I was trying to be circumspect in my annoyance at the div/p switcharoo, but I'm gratified to see I'm not alone in this.
@Sarmat89 - I'm curious what you mean by divs "requir[ing] less work to generate"... Are you referring to defining margins in css? If so, I suppose you're technically right, but I'd also think that typing "margin: 0;" isn't all that onerous. Or is there some other idiosyncratic workflow to which you are referring?
@Quoth - archive's epubs are a travesty. It doesn't help that their OCR is pretty crummy to begin with, but the epubs are basically just a straight pdf to epub conversion, complete with headers, footers, about a gajillion forced line breaks, and more OCR-based artifacts than you can shake a stick at. Without an ounce of exageration, the end result is flatout unreadable. I don't know why they even offer the epub file.
But at least it's not all <div> tags, I guess...
@Jellby - that's EXACTLY it... I've seen this now multiple times: the whole document is just <div>s and <span>s! I thought that maybe someone out there had a really unique workflow; I didn't realize that this is autogenerated. Although it invites the question: what kind of sociopath would design software that exclusively uses those two tags for text conversions...?