– DRAFT –
AGWG Teleconference

<Dirk> all the time!

Intros and Annoucements

alastairc: Asks for introductions or change of affiliation

<kevin> https://www.w3.org/news-events/tpac/2026/

<Patrick_H_Lauke> TPAC 2026 dates: 26-30 October

<AWK> When is W3C releasing hotel reservation info?

alastairc: None. Next: Save the date for TPAC. AG meetings are on the Monday/Tuesday of the week. October 26-30th. If intending to join, the 26 and 27 with be the main meeting dates. We are also booking crossover meetings with groups like APA, ARIA and ACT.

<LoriO> I just got booted off irc and had to rejoin

WCAG 2 updates https://lists.w3.org/Archives/Public/w3c-wai-gl/2026AprJun/0035.html

alastairc: Hotel reservation information is not out yet; will be released to several lists, and will be announced when available.

alastairc: WCAG2 updates - handing it over to Patrick

Patrick_H_Lauke: Repeating announcement: we're in the first week where we send a list of potential updates to the working group. This was done last week. One more week to comment. Pre-CFC. There's a list of potential changes. It has a 'special' request, meaning an either/or. There's a wrinkle in 1.4.11 and includes an exception for logos, one for
… graphical objects and one for user interface components. The exemption is only in the graphical object parts. The question is does the exemption bubble up or get inherited if that graphical object is also used as the primary content of a user interface component. Apart from that the other pull requests were replies to the comment-only ones where
… they are straightforward as they normally are.

AWK: Suggest that this is something that might be worth discussing this on this call because doing this in GitHub is a big onerous. It's important for people in the working group to read and understand this.

<bbailey> https://lists.w3.org/Archives/Public/w3c-wai-gl/2026AprJun/0039.html

<Zakim> AWK, you wanted to ask how the actual CFC will go? (e.g., inidividually vs all together)

alastairc: there's asynchronous work in GitHub threads to get everything down and discuss things. If we get stuck, we could pull that back to a future meeting. I encourage people to check the GitHub conversation in the meantime. To describe this for new members, we don't run calls for consensus that often. A pre-CFC (pre-corporate consensus), we
… are just about to get a call for consensus because they are a normative update, or they are things that change the WCAG spec. In this case, the criteria is if in the future, we did an update to WCAG, this is a way to go through that process. If there's no objection, we'll go through CFC. If there are concerns it would be better to discuss that

first.

AWK: Wanted to ask how you expect the actual CFC would go? I do know there are some of these in there that, as far as normative changes, and I don't think they are good ideas. Do you anticipate the CFC is going to be all in one question or broken out into individual items?

Effective participation https://github.com/w3c/wcag3/wiki/Effective-Participation-in-the-AGWG-Meetings

alastairc: It depends. If everyone is happy with everything, we'd just do it in one. I'm not anticipating that to be the case. There's a bit of it depends, we'll try and group things together as best we can to give the email list the least amount of traffic.

<Patrick_H_Lauke> additional to what alastair said, there MAY be a survey/questionnaire before it goes to actual CFC to allow for further gauging of interest/pushback

alastairc: Circling to retrospective. We have a few people that have asked where some of the output from that is going. Hopefully everyone knows about the positive Work Environment Code of Conduct. Asks for people to scribe once every two months. [Above is a link that contains content reviewed]

<Zakim> Rachael, you wanted to give a bit of context after Alastair is done

<Zakim> AWK, you wanted to inquire about AI scribing for conversation and human scribing for resolutions

Rachael: We put this together primarily after the retro to address a lot of the questions and concerns that came up during the conversation. We wanted to make sure that people could see that the chairs will be managing to try to make the conversations more efficient.

AWK: I can appreciate the difficulty of getting people to scribe. I think there's a few people who like to scribe. Has there been any further discussions or thoughts with the W3C about allowing the transcript that's being created by AI to describe the scribe instead of just doing resolutions instead?

alastairc: There have. One problem is that there are members of the group who can't attend if there's some AI running that is sending their content to the cloud. There's a corporate privacy, security that is one of the blockers.

AWK: I think that is one of the things that's worth finding out.

<Zakim> wendyreid, you wanted to mention zoom's different transcripts

<kirkwood> with you 100%… never have been able to change it yet

<LoriO> kirkwood +1

AWK: It would make things much easier for people to agree to scribe, but also, I think some of the people who can't scribe is because there's a difficulty of sitting there typing continually for an hour and trying to pay attention.

<Stephanie> Given that W3C is intended to operate as a vendor-neutral standards body, it feels important for W3C to do what is best for the group, rather than workflows being shaped indirectly by the internal policies of individual member companies.

wendyreid: There are two zoom transcript systems. One is AI companion, which may be the one that is the blocker - it literally involves using AI to create the transcript. There's also a transcript feature that's been in zoom for a very long time; it's voice detection and processing and it's not summarizing. The transcript needs to be cleaned up
… after the meeting for the content. Time may be perceived as saved, but with the cleanup afterwards, it's something to consider.

Kevin: In response specifically to Andrews comment, yes this is a public meeting. However AI data processing and where geographically that it happens is relevant to internal policies. For some people, being unable to state exactly where the data is being processes, or being able to state it and it not being int he appropriate control, it precludes
… certain people participating in the meetings. Using a transcript and cleaning it up does create a specific burden. There's no simple AI solution that will solve this for us at the moment.

GreggVan: Back to Andrew's suggestion as I understand it. We have minutes that only have resolutions in it with timestamps. We accompany it with the transcript with a timestamp. You can see what was decided at the meeting. If you want to know how the decision came out, you can go back and look at the transcript, and that will accurately have what
… people said, versus the minutes. It catches half of what people are saying. People might try to read that instead of the long ones. There's no data processing on that. There's only the transcript and the resolution that's hand typed by a scribe.

<graham> I am watching the transcription as we speak - so if that is an issue we should switch it off.

GreggVanIf AI is only being used to transcribe exactly and not aprocess. All voice recognition does that and everyone accepts that it's analyzing.

alastairc: Not sure that is true, if so we'll have to find that out.

stevef: Wanted to support Alastair, to edit the transcript to be a wailful reflection. I don't see there's a lot of gain.

Policy document audience https://www.w3.org/wbs/35422/policy_audience/

alastairc: We'll keep an eye on it and it's a question that comes up often, and we can put something in the wiki about it.

<Patrick_H_Lauke> human transcription frontloads the work. machine learning /voice recognition transcription requires post-processing human work. no silver bullet...

alastairc: Survey: the question was about a Note for Policy Makers. Originally when proposed, we were thinking of regulators and on the legal side because of instances where WCAG has been incorporated into legislation or regulation and not in the best way. To check people's understanding of what policy means in this context. There were interesting
… answers in there. For example, Mary Jo Mueller's comment (see description in link). The main upshot I took away from the survey is which audience we thought should be supported as an audience by this policy note. Almost everyone said 'both.' There's a few comments around the 'how.'

GreggVan: We keep talking bout policymakers, but I noticed on there you said policymakers, or lawmakers, last regulators. It's important to remember that its not just the policy, but the regulation that can be as critical, and the regulation is typically where you find things like 'undue burden' And also think about how you enforce it that ends up

being really where the pain comes in.

<bbailey> +1 to GV point about Undue Burden being example of what’s in regulation versus what’s in The Law.

alastairc: On this topic, there were a few comments worth picking up. See Laura Carlson's comment. I thought where we had a skeleton of what topics would be included. Had a feeling it might be best to go topic by topic, but each topic could be threaded. You might have regulatory section and an organizational section, which are almost like two sides
… of the same coin. We can be flexible at this stage, and we'll be setting up a task force to continue this work.

alastairc: From Shad, most of the garnets is typically applicable across a wide range of different types of policies, so that was positive for keeping it together.

GreggVan: think about lawmakers, regulators, enforcement and organizations as four separate categories, they all have different roles. personally noticed that enforcement gets dropped off here.

<kirkwood> For example NYC (if useful): New York City's web accessibility policy mandates that all city agency websites and digital content conform to the WCAG 2.1 Level AA standards. Governed by Local Law 26 of 2016 and guided by the Mayor's Office for People with Disabilities (MOPD), this policy ensures digital platforms are fully accessible to New Yorkers with disabilities.

alastairc: Last question was on the kind of objections (see survey responses). Points out Jeremy Katherman's comment; Jeremy is not present in the meeting to comment.

alastairc: Calls out good comments from Laura, Mary Jo and Shadi. Comments on Shadi's comment. It wasn't about exclusion, but it has expanded the scope of the work for this task force. If one of the chairs can find a link to the survey for people doing that task force, that would be helpful.

<alastairc> Draft RESOLUTION: The AGWG will work on a policy guidance note aimed at policy makers, both at the legal/regulatory level, and organizational level.

<Rachael> Policy participationhttps://www.w3.org/wbs/35422/policy_subgroup_26/

alastairc: Asks if this is a reasonable resolution, or if there are any updates to that?

<Rachael> Policy participation https://www.w3.org/wbs/35422/policy_subgroup_26/

<SydneyColeman> present_

GreggVan: You have legislative, regulatory (which are two different levels) and it seems enforcement is left off again.

alastairc: What kind of organization would count as enforcement or role within an organization would encounter enforcement?

<alastairc> Draft RESOLUTION: The AGWG will work on a policy guidance note aimed at policy makers at the legal, regulatory, enforcement, and organizational level.

<alastairc> Draft RESOLUTION: The AGWG will work on a policy guidance note aimed at policy makers at the legislation, regulatory, enforcement, and organizational level.

GreggVan: Enforcement would be judges. Judges and police are enforcement. For most things, the FCC enforces for telecom, and there are agencies that have enforcement power for different aspects in the United States and Europe. If you don't follow the regulation, who comes to your door and does something to you?

<bbailey> +1

<Rachael> +1

<JeroenH> +1

<HaTheo> +1

alastairc: I thought that was legal; doesn't see a problem with separating that out.

<Stephanie> +1

<kirkwood> +1

<stevef> +1

<Patrick_H_Lauke> 0

<laura> +1

<sam-estoesta> +1

<LenB> +1

<Dirk> +1

<ShawnT> +1

<Detlev> +1

<graham> +1

<Monica> 0

<giacomo-petri> +1

<AWK> +1

<Charles> +1

<Francis_Storr> +1

<Adam> +1

<Jon_Avila> +1

<Heather> +1

<Makoto_U> +1

<GreggVan> +1 but not sure of the word level more like aspects for dimension

<LoriO> +1

RESOLUTION: The AGWG will work on a policy guidance note aimed at policy makers at the legislation, regulatory, enforcement, and organizational level.

alastairc: Rachael put in a link tot he policy participation task force at :37 minutes past the hour. If you're interested, please fill that in.

Scoring as a way of measuring progress not as a way of defining a conformance level https://www.w3.org/wbs/35422/scoring_question/

<kirkwood> sorry late: Department of Justice (DOJ): is the enforcer in US

<CClaire> 0

alastairc: Moving on. The other survey we had was on the use of scoring. This is a continuation topic that was talked about last week. On that scoring topic, to give background there were a few questions in the survey and responses. We've had plenty of work on this, and just this year, we had 7 proposals from the subgroup. Starting with breaking
… things down by feature. This is to help get us through and eliminate possibilities, which is as useful as getting agreement on possibilities. Things to remember: We wouldn't take paths are sufficient for conformance; wouldn't automate testing as a level, usability testing wouldn't be required, but could be a method of testing particular provisions.
… We agreed to test out the scoring and multi-level approaches. Last week we went through scoring.

<alastairc> https://github.com/w3c/wcag3/wiki/Conformance-Proposals-2018-to-present

alastairc: [Reading survey responses] Calls out Stephanie-Lee Steer's, Shadi Abou-Zahra's comments. Pushes back on Shadi's comment; we've had a lot of scoring proposals for conformance models that include scoring. Put the link above. One of the outcomes from last week is that we don't know how well a reasonable score could be used to line up with

the experience from a disabled user's point of view. That's what is harder to assess. Even at an aggregate level, it's difficult to see. I think there's a path to working that question out. The current question is that assuming that the score represented something meaningful, how would we use it? that's what the survey was trying to get to.

alastairc: Calls out Grace's comment. Again, devil's advocate. A lot of comments we've had would be having something that's standardized; like a standardized baseline.

alastairc: Calls out other's comments on the page. Pauses on Wendy's, and comments that he doesn't think it's necessarily any alternative view that conformance means meeting all the provisions. Certainly agrees with building on and am interested in how scoring could work if it was geared toward levels of performance o grouping provisions with
… specifically usability goals.

alastairc: Continues to read comments in survey responses. Severity was noted last week; we don't think we can use that.

alastairc: [Continues to read comments from survey link]

GreggVan: When we're talking about weight there are two kinds of weighting. One is contextual. For example, this error has more impact than that error. There are varying results from users. The other is if you have cognitive and physical and visual and hearing, here are the provisions that we have. Depending on the characteristic, there are some
… that are more critical than others. Once that is set, you would have inter-rater reliability because you're all using the same weighting factors. Not sure if the disability groups would think that one was good or bad.

<kirkwood> When explored before, and i foundationally agree scoring is not possible. except for progress over time (potentially). But that’s not our responsibility

alastairc: Trying to get a feel for what the group thought we should or shouldn't use. For most people, if they were in favor of scoring, have selected both. Thinks 3 out of 4 people who said scoring should be used in WCAG also ticked one or the other options, and is a little confused as to what that would mean. Continuing with exploring having a
… score I think we can tackle for now the assumption that it is feasible to do scoring and it provides that scores better generally to be more accessible to people.

GreggVan: If you put any scoring in WCAG 3 document as proper, you'll find that people may conflate it with conformance.

<kirkwood> scoring depends on what the ‘test’ is.

<laura> +1 to Gregg

alastairc: Let's tackle that with a poll.

<alastairc> Poll: If we do have scoring (subject to testing), it should be separate from the main WCAG 3 document, e.g. in a note

<GreggVan> +1

<laura> +1

<ShawnT> +1

<Adam> +1

<Stephanie> +1

<Monica> +1

<bbailey> -1

<Jennie_Delisi> -1 due to people not going there

<AWK> -1 we should explore

<GN015> Does it mean it is neither usewd for defining conformance nor for reporting progress?

<JeroenH> +1!

<stevef> 0

<Rachael> 0

<CClaire> 0

<Dirk> +1

<Makoto_U> +1

<LoriO> -1 need more information how it would be presented and how it would be used

<Detlev> 0 still not sure what we have actually agreed on reg. conformance

<bbailey> i am a 0 then

<LenB> 0 - people tend to 'dismiss' notes because they do not understand the role they play

alastairc: It's not used for defining conformance, but it could be used for reporting progress. Asks for the -1 responders to get in queue.

<jtoles> 0 ok with exploring both for now

<Charles> 0 until conformance model resolved

<AWK> +1 to Bruce

<Jennie_Delisi> +1 to Bruce

<Francis_Storr> +1 to Charles' comment about wanting the conformance model resolved

bbailey: Would like to see the scoring not necessarily part of the body of WCAG 3, but certainly endorsed from the body. Having a uniform scoring would be fantastic to promote accessibility. If it's not part of the body it may not be taken as seriously.

<Zakim> Rachael, you wanted to say that we could postpone this decision

<HaTheo> +1, scoring if it exists should be separate

GreggVan: We could list it at the front that we have this other document.

<bbailey> +1 to that scoring can be included without attaching to levels

Rachael: Chair hat on. One of the goals with this survey is to try to get a clear decision about whether scoring should be used to set levels because we seem to have some trending towards that in our last conversation and per our retrospective, making decisions about what we don't want to do is an important part of this as well. We could postpone
… exactly where we write it up. I think we could make a decision about the levels portion without necessarily deciding finally one way or another about whether it goes in the conformance section or whether it's referenced somewhere else. I can draft that up.

GreggVan: Are you talking about the progress toward conformance or using percentages and levels?

<Detlev> that was unclear Alastair

alastairc: Greg, your proposal on progress towards conformance wasn't to do with levels, however the current working draft that we have has levels from a percentage basis. People were also considering it as part of various conformance proposals. this is around eliminating that as an option.

<kirkwood> note scoring often results in gaming. also doesn’t work in a legal settlement

GreggVan: This seems like it's a different discussion. I don't think we can use percentages. People who use percentages to claim conformance negate the provisions that may be important to people to be accessible. Civil rights falls apart if you say it's just that people have to make an effort for the basis. The minimum should mean the minimum.

alastairc: It's hugely negative to use percentages in that way.

<alastairc> Poll: Scoring will not be used to set regulatory levels within WCAG 3

kirkwood: I think you were just about to say that it's very difficult in settlements when you have someone from a particular disability group that has sued someone because it's inaccessible to them, and this community, to meet our regulations at this level.

<JeroenH> +1

<GreggVan> +1

alastairc: Put the poll up.

<Rachael> +1

<bbailey> +1

<Detlev> 0 can't say without more clarity of what our conformance model will be

<ShawnT> +1

<LoriO> +1

<LenB> +1

<laura> +1

<Heather> +1

<Makoto_U> +1

<kirkwood> +1

<Jennie_Delisi> 0

<Stephanie> My concern is that a scoring model would give vendors something to hide behind by creating the appearance of partial or sufficient compliance, while offering little legal value because most compliance frameworks require a clear determination of whether the product conforms to the applicable accessibility standard. Given that vendors already use

<Stephanie> VPATs and ACRs to report conformance, I am also not convinced they would adopt or meaningfully engage with a new, more complex scoring approach.

<AWK> +1 to Giacomo

<GN015> publishing roadmaps might not be allowed for creators of business software

<Zakim> alastairc, you wanted to talk about vpats and scoring

giacomo-petri: The idea of the core requirements is to emulate WCG to AA. The idea of a score is useful to show progress towards an organization taking action as opposed to the pure black and white model we currently have to conformance.

Stephanie: From a procurement perspective, accessibility scores would largely duplicate information already provided through VPATs and remediation roadmaps, adding complexity without clear benefit.

<Zakim> Jennie_Delisi, you wanted to say not completing can be part of a score, and levels

Stephanie: feels that this adds complexity to an already complex and nonuniform process.

<Stephanie> Agree, should be separate

Jennie_Delisi: believes accessibility scoring could support procurement and policy enforcement by giving organizations a measurable baseline requirement alongside VPATs, making accessibility performance more visible and accountable.

<Detlev> is that documented somewhere , what you are just saying?

<SydneyColeman> of course

<alastairc> draft Resolution: Scoring will not be used to set regulatory levels within WCAG 3

alastairc: scoring will not be used to support within W3C

<kirkwood> +1

<Stephanie> +1

<LoriO> +1

<HaTheo> +1

<bbailey> i don't like "regulatory"

<GreggVan> +1

<Rachael> +1

<bbailey> but +1

<Adam> +1

<laura> +1

<CClaire> +1

<AWK> Explain that last bit again?

<Heather> +1

<Charles> +1

<ShawnT> +1

<bbailey> ...used to set formal levels within wcag 3

<AWK> So levels that are equivalent to what regulation currently doesn't pay attention to can use scoring?

<GN015> Any level might be used in regulations.

<stevef> 0

<Jon_Avila> +.5

<Detlev> 0 so we are back at what WCAG 2.0 is with all the issues.

<AWK> Still think we need something better than what we have now. -1

<giacomo-petri> +1 to detlev

<AWK> You're not alone, Sidney

<Stephanie> agree with sydney

<Zakim> GN, you wanted to ask whether regulatory level is equivalent to conformance level

<SydneyColeman> completely confused

<Stephanie> My brain is already seeing Bronze as Level A, Silver AA, Gold AAA, its like we are trying to reinvent what exists?

<GreggVan> instead of regulatory level --- perhaps say conformance levels we think will be used in regulation

<SydneyColeman> it is another way of requiring 100% conformance

<Zakim> AWK, you wanted to ask if this also means ruling out having more scoring levels to reflect the level of accessibility with better granularity

alastairc: clarifies that what we are discussing that scoring will not be used to determine the primary regulatory conformance level.

<SydneyColeman> 100% provisions to meet a level of conformance (bronze, silver, gold)

<Stephanie> This is my concern playing out in real time, this adds complexity to an already complex process.

AWK: proposes figuring out this issue and deciding on the levels, as there might be a dependency here.

<Zakim> Rachael, you wanted to suggest we may need to make this more granular

giacomo-petri: If scoring isn't used to define regulatory conformance levels I can live with it, but emphasized that WCAG 3 should clearly communicate that conformance alone is not a sufficient measure of accessibility, and suggested that WCAG 3 should include additional ways to evaluate accessibility beyond conformance status alone.

GreggVan: Agrees that no scoring system can perfectly reflect real-world impact, the group should focus on defining the most practical and defensible regulatory approach.

<Zakim> Rachael, you wanted to respond to sydney

<Zakim> Rachael, you wanted to say it sounds liek we need a definition of scoring

<alastairc> acl alastairc

Rachael: clarified that the group’s current goal is not to make a final conformance decision, but to identify and eliminate options they are reasonably confident they do not want to pursue, thereby narrowing the scope for future discussions on scoring and conformance.

<Zakim> SydneyColeman, you wanted to ask is scoring meeting a certain level / minimum accessible baseline? i.e. 100% provisions met so now you meet gold, silver, bronze

<Charles> i understood ‘organizational progress toward conformance’ (behavior, actions, etc) to be a different topic entirely than ‘what equals conformance’. i don’t care if they both use a score as long as they are disambiguated.

<Zakim> wendyreid, you wanted to say we can't separate these conversations

<Stephanie> +1

<Dirk> +1

<Charles> +1 to confusion with above note

<AWK> +1

<janina> +1 to confusion!

<ShawnT> +1 I'd like to see working examples

<AWK> Need to drop for doctor appointment.

<Detlev> you said that already Gregg

<alastairc> I don't think so, it's a different point on levels not scoring

<giacomo-petri> typo

<SydneyColeman> so bronze is the minimum baseline

<Detlev> +1 to Rachael

<GreggVan> not sure what we say is the base required. -- I should not have said bronze since we have not named anything yet

<Zakim> Rachael, you wanted to talk to quality

alastairc: next week we will be looking at level as our main topic and the WCAG EM.

Rachael: [Group expressed that there is some confusion.]

Gregg: Percentages don't work because they, by definition, say that any X% can be skipped. And if any provision (except the "four" you always require) can be skipped -- then one is admitting that none of the other provisions are important enough to actually require them. That is - a site can be minimally accessible with any

<GreggVan> Percentages don't work because they, by definition, say that any X% can be skipped. And if any provision (except the "four" you always require) can be skipped -- then one is admitting that none of the other provisions are important enough to actually require them. That is - a site can be minimally accessible with any (other than the "four") not done. So there is no real basis for saying that ANY of them are critical for accessibility since ANY

<GreggVan> of them can be skipped and still pass

(other than the "four") not done. So there is no real basis for saying that ANY of them are critical for accessibility since ANY of them can be skipped and still pass

Rachael: [chair hat off] It isn't just percentages. Part of the scoring problem is ambiguity. If we have scoring that is based on whether something passes or doesn't pass that isn't ambiguous. Also, some of our provisions are based on whether something is present but others are based on quality checks. Percentages of quality checks may accurately reflect additive issues