Meeting Minutes
Ned Zimmerman: https://docs.google.com/document/d/1--4RetwmMJwNmO10lFZM77iTmWUpNhfdQWSbUsH2_T4/edit?tab=t.ui17glvke3eb#heading=h.xzunqhnjq68s
Wendy Reid: Yeah, Ned's fast.
Lisa Liskovoi: Okay, right, it is all coming back to me. so one thing we talked… and please, if folks can jump in to…
fill in any gaps. So we had a look at the definitions that exist in ATAG and started to break down the things to define into definitions that need updating. So things that have existing definitions and definitions that need to be added.
We talked a lot about the roles. So one of the things was for some of the bigger roles, adding examples of what they mean. So particularly the platforms, the two types of platforms.
We talked a little bit about other things that need to be documented beyond the standard, the use cases, the roles and the definitions.
And then we had a bit of a discussion about who the audience for ATAG is. And we talked about how there's a primary audience, so the actual authoring tool developers, but that there's also a secondary audience. So similar to WCAG, the primary audience would be the people who are creating content, but we often see people using WCAG to evaluate tools as well. So that's kind of a secondary audience there.
Anything that I'm missing?
Miriam Fukushima: yeah, we also talked about that, it… just for defining the roles and for understanding and seeing what's actually necessary to put in the standards, that it's maybe good to, put all the examples for all the roles in. We didn't get through all of them, but,
also document the examples that we have so far so that the definition is based and that we don't reiterate that over and over again.
Wendy Reid: Right?
So, why don't we continue on the,
Definitions, roles…
discussion, since I think that is, really important. I know it looks like, we talked about content generation, and I guess,.
the second definition, the one, and I think we talked about this a while ago, when it came up, is that ATAG 2 has a definition for prompt.
For the for prompt or prompting.
And of all the things that have changed since ATAG 2, that has now come to mean something completely different than — well, not completely different, but definitely much more different than it did.
Back then.
And we probably essentially need to create a new definition for what ATAG 2 meant.
in the first place.
And then an updated definition.
for prompting, as in the 2026.
definition of the word.
Jutta Treviranus: Or a new term as AI has appropriated the old meaning of prompting, right?
Wendy Reid: Yes.
Because the ATAG 2 definition for prompt or prompting is any authoring tool initiated requests for a decision or piece of information from authors. The term covers both requests that must be responded to immediately.
Example, modal dialog boxes, as well as less urgent requests.
Underlining a misspelled word.
And I am immediately blanking on what could be attached.
Miriam Fukushima: Yeah, I don't… I can't come up with a one-word definition, but maybe just like that, requests from the authoring tool to the user to author.
And then give the examples that are already stated, like models, dialogues.
Jutta Treviranus: Yeah, I wonder whether we want to go more broadly to create a term that is… I mean, the intent of the use of prompting in ATAG 2 was to nudge the…
the author to do the right thing, right? To support and guide and nudge,
so it… even that… that definition, I… I… I should have re… reviewed it before, but the idea was, how do you influence the… the, author to…
do the the accessible thing versus something else. So it was providing that scaffolding support. But also the dialogue requests would be for to provide things that are necessary for accessibility.
Miriam Fukushima: Yeah, I think, the…
definition should be broad enough to not only reference accessibility decisions, but also other decisions that are maybe just in the function of the authoring tool itself regarding creating the content and.
Not just to make the content more accessible, but any decision that is prompted to the user.
Minus the prompting term.
Wendy Reid: Something that comes to mind is something almost interventions.
In the sense of, the tool is asking the author to intervene in some… the tool is intervening in some way on the author's…
actions. Those interventions are optional to deal with.
Ned Zimmerman: I wonder about that, though. I'm just thinking…
Like, the definition here…
requests that must be responded to immediately., one thing I'm thinking of is an action like inserting an image into a content editor.
Where, usually, when you have, a insert image button.
It puts in a placeholder and then there's a control or a button, a file upload or image selector that you then have to use to actually choose the image.
And so that's not a situation where it's optional.
It's more of a…
Like when you take a certain action, in some cases, the prompt definition refers to the subsequent actions that you have to take to complete the task you started.
But there are also situations where it means encouraging you to look at something or review something. And so I kind of feel there's almost…
There's two different contexts in which it's used in some ways.
Probably muddies things even further, but I'm just trying to think through some of the scenarios here.
Miriam Fukushima: Okay, I think if we want to keep it in line with the previous definition that also included things underline, if there's, a misspelled word.
I don't think that asking or non-optional is enough because I could ignore the underline if it's maybe a made up word.
obviously it gets underlined, but I know it, and it's fine, and it's my made-up word for something in whatever context, and that might be fine, but still the request.
Is there…
to the author. It's just.
Okay to ignore it in that case.
And, it's possible to ignore, whereas if I,
insert, some widget that has, non-optional fields for it to be completed, then I have to complete those fields for the widget to be inserted, and then it's non-optional.
and I can't do it any other way than to do it the way it was intended.
so, yeah.
Maybe just…
I don't know.
Request to author.
Jutta Treviranus: Yeah,
I don't think the definition captured it very well, but the original intent was, we actually created a tool called Aprompt, which was not an evaluation and repair tool, but it had both functions. It was prompting, it was influencing the author so that it.
that there was a proactive action of creating accessible content. So it was preventing.
Issues with the, the, rather than, than.
post-author evaluating and repairing, so it was an alternative to checking.
more of a prompting the author to do the right thing and to create born-accessible content from the beginning. But,
I wonder if a term influence or… no, but that, of course, takes it in that direction, as opposed to a function that could be used, as Miriam discusses, for a variety of different things beyond.
Just, helping the author or nudging the author to create something accessible.
Wayne D: I agree with this conversation, but I also just think that in a certain sense, we need to say that.
Producing accessible content is the default mode.
Of the authoring tool.
And,
So many of the um.
key features of accessibility like heading structure and navigation can be handled so mechanically, it almost doesn't have to be in the awareness of the author.
Miriam Fukushima: How about:
'implicit/explicit requests for authoring actions'
Wayne D: And that there's… are a few things where we know that it's beyond the capability of automation,…
Like, all text.
And at those, then the person is prompted when you, when you insert a photograph or a.
diagram. It immediately asks you for alternative text.
And may even prompt you for a more descriptive,
Long text, either in the form of a caption, or…
or something that truly explains the visual experience. But I think that, you know, we're…
We're now in a state where we're going to have authoring tools that are so aware.
that they're going to understand the… in fact, they know it, because already they're doing it. I mean, when I write things, they do it. And they know the logical organization of your paper, and can…
can implement the markers that are needed so that a person with a screen reader or whatever facility needs to navigate your document.
So I think we need a part where we talk about what's conscious and a part where we talk about what's just given to you as the default.
Sambhavi Chandrashekar: Yeah,
I'm wondering if there are two components. One is proactive and reactive.
In the prompting.
Previously, during the WCAG two times.
Everything had to be hard coded in the program.
Whichever way it represented it, even whichever way it appeared in the dialogue between the author and the system.
Everything was hard coded. But now with AI generated content.
there is more of…
Different types of exchanges.
I wanted to just illustrate, if you don't mind, I can share screen and illustrate something. This is old, it's not AI. This is not any specific example, but I'm just trying to show this is our authoring interface. So if I want to change the font color here, I use the color selector and then suppose I selected this color.
Jutta Treviranus: there seem to be two aspects here: the purpose and the action
Sambhavi Chandrashekar: it's going to tell me automatically that it's not WCAG compliant, and as I keep moving the slider.
I get to some point where it becomes WCAG compliant.
So where does this kind of.
Prompting fall.
Is it proactive?
This is not generative AI, but is it the same, model window prompts?
Wendy Reid: Could be.
Sambhavi Chandrashekar: I'm just differentiating as while coding.
the developer would have already created all this. It'll always work the same way, whereas with AI-generated.
authoring tools using…
AI to generate content, there could be a different kind of dialogue.
Where the author asks and the system replies.
Here, the system is directing the author.
To give some prompts, to give some responses.
If you see the difference. If you could explain it.
Wendy Reid: And this might actually be part of the definition as well. We might have separate definitions for this.
Possibly determinative.
nudges.
Or prompts.
Like, you know.
misspelled word or wrong color combination.
But then there's ones that are maybe contextual or non-determinative. We'll have to come up with a better word.
Evelyn Wightman: There are only a few places in the actual ATAG text that “prompt/ing” is used - it might help to look at those usages and think how to rephrase them rather than trying to find a direct synonym based on the current definition.
Wendy Reid, Miriam Fukushima, Ned Zimmerman:💯
Miriam Fukushima: Yeah, I, I don't know if it's, if it doesn't, if it does really matter, how, if we differentiate those two cases, because the essence of both is that either through, implicit or explicit.
means the author is,
Sense a device to kind of do something.
and in some cases it's optional, and in some cases not. But, it can be, if I see it more a spectrlike the really harsh one is a prompt that you can't click away unless you do something. And on the other spectrum you have, the underline.
That you might ignore,
And the example with the color slider is kind of in the middle.
Where, you…
are morally compelled to do it, because it tells you why it's important because of the WCAG standards. But the.
Essentially, you can still ignore it,
And, the, yeah, it's by you sliding the slider and not, um.
Well, you can also not slide it, and you can click it away. But, and it doesn't tell you to slide until it's green.
So it's very implicit. But the moral obligation is there, whereas, if it's underlined, it's it doesn't necessarily have anything to do with accessibility at all. And if it's a prompt, that could also mean that it's technically required to function at all.
and also not have anything to do with accessibility. so I don't think that it matters how it's programmed under the hood, whether it's generating those.
implicit or explicit prompts to do something, or requests to do something to get away from the prompt term.
The important thing is that it's an interface between the user, in this case, the author of the tool.
Oh, well, the user of the tool to author content, and not,
It doesn't really matter what the technology underneath is, just how it is presented to the user of the authoring tool.
And that it's different from the user typing in a chat interface or an actual prompt, AI prompt to request something of an AI or an LLM.
So that we differentiate those two cases. But how the other case actually looks under the hood doesn't really matter.
At least.
Sambhavi Chandrashekar: Can I say something just to close this off?
I mean, I'm seeing a tension between…
allowing creative use, creative content generation by the author and regulating them very strongly. That's where all these discussions are pointing. So probably something along those lines.
Would be interesting.
To what extent will we compel the other to make things accessible, and to what extent we'll allow freedom?
Ned Zimmerman: I know the conversation has moved on a little bit from Jutta's last comment, but…
One word that came into my mind when Itu was describing the Aprompt tool and that sort of thought process is guidance.
and I think…
Guidance works more for the…
You know, the underline for spelling checks, the make sure you add an alt tag to your image. It can be more prescriptive or less prescriptive depending on the context, but it seems guidance fits into this in some ways.
So just wanted to add that into the conversation.
Jutta Treviranus: Yeah, I, I think it, it's become apparent there's multiple facets to this or dimensions to this, but I think, one that, certainly we had intended originally was this, this idea of what is the purpose of whatever.
Action and those actions could take many, many different forms and have many different.
means of…
I mean, the purpose was to support the author in creating accessible content, right? So for the actions that the author had a choice in, as opposed to what the tool had a choice in or determined. And so the…
I think perhaps in the definition, I mean, we might have multiple definitions, but.
I think it would be worthwhile to have a definition that that captures that purpose, and that support can be either, as you say, strong support or strong influence versus.
a simple, gentle nudge towards or… and the necessary supports, so that they… that the author does do the right thing as we define it. So guidance, nudging, influence, there's lots of, yeah.
Wayne D: I'm wondering, is there any reason ever to create a document that's not accessible?
Any good reason?
I mean, it's it seems like it's writing a program with a built in bug. If it's.
And if it's not accessible.
Sambhavi Chandrashekar: Ignorance.
Wayne D: But I mean, that's just, you know, writing a program with a bug in it is ignorance. So I…
Jutta Treviranus: Or, someone determines that their view of accessible is different from ours, or we've missed something that they feel they need to add, etc. I mean, yeah, it… if we… if we had a perfect notion of what is accessible, that…
encompassed everybody's requirements and all purposes of a document, then.
yes, I mean, I would agree with you, but I think we are also imperfect in our determination of what is the best way to author something to make it accessible within a whole variety of contexts.
Wendy Reid: Yeah, I think that's a really good point.
I didn't put my hand up for this, but I think it's applicable. There was recently some online discourse around.
in the book publishing world around authors using big words.
being ableist.
and the idea being that, to ask someone to look up a word in a book was too, too big an ask for someone who might have a learning disability or, you know. And it, it really, it was an interesting thing to, to witness because.
It is one of those things where it's, well, yeah, it is entirely possible that there is no such thing as a piece of content that is perfectly accessible for absolutely everyone because there is always going to be nuance and there are always going to be unique needs. And I can think about this, especially from…
the alternate formats world, where in a lot of cases, because it's an individual requesting an alternate format.
sometimes you can essentially get a very custom.
format produced for you. and it may be completely inaccessible to other people, because it's for you, and it's for you and your specific needs.
So, I think there are cases where there is… I don't want to say, intentional inaccessibility, but maybe it's targeted… targeted accessibility, which risks inaccessibility to others.
what I…
put my hand up to do is… Evelyn had a really good suggestion in the chat, which was to, instead of trying to reverse engineer prompting.
From the definition, look at prompting and where it's used in the…
Lisa Liskovoi: Another reason is that some accessibility needs conflict
Evelyn Wightman, Ned Zimmerman:👆
Wendy Reid: In the document itself and see if whether.
We can actually just address.
Where it's used, and that might help us either come up with new definitions or even new terminology.
for that, but I'll let Miriam go, and then we'll, we'll, we'll look at that.
Miriam Fukushima: Yeah, I just wanted to add, recently I've been involved in a couple of projects that revolve around the Second World War and Germany's role. I'm from Germany and, also about the victims and, in.
conjunction to that with some recent, victims of,
Yeah, that kind of ideology and navigating that to be represented digitally.
Sambhavi Chandrashekar: Asking authors not to use big words is what I call reverse discrimination. If we consider inclusion as an idea grounded in ‘differences’ rather than ‘less vs. More’, then we will try to design to balance every need.
Miriam Fukushima: In all its colors and edges and photos. Yes. Photos. No. How photos, black, white contrast and whatnot.
And it's a project from the government, so it needs to be accessible by law.
But at the same time, the, wishes of the family's victims come first and how they want to be represented.
And navigating that has been a two-year effort.
And I think we got it kind of accessible, but still there are some edges that I know can be.
refute it. If.
certain people were to press charges. And, it's been a very delicate.
two years. And yeah, not everything can always be 100%.
Wendy Reid: Replying to "Asking authors not to use big words is what I call...":
It was definitely interesting when you think about where the burden of access lies, is it enough to have tools to make things accessible or does it need to be the content itself.
Sambhavi Chandrashekar, Miriam Fukushima:👍🏽
Miriam Fukushima: accessible, and then you have to kind of wager what comes first.
Wendy Reid: Okay.
at the document.
And our use of the word prompting or prompts.
So it's…
In reference to a fully automatic process, so prompting during the automatic generation process, authors are prompted for any required accessibility information.
I think this is partly where my, idea of,
Intervention came in because essentially it's an automatic process is like.
requesting an intervention.
Or performing an intervention.
But then we also have…
Checking suggested, so after automatic generation process, the authoring tool prompts authors to perform accessibility checking.
Let me use that term a little bit.
We also have how to respond to prompts for.
accessibility.
We have author actions preventing generation of accessible content.
Ignoring prompts.
And the definition itself.
And then, we refer… I guess,… so we have…
the action, and then we also have, the noun form of, prompts.
And basically we're requests from an authoring tool to an author.
for inter…, for a response.
I'm adding this to the to the Google Doc, because essentially we have 2 uses.
Two uses of…
The definition.
Well, we have a usage of the definition, and then we also have.
a description based on the definition.
Okay, go ahead.
Sambhavi Chandrashekar: Yeah, sorry, I was just going to raise my hand. So when you said from the authoring tool to the author, I thought I got some clarity about what I was trying to talk earlier about proactive and reactive.
So authoring tool to the author and author to the authoring tool.
Wayne D: It appears the term prompt as used in ATAG is unambiguous. It is a directive from the program.
2026-08-28 09:41:10 From Wendy Reid to Miriam Fukushima (direct message):
Are you free right after this meeting? I had something I wanted to ask you about :)
Sambhavi Chandrashekar: In the case of AI-based content generation, you have also authored to the authoring tool.
Prompting. Is that correct?
Wendy Reid: Yeah, actually, so this makes sense. So, ATAG 2, we had…
the concept of prompts in the direction of from the tool to the author, and it was one direct, one direction. The tool would.
Sambhavi Chandrashekar: One direction, yes.
Wendy Reid: We now have a concept today of bidirectional prompting, where the author can also prompt the tool.
Sambhavi Chandrashekar: Buy that.
can say, give me content with all headings marked up, or something.
Jutta Treviranus: That definition is now the more popular and understood definition, which is why we need to, avoid confusion with that… that use… that now…
popular use of the term prompt. And that's what generated this discussion.
Wendy Reid: Mm-hmm.
What is interesting, too, now, though, even in the AI context, is, a lot of… it's very popular…
For people to give instructions to their tools around.
Being prompted, being…
interventions on the tool side as well, still. So, it's very common. There's a… there's a really popular skill out there called Grill Me that a lot of developers like to use, because the skill essentially…
is… the point of it is to push back on the developer when they're asking AI to do things for them.
I definitely like when I work on mine, I often put a line in about, you know, if there is any ambiguity, please ask for clarification.
And so there's also still an element, even in an LLM context, of the authoring tool asking the author.
questions or prompting for more information.
It's just not determinative like it used to be.
Lisa Liskovoi: Wendy, is that also referred to as prompting, would you say, colloquially?
Wendy Reid: I don't know.
I haven't seen, a term for,.
When the tool talks back.
This is maybe also where that determinative, non-determinative thing comes into play.
Miriam.
Miriam Fukushima: Yeah, I would say, in case of the LLM, asking back.
or for confirmation is the same use case as the coded or hard coded decision trees, or.
notes, sites, etc, as… yeah.
as we had before, and that's what I meant with, it doesn't matter how it's coded underneath, or, what generated this.
Request.
We just have to differentiate it from the prompting, how it's nowadays used, and not, and find some new.
terminology with it, so it doesn't get confused, because prompting is.
Nowadays, asking the AI to do a certain task.
and not the other way around, and also not the tool…
coming up with some hard coded decision tree.
So.
I I'm not sure about the term intervention, because it it sounds very finite and.
Non-optional.
That's maybe why I prefer request for action or decision or guidance.
Or maybe all 3 because it can come in various facets. But.
Yeah.
Wendy Reid: Yeah.
And I do think depending on the context.
I'm with you. I think something like guidance, something like request, because I think there's, in certain cases, it's the tool requesting more information.
And in other cases, it's guiding the user to a better outcome that the user can choose to take or not.
Miriam Fukushima: And sometimes it's also requesting for actions, and there's no point around it. They're, it won't go further unless you do something.
Jutta Treviranus: Yeah, guidance doesn't encompass those things where there's actually an action. I mean, guidance is usually just advice that you can either ignore or not. So the end of that spectrum that we were talking about doesn't cover it.
I wonder whether we lean into the purpose of the thing, so it's, words like influence or nudge or,
And then, within that, there is, of course, a whole range of ways in which you can nudge or influence someone to author more accessibly.
Lisa Liskovoi: one other thing that I was thinking about is the…
or another scenario is…
clarifying or,
verifying, if the tool makes a decision. And this goes back to Wayne's point about defaults as well. I think, I think in some cases the.
decision that the tool makes in order to create something that's accessible by default, so, automatically generating alt text or even interpreting a heading structure from the content might not be obvious, and so there's also… the reason I'm putting it in the same kind of category is because I think.
I'm thinking of it as, what are the instances where there's dialogue required between the author and the AI in order to land on something that…
in order to to land on an output that is accessible.
So I'm thinking, do we want to say something…
Yeah, the author is also asking for verification or clarification.
Wendy Reid: This makes request, I think, stronger, too, because request is bidirectional.
The tool can request, the user can request.
Lisa Liskovoi: If we wanted to add a softer option to that list of updated terms, we could also say invite.
Wendy Reid: Mm, invitations.
I the, at least right now, I like the idea of keeping this, I don't think we need a firm definition because the other, I mean…
Realistically, we're accidentally backwards engineering some things in the sense that…
ATAG 3, we're gonna rewrite a lot of this content.
we don't necessarily… we're not going to be taking things necessarily verbatim from ATAG 2.
And so, we may come to something in the writing of this that makes sense., one of these words will make sense, but it's good to have an idea of,.
What words…
It's good to have a list of synonyms that we're going to want to introduce, discuss.
Lisa.
Lisa Liskovoi: This is a related, but maybe bigger question with Wcag. We've seen the sort of backwards compatibility of any updates. Is the intention for Atag to also be backwards compatible? Or do we have freedom to kind of do a.
rewrite that isn't.
Wendy Reid: I think we have the freedom to do…
like, the charter for the working group has not locked us into backwards compatibility. I don't even think we mention it, which…
Nice. And so I don't think we're locked into backwards compatibility. I think, and I mean, WCAG 3 is kind of taking a very similar approach. WCAG 3 will not be backwards compatible to WCAG 2. There may be, there's definitely through lines of.
themes and principles and there's some things that just don't change.
And I think the same thing is probably gonna happen for us where some things are just not going to change.
But we're not definitely, we're definitely not beholden to.
ATAG 2's statements are being backwards compatible.
Miriam.
Miriam Fukushima: Yeah, maybe. In this document we should also write down some frequently asked questions, because where we just write that down that this is now the case we won't necessarily be backwards compatible, or that it's open.
Wendy Reid: Mm-hmm.
Miriam Fukushima: And we also had questions, do we rewrite the whole? Do we keep the sections, et cetera?
For just,
to write it down and have it easy, researchable.
What the decisions were.
Wendy Reid: I'll let — I'll leave that if you want to fill in things that —.
Wayne D: There are cases where accessibility is deterministic (navigation). There are cases where a program cannot create the accessible content (alt-text). There may be ambiguous cases. (Example?)
Wendy Reid: All right. Do we want to talk a little bit about audiences with our last 15 minutes?
Is this roles of authoring tools?
This… this part?
Lisa Liskovoi: so that was… we had platform where the content is shared, distributed, and platform where the content is displayed, and our list of roles, and those were the examples from those.
Two roles that were added for clarification.
Lisa Liskovoi: Which this actually is making. Oh, sorry, Mariam, you had your hand up.
Miriam Fukushima: I just wanted to ask if we could maybe put the examples underneath indented under the road so we.
At least for me, that would be easier to recognize.
Wayne D: Okay, I think we need more on,
scientific and mathematical content. It really gets, oh, it's really too hard to do or something, but…
You know, basically, accessible content…
dies at about the secondary school level.
That is, there's almost literally nothing that's accessible.
from
the college freshman level to professional work.
at present. And I think that it's a really.
Major hole.
Accessible content.
you know, I spent my whole life trying to create accessible content that I could use, you know? I taught for 2 years, and I couldn't answer my student's question in the class, because I'd have to go home.
and look at my textbook using an assistive technology, and find the answer for my student, and come back the next day, and it wasn't in…
for several years when they finally came up with enlargement Xeroxing that I could answer questions for my students in the classroom. And I think that, you know, it's just an…
I think people look at it and they say that's too hard, I can't do it, but I think we need to address it.
And I'd be happy to I mean I have lots of ideas on it but.
I think it's needed.
Wendy Reid: It absolutely is needed. So, anything that you have, it's…
I know a little bit about that world, but only through the LEMS VPUB.
Lisa.
Lisa Liskovoi: Well, this is to the in regards to the list of roles. I just feel the platforms are different than we sort of have more human roles and then more components of the operating flow. So to me, this feels like a bit of a mix.
And of course, I was thinking about that in that framing. At first I was, well, the rest of them are human roles. And I'm thinking, well, is the AI in some cases authoring? Is that the author? Is that how we want to split it? I don't know, but the platform does feel different to me.
Jutta Treviranus: Yeah, I was going to say something similar. Have we defined what we mean by an authoring tool, and the range of authoring tools? Because, of course,…
the mathematical system or mathematical system.
tools, etc.
would be a type of authoring tool that we need to ensure.
Lisa Liskovoi: I think we've defined authoring tools, but not authors.
Miriam Fukushima: yeah, in regard to what, Wayne said, what that always makes me think of is, okay, considering the case, there is, for example, a math authoring tool.
And it adheres to all the things we have discussed that we will update in ATAG and ATAG is fully and completely perfectly applied in that.
theoretical scenario. Is there anything more?
Or missing. That could still.
make this math authoring tool inaccessible.
Even though ATAG is completely applied.
So
to kind of, narrow down what could we add for math tools.
Wayne D: Well, for example,
Most people are used to writing text and thinking of it as basically you read across a line and through a line. And so the basic.
structure of reading is one-dimensional. I mean, when you get to the end of a line, you go to the next line, you go through a line.
I mean, math is two dimensional.
So just the whole process of interpreting that.
is fundamentally different than normal language.
Just fundamentally different. You can't… If you try and… Furthermore, there's…
Probably, let's see, it was 1980 to the present day.
the infrastructure of producing and publishing mathematics. Well, for example, all of the libraries and.
And people don't write with MathML. Professional mathematicians don't write with MathML. They write with LaTeX.
And,
And so the libraries in LaTeX, for example, allow you to have.
Jutta Treviranus: Range of authoring tools: Learning management systems, social media apps, code generators….etc.
Wayne D: print sizes between.
8 and, between about 9 and 11 point. That's it. So all of the libraries were actually doing real work.
don't even possess the capability of producing large print.
And the concept of producing large print that actually reflows does not exist in the subject. So I think that,
It is just it is such a fundamentally different thing. We probably need to.
And maybe I'll do that. Contact Maple, which is produced in Canada, or Mathematica. I think that this is…
a level of authoring that just isn't addressed. And I understand why it isn't. And I worked with the…
MathML people, and know them, and it…
But we have these big authoring tools that are producing wildly inaccessible content. And students go into class and can't use their work.
So.
Miriam Fukushima: So does that mean that we need a section in ATAG where it, for example, says that authoring tool using languages or scripts other than the standard alphabet?
including science language,
need to be accessible to,
assistive technology, or is it something that should take, best practice or example or edge case, some articles, materials, so that that are adjacent to ATAG, but not in ATAG.
Like if we have articles and shortcomings and difficulties and current status of technology and what is possible and what should be made possible. If you have materials and a section about that and articles.
Is that enough to further that, or does it need, a sentence or a,
Like, a definition in…
The ATAG standard itself.
Wayne D: Well, I think that it actually needs to be an ATAG. I think that,
Science and science, technology, engineering, and mathematics are serious enough.
social entities at this time.
And the, the,
language of communication is so significantly different that it actually needs to be considered as a case in.
accessible authoring tools.
Because, you know, right now you have faculty that can't create.
accessible handouts for their students.
And, you know…
Just think about all the calculus classes that are taught across the world.
I mean, and there's nothing accessible.
Evelyn Wightman: I do think there's a lot in there that is worth.
Granular.
guidance. But the theme that's standing out to me is preserving meaning.
like, okay, sure, maybe I can navigate to all of the symbols in a… a math blob, but that doesn't mean that I'm actually going to be able to.
pull that together into a meaningful thing in my head.
And that applies to math and code and graphics.
It's not…
Not just about can you.
access each bit of it.
Jutta Treviranus: Yeah, I I wanted to clarify a previous point in terms of what I think it would be good to do, and that is to when I said defining authoring tools more. What? What is the scope of.
what we're trying to address. So, what are the range of authoring tool types and functions? And I think that, then.
Of course, would encourage us to…
to also consider these issues.
particular instances where someone may not immediately, or it certainly hasn't been covered before.
So, learning management systems, social media apps, coding, systems, etc. That… not… not that it needs to be,
complete, but at least giving people some sense of this. Yes, it does apply in these instances as well.
Lisa Liskovoi: Replying to "Range of authoring tools: Learning management syst...":
We have a discussion thread on this: https://github.com/w3c-cg/atag/discussions/12
Lisa Liskovoi: I just put a link in the chat, sorry, just to answer you just question. We have a discussion thread about this on GitHub, and a lot of the tools that you mentioned are on there. I think, I think all of them.
Wendy Reid: Mm-hmm.
Yeah, I will just pull that list in.
Miriam Fukushima: Yeah, I… what I get from this is, we need to do all three, putting it in the definitions of what we're addressing, but also as a sentence in, how the authoring tool functions, or should function to be accessible, but then also maybe elaborate on it, on best practices, or on.
like appendages to ATAG to actually address this more broadly and also make it.
a bit more widely known what the problem is and maybe even further research and technology in that regard.
Jutta Treviranus: Well said.
I think we captured it well there.
Wendy Reid: Mmhm.
Yeah. All right. So we're at time. Thank you all for another really good conversation.
we will,
We'll continue this next week. These are all really productive and it's really helpful as we get our heads around all these new tools and everything.
And yeah, if there's anything, ever any use cases that we think we're missing or areas that we're missing.
you know, feel free to bring them or find examples and share them. this document's open, I think, for everyone to edit, so feel free to add things even between meetings.
It's very helpful.
So have a great week and we'll see you all next week.