Re: Presentation and Semantics debate

Jonathan Abbey (broccol@csdsun1a.arlut.utexas.edu)
Tue, 3 Aug 93 10:06:03 CDT


I don't understand. Why should presentation elements be excluded as
antithetical to semantic markup? This seems unnecessary. It seems
that the argument is that if there is *any* kind of presentation-level
formatting in a document, the document becomes somehow "unclean". Presumably,
this is because the document has had information added to it that is not
accessible by automatic document processing software.

So?

The information is (still) for human consumption. I can envision systems
which would allow wonderful things to be done in an automatic fashion by
computer information systems, but then we would all be writing in something akin
to Lisp. When I write a document, I can throw in puns, jokes, statements
that make no sense without cultural referents, etc. This may obscure some
semantics from those who do not share the culture (or the language!), but
that's life. Things are better than that in HTML[+]* at the level we're
discussing, because the added information need not obscure any semantic
tagging that would be present in the DTD and the document.

I think that HTML[+]* should have as rich a semantic markup schema as is
possible. Put in there every conceivable kind of document element you
can think of. Put in links so that a semantic net can be created over
a set of HTML documents. Put in a powerful equation system. Go to town.
Then, when you are done, allow all kinds of formatting "hints" which everyone
is free to ignore if they want, but which will allow for professional
appearing documents in something like NCSA Mosaic.

The alternative will be for people to start writing their documents in
somewhat less structured formats. Like PostScript. Or GIF. Don't make
us do this. If HTML[+]* does not provide for flexibility in this area,
it should *not* be the global hypermedia system. Something like Acrobat
with hyper-extensions might better serve the need for a rich universal
document structure.

Attitudes like "no formatting information unless there is a defined
semantic purpose for it (and it had better be in the DTD)" are understandable,
they really are, and I applaud the intent. But why is it necessary to be
so rigid? As long as semantic formatting information is embodied by HTML[+]*
constructs to the best extent possible, why not allow those constructs to
have a wide range of optionally observed formatting attributes?

Baby for the bath water, etc.

Jonathan Abbey
broccol@csdsun1a.arlut.utexas.edu
Applied Research Laboratories
The University of Texas
Austin, Texas