Re: LYNX-DEV WAI launch notes: pt. 1

[Al Gilman]

Thank you, Chris.  I got a lot out of your report (pt. 1, even..).
Most of the points raised richly merit the grains of salt you 
applied to them.  I want to rattle on just a bit on the matter
of "Web literature as knowledge representation" because I don't
think saying that this is called "generic markup" tells enough
of the story.

Improving the integrity of Web literature as knowledge
representation is an important performance variable or strategy
for achieving more robust end-to-end communication in the face of
ability-diversity in the users at both ends.

It is easy to define methods of knowledge representation and
capture which are rich enough to support multimode presentation
at the client side but are too burdensome to be enforceable at
the author/server side.  Finding the right lightweight set of
requirements is going to be risky and take time.  There is still
hope.  Some of the new technologies are part of the reason for
optimism.

  From: "Christopher R. Maden" <address@hidden>

  Gary has an innovative idea: "Modality-independent means for
  representing knowledge; (prior to its expression in a particular
  modality so as to facilitate alternative displays and robust
  multi-modal understanding, and after its expression to facilitate its
  to [sic] translation to alternative forms or its reduction to summary
  forms in channel capacity-limited contexts.)"  He calls this
  "Intermedia".  Jon Bosak (XML chair) and others whacked him over the
  head after his presentation.  What he's describing is *not* a new
  idea; it's called "generic markup" and has been around for at least
  thirty years.  (Besides, Intermedia was a hypertext system at Brown
  University.)
  
The dream hasn't changed much.  What we know about how to get
there has improved a little.  There have been some changes in
programming in the intervening thirty years that increase the
prospects for success with this idea as compared to the forms it
has taken in the past.

The most important change is the practice in OO technologies of
classifying things according to what you can do with them.  This
helps in providing sifting and sorting that preserve their
validity over a range of category finenesses.

Prior attempts have been saddled with operations that were
equivalent to assuming atomicity of meaning.  This is why I keep
speaking that inscrutable grunt: "the metalanguage has to support
rules."  Once we start classifying information resources on the
basis of what use scenarios they support (not what tokens they
contain), we will start getting somewhere.

--
Al Gilman
;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;