Re: LYNX-DEV Internal MIME types

[Al Gilman]

  From: Klaus Weide <address@hidden>

  On Sat, 26 Apr 1997, Al Gilman wrote:
  
  >   From: Klaus Weide <address@hidden>
  >   
  >   How does [relaxing the parse data structure beyond being a
  > tree] further the goals of the Web *Access* Initiative?
  
  > [...] different dimensions of the content [...]
  >   [...] solution to the access problem[...]
  
  Okay, I think i have got an idea what you mean.  Now, today there are two
  dimensions, for most Web pages: images and ALT text.  Most people don't
  seem to care about the ALT TEXT, although that should take much less
  effort to provide than sophisticated images.  So what makes you think
  that, if there are many more dimensions to fill, people would care any
  more than they do now, about those things that are not as flashy, not as
  new as the latest toy?

Better, say images and text are the dimensions.  But these are not
independent enough the way they are used now.  

This is a very serious question.  And the process to a successful
solution is long and hard.  But I think that there is a win-path
and it is related to the following kinds of thinking:

        There is the 10-megaton version of combat, and there
        is ju-jitsu where you make yourself a fulcrum and use
                the energy of the adversary to defeat him.

        I think that ALT text doesn't work because it is purely added
        work.  

        In most writing where you use graphics and text, the figures
        have captions.

        I believe that the reason text interpretation of documents
        that use figures in HTML breaks because the language 
        specification does not capture the connection between
        graphic object, caption, and first-mention-in-the-text,
        for example.

        If you look at the actual text generated for pages without
        ALT text, there is usually
        text that would make the ALT unnecessary if it were but
        bound to the image by some linkage.
        If the web-authoring tools incorporated the creation of
        the necessary linkages between images and the _text that
        the author creates anyway_ which explain what the image
        tells you, you would rarely need to create additional
        text to make the text-mode browse whole.

        Read up my discussion on what we should do instead of this
        failed ALT mechanism for text-view presentation of links
        which have image content in graphics mode.

http://www.access.digex.net/%7Easgilman/web-access/

        That also illustrates how we need to be free to trade
        off among data, computation, and communication in the
        way we provide (variant forms of) various services.
        In that proposal I suggest we use more comm and not
        extra data.

        Similarly, suppose there is text or audio description
        for an image.  We probably don't want that embedded
        in the page as routinely distributed.

        There are lots of ways to combine services to keep
        the end product from being a bear.  There could be
        a MUD-like drop-in center where sighted chatters
        interpret images for blind chatters.  This could
        be just one of many translation paths served by a
        translation chat net.

        We need to engage in a free and open negotiation among
        content providers as well as blind accessors and server
        and browser developers.  There is no quick fix.  We can't
        take the first proposal.  The answer has to be extremely
        lightweight in terms of the overhead it imposes.

ALT text is a clumsy and burdensome solution.  It won't sell.
We need to understand the problem better to design a solution
we can afford.

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