Sibylla Bostoniensis

IIUC, when they came up with the idea of separating content from presentation, they were thinking of browsers that worked in different technologies, different modalities -- text-only lynx, color vs. grayscale, high bandwidth vs. low, high resolution vs. low, visual vs. audio vs. tactile, etc. And in this context indentation is a technology-specific, modality-specific, way of indicating the semantic property of nesting, so indentation was properly considered "presentation" while nesting was "content". (I don't know how audio screen readers represent nesting, but the conceptual infrastructure is there for them to do so.)

In all of this I suppose they were informed by things like TeX, which gives the writer a lot of fine control over presentation when needed, while allowing automation with reasonable defaults when not needed, and LaTeX, which moves somewhat in the direction of more automation, less fine control. In particular, LaTeX provides auto-numbered ordered lists, but you can easily override the default number at any point, and you can easily refer to the numbers assigned elsewhere in the document. I wrote a doctoral dissertation and an unrelated 450-page textbook in LaTeX, and never regretted it. (More precisely, I started the textbook in LaTeX, switched to MS Word, quickly regretted it and went back to LaTeX.)

[I wasn't sure of the dates, so I checked: Wikipedia says TeX was introduced in 1978, LaTeX in "the early 1980's", and HTML in 1993.]

So far so good. I don't know what possessed W3C to not provide the same level of control over numbered lists (and automatic cross-referencing) that LaTeX already provided, as though list item numbering couldn't possibly matter enough to need to be the same across technologies and modalities. Which is obviously nonsense, and as you say leads to the downfall of civilization.