Meeting minutes
<chiace> + present
Agenda Review & Announcements
matatk: Some announcements, actually.
Baseline and accessibility
matatk: Baseline is the benchmark set of features that are supported in recent versions of major browsers and so it is regarded as safe to sue features. Nothing too new or experimental, but stuff that is expected to work reliably across browsers.
matatk: It doesn't currently take a11y support into acccount.
matatk: Baseline featuers need not, at this point, necessarily work with AT.
matatk: The goal is to include a11y information in what is regarded as baseline features.
matatk: It is tricky to gage what is actuqally supported: there's a lot of variation and a lot of patchy data.
matatk: It is hard to translate between baseline features and which parts are actually supported.
matatk: There has been some work done in this area trying to make thigns happen.
matatk: There are two upcoming meetings where this is going to be discussed.
matatk: Here's signposting!
matatk: There is a WebDX commnity group meeting.
<matatk> WebDX CG meeting on the 19th: https://
matatk: Anyone can join the WebDX CG.
matatk: There is also another one with the ARIA WG.
matatk: Further information about that is forthcoming.
matatk: Maybe we can get lola to come to this meeting too, who knows?
TPAC 2025
matatk: TPAC time! We need information from everyone in APA about which *Gs or TFs you want to meet at TPAC so that this can be finalized for the TPAC planners. This has to be finalised this week.
matatk: We have at least three thus far.
matatk: Very important stuff.
matatk: This is not the same as breakouts: this is working meetings with dfferent groups and task forces.
Update on horizontal review process improvements
matatk: A couple of weeks ago, Ananya showed the work she has done on the check lists.
matatk: See that meeting minute collection for all the lowdown on that.
matatk: The TATG one is close to finished; the FAST one is a bit further off.
matatk: THe check lists are going to make horizontal reviewing far, far easier fror people.
matatk: The bit that i have been doing is carrying out horizontal review for specs.
matatk: This is effectively copypasted from i18n's process, but theirs was the pattern for all other groups' reviews.
matatk: I'm tailoring it for APA.
matatk: Have a look at the wiki here.
matatk: Other than documenting the processes, we konwn that our repositories are named confusingly.
matatk: We are workign with W3C management to change our a11y review to make it more understandable.
matatk: That will hopefully be a transparent thing.
matatk: The URL will change, but the rest will stay, well, the same.
New on TR
Spec review requests
matatk: I have some stuff to paste in.
There is some humblebragging going on. ;)
<PaulG> I need to drop. The only CSS issue I'm tracking is w3c/
Horizontal Review Request for DID v1.1
<matatk> w3c/
<matatk> Due: 2025-05-25
@PaulG: it appears to be a major improvement if devs use it.
matatk: It looks relevant to stuff TAG is working on. We'll pick it up next week.
Fredrik: Leaving scribing duties in the capable, steady hands of the incomparable Jenn Strickland.
matatk: DIDs, decentralized identifiers, @Niklas raised this in mid-April, but we didn't close the thread request, so I think we just need to do that, I don't see any problems with it.
@matatk: Any concerns?
matatk: Then the one that Jara contributed to, …
Web Neural Network API
<matatk> w3c/
<matatk> Due: 2025-06-20
<matatk> APA internal thread: w3c/
matatk: Web neural networks, Jara looked at this one recently, and there's a recent thread:
matatk: We have several good suggestions from Jara, we did discuss briefly… and one of the cochairs of the group found this thread and was pleased with the comments. Are we happy with the way things are worded? I have a couple of thoughts and wanted to open to suggestion, as we don't necessarily have text to propose.
Note: all previous mentions of Jara s/b chiace.
matatk: The documentation looks good, there are mentions of ethical considerations including accessibility. We don't need to emphasize it in the first one but certainly do later down.
matatk: We have a number of comments about use cases, and a comment that chiace raised that the captions need to be exposed to assistive technology with a couple of specific examples -- lang is very important and the aria-live is situation dependent.
matatk: if the live update happened after a user activated a button it would make sense.
matatk: and also direct them to the WAI tutorials on how to expose alt text.
matatk: next is 2.1.11 emotion analysis. I think this is great. These days the tech isn't good about exposing confidence scores. There's a wider topic there, where the accessibility side of things would be covered if it did expose scores.
JenStrickland: Yesterday I was grateful to attend Credible Web CG and they're working on such a thing - there's a browser extension from MIT's CSAIL called TrustNet that exposes confidence ratings for web content. Maybe a good thing for this group to refer to.
… To check out what options they might recommend. I don't think that anything is quite 'ready for prime time' - implementation - but there is work happening in that direction. There are opportunities and if we can provide implementer that guidance then that would be helpful.
matatk: they are looking at content that exists on the web, where machine learning content can't tell you the confidence of its content because it just generates it. I think the project is worth everyone's interest as a whole.
matatk: 2.1.13 noise suppression. chiace's suggestion seems reasonable to me. We have two more that may have discussion.
matatk: 2.2.5 integration with real-time video. background blur can introduce too much delay and break lip reading and live captions.
JenStrickland Can we suggest not having background-blur, i.e. vestibular conditions?
matatk: There are different needs where it can be helpful. There doesn't seem to be a good technical suggestion, perhaps the solution is a social one?
matatk: Perhaps the accessibility of remote meetings document from RQTF?
Accessibility of Remote Meetings: https://
matatk: there might be something in there we can refer to. And the last one…
matatk: 6.2 device selection. Not sure what to suggest. There's a general concern about power and impacts on performance. Is this something that can be worse for some with particular disabilities?
chiace: My main concern was that people cannot access and access to real time captions, and the ability for certain hardware to connect. Also, socio-economic access is uneven. There is also an issue regarding privacy and fingerprinting. There are particulars with low latency hardware that can contribute to issues with fingerprinting.
chiace: also the battery tradeoff, where it can reduce battery life, degrade device and battery.
JenStrickland: this may also be a time to loop in the Web Sustainability Guidelines regarding device and battery degradation.
Ethical Principles for Web Machine Learning: https://
matatk: If a web conferencing tool realizes its using too many resources, and if it chooses to drop captioning, then the solution is that a user can pin a stream — no need to say why — and it depends on the app. There are similar unforeseen consequences of this tradeoff, and an opportunity for us to surface those.
matatk: feel free to expand your comment to include those and we can add the Ethical Principles for Web Machine Learning.
matatk: I think it was helpful to go through it, we understand it much more. I think we should propose the wording you've used. For the last couple we can come back to them.
chiace: I can add this when?
matatk: the deadline for this review is the 20th which is next Friday, but it would be best if we do this sooner.
chiace: That helps, and this has been an interesting review.