My understanding is that nobody should have to type
"naked" SGML (or HTML or Mime-language) anyway.
We should have programs like WYSIWYG editors
manipulating the markup for us. (Now of course
at present we do have to type HTML, at least I do
here, but hopefully this will not persist). If
that's right, then the more explicit and simple
the document structure is, the easier to parse
and manipulate by programs, the better we are.
One thing I like about Dan's proposal - it makes
it possible to collect a hyperdocument into a single
file (by embedding the docs within one mime file)
which will make transporting easier