Hacker News Reader

Comment by impendia | Hacker News Reader

> Use HTML rather then PDF.

The PDF is the original paper, as it appears on arXiv, so using PDF is natural.

In general academics prefer PDF to HTML. In part, this is just because our tooling produces PDFs, so this is easiest. But also, we tend to prefer that the formatting be semi-canonical, so that "the bottom of page 7" or "three lines after Theorem 1.2" are meaningful things to say and ask questions about.

That said, the arXiv is rolling out an experimental LaTeX-to-HTML converter for those who prefer HTML, for those who usually prefer PDF but may be just browsing on their phone at the time, or for those who have accessibility issues with PDFs. I just checked this out for one of my own papers; it is not perfect, but it is pretty good, especially given that I did absolutely nothing to ensure that our work would look good in this format:

https://arxiv.org/html/2404.00541v1

So it looks like we're converging towards having the best of both worlds.

View on HN

> In general academics prefer PDF to HTML. In part, this is just because our tooling produces PDFs, so this is easiest.

The tooling producing PDF by default absolutely makes the preference for PDF justifiable. However, tooling is driven by usage - if more papers come with rendered HTML (e.g. through Pandoc if necessary), and people start preferring to consume HTML, then tooling support for HTML will improve.

> But also, we tend to prefer that the formatting be semi-canonical, so that "the bottom of page 7" or "three lines after Theorem 1.2" are meaningful things to say and ask questions about.

Couldn't you replace references like "the bottom of page 7" with others like "two sentences after theorem 1.2" that are layout-independent? This would also make it easier to rewrite parts of the paper without having to go back and fix all of your layout-dependent references when the layout shifts.

HTML has strong advantages for both paper and electronic reading, so I think it's worth making an effort to adopt.

When I print out a paper to take notes, the margins are usually too narrow for my note-taking, and I additionally have a preference for a narrow margin on one side and a wide margin on the other (on the same side, not alternating with page parity like a book), which virtually no paper has in its PDF representation. When I read a paper electronically, I want to eliminate pagination and read the entire thing as a single long page. Both of these things are significantly easier to do with HTML than LaTeX (and, in the case of the "eliminate pagination" case, I've never found a way to do it with LaTeX at all).

(also, in general, HTML is just far more flexible and accessible than PDF for most people to modify to suit their preferences - I think most on HN would agree with that)

HTML still lacks one key feature: a way of storing the entire document as a single file that remains fully functional offline and can be reasonably expected to be widely supported for decades. Research papers are used both for communicating new results and for archiving them. The long-term stability needed for the latter has never been a strong point of web technology.

Indeed, I posted my first paper in 2006. It is still live on the internet in exactly the same format, and I've done absolutely nothing to maintain it.

I'm guessing there are few web pages of any significance which need to stay exactly the same for a long time. Here is one example which I've seen trotted out from time to time on HN:

https://www.dolekemp96.org/main.htm

This is clearly the exception. It seems that maintainers of web pages usually expect that they'll need to maintain and update them for as long as they want them to be accessible, and that's definitely not something I'd care to do for research papers.

You can make an HTML file self-contained by embedding CSS in a `<style>` tag and converting images to Base64, embedding them directly in the `<img>` tag as data URLs. This removes the need for external files, making everything contained within a single HTML file.

I agree that PDF is better than web technologies in terms of stability. I'm not objecting to PDFs being available (like you said, for archive purposes you want them provided by the authors), but to PDFs being the default, and oftentimes only, format available.

Note that ePub is basically just a zipped HTML file, and has become quite common for ebooks. I don’t know how that might be for archiving purposes?

I generally stick to PDF myself, but I do sometimes wish it would be more ergonomic to reflow a 2-column paper for reading on mobile on the go, for example. Also, ePub is easier to read in night mode than PDF recoloring, and seems easier to search through (try searching for a Greek letter in a PDF…).

EDIT: How is the math support in ePub though? Are people embedding KaTeX/MathJax or just relying on MathML, and how is the quality compared to TeX?

> Couldn't you replace references like "the bottom of page 7" with others like "two sentences after theorem 1.2" that are layout-independent?

Yes, but I think such references are inherently harder to locate. Personally I try to just avoid making references to specific locations in the document and instead name anything that needs to be referenced (e.g. Figure 5, Theorem 3.2).

Yes, I absolutely agree - I just figured that there had to be a reason that someone would want to do that. Chesterton's Fence and whatnot.

> This would also make it easier to rewrite parts of the paper without having to go back and fix all of your layout-dependent references when the layout shifts.

Just thinking about having to change layout-dependent references, every time I add two sentences to the introduction, gives me a migraine.

I never do anything like this in the paper itself, nor does anyone else that I'm aware of. I'm thinking of informal discussions, where I ask another mathematician about something specific in a paper.

I increasingly recommend against the Arxiv HTML version. I thought it had an acceptable start and they would fix the remaining problems and rapidly become on par with the PDF, but that seems to not be happening.

The HTML version is seriously buggy; and the worst part is, a lot of those bugs take the form of silently dropping or hiding content. It's bad enough when half the paper is gone, because at least you notice that quickly, but it'll also do things like silently drop sections or figures, and you won't realize that until you hit a reference like 'as discussed in Section 3.1' and you wonder how you missed that. I filed like 25 bugs on their HTML pages, concentrating on the really big issues (minor typographic & styling issues are too legion to try to report), and AFAIK, not a single one has been fixed in a year+. Whatever resources they're devoting to it, it's apparently totally inadequate to the task.

I think development on the TeX-to-HTML compiler has slowed down at some point, and it's far from perfect yet. Some of the issues are probably HTML5 limitations, unlikely to be fixed any time soon (unless one wants formulas to become graphics).

But there is another problem: It takes too long to load on mobile and doesn't reflow. I thought mobile was one of the reasons people wanted HTML in the first place!

> Some of the issues are probably HTML5 limitations, unlikely to be fixed any time soon (unless one wants formulas to become graphics).

You can convert a lot of formulas into either Mathjax/Katex-style fonts or MathML, or even just HTML+Unicode. (I get a very long way with pure HTML+Unicode+CSS on Gwern.net, and didn't even have to write a TeX-to-HTML compiler - just a long LLM prompt: https://github.com/gwern/gwern.net/blob/master/build/latex2u... )

But that's missing the point. Who cares about all of the refinements like reflow or pretty equations, when you are routinely serving massively corrupted and silently incomplete HTML versions? I don't care how good the typography is in your book if it's missing 5% of pages at random and also doesn't have any page numbers or table of contents...

In PDFs on arXiv, syntax highlighted codeblocks are graphics.

I think that's essentially only true if they are that in the original source. You can check for yourself, most papers have the TeX source available on arxiv.

> That said, the arXiv is rolling out an experimental LaTeX-to-HTML

Some history: https://www.arxiv-vanity.com/

I’m ok with the PDF but the title should be in HTML. The pdf failed to load for me due to tracker blockers (also why?!) so I was confused because there was no title but had comments

Reply on news.ycombinator.com