WCAG F2F Telstra, Melbourne Australia. 13 November 2001


On the Phone

Notes from morning session

Note that we had Matt May on the phone and Matt Mirabella in the room. Therefore MM is overloaded. We apologise for the confusion. Jo Miller minuted on IRC for those attending by phone or IRC only.

/* Cynthia Shelly minutes */

JW Email from Greg summarizing axes for summarizing checkpoints. What criteria will we use to decide what goes in the minimum set, and how the conformance will work. Can the user access the content Ability of user s technology What the author should be able to do in terms of selecting audience and prioritizing implementation Temporal dimension. Some requirements exist because of software limitations which will disappear over time, and new opportunities will come up over time. How can we formulate these into a coherent conformance scheme

LN IMS takes an approach where you separate the user issues from the provider issues. They use a learner profile, and allow learner to add to the profile their accessibility needs. IMS is interested in how you link resources together to make a composite resource. Use a user profile of the user to automatically link the resources together. IMS is all about making standards to which members will comply educational sector. IMS definition. Trying to create a world where you can plug a student in and have them do all there work and be evaluated. Or, allow teachers to construct customized educational tools. Consortium of commercial software companies working on voluntary standards for educational software

Wendy: Conformance Introduction. We re trying to decide what kind of priority scheme we should use for WCAG 2.0. It s very different than WCAG 1.0. There is some contention about how uniformly we applied criteria in 1. Many people also thought it was too rigid

Jason 2 ways we can approach discuss what are the axes try to apply the ones we have

Liddy Perhaps Dublin core can be extended to include an accessibility element. Add accessibility metadata. Dublin Core has done a lot of thinking about how to represent accessibility in metadata.

Lisa There s a practical consideration people have when trying to make conformance claims, where it should be doable. That goes against the grain for people like me who think it should be based on how hard it is for people to access. We should have a minimum set for retrofitting, and a different one for people who are building a new site. This might make a good compromise

Andrew What level should we achieve? Levels don t necessarily map to what you re trying to do for your users. Priority 1, 2 and 3 are too course. Some way to mix and match according to their needs, according to what their audience is and what they re trying to achieve.

Wendy We have consensus that we need to allow authors to make a more granular claim. But, there is some concern about people only doing a minimal set.

Lisa If we label minimum set as retrofitting,

Liddy This is the same argument Dublin Core has been having for 5 years. How do you get from the core set to a very granular thing for a special situation. We should learn from their work.

Wendy How did they get there?

Liddy Core must be doable, recognizable. Special domains have added extensions. Education wants to know about the audience for whom the resource is intended, but there are 2 audiences students and a mediating teacher or some combination. Allows for refinements of extensions, for example adding controlled vocabulary. Dumb-down rule to allow refined set to be translated to core. There is a registry for cross-mapping.

Wendy How would you apply

Liddy Dublin core is about discovery. We could add a domain specific element recommended for use by everyone in Dublin core that would describe accessibility. Could use application profiles to look at [508 vs. WCAG vs. New Zealand, etc]

Charles We have technical tools for linking different requirements. WCAG has a list of stuff that you can do and ways to decide if you have done it. In WCAG 1.0 we had must should may, which is well understood. One way to approach is to get rid of priority, and give information on which users are effective so that different organizations can make proper cost-benefit decisions.

Liddy Could do that w/ Dublin core using application profiles.

Andy On the call last Thursday, we talked about the pros and cons of having and not having priorities

Wendy Was there consensus?

Jason There was recognition that there were pros and cons both ways

Andi Drawback of bare minimum many orgs will do just that and no more. How can we express or present as Phase 1, so people will continue with the next phase. Kynn compared this to transitional and strict HTML standards.

Jason Kynn also suggested that trans/strict analogy worked for tech change too. The requirements that will go away with tech innovation would go into transitional. Dimensions Possibility/impossibility for use W/ w/o tech What authors can do with their resources / needs What s available today, what can be done when there are new techs. The challenge is to examine how they interact and what kind of scheme we need

Lisa Something I m worried about. When speaking to corps about learning disabilities, people often complain against the policy of self-identifying. In the adult population, there are people who don t know.

Liddy IMS profile is based on what you need not what you have

Lisa For cognitive disabilities that won t work, because you have a lot of denial. The opposite of self-identification is for someone else to identify you, and I don t trust others to do so. I m hesitant to let authors identify their user s needs. At the same time, I think Charles point is important, that each and every checkpoint is essential to some user group. Single A is only about physical barriers, comprehension moves to double A. If the data is not comprehensible, has accessibility been achieved? Taking a solution that is almost asking people to miss use it for the same groups is not a solutions. Similar issues with Phase 1, Phase 2. A lot of things that are in AA, are things that can only happen in phase 1 development can t be added later. A better representation might be who does it and at one point in the process (general architecture, updating, etc).

Charles Language for motivation is useless by the time it gets translated. We aren t requiring conformance, we are explaining what is required to ensure accessibility. Pro dropping priorities: we can do the work that we need before we can make sensible distinctions Con: ATAG will be a nightmare. There will be a big list of things to do. I think we need to do more work on techniques and how to make this happen, and less on conformance schemes.

Graham Small places are looking for guidance and leadership from the guidelines, so having a conformance scheme is important there aren t enough people for them to be able to split out the requirements.

Gian Having a retrofitting thing that seems easier will be bad, because that is what people will do instead. Splitting things by disability is tricky. It is important to justify that a requirement is important for somebody, but we should not incorporate that in a way that encourages people to try and match their target audience. Splitting conformance. There are 3 types of site that attempt accessibility general public sites, (should aim for AA), disability service orgs they are going to go as far as they can, and there are sites with targeted audiences they assume that there are no people with disabilities in that audience. Question: I am a bit hazy on how the technical environment fits into the group. Why do we assume that people don t have JavaScript because of their systems? Economics? &

Gregg People look at the guidelines and say where do we start? . Maybe there is something other than importance to people as a way of prioritizing. If we have the full set and let people pick where to start, people risk starting at a point that is not useful doing a lot of work that relies on things they haven t done yet. There are things that are critical, the minimum set. I think we need to have some type of prioritization.

Loretta Charles proposal appealed to me, but what would it mean to claim conformance. In addition we discussed whether conformance could be related to disabilities I thought

Charles Conformance would be doing the lot. Rare and highly valued, but we would have to make it clear that specifying what has been done is really important in the real world. It should be impossible to claim conformance based on helping one group or other of people with disabilities

Jason We need to have information for our checkpoints who benefits, what the technical requirements are, how to do it, & We haven t got an agreed set of this information, but these are consistent themes. We should have a task to work out what we need from this information. There are possible dependencies among checkpoints these should be identified. Maybe we should go through the checkpoints, categorize them so the extra info is in the document. Then we can go and consider conformance again.

Cynthia This group can define what things can be done to help people that s what we are really good at. The next thing is to help figure out where to start. There are loots of different audiences. If we give implementers the things they need to implement without worrying about which is more important, then people can find it and use it. Then we can give policy makers the information they need to make a policy. They do that anyway we otherwise make it hard for them to relate what we are doing to what they are doing. And then we need to make things discoverable EARL, DC, or some other metadata-based system so people can find it. If we do those, we will do a lot of good. If we try to write regulations, we will help fewer people if there is what my boss wrote and what W3C wrote, I am going 5to follow my boss.

Lisa I like the way we are going. I disagreed with the premise that Charles made, that our job is just to give information. Perhaps our job is to do what we can do to make the web accessible. Looking at the situation in New Zealand is very important I don t think everybody is going to make up their own policies that only applies to big countries and big companies. I wanted to suggest we have the complete requirements, add a level of burden in different situations when you are creating something, when you are fixing. Also, demonstrate the spread effect the dependencies. Then we should look at ways of implementing accessibility policies.

Liddy: IMS are not here because they are meeting tomorrow morning (days are different on different continents).

Break 15 minutes. Q == Andrew, Matt, Charles, Graham, Liddy.

/* Charles McCathieNevile minutes */

AA In theory support not having prorities, but people want them. So they will start creating them. maybe the idea of retrofitting / from teh start is a good approach. I am finding a lot of interst in bulding it in, as companies become aware of the issue - there is still a lot of ignorance. I support the metadata for people providing details, but we need to have a more visible elephnat stamp approach of some kind.

MM Would be keen for an assumptions discussion. Re laundry list, and people looking for guidance, it is important to make sure that it is difficult to get it wrong with the tools and languages. That should be long term. In middle term we should look at getting there as a process, and now we need to look at getting there from broken things. We shoulld be looking at several years down teh track rather than right now

CMN need to work out dependencies. There will always be different starting points and priorities, and we should be clear about who is losing for each one - maybe we should not write the policies, but identify a few different ones. EO is working on implementation policies already

Graham Curb cuts, when they were introduced, helped people in wheelchairs, but for blind folks thhere was no longer a distinction between walking on the footpath and walking on the road. I strongly favour using a cross-disability approach. I see temptations arising from client approaches to try and classify a target market. If we bow to that we will not be doing favours. People with disabilities have come together to solve the collective problems, and I don't think we shoulld be trying to factionalise or re-divide them. In an intranet the users are defined, so some things will be left out - this is a classic example of a situation where if we serve that need we put barriers up in the general case.

Lisa So in that case a person with a disability (or getting one) faces a higher barrier in joining. It is also difficult for people who don't want to identify their disability.

Liddy Metadata doesn't do stuff for you. It is only data that can be used. Whether it is a badge, or a detailed bit of code it is available. The metadat community talk about metadata as never being finished. There isn't a fixed quality of "having metadata" - it can be extended and added to and adjusted... The goals here are how to make the guidelines most effective. I think it would be usefu to put everything our there and say what is there.

Wendy Chaals says there are different starting points - if you start with a different disability the start would be different. Designers look at their site and say "where is it most broken?" and start with that.

Gregg Most people looking at their websites are doing it because it is good, or generally helpful to users, but most people I know are doing it becausee there are a set of rules. If we start talking about disabilities there are tw problems. 1. People will match consumer groups or perceived targets. 2. If you only think of one disability you can reduce accessibilty for others. Lets look for the underlying principles - if you follow those you make things more accessible. The first guidelines were a patch. Don't do this, do that, ... We have tried to draw out patterns and principles and today we have guidelines along those lines. I think we are close to where we can give ourselves some tools. Someone talked about making guidelines that people can use to do what they need to do. We could say "there are 5 basic principles - these are the guidelines for achieving XYZ". Then groups can look at those and see what they can achieve. This group needs to lay out an order for people to tackle things in - otherwise how does a manager give some instructions.

Gian Intranet - it is about convincng the business manager that someone with a disability will need to do something in the future. One site is an intranet for scientists who work on communicable diseases - they claim that it is a prerequisite not to have a disability, in a small group of people. We still need to convince the people who think they are in a closed world. To convince them, it is important to highlight how this is relevant for the general public. If corporations are going to focus on accessibility a lot of the work goes into PR and not accessibility itself. With retrofitting, it might be difficult to write a list of guidelines, because it depends on the situation. I keep getting people wanting to minimise the amount of work - any amount is too hard - give me 3 major things. People will stop at a minimum, so the minimum needs to be a maximum.

Wendy Breaking conformance by disability - is that an open issue?

Jason There is a current consensus that we won't do it, and nobody has sugested we open it, so it isn't at the moment.

CMN agree that the process of how to fix things is good, but that relies on knowing what things to fix first, and second on priorities. We are not in a position to set policies. THEY are. This is not a case of doing it by disability, just taht there are too many starting points.

/* Loretta leaves

Jason I would like to establish where it seems there is agreement: We want a good benefits section. There are dependencies among checkpoints where benefits are only achieved if some other requirements are also satisfied. We should capture that information. Difficulty of implentation is dependent on many "outside" factors. That doesn't seem to be a useful axis for us to make statements. The issue of what will become obsolete or possible - we can probably make reasonably informed statements about that. So for a checkpoint by chekpoint basis, we can introduce some categories - perhaps that would be a useful exercise in any event - who benefits, what checkpoints it depends on, what checkpoints overlap/obsolete, will the technology change?

Wendy Can we categorise what willl change at the checkpoint level?

Jason Well, you did a good job of it recently..

Wendy Propose that we start capturing this information this afternoon.

Mat Agree with charles that we need to provide the information, but I don't necessarily agree with dumping priorities in terms of A, double-A, triple-A. I think  we might get something like that that will work. If you look at different groups and provide info about how to make things good for one group, that raises the possiblity of segregating. Maybe one approach is to highlight benefits that can be applied to help an identifiable group beyond (for example) single-A. Provide information about the level of impact for different groups - e.g. one checkpoint helping several groups, or only one, at different levels. Have a "strong compliance" by conforming to single A plus the things that leverage those single A requirements.

Cynthia We can't figure out a scheme for situations that works - we aren't the best people to set policy. We can give them data they don't have - there are things that we know which the government or manager of a company or site manage doesn't know. We can allow them to improve the cost-benefit analysis they are already making

Jo What Cynthia said. There is a tendency to assume resistance and predisposition to do the bare minimum. By the time WCAG 2 is out there will be a different situation. Selling the benefits isn't our mission - it is for EO and for other outside people. Our strength is information and data that people don't have. Focussing on that will help us get our work done. Get back to guidelines and success criteria before assigning priorities to hypotheticals.

Lisa on what Mat said I think the problem with the priority levels is there is no fair way to take much out of priority one. Having a three level priority scheme is pretty much impossible because there is so much that belongs in P1. I think there is a need for an approval stamp somewhere - maybe through testing.

CMN To be a recommendation there does need to be at least one level of conformance.

Jason The problem isn't whether there will be one - there must. The question is how to distinguish if we want more than that.

Graham thinking about the idea that we're not qualified to prioritise guidelines. First we need to have some humility about what we do and don't know. Second, let's have lots of people with disabilities involved. Third, there are people in the bricks and mortar world who have been coming up with ways to prioritise this kind of stuff and we need to learn from them. Jakob Nielsen's "beyond alt text" is essential reading on this.

Lisa Some groups are making policy anyway, but in other places they aren't.

Gian People in Australia don't want policy rewritten, they maybe want it interpreted. Maybe we have something based on where you start based on an audience, cross-referenced with whether it is public, private, disability service, etc.

Liddy I disagree about people not rewriting policies - people don't take on stuff without rewriting it in their terms. So how do we make it easy for people to do that and get the results that we are aiming for. There are things that can be done easily, and it makes people feel good - getting the low-hanging fruit is good.

Wendy How about "you have to do 5 things - choose them". Then say what you did.

Jason This afternoon: One group will work on providing some of this information by checkpoints. I don't know that we will make much more progress in this discussion. Tomorrow morning we will work out what we are doing tomorrow.


Additional notes from morning session

In addition to the minutes taken by physical attendees of the face to face, these are Jo Miller's notes from the IRC log.

Jo, Andi, Matt May, Gregg, Loretta, Paul, Tim joined by phone/IRC. Jo minuted for Matt who was on IRC but not phone.

Jo: Jason refers to GV's message re. axes, conformance. Extent to which user can access the content with AT and client support. On other side there are questions of what author should be able to do in terms of prioritizing implementation by difficulty, etc. Temporal dimension: some requirements address software limitations which will disappear, new software will bring new issues, etc. Make sure guidelines can outlive medium-term developments in tech. Formulate all these into conformance scheme? IMS has separated user issues from provider issues. Accessibility group of IMS has added accessibility to learner profile.

Gian just joined. Wendy suggests re-doing introductions.

Jo: Matt, want to introduce yourself? They're asking how long you've been with WCAG and what you do.

mcmay: I'm a consultant in Seattle, and I've been a member for a little over a year. I'm working on the techniques document for HTML/XHTML, and some of the CSS techniques.

Jo: IMS will be joining in about 5 minutes and then they'll get into conformance issues...

Jo: Wendy is raising sticky issues like baseline capabilities.

Paul joins.

LS: Conformance should be "do-able." Thinking of adding different dimension to grid of conformance. Author's perspective. Minimum conformance standard that makes sense for retrofitting only. Completely different minimal conformance list for people building site from scratch. Addresses practical ramifications. For people who otherwise won't bother retrofitting.

Andrew: Corporate and gvt ask what level should we achieve? Levels don't necess. apply to what you're trying to do for your customers/audience. Some way of being able to mix and match checkpoints according to their needs will be more successful than offering A or B or C.

WC: We've been saying authors can make a more granular claim. Using Earl is one way. A few proposals: a minimum set, make granular claims based on what you've done above that. But if there's a minimum set, there's concern that the minimum is all they'll do.

LS: If we label minimal set as retrofitting...

WC- you think minimum set only applies to retrofitting?

LN "Core" are the things that are common to everyone. "Main specific elements" extend the core. For instance, for the domain of government. 1. Who is ultimately using it? 2. Is there a mediator (as in a classroom situation with a teacher) Different groups add further refinements to core extensions based on their interests and needs.

WC- How would you apply? What's the equivalent of domain in WCAG?

LN- Domain is accessibility. Within that domain there are different application profiles (WCAG, 508, New Zealand), that can be cross-mapped to each other.

CMN- We have the technical pieces in place to do metadata schemes. But in building new vision of WCAG, what does it mean to conform to WCAG? Which of those pieces are compulsory and which are useful things you can also do which go beyond some level. In WCAG 1 the three levels (must, should may) If we're changing that, then we need an alternative. If we keep it, then let's look at requirements and check if priorities are assigned at reasonable level. For accessibility as a whole, there is a group for almost any given requirement for whom THAT requirement is a MUST (essential). One approach: drop prioritization. Here's the whole list. To build a policy that some gvt. is going to implement, show them what group they're cutting out. They are in the biz of excluding people. This is what they do. Different orgs will come up with different CBAs. Degree of difficulty plays a large part for them.

LN- suggest prioritizing by applications?

CMN- If you're happy to exclude x group of users, here are the things you can drop off your list. That information is reasonably available.

Matt, we're recapping last week's telecon

Jo: Pros and cons of having levels vs. having no levels.

mcmay: k.

JW- 3 or 4 different dimensions along which we could make classifications. Examine how those dimensions interact.

LS- User self-identification.

LN- IMS goes by what you need, not who you are. One person may have many profiles.

LS- Flip side of self-identification is allowing someone else to identify what your needs are. Cognitive disabilities, learning disabilities. Not leave it to web authors to identify their end-users' needs.

every checkpoint is essential to one user group. Single-A tends to go with physical barriers. Comprehension goes to double-A. Same applies to phase 1, phase 2 representation. Anything left to AA or AAA is something that must happen in phase 1. Better representation: who does it in the process and at what point in the process can it be put in? People can choose according to where they are in their web development. Planning, updating, architecture, photographics department...

CMN- First point: using the way we talk about conformance levels as motivational tool is useless. If there are 3 levels people will go for 1st one, maybe second. Second point: we're not writing laws. If we say these are the requirements for accessibility, we're not forcing anyone. People aren't necessarily going to do them all at once. Kynn says we shouldn't be setting implementation priorities. We should be setting out information that people could use. Argument for dropping priorities: we could get on with the work. We're not in a position to take what we've got and assign priorities. There's another related rat-hole: what do we assume about users? We assume they can identify and follow a link. WCAG 1 does not assume access to javascript. We need to start out with process for deciding what our assumptions are.

mcmay prays for a short user-assumptions discussion ;)

CM is not keen to drop priorities because AU work assumes priority scheme. Also, we'll end up with a big laundry list. Well, we will end up with a big laundry list. There are lots of pieces to consider. What's the most effective way of implementing checkpoints so that you cover 7 or 9 of them in each piece of work you do.

GO- Request that we maintain one level of conformance at least. In smaller jurisdictions (NZ, etc.) we're looking for leadership from guidelines. In America there's a huge legislative environment for coming up with appropriate guidelines for various sectors.

mcmay: The most effective way of doing that is to make sure that implementing things in an inaccessible fashion is difficult using the languages and tools.

GSW- Transitional and retrofitting...

mcmay: I think that's what should be the long-term goal of WCAG. In the short term, I feel we need to work on a transitional approach to bring current sites to more people...

GSW- Useful to explain to business managers why you need to conform to a certain checkpoint. They need concrete explanation with reference to user groups. But this shouldn't be incorporated into the checkpoints. Allowing people to exclude groups deliberately.

mcmay: And in the middle-term, we should focus on how to do things right from scratch (WCAG Strict). But in the end, we need to realize that UAAG and ATAG need to rejoin WCAG at the end.

GSW- Three types of sites attempting accessibility levels. 1. public site, general. should aim for AA 2. Disability services organizations. Willing to go far and above AAA. 3. Site with targeted audience. Small, known audience. None blind, deaf, but maybe color-blind. How does technical environment fit in? Baseline assumptions? People with disabilities are usu. from lower income? We may be setting baseline too high.

GV-History of prioritization schemes. Layers. Where to start.

mcmay: I think I'll hold with with that. The idea is that we should be looking several years down the road, rather than trying to pick apart current capabilities.

GV If we have just the full set, letting people pick where to start, what we often find people doing (to use hotel analogy)... Is that they're doing the carpet on the second floor before they've installed the elevator. Where do we draw the line, how do we figure that out? Give people a way of attacking the problem in terms of grouping, layering. Also flag the ones where if you don't do this, no one's getting anywhere. Minimum set or whatever. I feel we need some type of prioritization on the items.

Loretta- Charles's proposal appealing on many levels. But what would it mean to claim conformance? This came up at last F2F. Discussion about whether conformance could be related to different disabilities, consensus seemed to be that it could not. Are we back there?

CMN- Conformance would mean you did the lot. Means conformance is rare, highly valued, reasonably measurable. What I would expect people to do a lot of is claim how far they'd got. Identify each thing they'd done. Not claiming "i conform." But produce detailed information about how far along they are, what they've done.

Jo: BTW, Matt, this also came up in the telecon last week. Stages on the way to full conformance.

CMN- Getting real conformance would be a big piece of work. And stating how far you'd got would be A CRUCIAL PART of implementing the guidelines. And no, we shouldn't let them claim conformance by splitting people by disabilities.

JW- We have some agreement that there's some information that we want to provide with guidelines. Who benefits, who would be excluded without them, some implementation information. Where it is best done in implementation process and how it is best done. Areas in which we want to at least provide information. This seems to come through pretty clearly in discussions. Differences of opinion on whether there should be differing levels of conformance. Need among implementors to have suitable starting point. Some dependencies among them, as GV mentioned. Maybe what we should do is go through checkpoints at some stage, categorize them in some way, so extra information is in the document. Then move on to prioritizing or categorizing in some way. Start adding the additional information first. Consider conformance questions later.

CS- We need to define what are the things that can be done to help people. This is what we can usefully do. Next thing is to help implementors know where to start, how to do it. Many audiences for that. If we set it up so we give implementors info they need to implement, without worrying about what's more important. Just write it down and get it out there... Then give policymakers the info they need to make policy, which they're going to do anyway. Third thing: provide info to users so they can provide content they need. In metadata. If we can do those 3 things, we'll have done a lot to help people in real world. If we try to write a set of regs, we'll help fewer people.

LS- Disagree with chaals that our job is to give information. Perhaps our job is to promote, to do what we can to make web accessible. Little groups aren't going to come up with their own policies, guidelines. Need to take into account practicalities of implementation. Suggest: have complete requirements. Add levels, grid. Burden in different situations, for example. Distinguish retrofitting from building new sites. Recognize different stages in web creation. Inaccessible navigation is a disaster, for example. So we could add something about how to make a regulation.

WC- You're talking about profiles for types of sites, stages...

/* break */

$Date: 2001/11/28 22:46:49 $ Wendy Chisholm