The real problem here is that, 20+ years on, printing to PDF is still a totally natural and easy-to-understand metaphor for a normal office desktop user; but producing HTML for the browser is still impossible for them.

If we simply had print-to-HTML functionality which resulted in a document identical to what you view onscreen while editing, PDF could die the death it deserves.

But HTML+CSS somehow manages to suck just as much for common usage, so it persists.

I wish epub would catch on for more than books. An epub is just HTML and CSS in a zip file, and a large part of the world population has a device than can load it and present it cleanly.