Plain text is great as far as it goes, but when it comes to structure you start from zero for every file. There’s always someone getting wistful about ad-hoc combinations of venerable Unix tools to process “plain text”, and that’s fine when you’re in an ad-hoc situation, but it’s no substitute for a well-specified format.
XML, JSON, YAML, RDF, EDN, LaTeX, OrgMode, Markdown... Plenty of plaintext, but structured information formats that are "yes, and". Yes, I can process them as lines of plain text, and I can do structured data transformations on them too, and there are clients (or readers) that know how to render them in WYSIWYG style.
If that’s our definition of “plain text”, sure. I would still rather our tools were more advanced, such that printable and non-printable formats were on a more equal footing, though. I always process structured formats through something that understands the structure, if I can, so I feel that the only benefit I regularly get out of formats being printable is that I have to use tools that only cope with printable formats. The argument starts getting a bit circular for me.
XML arguably isn’t plain text, but a binary format: If you add/change the encoding declaration on the first line, the remaining bytes will be interpreted differently. Unless you process it as a function of its declared (or auto-detected, see below) encoding, you have to treat it as a binary file.
In the absence of an encoding declaration, the encoding is in some cases detected automatically based on the first four bytes: https://www.w3.org/TR/xml/#sec-guessing-no-ext-info Again, that means that XML is a binary format.
[dead]