Re: Validity

On Nov 5, 2005, at 7:31 AM, Bailey, Bruce wrote:
> > Can you please provide a real example of an assistive technology
> > doesn't work as a result of invalid code, where all WAI  
> guidelines pass?
>
> I do not believe that the above is controversial as examples are  
> numerous.  Am I wrong about that Matt?

You're not wrong, but this is not sufficient evidence to support your  
position. Someone else within the last two weeks has outlined which  
components of validity directly affect interaction with assistive  
technology. If this is an issue of access to AT, then the least  
restrictive means of meeting our goals is to require that those  
specific issues be resolved. Further, a case was made at the f2f that  
those specific issues _are_ covered by other guidelines.

This is a big part of my argument: we cannot require content  
producers to change things we know are not required for accessibility  
simply because there is a convenient concept that also contains what  
we want. It is a tautology that a specification is _specific_, and a  
specification is what we are building.

> The most intractable problem is with nested tables.  Things get  
> better with each release, but I must have heard Jaws announce “not  
> in a table” (when cursor clearly is in a cell) literally thousands  
> of times.  I am a little burned out by this.  Repairing the html  
> fragment resolves the issue about three-fourths of the time.

Nested tables are not invalid code.

> I will repeat my often-made but never satisfied plea:  Please name  
> three mainstream site that nominally pay attention to validity --  
> but not accessibility -- that have non-trivial WCAG1 P1  
> violations.  Matt (and others) have offered compelling sounding  
> theories why this is so, but no evidence, and nothing that  
> satisfactorily explains the improbably high level of one-way  
> correlation.
>

Hmm. Someone seems to have studied at the Joe Clark School of Debate.  
But, no matter. The premise of the question is false.

First, it's hard enough to find a site of any real size that is valid  
in the first place -- which is yet another indicator that the Web is  
not ready for a validity constraint.

Second, you're asking me to somehow divine whether the content  
producers of a given site "pay attention to validity -- but not  
accessibility". I'm not a mind-reader, and we are fortunate enough  
that many of those who make their living evangelizing validity also  
care about accessibility. But I don't think it's hard to find  
nominally valid content that has images with alt attributes that say  
"spacer gif", or "9425 bytes", or "filename.gif"; or tables lacking  
headers; or pages that are unusable when you strip the style, or  
disable script or plugins.

Third, the plural of anecdote is not data. What you need to prove  
your assertion that validity is a minimum requirement is not three  
examples where valid sites otherwise fail WCAG 1 P1, it is the  
absence of any possible way for invalid documents to conform to the  
individual requirements of WCAG 1 P1 _and_ function properly with AT.  
I can disprove this in three steps on any valid, WCAG-conforming HTML  
document on the Web:

1) Open the document.
2) Before </body>, insert "&nbsp"
3) Save the document.

The document is now invalid, and no less functionally accessible.

And finally, a violation is a violation. I don't know what you  
consider to be a "non-trivial" P1 violation, but if you can shrug  
them off, then so can anyone else.

-
m