Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

ArXiv now offers papers in HTML format

blog.arxiv.org

Post reply on HN
Print view View original

131–140 of 325 posts

Re: ArXiv now offers papers in HTML format

#131

Post by murphyslab »

philipashlock wrote: post #98

30 years after HTML was invented to support accessibility and collaboration for research and academia and the same day the White House released their new accessibility guidance which happens to be the first time they've published formal new policy natively has HTML rather than PDF - https://www.whitehouse.gov/omb/management/ofcio/m-24-08-stre...

I feel surprised by how succinct, easy-to-understand, and sensible the policy (M-23-22) is:

> Default to HTML: HyperText Markup Language (HTML) is the standard for publishing documents designed to be displayed in a web browser. HTML provides numerous advantages (e.g., easier to make accessible, friendlier to assistive technology, more dynamic and responsive, easier to maintain). When developing information for the web, agencies should default to creating and publishing content in an HTML format in lieu of publishing content in other electronic document formats that are designed for printing or preserving and protecting the content and layout of the document (e.g., PDF and DOCX formats). An agency should develop online content in a non-HTML format only if necessitated by a specific user need.

https://www.whitehouse.gov/omb/management/ofcio/delivering-a...

Re: ArXiv now offers papers in HTML format

#133

Post by ahepp »

miki123211 wrote: post #46
tarboreus wrote: post #39

One of the reasons is to make the papers more accessible to people with disabilities, especially the blind. I participated in a conference they hosted on this a few months ago, I recommend taking a look at the recordings if you're interested in thinking on this. https://accessibility2023.arxiv.org/

Blind person here, can confirm this. Reading PDFs with a screen reader is bad, reading PDFs that come from LaTeX is worse, reading LaTeX math is pretty much impossible. All the semantic info you need is just thrown away. You can make decently accessible PDFs but it's lots of work, you need Acrobat on the producer' side and might also need it on the consumer's side. Free tools don't even come close. There's also the f…

Do you think there's potential for language models to play a role here? I know that AI can get tossed around as a buzzword, but hasn't it proved quite successful in fields like computer vision?

I'm not deeply familiar with the state of that art, but it seems like recovering the metadata from a PDF generated by LaTeX would be no more impressive than many other things we're currently seeing language models achieve?

Re: ArXiv now offers papers in HTML format

#135

Post by mastazi »

FredPret wrote: post #28
burkaman wrote: post #4

Earlier quoted context omitted.

And here's the PDF of the same paper for comparison: https://arxiv.org/pdf/2312.12451.pdf

The contrast is massive. I'm much more likely to read the html version; that PDF is deeply off-putting in some hard to define way. Maybe it's the two columns, or the font, or the fact that the format doesn't adjust to fit different screen sizes.

I wonder if perhaps it's a generational thing, I prefer the PDF because it reminds me of printed paper, which is what I used growing up.

(For reference: I am at the end of Gen X, people 3-4 years younger than me are considered Millennials).

Re: ArXiv now offers papers in HTML format

#137

Post by staunton »

ahepp wrote: post #133
miki123211 wrote: post #46

Earlier quoted context omitted.

Blind person here, can confirm this. Reading PDFs with a screen reader is bad, reading PDFs that come from LaTeX is worse, reading LaTeX math is pretty much impossible. All the semantic info you need is just thrown away. You can make decently accessible PDFs but it's lots of work, you need Acrobat on the producer' side and might also need it on the consumer's side. Free tools don't even come close. There's also the f…

Do you think there's potential for language models to play a role here? I know that AI can get tossed around as a buzzword, but hasn't it proved quite successful in fields like computer vision? I'm not deeply familiar with the state of that art, but it seems like recovering the metadata from a PDF generated by LaTeX would be no more impressive than many other things we're currently seeing language models achieve?

I'm absolutely positive a few million dollars could get you a system that can "read aloud" pdf math papers in no time. I guess people will wait for it to become cheaper though.

Re: ArXiv now offers papers in HTML format

#138

Post by wolverine876 »

murphyslab wrote: post #131
philipashlock wrote: post #98

30 years after HTML was invented to support accessibility and collaboration for research and academia and the same day the White House released their new accessibility guidance which happens to be the first time they've published formal new policy natively has HTML rather than PDF - https://www.whitehouse.gov/omb/management/ofcio/m-24-08-stre...

I feel surprised by how succinct, easy-to-understand, and sensible the policy (M-23-22) is: > Default to HTML: HyperText Markup Language (HTML) is the standard for publishing documents designed to be displayed in a web browser. HTML provides numerous advantages (e.g., easier to make accessible, friendlier to assistive technology, more dynamic and responsive, easier to maintain). When developing information for the we…

Hmmm ... accessibility is essential, but PDF is far better for static documents: There's no straightfoward, standard way to read an html document on another platform. Also, the html document may not be readable in 10+ years (unlike most PDFs), and updates are too fluid and hard to track.

I think the general problem is that the end-user doesn't control an html document, e.g., for annotation, as a local record, etc.

Post reply on HN