He's right about those generally not actually needing PDF for non-print purposes though - simple text and a few images is way more web-friendly as HTML. Sadly only very few repositories make HTML versions available as well.

Are you trolling? HTML was invented for scientific papers.

Try reading the HTML version of Fielding's paper and tell me it looks better than the pdf. It might have been invented for that purpose, but that doesn't mean it's still suitable for it. When you are writing a paper, you want to control the presentation, not just the content.

I'm pretty sure you have way more control over "presentation" when you use HTML vs PDF.

No you don't, HTML has massive variation between devices and what software is used to display it. Try setting your screen resolution to 640x480 and opening a webpage or, even worse, modifying DPI. PDFs on the other hand specify exactly where to place each glyph (admittedly there is still variation between software but it's much more consistent).

You sure do - the variation between devices is the manifestation of the ability to control layout depending on the recipient's display's resolution and size.

If anything, a PDF designed for A4/letter is going to be cumbersome to read on a (probably rather small) 640x480 display.

The sad state of PDF rendering on (most) e-book readers should be evidence enough.

More control over how it looks at the users' end with HTML than I do with Latex? Yeah, I don't think so. Why would everyone be using JS if you could do whatever you needed in HTML?

What makes it "scientific"? The fact that it is a PDF and the impersonal tone?

It was peer-reviewed and published in a journal.

You are probably thinking of the advertising whitepapers of scammy alt-coins and crypto-tokens. This is actually research.