The discussion reminds me a little of the one around LaTeX in the scientific community. Latex is extremely powerful but reading the raw source of a text is a pain. WISYWIG is just so much more comfortable when writing, but MS Word sucks in its own specific and manifold ways (as does LibreOffice).
HTML is still not nice to read, too much boilerplate tags required. Markdown source however is mostly well-readable in a plain-text editor, and offers the basic formatting that serves 95% of the needs.
But the truth is: the ideal text layout platform for scientific writing does not exist.
Markdown+HTML+CSS could be getting close, but it currently still lacks LaTeX's tooling for citations, support for pagination.
MS Word's "track changes" is extremely useful when collaboratively editing (and no, Overleaf's version doesn't come close, unfortunately). But references and citations in Word still are no match for the flexibility of bibtex, even with Papers/Zotero/Mendeley. Figure placing can be a nightmare, although LaTeX is not infallible, either. And you need a Mac or Windows to use Word, and on a Mac it feels like it's running in an emulator.
I have given up on the belief that one tool could solve all this. There are a few good attempts like overleaf and the less-known authorea (which I liked more), but these are cloud apps which complicates writing offline.
The only solution is to choose whatever tool works for the specific use case, learn to live with its shortcomings, and never forget that in the end only the finished manuscript counts.
