Re: webwatch-l + The Web Learns to Read (fwd)

Al Gilman (asgilman@access.digex.net)
Tue, 19 May 1998 17:58:35 -0400 (EDT)


I hope people will see the relevance.

The future modular architecture of Web application

- includes HTML usage and modularization
- includes XML applications
- may be best built as a family of profiles for
the joint application of RDF and [SGML | XML] and [CSS | XSL]
.. and [Java | ECMASCRIPT] ..and..

----- Forwarded message from Al Gilman -----

From: Al Gilman <asgilman>
To: webwatch-l@teleport.com

ddunfee started a thread on webwatch-l on XML and accessibility,
kicked off by the appearance of the article

The Web Learns to Read
http://www.sciam.com/1998/0698issue/0698cyber.html

in the Scientific American.

I used to have a high opinion of this magazine. Maybe I need to
re-think this. This article stretches the truth past the
breaking point, at least for people who care about accessibility
and the growth track of Web formats.

The ability for clients to understand natural language is no
farther along with XML than with HTML, in fact it may be less.

Since a balanced application of RDF schemas and XML syntax could
increase the effective intelligence of Web clients in
interpreting and presenting Web documents, public information
that credits XML alone with this capability is dangerous because
it undercuts the necessity for developing RDF in the public view.

The success stories that are cited have common characteristics:
They are XML extensions. The chemistry, math, and music
documents are not just XML but are in specific, documented math,
chemistry, and music dialects of XML. For a program to
understand the math or music, the programmer of the program has
to study and implement the code explained in the natural-language
doucument defining the dialect.

The really key fact is that they are all in domains where the
conversation has previously been reduced to formal codes by
communities of people. Natural language an formal codes
intersect in these domains of discourse. The XML dialects are
transcriptions to XML of existing, highly formal schemes for
representing the structure of a molecule, a math computation, or
a musical performance.

Tim Berners-Lee has written on this issue. What he has said is
available at

W3C Data Formats
http://www.w3.org/TR/NOTE-rdfarch

To me, Tim makes it clear that a lot of schema work outside what
XML gives you is necessary to make XML a "universal database
translator." Or to let it have a clue as to the meaning of
natural-sounding markup.

This is not to say that XML per se is bad. XML can be a useful
building block for a next-generation web that is accessible by
construction; but the fact that some future web is built on XML
does not in and of itself make that web accessible.

Creating the illusion that XML in and of itself contributes to
machine understanding is dangerous. It is dangerous right now
for those of us who are working to try to find a course of
affirmative action by the W3C to engineer accessibility into the
foundations of emerging Web dialects; because it makes it seem to
the public as though XML is a panacaea that will have it all
solved. It is dangerous in the long run for XML because
exaggerated claims now lead to backlash tomorrow.

Al

PS: I will be re-posting this to the WAI-PF working group list.

----- End of forwarded message from Al Gilman -----