- From: Tom Croucher <tom.croucher@sunderland.ac.uk>
- Date: Wed, 17 Mar 2004 14:57:02 +0000
- To: w3c-wai-gl@w3.org
Hi Everyone, Sorry for the length of this text, but this has been a bit of a hard concept to express. Tom --- This is a proposal developed by Tom Croucher and Michael Cooper to provide a normative mechanism for ensuring compliance with the WCAG without a requirement for us to provide normative techniques documents. We suggest instead a normative process for verifying techniques that have been applied. The problem we face is that the ability to claim conformance to the WCAG will be important to many consumers. Since WCAG 2 is technology-agnostic, there is no direct way for many of the guidelines that one can achieve conformance except via implementation in a specific technology. This has raised the issue that particular technology implementation techniques for WCAG need to be considered sufficient for conformance. One mechanism that has been proposed is that Checklists of technology-specific techniques be normative. In discussion with the Techniques Task Force, we consider normative checklists highly undesirable. The checklists are closely tied to the techniques themselves. The techniques should not be normative for a few reasons: * we would require authors to apply only the techniques we happen to have considered and could preclude other viable ways of conforming to WCAG; * we would have to go through a more stringent process to update techniques, dramatically increasing the likelihood that the techniques will fall behind the level of current technology, which is one of the things the technology-agnostic structure of WCAG was designed to avoid; * we would preclude WCAG conformance for any technology for which we have not provided techniques, whether as a result of lack of resources or because the technology is not an open standard. However, we are aware that the techniques as currently conceived are intended to fulfill some normative requirements of WCAG. For instance, the determination of the testability of a guideline is likely to be made through our ability to create techniques for the guideline, and test files for the techniques against which demonstrations of our testability requirements can be made (including an acceptable degree of inter-rater reliability). We intend that the techniques we create will fulfill these requirements (in spite of not being normative). The challenge is to provide a way to ensure that people using techniques not created by us will have the same assurance of WCAG conformance. The proposal is to make a normative conformance requirement that web sites use only techniques that have demonstrable proof that they have followed a sufficiently rigorous QA process to ensure that they do, in fact, meet the WCAG requirement. This is, in practice, asking web sites to show that they can prove all the techniques used for the accessibility of their content have been tested with due diligence. This, of course, does not preclude checking that all content can be proved to be accessible under the success criteria of the main set of guidelines, it merely adds a more manageable approach to checking conformance. This proposal would require the guidelines to provide a skeleton model of a QA process for technique development. This would likely be drawn from the methods already in use for validating the techniques and techniques documents. This skeleton model would then be used as a basis for QA methodologies used to validate techniques used in conformance claims. Sites claiming WCAG conformance would be expected to be able to prove on demand that their techniques have been tested and determined to meet our inter-rater reliability requirements. This process would not impose any specific additional requirements on the nature of conforming content and we do not propose to require there be a statement on the site indicating they have followed this process. But if the site claims WCAG conformance and that is challenged, documentation of the testing process would need to be provided. Of course, sites that choose simply to use our own techniques could simply refer to documentation we would provide as part of our techniques development process. This proposal is certainly not complete, and there are known challenges (#1: is it really ok to make a normative requirement for a process, not a state?). But it does seem that it provides a way out of the extremely undesirable alternative of making techniques and checklists normative.
Received on Wednesday, 17 March 2004 09:57:34 UTC