Re: Conformance claims - by success criteria?

Wendy A Chisholm writes:
 > In today's call we talked a lot about using metadata to make conformance 
 > claims.  Our assumption was that we would be expressing conformance to 
 > checkpoints.  What if instead conformance was to the success criteria?  We 
 > already have some criteria that are conditional.
An interesting idea, but too complicated? Also, if one hasn't met the
success criteria, one hasn't satisfied the checkpoint/resolved the
accessibility problem, so there seems to be no value in claiming
partial implementation of a checkpoint. If the success criteria have
been met then the checkpoint is satisfied and the access problem which
it addresses has been overcome; otherwise not.
> 
 > I still also wonder about what conformance will look like at the 
 > technology-specific level.

I will try to justify this briefly. My initial examination of the feasibility of
writing, for instance, HTML technology-specifics for WCAG 2.0
indicates that the  idea of "technology-specific checkpoints" is
seriously flawed and needs to be revisited.

Most of the success criteria in our guidelines do not give rise to
potentially normative, technology-specific requirements that could be
formulated as checkpoints. Technology-specific examples can of course
be given. These are valuable and illuminating, indeed, I would argue,
essential to implementors; but there are always several different
approaches which can be taken to meeting success criteria.
Furthermore, many of the success criteria (especially but not
exclusively under guidelines 2 and 3) do not give rise to readily
identifiable technology-specific requirements. The most that can be
usefully offered are examples and design suggestions.

The main areas where technology-specifics come into play are under
guidelines 1 and 4. Even here, however, they have a limited role.
Perhaps we should start thinking in terms of "technology-specific
success criteria" rather than "technology-specific checkpoints". Thus
for each technology we could supplement some of the success criteria
for particular checkpoints with technology-specific requirements which
would be normative. This might overcome my concerns, and the list
would probably be rather short. We could also give (much larger)
Techniques documents with explanations and code examples (these are
already under development).

To illustrate the problem, consider checkpoint 1.1 of WCAG 2.0.

<blockquote>
You will have successfully provided a text equivalent for all non-text content
if:

   1. all non-text content is explicitly associated with a text equivalent
     (images have alt-text, movies have collated text transcripts, animations
     have descriptions, interactive scripts have a functional equivalent such
     as a form, audio files have a text transcript),
   2. the text equivalent fulfills the same function and conveys the same
     information as the non-text content. > Note: Depending on the purpose and
     content of the non-text content, a short label may be appropriate, or a
     more thorough explanation may be required,
   3. where it is not possible to describe the non-text content in words or
     for text to provide the same function as the non-text content, a label
     identifying the content is provided.
</blockquote>

Only the first success criterion can be supplemented by
technology-specific requirements that can be considered normative. In
HTML for instance one would discuss the ALT attribute, the OBJECT
element and so forth; in SVG the DESC element, etc. However, the
remaining success criteria do not have consequences which are
technology-specific. Indeed, whatever guidance can be provided with
respect to them, applies irrespective of the technology which is being
used. Of course, one can and (I would contend) should provide
technology-specific examples to illustrate the correct application of
checkpoint 1.1, but these necessarily belong in (non-normative)
techniques, because they are only examples and clarifying
explanations, not definitive requirements.

Technology-specific success criteria for checkpoint 1.1 would
supplement the (generic) success criterion 1, and would leave the
remaining success criteria as is.

Checkpoints 1.3 and 1.5 clearly need technology-specific
supplementation, as do some of the checkpoints under guideline 4. We
could write technology-specific success criteria and provide several
different views of the document for different technologies.

Whether the idea of "technology-specific success criteria" works or
not, I think the concept of supplying "technology-specific
checkpoints" to supplement each checkpoint of the existing guidelines
is clearly mistaken, as the foregoing discussion has endeavoured to
show.

Received on Thursday, 11 October 2001 19:44:05 UTC