Meeting minutes
Accessibility of Machine Learning and Generative AI draft.
janina: my paper jumped to priority, still not done. Reformatting.
jasonjgw: when do you think it will be out of the way?
Janina: maybe today or tomorrow. Will try again for next week
jasonjgw: topic remains on agenda, unless you tell me otherwise. always at your discretion to move it a week
… work is ongoing and we will revisit it next time. Any other comments?
Game accessibility.
Jpaton: this was triggered because of the white paper - legislation in the UK, EU, and US. Recommendation we could potentially create game user a11y requirements, The paper points to having standards that helps us know what a11y looks like, partially informed by a11y dev kit that we've started a couple of years ago. I think that user requirements
can be useful. When we have collaboration and other people commenting and adding, can form a good basis for standards later. Intersted in starting and getting a first draft. What are the hurdles to go through to get that going? What do we need in terms of sign off?
Janina: we can start a draft in github. A repository for it, so Jason - we need to ask Roy for that right?
Jasonjgw: yes
Janina: might need APA approval to get started. Certainly need that to finish and publish and working draft and final...I suggested to Matthew and his initial reaction was do we have time/lots to do. And we have someone who can take lead, and he asked don't we have this already? So we need John to really respond...Maybe we have some convincing to
do. Get a repository created then all we need is to start writing.
jasonjgw: agree. and I'm one of those who needs to be convinced, as the case would have to be developed. There are a number of aspects to this, existing game related guidance and existing w3c guidance, and need to know what the gaps are in that, especially in respect to web content accessiblity guidelines, and aspects for w3c to get involved. At
least there aspects, maybe a fouth - technologies that haven't been developed or need to be developed further. Term gap analysis comes ot mind here, and that might be a good place to start here. What are the issues that haven't been addressed adequately elsewhere and inform a publication.
Scott: I think that people often ask (in their org) on accessible gaming and remakes are done but newer ones do...I do think that there is a place for w3c to have this guidance, but I don't think anything much comes up if you google w3c gaming accessibility...I do think this has huge potential and people do look for anything under w3c and having
that credible source of truth in this area. I don't need convincing personally, there have been standards work in this space, but to tie into accessibility user requirements I think it can be a powerful publication.
jpaton: yes, in the main ones able gamers has a toolkit, game accessibiltiy guidelines is the biggest one. It's quite hard to make the easy link between user requirements and user guidelines. Very good, and cover a lot of ground. In terms of ... the aim for us is a stepping stone to standards. And i don't think there's anything out there to create
the stepping stone. I recognize a proper documentation for this to debate this is needed. WCAG is the closest. It's used in Section 508, saying that WCAg is trying to move all friction where gaming requires friction, but WCAG has exceptions for if you need it. Like timers, IF you need it, keep it. Important point...trying to remove all friction
would smooth out the game so it's no longer game. Existing guidance doesn't quite fit. Future version s of WCAG would more likley be more applicable to gaming, but i think something bespoke is needed. Coming out fo this group won't conflict with WCAG but used to inform WCAg and check and make sure still meeting that. Scope - technology...I think
there's a lot of games that are online but that would be my big concern. Future technologies, in a way user requirements aren't pointing to "this is how you fix it" - but in terms of "this makes a good user experience"...setting up a roadmap for this is what needs to be solved. Not to point out the user needs themselves.
<janina> https://
Janina: I found some stuff...
I didn't read it, but we have something in WAI called gaming.html
<janina> aming Accessibility Workshop in 2018:
<janina> https://
Janina: my search string was gaming accessibility and refined site: w3c.org. So returned just w3c. Not having read them, here's what I'm hearing, John really wants to take a look at this. I think we're here to facilitate people doing good work. IF you're willing to take the lead, we shoudl go rhgouh the process. First is review of prior art. Don't
limit to w3c, but start there. Gap analysis, we don't need permission for that. Lay out what's out there, what's covered, what's missing, and what would fill that. I like everything that's been said...but thinking of WQCAG 3 and challenge laid down to 3 and how current conformance model isn't human so telling us to make websites perfect is a fools
errand (my words) so how do we define this for percentages ...maybe for instance, talking about astronomy and doing educational materials and making them accessible and can make them quite accessible but not for all disabilities. Can we build fences that way..anything more than just basic image descriptions would really be difficult, if it's image
manipulation and enhancement and discrnment and beyond what a scope of what a screen reader can take on. Music history and music appreciate - i don't know how ot facilitate that for a Deaf person. I can do a screen reader...maybe we write every rule as if it's for everybody but only appliues in certain circumstances overlooks appraoches. Gaming
takes a different approach. A screen reader won't take part in a shoot em up game per se...we may have a prorotype here for how to think about it.
jasonjgw: in a situation where not every game is going ot be usable by everyone, but might be rquirements you should/can meet. Some games or relevant games accessible to that individual and specify something looser that is technically feasible and socially desirable. In favor of gap anayisis. John in good position to start working on that, we we
can comment and assist on that. Start out on that, in what exists in and outside of w3c and sharpen scope of work.
jgpaton: sounds good
Scott: John, to follow up. IN terms of scope. I don't think it's an issue if the advice can be applied offline. If good user requirements, i do think there would be no reason they don't have broad applicability...question, when you look at other user requirements, do you envision what you want to put teogehter fits that model and same type of
structure?
jgpaton: I think so. Need to draft something and see how i'd do it. Let's say you like game...let's say you like sports. It's a broad category. Abstract it so you can have one document and not a whole internet written on it. Enough info to inform the choice without telling the user what to do. If trying to make a puzzle game like sudoku, try and
provide info about the other numbers without telling them the other numbers. Look to try and mirror that in the topic, that these are the user requirements and how to try and do that. Point to WCAg often when talking about tv interface design. It's parallel to web design like contrast ratios...and most tvs are web pages on scaled down browser. We
can go outside. whether we can link it well enough to web technologies well enough..
Scott: as the gap anaylisis...when one of the points on screen magnification and 36- degree view and a map in the zoom. that's directly related to game and first person shooter games. Lots to draw on. I think there's a lot of stuff we could draw into this. Great initiative.
jasonjgw: facilitator perspective. another link is with the XR user requirements and games are one of the principal applications of XR so some synergy between XR and game interest.
Janina: when you think of games, it's hard to define. such a variety. things you play by yourself, spectator, scrabble...an opportunity to talk about human function and to what degree we accommodate content function. We don't do that as nearly as well. Opportunity to try that out. Everything humans do is a way we game
Scott: we reference a lot of technologies in how we work, I don't know if we reference play. this can cover ground we covered before, but overarching concept of play and how that ties into accessibility.
… XAUR
Janina: Have in gap analysis. As much as relevant. the links i shared. anythign we find anywhere else that's been published. And we have something to say that hasn't been said, and be ok to say this is anexplpration of human functionality and how to accommodate and that we'll never fully accommodate that. Different angle, and very powerful.
jasonjgw: john - do you have a sense of a starting point?
jpaton: yes. i think so
… i'll come back and see if you see any gaps
jasonjgw: see how it's coming in a week?
Janina: start it in the RQTF repository?
… or the wiki...either one is fine.
jasonjgw: let's get that worked out.
… let's start getting it documented and work out logistics as we go. will put on agenda for next week
Scott: Thank you John. This is great work and excited to see where this goes
Miscellaneous topics.
janina: ran into an audio captcha i couldn't solve. Apple had an audio captcha. works as follow,s numbes you write down are two digits. actual numbers from 00 to 99. and they put them over a noise floor where the ratio is challenging especially the frequenqies and have a syntehtic voice wispering the number.
<janina> http://
Stacey: (asked info on CAPTCHA in general, Janina shared the publication link)
Janina: gotten to the level of the ink blot
jasonjgw: look at existing guidance and see if any advice to Janina for the publication
jasonjgw: i don't know if i told the group, have opp to try health related LLM tools. Came via service provicer that performs routine medical diagnostic tests and interact with LLM to gain explination about medical test results before discussing with a doctor.In my case, it delivered the correct info, and I would say it was moderately useful,
especially... not set up as a table but as a linear list of values, so having the AI tool available to summarize and highlight the important issues in the results was useful. if the original table was accessible, the AI still would have been somewhat useful. benefits - expected to discuss results with an expert human, so an expert human check on
that.
… everything was consistent with what the machine model had said
jasonjgw: i had a tech issue with a software update, asked one of the LLMs about it, all potential causes were wrong, but among them was a line of code that could be used to find a relevant module, and when i ran that, it showed it was looking in the wrong place, and how to fix it.