Meeting minutes
<alastairc> Revisit accessibility support sets https://
alastairc: would anyone like to introduce themselves? or update their affiliations?
Annoucements & introdutions
Monica: this is Monica, I'm new to the group and joining from Oracle
alastairc: great, we have had many great participants from Oracle in the past
Anton: hi, I'm Anton, I work for SAP!
alastairc: if things don't work feel free to comment in Zoom chat and someone will help
Revisit assertions https://docs.google.com/presentation/d/1qWuFM3fFgC_e1Jik05Os11O0Rl86HLDXu9dolwyWWtc/edit?slide=id.p#slide=id.p
Slideset: https://
alastairc: so… assertions are a new concept to WCAG 3. We want to test their feasability in practice, testing with organisations who would be on the sharp end of using these
alastairc: my summary of what an assertion is: it's something you've done, usually concerns process behind the scenes, rather than things that concern content of the website (like all requirements in WCAG 2 are)
alastairc: so an example of an assertion is a statement on what you looked for when building a video player
alastairc: comparing to supports/partially supports/does not support, with an assertion you include certain specified information
alastairc: we also specify 'informative supporting documentation', we recommend it but don't require it and it is definitely not required in the performance statement
alastairc: the part that would be public would be the date of implementation, date of claim and some other bits
alastairc: the possible internal documentation would be recommended but not be required
alastairc: what we're trying to do here is find out if your organisation would want to include this information, and whether they would want to use assertions
GreggVan: great question
GreggVan: we could describe the informative/supporting documentation as 'things you might include'
GreggVan: this way would be less risk that regulators start to require the 'recommendations'
alastairc: there is an update since last time. We'd like people to go through the questions, only had a couple so far
<Charles> +1 to careful use of ‘recommendations’ as ‘suggestions’ since it already means a TR document status.
alastairc: last week the conformance subgroup met with a lawyer recently who has a lot of experience in accessibility cases (prosecution side). We had a long conversation on assertions. From her point of view she said it'd be preferable to have outcome based requirements. I asked if the concept of assertions raised alarm bells, it did not for her, but something lawyers don't like is assertions _about_ outcomes. She wasn't sure if it sohuld be part of
basic conformance.
alastairc: if there was a legal process that sort of information could come into play but in the court process
<bbailey> Are U.S. .gov accessibility statements sufficiently shaped like a conformance claim?
<Zakim> bbailey, you wanted to ask if section508.gov accessibility statement is sufficiently shaped like a conformance claim ?
<bbailey> https://
bbailey: I'm curious if Section 508 are sufficiently shaped like conformance claims?
alastairc: not sure yet
GreggVan: the lawyer we spoke to is a strong disability lawyer
GreggVan: when you said no alarm bells re user needs in the cognitive space, wasn't this about assertions in the cognitive space?
GreggVan: when we asked her 'what level' we didn't get a clear answer back. This questionnaire you're putting out is a great idea
GreggVan: re public… once they're in the court, if they have been documenting, they have to produce it
<Charles> documentation is subject to discovery. there is now legal precedent where AI conversations are also subject to discovery.
GreggVan: lawyers hate documentation when going to court
AWK: a general comment… when we're talking with people about assertions, I'm struggling with some of the examples. Eg on slide 7, are we making an assertion that we provide a video player? that seems testible and more suitable for a requirement; is this instead of or in addition to?
alastairc: there are better examples
<Ben_Tillyer> +1 to hdv
<Rachael> +1 to this being one of many conversations so we encourage everyone to reach out to your contacts as well
hdv: Love that we've talked to one. But, don't want to fall into trap of using law/lawyers from one region. Would encourage people here to find people from their region to talk to.
<AWK> Completely agree with Hidde. And we get value from talking with attorneys. Suggest more ppl to speak with!
alastairc: thanks for those questions
<Frankie> +1 to hdv
<shadi> +1 to hdv
alastairc: let's talk about some of the responses we got, we anonymised them
alastairc: we talked to a travel sector company, they would not publish their statements as they worried it would make them a target
alastairc: to them, assertions seem easier to meet, as they're not dependent on the website. The person I spoke to had more influence on the organisation than on the website
alastairc: we can improve examples, if there are any questions do let us know
alastairc: first meeting of next month we'll get back to this
scott: I would appreciate making some of the updates you just talked about… speaking to people internally about this… reading through this presentation I'm not sure what about assertions would be required and what would not be.
scott: it is confusing to me, and people who were looking at it with me were confused by it.
<shadi> +1
scott: eg in the clear language one, the first bullet says to meet the requirement, the other bullets have more requirements but it is unclear if they are then _also_ required, or if they are some sort of optional
alastairc: we'll try and clarify that
bbailey: I took the plain language assertion to be an example, not the expectation
alastairc: good feedback, we'll need to make some of this clearer.
Conforming alternate versions https://github.com/w3c/wcag3/discussions/623
Rachael: we had a conversation around which requirements had equivalence in WCAG 3 and what would happen
Rachael: main question was: if we did something global with alternative versions and equivalence, what would happen?
Rachael: Gregg said conforming alternative version should work the exact same as in WCAG 2
Rachael: Patrick said there are other approaches worth looking at
Rachael: including thinking about different modalities
GreggVan: we can't have three URLs for different versions for differnet target groups and have them all be equally primary
GreggVan: there's always one primary
GreggVan: re mobile… if you're on a phone it will serve you the phone version
GreggVan: if it's auto-determined they all are primary
<Zakim> Rachael, you wanted to say how does this play with AI?
Rachael: I think we have different examples of versions we need to talk about
Rachael: if I have a website and the website is not accessible, I link to an alternative that is accessible
Rachael: that's different from the multimedia example where there's audio description
Rachael: but we have to future proof this
Rachael: if we get a website but when browsers include enough AI you can customise different versions, is that considered a conforming alternate version?
<Zakim> alastairc, you wanted to comment on examples of different modalities that can fall foul of SCs.
<Charles> note: equivalent and alternate are different concepts. and equivalent has different meanings in WCAG 2.x (at least information and purpose)
alastairc: chair hat off… in some areas I've noticed it being difficult, re regulators misunderstanding how WCAG 2 intended this. Something often discussed is images of text, like a banner with text in.
alastairc: on a video site, there's these horizontally scrolling cards with arrows on either side… but then have to choose implementing if users should use the arrows
<AWK> similar for a site with an embedded video and a "(Audio described version of welcome video)" link
GreggVan: nobody ever conceived of the conforming alternative to be plain text
<AWK> sure they did! That's why we have Plain text techniques...
GreggVan: there's nothing wrong with an optimised version
GreggVan: once all browsers are smart browsers the author doesn't need to do the work
<alastairc> Gregg - I meant https://
<Zakim> hdv, you wanted to react to GreggVan
hdv: conforming alternative versions in WCAG are scoped to web pages. Why are we talking about specific components and parts of images?
<GN015> Gregg mentioned a standard keyboard document. Can you add a link to it?
alastairc: Because they fail in that context
hdv: In our government, we talk about them failing as part of a page.
<Zakim> GN, you wanted to say compliance should not rely on AI
<Zakim> alastairc, you wanted to comment on an approach about achieving the same thing, rather than making a part of the interface accessible when that's not the best way.
GN015: re “smart browsers”: authors cannot know what an AI will do, and also users may not be able to use AI, be it because they don't want to or work for an organisation where they're not allowed to
<Ben_Tillyer> +1 to GB. Adoption of a self-rectifying smart browser by 100% of a user base is not happening
<hdv> +1 to GN015
<Frankie> +1 to GN015
alastairc: I wonder if we can have a more general approach
alastairc: it can be unusable in some UIs to try and make a certain part of the UI accessible but having another way for someone achieving that seems reasonable
<Jennie_Delisi> Ensure we address the challenge for those who teach (not just in education) or need to use vocabulary about what is onscreen for shared understanding
alastairc: we don't want loopholes or make it possible for it to be gamed, but if there's no accessible way of doing something that seems like a reasonable approach
<joryc1> I agree with GN015 and also would say unless there is a User Agents MUST provide automated Alt text in the HTML spec we couldn't rely on that and while I do think there will eventuaally not be many browsers that don't heavily leverage AI, but that seems pretty far off.
Jennie_Delisi: I agree with you Alastair. There is also a challenge for those who teach and need to use vocabulary to support what they're teaching… like there's a link style that is a button on a screen, to support someone without vision can understand
GreggVan: good point Alastair. We need to separate discussion between alternate pages and components on a page
GreggVan: re having a global alternate component rule… good to keep them separate, but the concept of something that is global and catches what we didn't think of, we create a specific alternate component
GreggVan: have to make it very clear what counts
GreggVan: all browsers today are 'smart browsers'
<alastairc> q/
<Zakim> Rachael, you wanted to respond to GN
<Zakim> AWK, you wanted to comment on small CAV's like videos with description
Rachael: as we write this, we're writing for 20 years from now, not just today's use cases, we have to keep that in mind
<Charles> note: the understanding conformance document clarifies why conforming alternate version was based on page conformance is “For a variety of reasons, it may not be possible to modify some content on a web page.” the intent does not seem to be replacing a whole page – just parts.
but the understanding doc isn't the norm
<Charles> agree. but the assumption was part of the page failed.
AWK: 'whole page' is what we have now. If you have a chart and link to another page that provides a table with the chart's data, that's part of the original page… we don't really have a term to talk about those things
AWK: it makes sense to bundle some of those concepts together
<Zakim> Rachael, you wanted to straw poll
<bbailey> +1 to AWK concern that CAV are overlooked, despite being referenced by 5.2.2
draft straw poll: For conforming alternative version should we 1) start from the WCAG 2 approach with the conforming alternative version as a fallback and equivalents on a provision by provision basis or 2) stand up a subgroup to explore a new approach with more flexibility?
<Charles> 2
Rachael: I'd like to propose this poll
<kirkwood> 2
<Zakim> GreggVan, you wanted to say +1 to AWK comments. and that would be handled by the conforming alternative components/elements
GreggVan: I think we should separate conforming alternative page from conforming alternate component
<LoriO> hdv +1
<kirkwood> adjacent?
Wilco: it strikes me as silly to make distinction between alternate page or something else
<scott> +1 to wilco's comment
<Ben_Tillyer> +1 to Wilco, need to think about the user journey more broadly
<Glenda> +1 to what Wilco just said
hdv: Can re-use what's in WCAG2, but refine it.
… what we should change is the page to page/view
… don't want to make it part of components, hard to define.
<CClaire> Sorry can't do it today
<Zakim> bbailey, you wanted to comment on the poll
bbailey: on the poll
… my main concern is we need a whole new term other than CAV
… I think WCAG 2’s model is great
… but want to change the name
GN015: we have 2 alternatives
… 1 is equivalent on same page, like text alternative for image
… 2 might be something bigger
… legacy technology that cannot be made accessbile
… we used to have this concept in WCAG 2
shadi: I’m not hearing an overall approach
… my suggestion is splitting decision
… let‘s get cracking on a subgroup
<bbailey> +1 to draft poll as most recently proposed
shadi: new alternate for components
… and then see if it fits in to CAV, or replaces it, or...
… we’re jumping the gun by trying to make this decision now
<Zakim> GreggVan, you wanted to say +1 to alternate conforming components being on or off the page - -same thing. Do do that for specific things and we should do for globally but don't confuse with alt conf pages.
alastairc: I’ll take that as a positive signal for #3
GreggVan: +1 to shadi
… the problem we’re trying to solve is conforming alternate _part of page_
… we’ll talk about element, component, etc., huge discussions around that
… could be a whole part of the page
… should be separate from CAV for the whole page
… for the _part of page_, there should be no difference whether it’s _on_ the page, or it’s linked from the page
… we already say that in some provisions
… but you can’t say that for the CAV for the whole page
… it is by definition a separate page
… don’t confuse them
… but good idea to stand up subgroup to look at conforming alternate _part of page_
<bbailey> FWIW, I'm not convince that CAV "part of the page" is the main stumbling block
<Zakim> alastairc, you wanted to ask whether we can agree that we need a draft of CAV + another within-page
alastairc: it sounds like we’re generally agreeing on the approach needed
… to update WCAG 2 version and try and draft in something that’s kind of within page
… chair hat off
… having something about being able to achieve the same goals from the same starting point
<Zakim> Rachael, you wanted to say that would be what a subgroup would do
alastairc: we have something, but it needs to be overhauled for understandability
Rachael: another way to think of this straw poll
… do we feel WCAG2 is solid enough without redrafting?
… or do we need more exploration
… and today’s discussion suggests we need more exploration
alastairc: if anyone is interested, please get in touch with the chairs: group-ag-chairs@w3.org
Charles: I understand the point of view of differentiating between page and component
… but what confuses me about that differentiation
… is the draft of the scope of conformance
… the scope can be pages and view
… and processes
my 2cts are the WCAG 2 approach is solid enough (updating to modern age by updating page def), based on how we use it as a monitor/regulator
Charles: so should there be conforming alternate page and conforming alternate process?
(but am happy for others to go and explore better options / improvements)
alastairc: part of it is the “within page” aspect
… avoiding the term “component” for now
… alternatives within the same view
<Zakim> GN, you wanted to ask whether 'modes' come closer to what is meant by alternative version
alastairc: but as AWK pointed out, kind of already covered by WCAG 2, but could be clearer
GN015: high contrast mode is an example
<shadi> +1 to exploring something along the lines of "mode of operation"
GN015: or in a support call, where sign language is provided as an alternative
… does that come close to “alternative version”?
alastairc: good point
… could be at same URL, but also kind of a personalization thing
… so yes, we also need to consider that
… not the same as the scenarios we’ve discussed, but should be included
Jennie_Delisi: related
… one issue I’ve seen is where there _is_ an alternate version
… sometimes when you copy & paste from that alternate into the “main” area where people are interacting, it doesn’t paste well
… it has been problematic
<Zakim> Rachael, you wanted to wrap up
Rachael: wrapping this up
… let‘s move to subgroup and then bring back to the main group
Anyone interested in joining that group?
Rachael: thank you all for the conversation
… please let us know if you’re interested
GreggVan: question for Jennie_Delisi
… want to carry your idea into the subgroup
Jennie_Delisi: if we had someone using a high-contrast version in a collaboration tool, when they copied & pasted, the paste was wrong, so they couldn’t participate
alastairc: can move on to last topic
Revisit accessibility support sets https://github.com/w3c/wcag3/discussions/621
back on a11y support sets
… back on a11y support sets
alastairc: in GitHub discussion
… a11y support sets is a mechanism to try to make explicit what we need to test with when creating methods
… methods should work, basically, with a11y support sets we define
… we can’t talk about specific products
… so need to define in a way that leads you to the right answer
… in last meeting, there was concern with straw man text I’d proposed
… support set _per platform_ was suggested
… still need to find balance between “meeting the standards” and “does it work for the user”
… different voices in the group supporting each perspective
… need to be careful to accommodate AT that work differently by design
… e.g., VoiceOver ignoring lists when bullet styles have been removed
… also need to define when “it works” is sufficient
GreggVan: cost is also a factor
… the most popular one might also be the most expensive one; out of reach
… free or built-in should be considered
… also, some people don‘t have their own computer
… they have to use one at the library, for example
<Jennie_Delisi> +1 to Gregg
GreggVan: and they can’t install anything there
… so their only choice is to use what’s built in
<Jennie_Delisi> (re cost_
GreggVan: otherwise we’re excluding people
alastairc: agreed, I’ve updated the discussion
… defining cost is tricky
… if you’re buying the AT alone, it’s clear
… but if you’re buying a phone or other device, how do you calculate
AWK: in the past, we’ve considered cost
… don’t think we’ve ever considered mandating built-in support
… and that’s scary
… especially because of the differences that exist
… the built-in solution could be worse than another one
Ben_Tillyer: someone could take your product and put it on a closed technology platform
… can’t control the environment
alastairc: as in the website you’re making?
<kirkwood> Cannot depend on proprietary technology?
Ben_Tillyer: if you’re buying third party equipment and they have their own list of AT they support
GreggVan: thinking about AWK’s concern about bulit-in
… pro: if they rely on it being there, what if it’s not there?
… on the other hand, if you’re going to rely on AT to solve the problem, then you need to be able to rely that it’s actually going to be there
… if they can afford it, if they have their own computer
… the fact that it may not be the “best” experience, that’s not an issue
… we’re not saying you have to have the best experience to meet WCAG
… just that you have minimum accessibility
… access to the information is good enough
… but I‘m sympathetic to the wariness
<AWK> So if narrator didn't support aria attributes the way that NVDA does... you can only use the features that Narrator supports
GreggVan: I’m spending a lot of time in communities that don’t have their own AT
<Zakim> Jennie_Delisi, you wanted to discuss freeware and security
Jennie_Delisi: I agree with GreggVan in that having an option to test with a free AT is important
… but also don‘t want to _require_ orgs to test with those ATs
… because from a cybersecurity perspective, some orgs may not be able to install
… need to recognize that both might be needed in certain circumstances
<Zakim> alastairc, you wanted to comment on the balance between meeting standards (e.g. html), working for the user, how that impacts requirements.
alastairc: on the topic of what AT supports what
… we’ve got a balance of “meeting the standards”; that’s your job as an author
… the rest is up to the user agent
… last time I tested Narrator, it failed on some fairly basic things, although that was a while ago
… but imagining a situation where the quality of the built-in AT is well below what we‘re used to
… it suggests we almost need to write a loophole or lower requirement
… for example, if something doesn’t support heading structures, or something similarly basic
… then we almost need to tackle it as an accessibility support set issue, and not include that AT
… either we don’t include certain AT that aren’t up to the general standard, or we have to lower the requirements
… acknowledging that this is somewhat provocative
<Ben_Tillyer> Think what I was trying to say was - web products that are designed to be sold with a view to be installed or delivered by other people (e.g. a museum wayfinding software expected to be put on a tablet in a museum), do we need to be concerned that the vendor can't be certain that the implementer/purchaser is actually able to use the supported set
<Ben_Tillyer> of AT that was tested by the vendor.
shadi: might be a tangent
… but there have been several attempts by w3c staff and other groups
… to have an a11y support database
… and it never really worked
… because there were too many combinations
… just wondering in the era of AI
… how many of these tests might be automatable
see also Lola's accessibility compatibility data project: lolaslab/
AWK: is that you volunteering, shadi? ;-)
… but yes, we really wanted to know all this during WCAG 2.1
<bbailey> +1 for shadi being voluntold
AWK: whatever we decide
… we may need to accept that people are going to have to work around doing the best they can
… a very hard problem to solve
… have we actually had substantial problems in the last 20 years?
… if so, then let’s examine those?
… rather than just update something
<kirkwood> a problem statement would help
alastairc: big internationalization issues in terms of techniques we’ve provided not being supported in other regions
… and they’ve had to work around that
… basically trying to make “what is a11y supported” clearer and more explicit
… and give people the freedom to choose different ones
Previous slides are at https://
Key problem for any cross-platform accessibility guidelines: How does an author know that meeting a guideline will work in practice with user-agents that are used by real people?
alastairc: in terms of next steps
… we need an updated draft
… I can try to take in the conversation so far
… it’s more than 1 AT per platform
… more than one support set per platform
… which might make it clearer
… haven’t really got any better guidance on the balance between “meeting the standard” and “does it work for the user”
… but in terms of how we select the things we test with
… our default a11y support set
… I’ll try an update
… I’ll update the strawperson text
<bbailey> s/<kirkwood> thx//
GreggVan: you should add bullet: built-in AT for those who rely on public access computers
GN015: does it make sense to give qualification rules?
<Ben_Tillyer> +1 gn
GN015: like what a screen reader must be able to do?
… does it make sense to have a qualification
alastairc: that’s an interesting idea
LoriO: +1 to GreggVan
… we all need to remember that the disabled population not just in the US but around the world are some of the least served
… having to use a public computer is a big issue
<kirkwood> +1 to public computer
LoriO: many are unemployed, don‘t have reliable transporation
<scott> sometimes a perceived AT "cannot" is more of an AT "purposefully will not"
LoriO: so we must make it easy for them to access
GreggVan: the fact that you can’t get a hierarchical list of titles
… that’s usability; not accessibility
<Frankie> +1 to Gregg and Lori's point. Reminder that many governments require abject poverty for disabled folks to access disability benefits.
GreggVan: we shouldn’t be requiring those as the base
<Charles> note: accessibility supported applies to more than screen readers and their features
GreggVan: we have no control over the AT; we can’t put requirements on them
… authors can’t be forced to scratch off the screen readers that don’t support features because some regions _only_ have ATs that don’t have those features
… must define the bare minimum for things to be accessible
alastairc: any final thoughts?
giacomo-petri: I think the minimum bottom line is quite tricky
… we have a lot of requirements that ask much more than that
… I’m thinking about my All Steps Listed provisions
… which I’m considering moving to an assertion
… depends on user needs
… not easy to define this “minimum” level
alastairc: I‘ll take this away and create an updated draft
Rachael: we’ll send out a survey for conforming alternate versions
alastairc: alright, we can adjourn
… thank you, everyone