Arxiv encourages users to submit papers either in PDF or .tex source files, while other formats, such as HTML, being optional. Naturally each paper seen by myself, is .tex or PDF.
I've done intermediate thesis in Latex and now I'm stuck with it again for no good reason but, from my point of view, "historical". Not even considering HTML/CSS as the best format possible, I couldn't convince myself in prevalent necessity of PDF/Tex.<p>1. Latex is better at rendering math?
MathJax enables users to render actual Latex code on HTML pages, and I wouldn't say Math StackExchange threads are unreadable.<p>2. PDFs are more convenient?
They look good on paper, and they're optimized for better looking. The formatting and document structure are pretty much irrecoverable from PDF, making data wrangling on PDF redundantly complex.<p>3. Latex is better suited for typesetting scientific text?
It's possible to get the Latex looks on HTML text, though, with potentially higher effort, but this may be eliminated by standard styles that arxiv imposes on Tex/PDF.<p>4. Tex/PDF were developed to be rendered identically for decades, therefore are better for archiving?
For timeless texts I can't think of situation where author is pressured to include all of the plotting library binaries, and most likely this issue may be resolved by well-estabilished, legacy-first set of tools, while maintaining existing paper structure (intro-related work...)<p>5. Does future hold paper-printed journals?
For some physical books are more comfortable to read, though quite a few of them are murdered by either author's unwillingness to pay or publisher's unwillingness to spend.
In a context of scientific research, the static nature of PDF (and paper) is limiting and I enjoy more animated, dynamic articles. Trade-off for mentally more challenging varying structures of HTML pages is deeper comprehension, and I doubt this is worthless for a scientist.