There's still the issue of displaying unrecognized markup. Three
possibilities that come to mind for dealing with that are
1. Displaying the text content without the markup, with instructions in
the users' guidelines to the effect that any inline markup scheme should
leave intelligible text when the tags are removed.
2. Converting the markup to text and displaying it, e.g.
is converted to
for display on a Web page.
3. Alternate text.
I'm partial to #2, which is the simplest to implement (no new
attributes/elements for denoting the alternate text, and no restrictions
outside of schema validity). It's also more or less what people have
been doing all along with BibTeX--putting the records on the Web with
embedded TeX markup unmodified.