While this doesn't go into as much depth as draft-murata-xml does, the
HTML WG believes, despite the DOCTYPE/xmlns/HTML-header preamble, that
the bulk (i.e. body) of most XHTML documents will useful, to "some
extent" (per above), to casual users.
I think the general consensus of the MIME community is that making HTML
a subtype of "text/" was a mistake. While it is possible to write HTML
which is readable "to some extent" as plain text, the HTML that is
generated by a typical MUA or HTML editor is so full of useless cruft
that it doesn't qualify. Perhaps a determined human being can read the
text "to some extent" but the typical human gives up.
So IMHO we should learn from this experience and make XHTML and other
XML-ish things subtypes of application/.
I completely agree with Keith. Text/html was a mistake.
Ned