W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > July to September 2002

Re: DC2002 and Accessibility Metadata

From: Jim Ley <jim@jibbering.com>
Date: Mon, 16 Sep 2002 13:03:44 -0000
To: w3c-wai-ig@w3.org
Message-ID: <am4ktj$ael$1@main.gmane.org>


"SHARPE, Ian" <Ian.SHARPE@cambridge.sema.slb.com> wrote in message
news:FA94B04D5981D211B86800A0C9EA2841011422E3@cames1.sema.co.uk...

> What information should be contained in your profile? This should
contain
> any access technology installed, browser configuration (eg colours/font
> sizes, user stylesheet in use), OS display/screen configuration, and
user
> preferences. This could be created by any or all of the browser, access
> technology, manually by the user or even the OS. A simple text file
(XML)
> could easily be maintained for this purpose. Since there is no personal
> information I can't see any problems with storage or abuse of civil
> liberties. This actually tells you nothing about the user themself,
simply
> how the system is configured.

I can't agree that it tells you nothing about the user, it tells you a
huge amount about the user and almost all of it is things that I do not
wish websites to know about me.  Even if I'm completely anonymous when I
first visit the site, as soon as I do any interaction (register/purchase
etc.) then I'm immediately "known", and the AT technology installed would
appear to very well indicate information I may not want people to
automatically know.

Also your examples above are all about the configuration and the browser
environment not the preferences of a user, how do I configure my browser
to say I have red/green colour blindness, how do I say I find flickering
at 3hz annoying but 10hz fine, how do I say I prefer text content unless
it's the image is photographic and what is the server supposed to do with
all this fine grained information.  If we give the users access to all
the content we've got, the user can decide  (by configuration of their
UA) how they access that content - How can some server who's never met
the user, knows only a tiny amount about them, know what's best for the
user?

How is a server supposed to know what to give a UA/AT combination it
doesn't know about, do UA's have to be registered, or were you thinking
of a complete UA description language - if not we'll run into the exact
issue we've already got with the server making decisions on what the
browser claims to be - currently in the HTTP_USER_AGENT string - the UA's
will be forced to lie otherwise they won't get content appropriate to
their UA.  This degrading of the information will make the information
worthless as HTTP_UA already is.

Also, far from using the information to provide accessible content it can
be used as an argument not to provide accessible content online.
Providers can choose to build for a single environment, safe in the
knowledge that they can reliably identify people and tell them "Phone ...
for the service" or however they wan't to (claim they) provide the
service to users.  Accessibility also becomes a much harder "sell",
because the management can genuinely claim that all their users can see
images (why bother with ALT and LONGDESC, look we know they don't need
it)

> The use of meta tags for this purpose could be included in the
guidelines
> and even used to possibly promete the above desing model as best
practice?
> As government policy throughout the world adopt similar approaches to
the
> US, rather than "gettosing" accessible sites I this "categorisation"
could
> actually do the opposite. It could almost become a kind of kyte mark
and a
> sign of quality. All users prefer to use clear and clean sites and
would use
> them in preferrence to cluttered and unclear content.

This appears to be describing something very different to previously,
this is about their being machine-readable claims about the accessibility
of a resource, and in what way it fails, as we know most users don't care
if a resource is WCAG AAA or not, they care if _they_ can access the
content, and a fine grained vocabulary for describing how accessible a
document is, is very useful.

Jim
Received on Monday, 16 September 2002 09:05:13 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 July 2011 18:14:06 GMT