Re: draft guideline (R2/2)

Al wrote:
> In terms of what works now, you have to look at what Jamal Mazrui
> is doing to take multi-file web documents and make them available
> as single-file text documents for blind people.  The offline use
> of the screen-reader is their analog to our dumping things to the
> printer.  I'm not sure that yet more separate files would deliver
> a net usability benefit in the end.

I just a got a mail from Mike Paciello forwarding a request from Jamal
to get HTML4 as one single text document.

As I said in my answer, before doing the HTML to text translation, a
single HTML document must be created from all the little documents
that make up the specs today. A simple concat will not do it since the
multiple HEAD/BODY would create parsing problems.
 
So a script must be written than does just that, in the order of the
TOC probably. Any summer programmer available ? (if not, I'll try to
get someone to do it at W3C, but our resource are scarce in this
area).
 
Al again:
> One of the issues that I realized is floating in my mind on this
> objective/subjective issue is the question of who is doing the
> describing.  NCAM and RFB&D have been in the situation of
> furnishing descriptions as a third party.  These groups have
> developed a "just the facts" ethic to a high level because
> they are systematically providing third-party transcriptions in a
> lot of cases.

Funny you mention this third-party description since (and I'm not sure
I mentioned it to this forum already) this is also something I thought
about for regular ALT (out of the objective/subjective focus) a while
ago.

The idea would be for some organizations to maintain an "ALT
description server", or AltServ, and for users looking for ALT text to
go ask this server if the description in the source is not present (or
useless).

Sighted users willing to help could fill in the database (one database
entry: URL of image, ALT text) when they browse the web and hit IMG
with no ALT (in the source or in the database). Different ways to fill
the database could be via an email procedure, or some Web form.

AltServ-aware browsers would automatically fetch the ALT text from the
server (or its closest mirror) when no ALT attrib is available: it's
transparent for the end-user. If there is no ALT in the database, one
could imagine an additional service where the images with the most
no-ALT "hit" could get prioritized and "ALTed" first by the sighted
friend circle.

Anyway. I didn't exactly invent this architecture: this is pretty
close to a third party PICS rating system where the PICS label is the
ALT description.

The biggest non-technical drawback I see is that it doesn't encourage
self-ALT-description, which is probably the best in the long term.

Received on Monday, 21 July 1997 11:28:45 UTC