Meeting minutes
scribe PeterKorn
Agenda Review & Administrative Items
Sajka: reviewing agenda
Sajka: Got a new GitHub issue; worth going through those soon?
Sajka: expect we could reply to one or a few Github issues with the next public working draft.
Sajka: reminder, we aren't meeting next week - we have a joint meeting with ePub at this time next week during TPAC. Also Silver meetings this Friday and next are likewise cancelled.
WCAG3 Timeline and Conformance Options https://docs.google.com/spreadsheets/d/1yzR1H0SnNFRELGchb_BJr4Necsrj6xVjDF1n7Tc0kTc/edit?usp=sharing
PK: row 8, description not fully there
… seems it needs updating
… also wondering about moving work on sampling to later?
… seems other things need to be resolved earlier
… finally, not seeing anyone assigned to conformance levels and wonder if we want to sing up for that
<PeterKorn> PeterKorn: scribe
Jeanne: Cell c8 doesn't seem to match with 3rd party content at all
(thanks Shadi)
Wilco: Want to put as much in parallel as possible, hence Representative Sampling where it is
<jeanne> ack
Jeanne: Representative Sampling is a gating toward scoring
… hard drive to get to scoring in larger group
<Zakim> jeanne, you wanted to talk about Representative Sampling gating scoring
Jeanne: multiple scoring proposals have been hit by lack of sampling
PK: maybe have different view point on sampling
… think it is a mechanism to gain confidence that the content behaves as you think it may behave
… it doesn't tell you if you pass or fail
… or if there are 0 errors
… just gives you a certain level of confidence
… helps backup a claim you make
… but not really able to do much more than that
<shadi> ...concerned how important sampling has become
Janina: feels similarly. I think there is an expectation of further refinement in the concept of sampling (sample from different portions of site)
… process sampling for example
Wilco: curious what others' perceptions are. What Peter said matches Wilco's understanding. Approximation of reality.
… example of how this might work with scoring; in Netherlands there is a sampling methodology in WCAG-EM. It would allow some number of "minor issues" (e.g., 5 out of sample)
… that is how scoring might work. Based on some number of your sample. At a meta-page level (vs. within a page)
Shadi: hearing Jeanne's explanation, relationship between sampling & scoring, it strengthens argument for signing up for conformance levels.
… several on this call involved in WCAG-EM development. Has shortcoming for single page applications, especially complex apps (e.g. web-based word processor)
… other shortcomings as well, which we weren't able to solve in WCAG-EM. Hence WCAG-EM focus on website.
… increasingly common cases in WCAG 3, which WCAG-EM doesn't solve. Not trying to shoot things down, everything late & in a hurrt
… a focus on sampling right now takes away some decision of what conformance will even look like
… feels conformance needs more maturity before we can look at sampling, scoring, and conformance levels
Bryan: How important that we understand relationship between sampling & scoring to do this?
… can we do sampling, and then have scoring sit on top of that
Jeanne: Bryan's understanding matches her's. Proposal for sampling large, dynamic sites will inform scoring. That is why feels sampling should come first.
… for levels, lots of people with various ideas on levels. If we do that first, fears we will tie ourselves in knots and have to do over
… feels order is sampling, then scoring, then levels. And whole group wants to work on scoring.
Janina: are we still expecting a table-like scoring approach?
Jeanne: don't know. Lots of different ideas; don't know how group will go
Janina: Feels like chicken-and-egg. Maybe early sampling thoughts; drafty; from which we can then look at other things, vs. fully mature sampling.
<Bryan> +1
Jeanne: thinks that is the expectation for sampling at this point
… thinks a logical starting point is use cases.
… this group has good experience with starting with use cases, then getting to proposals coming from those use cases
Janina: what was your question PK?
<shadi> PK: should we sign up for conformance levels?
Wilco: would not want us to take on conformance levels.
… that is a meta question, feels we need to figure out the individual pieces.
… feels levels is better later, and so many opinions, feels larger group needs to work on this
Jeanne: +1
<maryjom_> +1 to Wilco
PK: maybe misunderstanding but isn't everything that the sub-group works on starting point?
… not only necessarily voted up or down, could be taken up
WF: already a document with 15 or 20 different proposals
… all depend on how outher things fall ot
… meta-level topic that touches on everything we are doing
Sampling & Reporting -- Initial Use Cases https://docs.google.com/document/d/1YgiOg3CZz-LAVxRT0CWUTWHzyVa3UrjqdU4NvoyUZ_8/
Shadi: Sees a lot of "catch-all" items in the google doc.
… thinking about this a bit more systematically, what are the issue behind the challenges
… what is the responsibility of a technical standard, and what belongs in policy
… example: user-generate content / 3rd party content - is it the nature of the content, or is it a feasibility issue?
… if user generated wasn't a large volume, would it be so much problem?
… looking again at user-generated, what belongs in technical standard, what belongs in policy?
… would like to map these out, so we can develop solutions for what belongs in technical standard, what fits better into guidance for policy makers on how to take this work up
… don't have a concrete proposal to put forth right now
… hair salon example from Google doc. WCAG too complex for typical "mom & pop" store to digest.
… what is minimum level of expectation for people to develop things, put them out there
… just as we expect user to have some knowle
… knowledge of their AT. One thing to build an Ikea cabinet, another to be a carpenter
<sajkaj> bry
Bryan: looking at the Google doc., can we help a content provider understand if they are large/medium/small?
<shadi> +1 to Bryan on getting more specific
Bryan: definitions out there for these. Would that help with sampling & scoring?
maryjom: so much is focused on websites. There are large web apps, they get updated rapidly. Want to make sure we incorporate that in.
… can't do full a11y testing for every small iteration. How would sampling work for that?
<jeanne> ack
<Zakim> jeanne, you wanted to talk about the separation of policy and representative sampling as related to scoring
jeanne: (to Shadi) representative sample isn't about policy.
… goal: allow people to pass when they aren't perfect. Trying to move away from 100% perfection.
… in order to do that, we need a way to determine if a site has chosen a good path through site
… to enable us to score "how much is good enough?"
… to get consistency, so it can be replicated
… not going into policy area about it
Wilco: WCAG 2 doesn't consider more reasonable measures for different types of people putting things on web
… 9 yo cousin puts things on web; he won't put language tags. Too high an ask for him.
… he can put video on YouTube. Can we expect captions from him? Should we expect YouTube to caption it?
… browsers aren't required to make all content they serve up be a11y. Larger the org is, more responsibility they have.
… degres and levels in here that matter, and WCAG 2 doesn't go a good job of addressing those.
PK: responding to Jeanne on samping
… absent some scoring mechanisms and absent some levels
… that say this is an amount of imperfection
… but sampling alone doesn't say what is acceptable
<Wilco_> +1
PK: when I hear you say this is the route to that, this concerns me
… it is a yard-stick not the bar itself
… exercise in statistics for grad students
… see the need to develop use cases here
… for example large vs small, static vs dynamic, etc.
… sampling gives you an idea
PK: then getting to the 9 year-old
… or moms & pops shops
… what of that belongs in the technical standard
… vs what belongs in the policy that sits on top of that
… making the distinction how the content is created
… saying which parts are user-generated vs not
… encouraging the 9 year-old to provide captions
… providing automatic captioning to support them
<PeterKorn> sajka: Critical errors are also important here. Submit buttons (for example) are still a headache, 20 years after web a11y began.
<PeterKorn> ...tolerance for critical errors should be much lower than 9yo uploading YouTube video
<PeterKorn> jeanne: agree with Peter on place / role of sampling.
<Zakim> jeanne, you wanted to agree with Peter, but our group can set the rules for sampling
<PeterKorn> ...but need to work on rules for sampling
<PeterKorn> ...a big task, important task
<PeterKorn> Shadi: +1. Concerned with the google doc - putting it into conformance challenges. Do we lower requirements?
<PeterKorn> ... (extreme not being argued for) since 9 yo can't do it, shouldn't require it.
<PeterKorn> ...where do authoring tools fit in (eg., for user generated)
<PeterKorn> ...need to look at what conformance means, what technically the developer must do, what policy maker must do
<PeterKorn> ...hence looking at sampling a bit later, after we work on conformance more generally
<PeterKorn> ...not just user generated becomes a large flow of content.
<PeterKorn> ...and that hits feasibility.
<jeanne> We need use cases across the spectrum of small medium large, dynamic vs transactional vs static, should we allow multiple types of declarations of sampling, etc
PK: sampling static site may be different from sampling an app
… gut feeling is that representative sampling of apps by 3rd parties will be a difficult nut to crack
… so maybe some reliance on the author is important
… need to focus on the use cases