On Validity and Accessibility

Hello,

There seems to be a fierce discussion on Guideline 4.1, in particular
on "validity" (whatever it means).  I communicated my opinion through
internal channel 4 months ago, but it seems it was not reflected to
public discussion, so I will resend my opinion for the (public) record.
This is my personal opinion, not representing anyone else.

Short summary:

  - I believe WCAG 2.0 should entirely drop "validity" (whatever it means) 
    as a success criterion at any level.

Rationale:

First of all, I do feel that this discussion has gone a bit out of context.
"Validity" (whatever it means) itself is not a goal, it is a possible
means for making Web content more accessible.  It needs to be evaluated
in proper context.

The basic Principle "validity" (whatever it means) is being discussed
is as follows [1]:

    Principle 4: Content must be robust enough to work with current and
    future technologies.

The Guideline in question is as follows [2]:

    Guideline 4.1 Use technologies according to specification.

I think these are reasonable principle and guideline (although, ensuring
robustness for "future" technologies sounds a bit hard to prove), and
"success criteria" should be written to "express what it means to follow
the guideline" [3].  WCAG 2.0 further says [3]:

    The principles, guidelines, and success criteria represent concepts
    that apply to all Web-based content. They explain what it means for
    web content to be accessible, regardless of the technology used.
    They are not specific to HTML, XML, or any other technology. This
    approach makes it possible to apply WCAG 2.0 to a variety of
    situations and technologies, including those that do not yet exist.

And this is why I believe "validity" (whatever it means) should be entirely
droped as a success criterion at any level.  I don't think the concept of
"validity" (whatever it means) is universally applicable to all Web-based
content, and requiring it does not always promote using technologies
according to specification.

Let's take xml:id as an example.  The whole point of xml:id is "to identify
IDs in the absence of validation" [4].  It can be used in valid XML
documents as well, but document conformance to xml:id doesn't require
"validity" [5].  Requiring "validity" in order to use xml:id in
an accessible way effectively undermines the usefulness of the xml:id
specification.  "Validity" should be required for WCAG 2.0 conformance
if and only if "validity" is required for conformance to a particular
specification in question, i.e. to "use technologies according to
specification".

For similar reason I would be against introducing "well-formedness"
into WCAG 2.0.  I fully agree that well-formedness is essential to
XML (after all, if it's not well-formed it ain't XML), but still,
it is an XML-specific concept, and trying to apply "well-formedness"
to non-XML technologies would only lead to confusion.

"Validity" may well play a role in "HTML Techniques for WCAG 2.0" [6],
for example, and should there be an "XML Techniques" document, "well-
formedness" would absolutely be a requirement, but technology-specific
requirements should be better addressed in technology-specific techniques
document, not in WCAG 2.0 itself.

Instead of arguing over technology-dependent concepts like "validity"
and "well-formedness", I think WCAG 2.0 should promote "conformance" to
specification.  In many cases, even if "validity" is a requirement for
a particular technology, it is most likely not the only requirement for
conformance.

A good example of "valid but not conformant" case is the 30 June 2005
Working Draft of "General Techniques for WCAG 2.0" itself.  All documents
lacked the required namespace declaration for XHTML on the root element,
thereby violated the third requirement for XHTML 1.0 conformance, even
though most of them are DTD-valid (see [7] for details).  Omitting
namespace declaration can sometimes cause more serious trouble than
mere validity errors.  I don't think WCAG 2.0 should arbitrarily "subset"
conformance requirements and "exaggerate" one of them (such as
"validity") as if other requirements were less important.

I admit that it is not always easy to fully evaluate conformance, but
working around it won't facilitate accessibility.

[1] http://www.w3.org/TR/2005/WD-WCAG20-20050630/#robust
[2] http://www.w3.org/TR/2005/WD-WCAG20-20050630/#use-spec
[3] http://www.w3.org/TR/2005/WD-WCAG20-20050630/#overview-design-principles
[4] http://www.w3.org/TR/2005/REC-xml-id-20050909/#abstract
[5] http://www.w3.org/TR/2005/REC-xml-id-20050909/#xmlid-conformance
[6] http://www.w3.org/TR/2005/WD-WCAG20-HTML-TECHS-20050630/
[7] http://lists.w3.org/Archives/Public/public-comments-wcag20/2005Sep/0002

Regards,
-- 
Masayasu Ishikawa / mimasa@w3.org
W3C - World Wide Web Consortium

Received on Thursday, 10 November 2005 11:24:40 UTC