@aartaka two main reasons: you can honestly use a class on a div to convey the same thing without the extra tags, and (probably more important) a browser like Dillo doesn't support then properly, e.g. by default they would probably be rendered inline.
I could probably modify the guideline and consider semantic tags as something that could be OK as long as they are handled properly (e.g. default styling is provided for Dillo & Co., assuming they recognize a styling on an "unknown" tag).
@h3rald if we’re only talking styling, divs with classes (and maybe ARIA attributes) can express anything. But classes are not structured and require consensus. While elements are optional yet semantic and express the consensus nicely. If I wanted to make a user script that removes the <nav>, parses its content, and presents it as searchable menu, I’d have a better luck with <nav> than <div class="ad-hoc-nav">. Accessibility too.
@aartaka @mauve Sure, good argument. Again, semantic elements add a little bit of complexity and so do aria attributes. I wanted to place SubWeb just above (well, a good deal above) Geminitext, and also be compatible with older browsers for maximum compatibility.
In my mind, I was also a bit nostalgic about "old" HTML and CSS... things were much simpler 25 years ago than they are today: I just wanted to prescribe a minimal subset that can still be used to make pleasant looking websites. Let's just say it doesn't *require* you to use semantic elements, and they are not explicitly forbidden like <object> or <video>.
@mauve for sure, styling is quite inconsequential in these. That’s why I mentioned parseability and accessibility in my comments—these are more salient if we’re talking semantic elements!
@h3rald