Awesome, hadn’t seen that. I’m sure there will be a zillion edge cases where it won’t work properly for specific documents because of odd latex quirks.
If the pdfs were served by some of the megacaps who posts tons of papers (e.g. Google or Facebook) then it would be an order of magnitude faster. And said megacaps would end up spending peanuts relative to the value they get from arXiv.
An order of magnitude faster sounds unimportant. Is all the time spent by researchers sitting around waiting for 0.4s for a paper to download really an issue?
They probably could make it faster by rendering server-side a preview of sorts. I don't think that most papers are large enough for this to have a major impact unless you have a very slow connection.
SumatraPDF on Windows is lightning fast with the kind of PDFs you get from the arXiv (true PDFs without heavy graphics). I don't know what is fast on Linux, but probably some MuPDF-based viewers are (the MuPDF PoC viewer is fast even on WSL).
https://www.arxiv-vanity.com/
I wish they would integrate something like that directly.