RE: Validation as test for basic accessibility

> For accessibility?    How do you figure?

I was thinking the other way round, and though I am sure you and many here have deep sympathy for the promotion of W3C standards, and better empathy than I for the potential far reaching beneficial effects of such.  Still, I will expound on some reasoning for the direction you took my assertion.

Please keep in my mind that my perspective is skewed by working for the U.S. Federal Government for the last five years.

I submit that the U.S. Federal Government treats accessibility with a fair amount of seriousness.  In support of this assertion, I would challenge the skeptic to find three .gov homepages that are not compliant with the Section 508 1194.22 Accessibility Standards.

I submit that the Federal Government treats validity with significantly less seriousness.  In support of this assertion, I would be significantly indebted to the skeptic who could send me the URLs to three valid .gov homepages.  (I haven't looked for this lately, maybe things are better?)

My personal observation, not at my agency, is that much of the attention to accessibility is quite superficial.  That is to say, things that are easily machine testable (e.g., missing alt attributes) get the bulk of the attention and are in actual practice fairly likely to be remediated.

Validity is readily machine testable.  If validity were a requirement, even as merely one subordinate point of accessibility, it stands to reason then it would be addressed, merely as a routine cost of doing business.

There have been offered several examples of broken code interfering with accessibility.  Valid code is, by definition, not broken.  The logical conclusion is that some accessibility problems -- which would otherwise only be caught by motivated careful human-driven testing -- would be readily identified by unmotivated lazy machine testing.

>> Does anyone care to argue that, in actual practice, the one-way
>> correlation between validity and accessibility is less than 99.9%?

> Yes.  The notion is absurd.

Okay, if that is so absurd, this challenge should be easy:  Please site three non-trivial pages that validate, but are not WCAG1 Single A.

> Accessibility is not a subset of validity, just a related 
> property.

The implication for correlation only works in the other direction.  No one is arguing causality.

> Both accessibility and validity are indicators of higher-order
> proficiency on the part of the developer.

Agreed.  It is time to promote that proficiency.  If WCAG2 fails to include validity at the P1 level, I predict that we will be having this same discussion for WCAG3 in five years.

> Maybe I should start with the Department of Education homepage?

I am quite interested in all reports -- preferably off line <smile /> -- of any accessibility problems on the ED Site, particularly the homepage.  The validity issue, sadly, I have been powerless to influence.

> The DRC report from late last year said that 81% of their sample of 
> UK sites failed WCAG 1.0 level A.  They can't all be invalid.

I am sorry to say I find that quite believable.  Has the DRC or W3C (or someone else) sampled UK sites for validity?  I would guess that the validity rate is a fraction of a percent.

> a validity requirement would be the first thing taken out.

You are correct that just because validity is P1 in WCAG2, it does no follow that it would make it into a revision of the 508 Standards.  If validity is not P1, however, the chances are zero.

If concerns for validity was wide spread, at least as wide spread as concerns for accessibility are now, this would be obvious nonsense:  accessibility would be clearly distinct validity.  (Or perhaps, the interaction would be more clearly understood.)  From my (admittedly skewed) perspective their have been satisfactory gains made towards wider accessibility.  I have not observed similar progress on validity.

The problem is that lack of attention to basic fundamentals, like non-broken code, means that the work to provide accessibility requires that much more time and effort.  Valid code is much easier to make accessible.  If the mainstream tools created valid code by default, accessibility would be improved if only because time could be spent fixing real problems rather that those secondary to broken code.