RE: conformance

I have discussed Cynthia's message a bit below (basically I agree with what
she says, and added some comments) but I would like to make a proposal that
has four parts:

1. There are two ways of recording conformance.

2. One way is simple with at least one, and at most 5 possible levels of
claim, with associated icons and human-readable text.

3. The other way is designed to be machine readable, and to provide as much
granularity as possible.

4. Expressing a claim of conformance to the specification is only considered
valid for the levels provided by the simple version.

Rationale:

For the average person looking for a target to meet in a policy (especially
a long-term target), or for a level they want to find content at, it is
helpful to have a small number of options to choose from. It is also easier
for a W3C Recommendation to pass its review requirements if it is clear about
what needs to be done to implement the specification - if nothing at all is
required, then it need only be a Note, and if there are too many possiblities
then interoperability will fail.

It is useful for all kinds of scenarios to have a more granular,
machine-readable reporting format. In addition, one already exists (EARL) and
other W3C work is looking at how to improve on its usefulness and its
incorporation into tools (Evaluation tools, Authoring Tools, and things that
have nothing to do with accessibility).

In practise, a granular reporting format will enable the re-use of reporting
done for conformance to other sets of requirements such as Section 508
purchasing requirements in the US. For international development this is
important - the US Federal government has differnt requirements from those in
most of the rest of the world, and in some organisations there are in fact
508 requirements and WCAG conformance requirements operating at the same
time. Doing one set of testing for things that the requirements have in
common is helpful to developers and purchasers alike.

Charles

On Thu, 11 Oct 2001, Cynthia Shelly wrote:

  On the issue of levels of conformance:

  Speaking as someone who implemented WCAG 1.0 on a major site, I would
  have to say that my issues with the conformance scheme weren't that it
  had three levels, but that it had ONLY 3 levels.  There was no way to
  say "I did this one, and that one, and I kind of did this, but I
  couldn't do that and here's why".
[snip]
  If we were to publish a metadata scheme for conformance claims today, it
  would be immediately useful.

CMN: http://www.w3.org/2001/03/earl (I could delete the page and republish it
today.. would that help? We are about due to update the page anyway <grin/>).
Actually this is a draft proposal - there is some implementation to the 0.95
version, like there was some for the stuff that was being done a couple of
years ago <http://www.w3.org/1999/11/conforms/>. But it should give you the
idea.

CS
  1) A search engine could start using it tomorrow. The search engine
  could do this entirely in server-side code, and users would not need to
  download or install anything. Search engines use metadata for a variety
  of tasks already. This is not vaporware.

  2) An entry-level programmer could write, in about a week, an
  accessibility aid that filtered on conformance to checkpoints.  Users
  would have to install that, but they have to install screen readers too.

  3) Browser manufacturers could incorporate filtering technology into
  their next releases.  They can't do that if there isn't a spec for it.

CMN This is also being used by developers of accessibility testing tools.

CS
  While it's true that the PICS scheme for accessibility was not widely
  adopted, I think that was primarily an issue of it not being evangelized
  (Charles, correct me if I'm wrong here).  I implemented 1.0, and read
  many, many documents, but I was not aware of its existence until
  Charles' recent post.

CMN Correct. At least I agree with you here <grin/>.
CS
  On the issue of understandability by users:
  Perhaps it would be useful to give an example of another metadata scheme
  that is in widely adopted - content ratings.

  There is a PICS scheme defined for rating the naughtiness of a web site.
  This scheme is used by a variety of filtering programs, including one
  built into Internet Explorer.  If you go into the tools menu in IE, pick
  Internet Options, and click the Content tab, you will see something
  called "Content Advisor".  This allows you to tell your browser you
  don't want it to display anything dirty (using a 1-5 scale for language,
  nudity, etc).  How does your browser know something is dirty?  It looks
  at metadata.  You, as a user of this software, don't have to understand
  the metadata.  In fact, unless you've read the spec for the RSACi PICS
  tag, you won't understand it.  As tags go, it's pretty cryptic.  Most
  Web authors (even experienced ones) have to look it up.  However, that
  in no way prevents it from being useful to the end user. Some sites also
  display a logo, and post text describing how squeaky clean they are, but
  that is NOT how the browser knows.

Received on Saturday, 13 October 2001 15:40:49 UTC