W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 2005

Re: working definition of baseline

From: Jason White <jasonw@ariel.its.unimelb.edu.au>
Date: Sat, 7 May 2005 13:45:51 +1000
Message-ID: <17020.14831.915145.95870@jdc.local>
To: lguarino@adobe.com
Cc: w3c-wai-gl@w3.org

lguarino@adobe.com writes:
 > 
 > What are we trying to communicate here:
 > 
 > 1. WCAG needs authors to use technologies for which accessible user 
 > agents are available to the users. We refer to such a set of 
 > technologies as a baseline (and we need a better word than baseline 
 > for such a set of technologies).

This is where we differ over the appropriate meaning of "baseline". I
don't want to include anything normative in the concept. I would
rather say that we want authors to choose a good, appropriate baseline
that takes into account the user agents available to a given
population at a particular time, and supports accessibility as far as
possible. It is equally possible to choose a bad baseline - one that
makes inappropriate assumptions. The guidance we offer helps authors
to choose good, suitable, relevant baselines. However the concept of
baseline itself implies nothing about accessibility or suitability -
it's just a set of technologies which someone has, for good or ill,
designed content to rely upon.

 > 
 > 2. Because the properties of user agents change over time, and because 
 > the set of user agents available to users differs for different 
 > populations of users, WCAG cannot define the baseline in the normative 
 > part of the guidelines.

Correct - WCAG can't require that authors use any particular baseline.
 > 
 > 3. A baseline can be defined for a given population of users at a 
 > given point in time, that is, it is possible to analyze the 
 > accessibility properties of user agents and to assess what user agents 
 > should be available to a set of users and come up with the list of 
 > technologies supported by accessible user agents. 

I would characterize this by saying that it is possible to carry out
such an analysis and thereby come up with a good, well researched and
appropriate baseline. At the other end of the scale I could just write
content that relies on my favourite list of technologies, which happen
to be unavailable to much of the user population that will be
accessing my Web site. If I still meet WCAG 2.0 success criteria at a
given level, my content conforms, but I've chosen a baseline
inappropriately - so the criticism of my content that ought to be made
is not that it's non-conformat, but that I've chosen my baseline unreasonably.
  > 
 > 4. A WCAG conformance claim is always relative to an identified 
 > baseline. Implicit in that claim is the assumption that the identified 
 > baseline correctly reflects the audience of the web content at the 
 > time of the claim, that is, that the analysis of user agents in step 3 
 > was correct at the time of the claim.

We could add an extra element to conformance requiring that the
baseline be chosen on reasonable grounds, that is, with due
consideration of what is available to the intended audience - but this
gets tricky as I can just define my audience as people who have access
to such-and-such technologies. I would rather not build any element of
reasonableness into the conformance requirement and provide guidance
on choosing good baselines. A fortiori, I don't think any requirement
that a baseline be good/appropriate/well researched should be built
into the concept of baseline; a baseline is just a description, it's
not a rule - it's not normative.

 > 
 > 5. An author can use technologies outside the identified baseline as 
 > long as the use of those technologies degrades gracefully to the 
 > baseline with no loss of information or functionality.

Again this treats the baseline as somehow normative. An author can use
technologies outside the baseline in a way that degrades gracefully,
without thereby changing the baseline. The baseline is still accurate
because it's the minimum set of technologies required for the content
to be presented/operated at all by a user agent.

Does this help?
Received on Saturday, 7 May 2005 03:46:40 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 5 February 2014 23:39:37 UTC