, they can contradict eachother. I see no problem with the question of whether the continuum hypothesis is true, given ZFC as a precondition. All we are asking is if the axiom contradicts ZFC. Axiom independence is very similar to operator commutation in quantum mechanics.
Re:How many real numbers exist? New proof moves closer to an... by dwohnitmok on Thursday July 15, 2021 @04:02PM (#27849414)
The Continuum Hypothesis and its negation are both proven to be consistent with ZFC (this is what the article is talking about RE forcing and Godel's proof of the consistency of CH). There is no contradiction to assume one or the other alongside ZFC. The article is really talking about Platonic truth when it talks about something being true or false.

> Individually, or independently, axioms have no truth value.

Indeed, you are not a Platonist :).

Re:How many real numbers exist? New proof moves closer to an... by IshKebab on Friday July 16, 2021 @08:54AM (#27855715)
Isn't that what an axiom is though? Something you have to define as true? The way you've described it, Platonism makes no sense. Maybe there's a missing part of the explanation...
Re:How many real numbers exist? New proof moves closer to an... by bonoboTP on Friday July 16, 2021 @11:38AM (#27857414)
Platonists think you can define them wrong if you're careless.

Imagine having a spreadsheet with the population of each country. I can go and enter 1 billion for the US, and the spreadsheet will happily recalculate all the cells about, say, GDP per capita etc. based on my incorrectly modified input. Clearly just because I've set the US population value in my spreadsheet to 1 billion the actual population did not change. There is a real population count and if we don't use the correct one then all our output will be garbage too: garbage in, garbage out.

Mathematics is like this spreadsheet, its input are the axioms and the outputs are various interesting theorems. Feed in wrong axioms and you get wrong theorems.

Platonists believe in a realm of ideas, that math talks about something real, just like the spreadsheet talks about real human populations. Even though within the spreadsheet everything remains fine and consistent if I tweak the US population. It just doesn't correspond any more to reality.

I myself am not a mathematical Platonist, I more of an engineer/practical person and see math as a survival tool of a bunch of primates on a rock floating in space, not something that taps into anything mystical.

But the Platonist worldview also seems coherent.

Re:How many real numbers exist? New proof moves closer to an... by dwohnitmok on Friday July 16, 2021 @01:20PM (#27858646)
bonoboTP gives a nice high-level description, I give a slightly more ranty description here: https://news.ycombinator.com/item?id=27857728.

Basically it seems like no matter what axioms you have, there's a bright line as to whether they accurately describe the natural numbers as we observe them in our universe, as that has physical ramifications. That bright line seems like "whether these axioms are true in our universe" and that seems to lend an objective standard by which we can measure our axioms.