WCAG F2F Telstra, Melbourne Australia. 14 November 2001


cynthia, mathew, lisa, rob, gian, graham, charles, liddy, andrew, emeline, jason, wendy

on the phone: andi, loretta, gregg, Annuska

Report from discussion on techniques and success criteria

Liddy, Emeline, Andrew, Wendy

/* who took notes for this?? */

Report from discussion on comprehension success criteria

Rob, Lisa, Graham, Charles

CMN We worked on success criteria of checkpoint 3.3. 21 sentences that we have varying degrees of agreement.

CS similar to 3.4?

LS Semantic/pragmatic, not a form of autism, but processing non-literal info doesn't work, instructions that are non-specific and rely on basic memory of what is involved. e.g. ask a child to put the toys away. no visual reference to "away" not sure what to do. verus "put in toy box" they get a visual reference and can understand.

CS content that are directions?

LS Cog dis - we put some things into instructions, we were more rigorous than what requied.

CS Can't do w/poetry.

LS If have a poem, doesn't count. But, instructions, would help.

CMN Rat hole - "but this doesn't apply to artistic expression." clearly. artistic expression - people will break all kinds of accessibility rules. it is not universally accessible.

LS Important to also say that snce not talking about conformance, there are sites that will want to do everything they can to aid comprehension. Put it all up there. Limit ability.

WC Have not ruled out conformance.

LN All kinds of testing going on. If make it too explicit, give the answer away. How do you make it understandable but not break testing. another rat hole.

CMN The fact that they won't be perfectly accessible, therefore don't bother is not right. Let's try to do as much as possible to make as accessible as possible.

..Looked also at grammar ideas. Success criteria will vary from language to language. Tech specific from language to language - i.e. human languages.

.. success criteria: Substitute common words for uncommon words does not change meaning.

LS That is testable.

CMN Machine testable.

.. other criteria. this list is on the web.

LS Non-literal text is a big problem for those who don't understand it. there are techniques you can use. In XHTML 1.1 you can use Ruby. You can have one text, but different displays of it. Sometimes display colloquism othertimes the literal meaning. Different words displayed. People then set up the browser display options. If have a complex idea say succinctly then go into detail. Aids comprehension. Triangular strucuter. Also want on document level. A form of summary, e.g. exec. summary. Get people to the point of the document.

LN On the other side, I think you make sense. This is what good writing about. But it is a purely semantic activity. There are few writers. It is difficult to say to people, "it's likely you aren't a good writer, but you must become one now."

AA Journalists do this.

LS A lot of site owners will not know they need to look for someone with this skill.

CMN We're telling people to be decent writers. The point of the exercise is to develop techniques and instructions for how to do it. And let them figure out if they have done it. the alt-text comparison: be good interpreters of visual info. People write all sorts of rubbish. Some techniques for testing it so that they think about it so it is effective. That's where we're trying to go with this exercise.

GSW We have to help people who have varying requirements. We had a site, "on this page you will find..." but 1/2 way down the page before real content. Pages were cluttered. If it means that the document becomes longer that might not be very helpful for others.

LS How would one idea per paragraph not help?

GSW A question: these assistive technologies for helping people to read, how common are they? e.g. ruby.

LS It's markup.

GSW But, those who can be most heloped by it, do they know about it or do we need to educate people about it?

LS We need to make sure covered by UAAG. needs to be in new browsers.

CMN re Ruby: It's markup, a spec. Transforms gracefully. Reasonable presentation. Implemented in browsers like IE and Amaya. We should publish the URI. the 4 of us should take an action to go through it.

LS Also discussed instructions. Those should be as accessible as possible.

/* gregg joins */

LS instructions should be active.

CS Happen in other languages?

CMN More true in French.

LN The open university in england has worked out these techniques and why you use active tense. I applaud what you are saying, but instead of rewriting, find the sources.

LS I have got some stuff from the autistic association.

CMN THere is established research.

LS The problem w/established research is that it centers around children. Will their target audience deal with this type of disability.

LN They do provide w/their students to be any person you can think of. they've used these rules for 25 years.

Rob also has a pointer.

LS The autistic info sheet that i've been reading says you must not disorient people. Translating that to the web, i was imagining a page...a box in different colors may not have much to do w/the content of the page. it could be irritating for someone that doesn't understand it is meant to be out of context. On the one hand, the content may not have anything to do w/the content around it. the other side, you might have a diff form of presentation where the presentation is in context, but they want to be funky. Only have changes in presentation that reflect changes in model, not just for fun. As LN says, this does need research.

CS there is a lot of research about placement and offsetting. Info about why things are done that way.

CS has more info - a pointer.

Discussion with IMS

Introductions around the WCAG WG meeting

Present at IMS meeting

MN IMS global learning consort formed a few years ago to promote tech learning solutiosn. It is a spec dev org, similar to W3C. Not standards, but consortium to make recommendations to govnt, industry, unis, etc. Focused on learning tech, training, teaching systems for higher ed, etc. We've been working on specs for 5 years, have several in place.

.. little over 1 year ago, people in IMS saw need to forsee to address accessibility. We are a collection of people who work on scope, and ratified by board. Passed in April. It serves as our charter. Defines 3 activities:

  1. investigate current IMS specs: metadata, testing and question, learner info profiles, etc. Determine which changes should be made to make more accessible.
  2. advising other WGs creating new specs: learning design, digital repositories, etc. help guide those specs to introduce accessibility requirements now - in dev cycle.
  3. collecting resources, info, etc. to create a set of guidelines to help devs creating tech based learning solutions. to introduce accessibility as they are being designed and built. 1st version is complete. 80 pp. Draws on work before, including WAI work, 508, NCAM. Not intended to replace other info, but central collection. "What are diff types of disabs? How do people use tech to accomodate? Defn of terms. etc."

MR Tackle topics for distance learning. Section on synchronous vs. asynch comm. ATAG like stuff. Forward looking sections: testing, assessment, etc. What do we know about math and other special topics? Similar terrain, but sliced in a diff direction. What kinds of tools and learning environments are people using.

MN compatible documents.

MR Next round: point to more tech specific solutions. Point to stuff you're working on.


WC These are being minuted.

LN Our section in 2 stories. JW will tell you about this group, then CMN will tell you what the other groups are doing and how it compliments it.

JW You are familiar with WCAG, so my task is easier. WCAG goes back to 1997 when Trace submitted guidelines to W3C as part of WAI. The WG developed guidelines over next 2 years. Became W3C Rec in 1999. Included guidelines doc and checklists and techniques. The WG continued development of guidelines for advanced version. to be backwards compatible. Briefly outline issues of current work:

abstract general principles of accessibility of tech-specific details. current draft of 2.0, draws distinction between general requirements and tech-specifics.

.. success criteria - determine difficulties in determining when particular content actually conformed to checkpont. 2.0 contains more precise and testable success criteria. testable - determine in some objective test, human or machine testable. issues - technology-specific success riteria and conformance is what we're working on now.

.. tasks before it:

CMN the rest of WAI - there are 6 WAI groups, I'll run through them. AUWG produced ATAG. Working on techniques for how to implement the requirements of accessibility in authoring tools. JR is an editor, JT is the chair. I presume you have good info about that group.

MR JT wrote our doc, it points to ATAG. A brief description would be helpful.

CMN ATAG - designed to work for a variety of tools. Primary work now, developing techniques, interface ideas, ways of using the work about making content accessible to figure out how to make it happen when someone using a piece of software and is not an accessibility expert. How do you evaluate and test for accessibility? How get the author to tell you things that requries a human to test?

.. UAWG works on browsers and plug-ins. Working on finishing version 1.0. directed at common browesrs and multimedia plug-ins (explorere, netscape, flash, etc.). more specific than the work of ATAG. Lot sof implementation testing at the moment.

.. EOWG produces documents on how to use. More from AA in a moment.

.. PFWG looks at all the specs that W3C produces and ensures support accessibility. Working on guidelines for markup languages. If you produce a new XML dialect, use this to help you ensure it will be accessible. Groups developing technologies find it very useful.

.. ERTWG used to work on how to fix problems now that could be folded into ATAG. Medium term plan to fold into AUWG and UAWG and QA. Working on EARL.

AA EOWG - working on education and promotional materials to get message out to broader community. Revising document about getting started. Draft about evaluating web sites for accessibility - concrete suggestions about checking for conformance. Ongoing work on biz cases. Sample biz cases and implementation plans. Auxiliary benefits to organization if make work accessible.

LN Any questions of either group. Next discussion would be overlaps and how to work together.

CMN How public is the IMS work?

MN 1st two are internal harmonization efforts. made public after new drafts come out.


.. make available as soon as possible.

LN I have copies here to hand out tomorrow.

CMN What about the working process?

AA public minutes of work?

MN 4 rounds of documents: scope, base document (internal - captures results of WG activities. presented to IMS tech board.), draft doc (public comment), final doc (also approved by tech board and becomes spec). common process.

LN Added strangeness - IMS Australia is member. They are open w/what is going on. Some are commercial entities. We've talked about what each are doing. I want to represent dublic core for a moment. We think of them as 3rd player. We are just getting IG together, expect a WG after that. Expect an element on accessibility at core of dublin core profiels. That will be open since DC is public. Yesterday we talked about a meeting where the 3 groups come together.

MN From high-level, IMS is working with OCLC. They are active in repository, DC part of.

LN There is a difference as well. I'm on DC init. group. OCLC is large org. DC is a separate thing. supported by OCLC. LS asking what DC is. it's an open group developing metadata elements for discovery. Very obvious one: discussion here there has been level 1, 2, 3 compliance is that the way to go in the future or look at specific things and have a lot of things can comply w/or not. 2nd is closer to work IMS has been doing.

JW W/2.0 we are trying to work through questions of how to define a conformance scheme for the spec. What levels possible, criteria, which checkpoints fit in diff categories of conformance scheme. etc. many dimensions and consequences. If an overlap, then interested in that. WCAG WG interested in techniques and strategies for implementing accessible solutiosn.

MC ??

MN Also struggling w/conformance and what it means. Not ready to publicly state how it will go, but getting closer. Degrees but absolute is a debate.

??, Apple raised issue of conformance about year ago, one form of conformance for all ims specs.

CMN Another angle, W3C has recently launched a QA activity. We have a conformance manager to work out what it means to conform. Working w/open group and working on test suites as important part of work we're doing. More important in some ways than strong approach that MC discussed. Good to have reliable answer about what that means - as an org, not just in WAI.

P, Apple. Relying on experience of open group. SIF. Test suite also. Looking for entity who can administer. Try to provide value about how conform to specs. Need to be clear, simple, and direct. levels of compliance will do nothing.

LN Want to test out - any time require something to be done, give example of how to do it and a way to objectively determine if have been done. Diff context place diff significance on. the model of being able to choose your general compliance level best left to orgs trying to work w/the standards.

CMN I mentioned that WAI work includes EARL. It's a vocabulary to report conformance. Started life in AUWG to track conformance. Developed a lang for people to use to track those results. Lots of guidelines, requirements, etc. One thing that is useful for folks at the end of guidelines is for people to esaily say what they are doing.Talk about conformance in smaller chunks. Do a test once and score in a variety of places.

CS Another thing that is useful about that - if you have that metadata associated w/the page, then searching and filtering on accessibility info can be done through search engines.

LN Another issue - the role of LIP - learner information profile.

MC LIP as part of profile in IMS. We have 2 use case scnearios want to address:

  1. system adapt content for user, how they use computer
  2. search across separate metadata - describe course, module in course, learner objects

.. are they accessible for particular individual. Schema that categories various ATs and OS adaptations, other features like CSS to make accessible. Get the corrollary finished. If provide alt in audio, then labeled in metadata.

LS user profile and how use computer, does it cover cog and learning disability. will they be willing to self-identify.

MC built on self-identification. user fills out a form - registration.  Competency w/language - specificy base language, what level reading/writing at. Under discussion about how you document.

LS Yesterday discussed using Ruby to make alt content - literal words translated from non-literal. May want to consider that self-identifying may not work for large groups of learning disabled audience for a few reasons:

    1. embarrassment
  1. do not know

.. especially w/people over 30 (opposed to children). e.g. I had a worker who had ADD but who resisted identifying that he had a problem. I changed by behavior to help him.

?? can not build in the management issue that you bring up into our specifications. Within the educational scenarios we've been looking at...there might be an exception there.

JW Interested in the categories you have developed and the ways in which those can be matched against the content that is actually delivered. Think about those in our thoughts on conformance.

MC We can make it available.

GO How are you dealing w/the production of education material to be put up on intranet. Particularly, the tools in the open university.

MC Happen outside this meeting. I head up research gruop, we feed expertise into course develoeprs and web developers. Not a trivial issue. continual revision issue. contact me separately.

JW Action to contact you in regards to creating guidelines to make content more accessible to people with cognitive disabilities. we are trying to develop success criteria for cog and ld.

MR Learning design group might have more of that info.

MN Trying to find balance between corporate and educational activities. Close to the end of the hour. What are the next steps?

LN WWW2002 conference - have a summit? IMS, W3C, and dublin core? Work together. It's the 6-11 of May. We would have to make w/in the conference timeframe. If we get a day, would that be of interest?

MN Ongoing problem of overlap of WAI and IMS meetings. IMS will be meeting in Boston that week. Many of the IMS Accessibility folks might be willing.

LN What about attend by phone? Just one day.

MR 6 hour time difference.

?? Probably some can attend, but not all could attend.

MR I would suggest we consider that date, but if there other. What about CSUN?

LN Agrees to be liasion between IMS and WAI.

Action CMN: send info about EARL to IMS

Action LN: follow-up on dates and coordination.

/* break */

Report from discussion about WCAG 2.0 dependency tree

Jason, Cynthia, Gian

JW Can some checkpoints be implemented independently. there are some dependency. Started resolving various issues. Draft dependency tree which we will make available soon. Had to be careful to distinguish dependence: to get benefit of checkpoint x need checkpoint y from inheritence. we looked at dependence. Especially under guidelines 2 and 3, most of them can be independent of other checkpoints.

GSW 4.1 had 4 dependencies. /* describes, as well as rest of dependencies discovered. these will be written up and posted to the list. Also several that have no dependencies. */

CS We discussed that it is good that checkpoints are independent.

JW Grew out of yesterday a.m.'s discussion. new type of exercise.

GSW One main conclusioN: there are guidelines that are independent. If we can highlight that, makes accessibility seems less scary.

CS Allows for iterative approach.

WC shed light on conformance discussions.

GSW doesn't make sense to have leaf at P3 and high level at p1.

CS since more w/out dependencies than with. allows better that we can say conformance by checkpoint.

LN How many people does that help? not flickering, a whole bunch of people would be helped. low-hanging fruit. good, independent. good affect quickly.

GSW 4.1 isn't obvious, not as clear cut as flicker. if you choose a particular tehnology not as obvious. was a p2 before. problem w/the checkpoint b/c talked about "what is appropriate?" you can have alt-tags rendered appropriately in HTML 2. Choose the tehs that support certain checkpoints rather than all of them.

CS don't want to say, don't add alt-text if use bad tech.

GSW if don't have right tech for all the checkpoints, but you can go down some trees. go down the tree that you tech allows you to do.

JW As far as conformance goes, dependencies show which cp's can be implemented independently.

4.2 (according to spec), 1.1 (text equiv), 2.5 (control over action), x.x (sep content from presentation)

WC Final form.

JW The result puts some constraints on conformance at a basic level. Indicates dependence.

GSW Annotate before write clearly and simply. Use device indie and design UI compatible with AT - do that before graceful transformation.

CS Whole chunk that you have to do together.

CMN Doesn't lead to prioritizing individual checkpoints. Don't start on the 2nd floor if you can't get there. Gives people idea of second floor.

CS Defines the term "lift."

JW We know this will be a W3C rec and that there has to be a conformance scheme. there seems to be some support for multi-level scheme. Haven't agreed on criteria.  Will be at least one level. Know we want to provide info about interdependencies as well as how benefit diff groups of users. We need to examine the impact on individual groups of users. Each checkpoint in itself is making accessibility and inaccessibility to some group. That point made but seldom argued in detail. Another way to analyze checkpoints, which checkpoints will have the greatest impact on the largest # of groups. Which ones will make content easier or more difficult to access, but not accessible or inaccessible. Effectively an impact matrix. Need to give action to someone to come up with.

.. IMS - need to analyze that info.

LS I want to identify where we had agreement and identify areas we want to develop further. I felt that we are very close to consensus. I thought we had agreement that we want to remove profiling from conformance. Profiling: this disability is good for this checkpoints. this be an informal document.

LN Not clear when the LIP document is available. I can give it to us for this meeting.

JW could you summarize.

LN The way it is structured.

JW I thought I heard clear statements that they wanted it to be clear who benefited from each checkpoint but not claim conformance based on disability. Info be there but not a way to segregate checkpoints for conformance.

GO I completely agree with that.

/* Everyone around the table agrees */

LN Agree with that statement. But, make it difficult for other people to do that? or do nothing.

CMN We dont' support then doing that to give an elephant stammp. Don't go out of our way to make it impossible.

AA Many apply to many disabilities anyway. Won't give them the stamp, but they are making headway.

Emma in reality, people will use cost/benefit for what they do. We need to provide multiple examples.

Matt If you have conformance scheme, this info not part of. It's informative.

GO AA point is good. Reason that we need conformance. If din't have any, then they will claim by disability, invent their own.

CS What's wrong w/inventing own.

GO danger is that we end up with by disability type and we are trying to avoid.

CS Depends on who "people" are - i'm thinking HR, govnts, accessibility group at a uni, not the page author. might be same person, but important to recognize tha tin the real world, all those people will be making own policies.

CMN What's wrong w/people inventing their own? Starting down path is better than not starting. But if only helping a handful of people then only helping that handful.

WC Standards for a reason - help us all get along.

Action LS get info about open university. Someone should chase that up.

LS Clarify or suggest way to move on - strong argument for both get rid of conformance since all of them will affect accessibility. Only way to exclude is to say that some disabilities are less important. We want to have w/each checkpoint additional info. e.g. burden (in a retrofitting situation vs from scrath), to help people when making own policy. Then we can focus on what can be there.

Emma need standards for interoperability. Interpretation of those vary considerably. But have standards - a matrix of some sort that we can tick off. If have more than 1/2, give you a pat on the back. Repositories - examples and techniques. to help you accomplish each point.

CMN Having who benefits is important. Don't like burden info included - it depends on state of the art and is difficult.

CS When I was saying that people make up their own policies is not a bad thing, wasn't saying that each door should be 2 or 3 feet wide, but that "we care about 3 feet wide doors but we don't worry about elevators." 1st floor accessible better than no floors. Defining based on who in user set. Perhaps give an example set.

Rob suggest there is a profound desire amongst policy makers to make own. key question, does WAI have knowledge to provide a better one.

JW There is a certain degree of confusion between policy and conformance scheme for guideliens. conformance is there as part of tech spec to give people guidance. Offer way to measure progress or achievement. Diff from policy. Might be basis of policy.

.. idea for conformance scheme. interesting to see if other standards have been in same way. in 1.0 we said p1 if it removed barriers. the argument has been made that almost any checkpoint will make diff between some groups be able to access and being unableto access. add second dimension to concept of impact: what is the # of diff types of needs that are supported so that breadth of x-disability coverage.

LN It isn't going to be wise to come up with 1,2,3 levels, but we could solve the problem by having 2 things: we need metadata available to devices to make decisions (search engines, access devices, etc.). "this things will do this..." a diff kind of conformance test: assign numbers to checkpoints then have a barometer. Assigning numbers would be difficult. I can go through a lot of little things, or pick one major one. Immediately have way to show that people have done something. By at least doing one thing, you get off the ground. I think you could do that in a way that would recognize the benefit of the work put in. That would be the elephant stamp. Diff from EARL statement.

Emma like pics rating?

LN No.

GO clarification - what is the process?

JW Work out some action items.

/* eating lunch and continuing discussion */

GSW I like the barometer idea. Only problem is "how do we say one is more important than another."

/* phone drops */

AA support JW's impact matrix. How much help diff groups? Helps people realize who helping rather than generic "PWDs". Encourage to come up with conformance levels. People need to know where to start. We don't have the resources or the time.

WC Perhaps we are more like the UN, each gov has their own. What is their impact?

.. In looking at action items - we do have benefits...what is missing? more research, more detail? Specifically - AA and LS - speak more about? You seem to be raising a lot.

.. Still like the idea of "select any 5 that you want" or "do at least 50%" and let people decide which ones they choose. But, that might mean all that you do is: low hanging fruit. also have the "x-disability" scheme" allow a few schemes.

.. Barometer - allows us to weight, but not necessarily say "which are more important than others." A test is the only experience I have with that. GV argues that could have high rating but not an accessible site.

LN Assessment field would have experience in that.

LS disagree w/JW. fundamentally diff from elephant stamping. metadata is constructive tool. not a certification. not a pat on the back for missing peoople. barometer where page is not accessible. depends on how define accessible. w/barometer: people will not be doing everything. don't throw out the idea, but start w/it. problem w/P1 a lot of people don't feel it is enough, that it is still not accessible. if barometer, each checkpoint that you fulfill adds to it, that's great.

CMN disagree on burden: examples of changing table layout being difficult - in some environments it is a snap. In other environemtns it is more difficult. agree - weight on checkpoints. we tried that in AU. It is difficult and morally dicey. Whichever way you go, you make policy-type decisions about which people are important. Have to profile checkpoint benefits and who is important. Whichever way you cut it, it's a policy decision that we should not be making.

CS Conformance and policy don't have to be the same thing. 1.0 w/three levels is a policy. "I've conformed to checkpoint 1.2" is conformance. "You must conform w/x, y, z" is a policy. Letting people make own decisions should be done.

GO People are excluded in the process of creating bricks and mortar accessibility. PWD accept that in the creation of an accessibility code of practice that people are excluded. It is accepted. We will never create guidelines that will work for everyone. There will be compromises. I feel strongly that ... are we working for or with people with disabilities? I want to be working with rather than for. It's about "what are the problems." Are we sufficiently qualified to know the problems? Can we produce real solutions rather than what we think we ought to produce. I've been uncomfortable. there is a huge body of knowledge we can draw from - in bricks and mortar. Why aren't we?

GSW People may choose low-hanging fruit. perhaps have min conformance (like w/A-level) with barometer on top of it.

JW The proposal for a combined impact rating is not something I'm supporting, but something to discuss. Interesting that AUWG tried and ran into trouble. One area where we might have agreement, is that it should be possible to make a cp by cp conformance claim using metadata. One approach to conformance is to say that is the only way to do it. Would not give priority to diff checkpoints.

straw poll: priority versus weighting?

CMN against a wide range of possibilities. If we try to weight checkpoints, spend a lot of time chasing cats.

Action WC: summarize discussion

CS Narrowed controversy to a few points.

/* end of session */