Hacker News new | past | comments | ask | show | jobs | submit login

> As for "using the properties I've mentioned", presumably in reference to the integers being a well-ordered ring: sure, being a unique factorization domain follows from being a well-ordered commutative ring where the ordering interfaces with the ring structure in the expected ways... because the ONLY well-ordered ring is the integers, and the integers are a UFD

In my opinion, this is the heart of the issue.

Most people take the ordered structure of the integers for granted to such a degree that they won't think it's necessary to isolate it as an axiom of any attempted proof. They won't bother thinking about whether each step does or does not make use of this structure.

This means two things:

(1) they're very, very liable to produce a "proof" that appears to erroneously apply to other more exotic structures because it will implicitly make use of the structure of the integers.

(2) they're unlikely to agree that something like Z[sqrt(-5)] is "similar" in any way to Z.

Here's an example of something I take issue with from Gowers's post:

> Here’s an example of how you can use \mathbb{Z}(\sqrt{-5}) to defeat somebody who claims that the result is obvious in \mathbb{Z}. Let’s take the argument that you can just work the factorization out by repeatedly dividing by the smallest prime that goes into your number. Well, you can do that in \mathbb{Z}(\sqrt{-5}) as well. Take 6, for instance. The smallest prime (in the sense of having smallest modulus) that goes into 6 is 2. Dividing by 2 we get 3, which is prime.

From the perspective of a layperson, what's this "modulus"? Is a layperson really going to just unthinkingly agree that this is the "correct" way of finding the "smallest" prime that goes into 6 in this other ring? I seriously doubt it. I think they're going to immediately feel that there's a very natural and powerful order intrinsic to the positive integers, and the same is just not true of Z[sqrt(-5)], even if you can define a modulus or a norm. That modulus will feel arbitrary and meaningless to a layperson. It's going to immediately shoot up red flags that this is a completely different domain.




Yes, I agree that a layperson is unlikely to feel Z[sqrt(-5)] is similar to Z. That's empirically just true. On this point, we are in no contention.

But I disagree that laypeople are correct to consider unique prime factorization (or Euclid's lemma, or any such thing) for integers obvious, or have correct reasoning for it latent in their head.

You gave for example a perfectly correct argument in https://news.ycombinator.com/item?id=11957549. I disagree that laypeople have anything like that in mind when they are claiming these facts to be obvious.

The process that leads laypeople to consider these facts obvious is not based on their having any special intuitive understanding of the multiplicative structure of the integers. It's just the process of "I've never seen it fail, and I keep hearing it's true, so... yeah, how could it be otherwise? How COULD it be otherwise?", the same process that leads them awry in other cases where they draw actually wrong conclusions.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: