24 Oct 2003 - Meeting between AUWG and WCAG WG Teleconference Minutes

IRC log


Wendy Chisholm, Doyle Burnett, Roberto Scano, Rob Ellero, Mauritzio Vittoria, Greg Pisocky, Loretta Guarino Reid, Jan Richards, Sailesh Panchang, Jason White, David MacDonald, Bengt Farre, John Slatin, Tim Boland, Matt May, Jutta Treviranus, Gregg Vanderheiden, Kerstin Goldsmith, Karen Mardahl, Francesco Fedele, Phill Jenkins


Yvette P. Hoitink, Ben Caldwell, Roberto Castaldo, Patrizia Bertini, Tom Croucher, Lisa Seeman, Andi Snow-Weaver, Charles McCathieNevile

Action Items


Agenda as sent to mailing list

  1. How ATAG and WCAG reference each other. Where does WCAG reference ATAG? Where does ATAG reference WCAG? What is needed in the organization of the documents to make the references as straight-forward as possible? What can we do in WCAG to decrease the complexity of ATAG's relationship with WCAG (relative conformance)? More thoughts.
  2. Scope of the documents. Basically, ATAG covers the process of generating the content and t thus includes Content Management Systems (CMS), authoring tools, etc. WCAG covers the end result of the authoring process (thus does not cover CMS). However, what about web applications that generate content? ATAG and WCAG could produce a shared techniques document "techniques for application that generate content" that would refer to both ATAG and WCAG checkpoints.
  3. The suitability of requirements to use W3C recommendations. ATAG 1.0 only discusses producing content for W3C formats. WCAG 1.0 says only use W3C formats. How should the 2.0 versions of WCAG and ATAG reference W3C technologies? In WCAG 2.0 we had checkpoint 4.3 (this has evolved since this was published), although all of Guideline 4 relates to technology.
  4. Development of test suites - How do we coordinate development and reuse? How do we design a test suite that is beneficial for both ATAG and WCAG? What will be the differences between our test suite needs? We also need to coordinate with UAWG and QAWG.
  5. If we have time -- Metadata in WCAG and any other topics that are raised.

NOTE: AUWG should be aware that the WCAG WG recently adopted a new organization of the document and we will be talking from this perspective. We are moving towards 3 levels of conformance (which should help address issues raised during the AUWG F2F in September in Seattle). This is part of the 1st agendum.

ATAG-WCAG Relationship

AUWG Question: Many ATAG techniques have to reference WCAG. ATAG has had to make changes as the WCAG (2.0) structure has changed. ATAG has relative structure. if p1 for WCAG 1.0, then p1 for ATAG. How should we link into WCAG if WCAG has only 2 priorities/conformance levels? We would like to encourage readers of WCAG to use ATAG-conformant authoring tools.

WCAG response: In the very latest WCAG 2.0 draft, there is no longer core/extended there are 4 principles. each principle has guidelines. each guideline has success criteria that are labeled level 1, 2 or 3. this structure is more like wcag 1.0. we aren't using the same terms as wcag 1.0 since p1 meant "important" and p2 meant "good to do" however we realized that some p2 items are important for access to some groups. thus in this latest reformulation we define the levels differently.

AUWG Question: auwg is working to structure atag in a way that is separate from wcag so that wcag can change freely. and atag won't have to change w/it. what is the rationale to assign priorities on a different way?

WCAG response: there are no longer checkpoints - only guidelines and success criteria. all criteria need to be testable.

AUWG Question: what if you have something that is important but not testable?

WCAG response: it would be a technique. we have a techniques gateway and a place for "other advice." - we're trying to figure out testable versions of things.

AUWG Question: What are the requirements about what makes something testable or not?

WCAG response: machine-testable is a no brainer. hirr (high inter-rater reliability) - if it is the judgement of the wcag wg that it would have a hirr then it does.

AUWG Question: test results can change based on changes in testing instructions.

WCAG response: HIRR assumes well-informed readers. work with EOWG to train them to become well-informed reader. andi, cynthia, kerstin, and wendy are going through recent draft looking for statements that are not testable. primarily, looking for subjective terms like "appropriate" and "important."

AUWG Question: testability is good news for authoring tool developers. examples and test suites are also interesting. auwg needs real, coded examples.

WCAG response: wcag wg is creating technology-specific checklists. real examples are part of test suite - that's the goal of chris and michael's test files. wcag 2.0 implementation testers will also need test files and test suite.

AUWG response: need to use the same terminology in test suite development in both groups. auwg is also using term "success criteria" auwg considering the option of having a module for a checkpoint that could be added to a usability test. how do you test if an authoring tool appropriately influences a user to "do the right thing." it sounds like both groups (auwg and wcagwg) are grappling with similar tihngs. the ranking of wcag criteria will likely influence atag 2. 0 devleopment, but that we can likely use exisitng strategies similar ot atag 1.0/wcag 1.0.

WCAG clarification: testability is not a factor for determining which level a wcag 2.0 criteria is assigned to. all wcag 2.0 items are testable. level 2 are things that constrain how the author expresses him/herself diff between levels 2 and 3 is reasonableness for all sites. if you have levels in checkpoints and then in success criteria, you will end up with a 2-dimensional structure. wcag 2.0 used to have that, but got rid of it because it was too confusing. wcag 2.0 has some guidelines that only have level 2 or level 3 criteria.

AUWG request: determine in criteria for level 1, 2, 3 if clearly an author responsibility or tool responsibility. any output that could help us in atag.

WCAG response: level 1 are things that the authoring tool should do. level 2 are those things that should help with. however, there are things the tool has complete control over and there are things the tool gives choices to the author.

AUWG request: It is important that wcag not specify how the content be created but what needs to be created. - whatever path that someone takes to add alt attribute, it doesn't matter. all that matters is that they have done it. i.e., wcag should not insist on how done. wcag shuld focus on outcome not process.

AUWG request: encourage wcag wg to read atag. atag covers evaluation and repair. don't tell developers how tool should conform, but have checkpoints that ask the tool to do the things that are needed. if you are using absolute rather than relative, the tool should say "use relative." atag says: the accessible choice should be the easiest to do. anything that the tool does itself, the tool should do the accessible thing. if author given a choice give the autohr the accessible choice first. make it easy. if author still chooses to do the inaccessible thing, nothing we can do. it is the author's choice.

AUWG request: auwg would like to request that wcag raise awareness of atag. an authoring tool that is atag compliant will have a great effect on wcag proliferation. if wcag promotes atag, it will promote more accessible authoring tools which will promote a more accessible web. thus, good to mention w/in wcag atag conformance.

Joint question: what about CMS (authoring tools embedded as object inside the page) ?

Joint proposal: joint techniques document between AUWG and WCAG WG that is "web applications that generate content" i..e, the authoring tool is web content itself.

suitability of using w3c recommendations

WCAG summary: wcag will produce techniques docs for w3c technologies. ideally, wcag would reference xag (to describe what makes an accessible format), but XAG is not a rec. This is different from WCAG 1.0 where it said, "use W3C technologies." thus, use technologies that have accessibility features and use the accessibility features. We plan to summarize basic points of xag and point to as informaitve reference. however, only addresses xml apps. not general technologies.

AUWG question: A tool could claim ATAG conformance if for the format(s) it produces, it also publishes in public a WCAG Techniques document for that format. in earlier part of discussion, time spent talking about testing and verification. thus, testing part of adherence to guidelines. if adobe and macromedia publish something, wouldn't they also have to provide ability to verify through testing that an accessible result was achieved?

WCAG response: wcag doesn't plan to have a test suite/techniques for every technology. wcag wg will only create technology-specific documentation for w3c technologies that are most widely used. it is up to other entities to create techniques for their technologies. however, idea of a "xag" test suite seems to be hinted at. i.e., how do you determine that a format is accessible?

AUWG question: WCAG needs to publish the criteria for techniques for a company to follow if they plan to publish a checklist. they have to say that it is testable. thus, wcag is following an agreed to process to say these criteria have to be testable to be success criteria. dtd is necessary but not sufficient. are techniques verifiable. like a requirments document for tech-specific checklists? add to the wcag issues list to document how to create checklists and techniques. will help atag. also help in standards harmonization.

action: wendy follow-up with auwg about documentation for "if you want to create a techniques document..."

Test suites and next steps

Not enough time in this call to talk about test suites. Have another joint call between wcag techinques task force, auwg, uawg and qawg? We ought to create a x-group task force. WAI CG has power (via W3C Process document) to create a task force of participants from multiple groups.

action: JW, GV, JT take idea of x-group task force to CG

$Date: 2003/11/07 18:29:01 $ Jan Richards and Wendy Chisholm