Re: Conformance and Implementations

A couple of comments and a clarification of my earlier comments:

At 10:51 AM 10/10/01 -0600, Alex Rousskov wrote:
>On Wed, 10 Oct 2001, Andrew Thackrah wrote:
>
> > The W3C currently promotes the use of an online HTML validator. If
> > a page is sucessfully validated the author is invited to display a
> > small 'conformance' graphic on the page.

Note that there are also several other icons in use or proposed now:  CSS, 
WCAG (AAA or AA or A, as I recall), and UAAG (proposed).  The latter has 
generated some issues and discussion in other venues, for example from the 
SVG WG.

> > This is a form of
> > branding. What is the thinking behind this service? What is the
> > W3C hoping to achieve by this?

This is certainly a topic which is ripe for a close look.


>I think we should distinguish the service from the 'conformance'
>graphic/icon.
>
>The validating service is a simple test suite. There were several
>remarks on this thread supporting test suite development/promotion as
>an important QA goal.
>
>The graphic itself probably has a minimal positive effect, but we can
>ignore it. AFAIK, anybody can:
>         - put an icon on an invalid page
>         - put no icon on a valid page
>         - put a "W3C sucks!" icon on any page
>with no significant consequences for W3C or the page. This is because
>W3C does not protect the "W3C" trademark. I have tried to argue that
>this "minimal positive effect" is the best we can do right now.
>
> > One problem it may have is the kind of negative 'halo effect' that
> > would come from a broken browser rendering a conformant page
> > badly. The page, although technically correct, would look broken.
> > If every page bearing a W3C brand looked broken (in that browser),
> > there is the danger of creating a bad association with W3C.
>
>Yes, there is a minor potential problem. My argument is that we should
>leave it "as is" before we make it worse and real. The world is
>imperfect. Not every problem can be solved, given the current
>mentality of an average human. However, virtually every problem can
>be made worse.
>
>All "solutions" we have been hearing on this thread involve raising
>legal status of W3C trademarks.

I hope that this is not in reference to an early comment of mine.  In that 
comment, I tried to identify the extremes which bound the spectrum of 
actions (or non-actions) around this issue.  At the one extreme, I pointed 
to the last section of a 1996 conformance document by TimBL, which 
suggested that there might be a basis to legally bind some veracity 
requirements to conformance claims, compliance statements, use of W3C 
trademark, etc.  I did not recommend this as a solution.

As I recall, this discussion started from a message by Karl.  I just looked 
at it again, and he suggests the utility of better-stated conformance 
claims (based on well-formulated statements in the specs), better 
product-feature information (possibly facilitated by forms or other tools 
released as part of or adjunct to the REC), etc.  I don't see any legal 
enforcement suggested.

For the record, I think our focus should be on improvements to specs (RECs) 
and associated tools and materials (tutorials, test suites, test suite 
reporting matrices, feature matrices, etc, most of which we are currently 
doing), with the goals of:  making good assessment tools (e.g., test 
suites) widely available and easy to use; and, making it easy to both 
generate and get access to conformance and implemented-feature 
information.  Legal initiatives look like a tar pit that we might do well 
to avoid.

Certification has a place, but W3C is probably not the place to carry it 
out.  The QA activity statements and charters state this, and everyone in 
this thread seems to agree (so far).  It is the business of industry 
application sectors that use the W3C technologies, and they should be 
responsible for the setup and execution of the services.  (However, it 
*may* be within our scope to design generic methodologies for running a 
certification service, which along with test suites could be picked up by 
third parties who want to initiate and run certification services).  There 
is a body of past experience which provides models for this approach.

>If there is any legal document
>attached to that icon or, worse, the test suite, I, as a small
>developer, am much more likely to stay away from them because I do not
>want to pay a lawyer for translating that document.

Still, there *are* some other interesting legal questions.  One involves 
"what sort of copyright" on the test suites.  W3C has a Software copyright 
(allows modification and derivative works), and a Document copyright (no 
modifications allowed).  The SVG test suite is bound by the Document 
copyright, for some very good reasons.  It is undesirable that people to 
take identifiable W3C test materials -- which hopefully are developed with 
some care and integrity -- and unilaterally modify them (e.g., so that they 
can pass a test that they otherwise fail).

(But this is another topic...IPR and copyright of W3C materials, including 
test suites.)


>On the other hand, if somebody wants to start a company that will sell
>legally binding "HTML/13.0 compliant" icons, that's perfectly fine.

Yes, like the model of Underwriters Labs.  Or slightly differently, the 
assessment and reporting model of the commercial Consumers Union ("Consumer 
Reports").

-Lofton.

Received on Wednesday, 10 October 2001 21:02:21 UTC