For me, it's an especially bad argument because the sloppy nature of HTML parsing is NOT a virtue... it's a source of bugs, vulnerabilities, and incompatibilities that provides (yet another) technical moat for existing web browsers. It's a huge tragedy that HTML5 beat XHTML.

Maybe, but after several decades of engineering, I became much more of an advocate for actual observed user behavior, which is chaotic, and as such nowadays preach “embrace chaos”.

That’s what HTML and the browsers did. They accepted humans are terribly bad at following instructions when you want to cater to a broad audience, and as such embraced error correction. The end result is that the early days were awesome because everyone knew how to build websites.

Perhaps in the current day and age, where people hardly write hand-written HTML anymore, this can be reconsidered. But a new, more restrictive format would have to show real benefits, because it’s precisely as I say: people don’t really write raw HTML anymore so it’s kind of a moot point.