<knights> (Eloisa Guerrero) present+

<dj> scribe+

Announcements

<dj> alastairc: CSUN next week

<dj> ... does anyone know how to join the WhatsApp group?

<dj> Jaunita_George: yes; I'll share the link in chat via /me

<dj> ... add me to WhatsApp and I'll add you to group

<dj> alastairc: announcements?

<dj> Rachael: next week is a working meeting

<dj> alastairc: two high-level topics today, and then next part is working on a document

<dj> ... we'll continue that next week

Next Publication - Updated Placeholder Guidelines and Outcomes https://github.com/w3c/wcag3/discussions/54

<dj> ... not a decision-making meeting

<dj> ... https://github.com/w3c/wcag3/discussions/54

<dj> ... since we've started, we've mostly been working on outcomes

<dj> ... options:

<dj> ... 1. update editor's draft

<dj> ... (very much a draft in w3c terms; doesn't even need concensus)

<dj> ... 2. update working draft

<dj> ... (public)

<dj> ... we're currently at Placeholder level

<dj> ... 0. not making an update

<dj> ... there is some pressure from W3C to publish something

<dj> ... publishing also allows us to get early feedback from a11y community

<dj> ... [reads github comments]

<dj> ... anyone have any other comments before decision?

<dj> Rachael: clarification: this isn't a decision to publish

<Chuck> dj: You said there is pressure to publish something. What happens if we don't?

<dj> alastairc: the longer you go without publishing, the more W3C management will ask you what you're doing

<Zakim> Rachael, you wanted to say we'vd been asked for quarterly

<dj> Rachael: we were being requested to publish quarterly

<dj> ... we set an informal goal of twice a year

<dj> ... that seemed reasonable

<dj> alastairc: we've also had some big tasks to get out of the way

<dj> ... recently working through scratchpad and organizing them

<dj> ... so seems like a good point to publish

<alastairc> Poll: We should update (1) Nothing, (2) The Editor's draft, (3) the Editor's and Working draft (4) The editor's draft and a filtered set of outcomes for the working draft.

<wendyreid> 2

<dj> 2

<Rachael> 4, 3

<Chuck> 4,3

<mike_beganyi> 2, 3

<ljoakley> presetn+

<dj> GreggVan: the working draft is more refined, right?

<dj> alastairc: yes; it is the more public one

<Azlan> 2, 3

<alastairc> 3,4

<dj> GreggVan: isn't the editor's draft updated all the time because that's where we work?

<laura> 4, 3

<dj> alastairc: it hasn't been yet; we've been working on Google Docs mostly

<Frankie> 3

<Chuck> 4 two's, 4 fours, 1 3

<sarahhorton> 4,3

<GreggVan> 4

<Chuck> 4 two's, 6 fours, 1 3

<dj> ... in general from this point we should be mostly working on the editors draft via PRs

<dj> wendyreid: question about option 4:

<dj> ... normally updating the working draft just takes the lasted version of the editor's draft

<dj> ... how would we do a filtered set of outcomes?

<dj> alastairc: we would probably start the editor's draft with the filtered set, then update the working draft, then update the editors draft again

<dj> ... or we could do a separate branch

<dj> Chuck: 4 & 3 are very similar

<dj> alastairc: people who wanted just 2: any specific concerns with working draft?

<dj> wendyreid: i wanted to see the state of the editor's draft first, then discuss

<GreggVan> +1 to that

<Azlan> +1 to @wendyreid

<Jennie_Delisi> +1

<dj> alastairc: that will be a natural part of the process

<dj> wendyreid++

<dj> ... we would go through the editor's draft before we get to the working draft

<dj> ... anyone opposed to that process?

<alastairc> draft RESOLUTION: We will bring a PR back to the group for the editors draft, then look at a refined update to the Working Draft.

<dj> +1

<mike_beganyi> +1

<sarahhorton> +1

<maryjom> +1

<ljoakley> +1

<Rachael> +1

<knights> +1

<Chuck> +1

<Azlan> +1

<wendyreid> +1

<GreggVan> +1

<ShaneDittmar> +1

<Makoto> +1

<Francis_Storr> +1

<laura> +1

RESOLUTION: We will bring a PR back to the group for the editors draft, then look at a refined update to the Working Draft.

<scotto> +1

<StefanS> +1

Universal Essential Exception https://github.com/w3c/wcag3/discussions/55

<dj> ... exception for when you cannot achieve something in an accessible way

<dj> ... we were thinking about a way for this to apply across outcomes, rather than saying it each time

<dj> ... [reads comments]

<alastairc> Suggested wording "Where the information conveyed or the functionality provided cannot be achieved in another way that would conform, the 'outcome' does not apply."

<dj> ... some outcomes you don't want to apply (seizures and physical reactions)

<dj> ... but we could also specify those separately (similar to non-interference outcome now)

<dj> GreggVan: strongly apposed.

<dj> ... as soon as you allow that, people will say "well there is no other way"

<jon_avila> I agree with Gregg's comments - I was going to say the same thing.

<dj> ... they could just use (example) Vander Script which does not provide any accessibility functionality, so there is no other way

<dj> ... we use the essential exception in just a few places where we know there will be exceptions (such as financial stuff)

<kirkwood> +1 to Gregg

<laura> +1 to Gregg

<dj> ... currently not applicable is better than blanket essential exception

<dj> ... this also gets into policy ---

<dj> ... we're saying that it's okay for some things to be inaccessible

<Zakim> alastairc, you wanted to comment on the technical not being a way out

<dj> alastairc: this is more for scenarios where (for example) scientific graphs or maps need some kind of graduated scale

<dj> ... there are potential alternatives, but you couldn't achieve that functionality in the same way while also meeting color contrast

<dj> ... also retyping a password is essential for security reasons

<dj> kirkwood: i don't fully understand your third point

<dj> ... but i strongly agree with what was said previously by others

<dj> ... i think this is very dangerous

<dj> Rachael: (chair hat off)

<dj> ... i agree this is both potentially challenging, but also provides benefit to plain language

<dj> ... i also think that creating it's own section around the exception has the potential for us to write it in a more specific way to prevent driving a truck through it

<Chuck> +1 we don't have to commit, but I'd like to try

<dj> ... i'd like to at least have the discussion of how we might do it, instead of just saying we can't do it

<alastairc> scribe+

<Chuck> dj: along the line of what Gregg said, this reminds me... ultimately voted in favor of not allowing technologies that don't comply to be considered accessible.

<Chuck> dj: Second thing, there might be a way that we can do it, but I fear that we might just miss the mark when we have the discussions. We could end up pushing something through that can be abused.

<dj> GreggVan: i don't think it's a good idea to pull out exceptions

<dj> ... example: changing Sans script (on a font page)

<dj> ... makes no sense on a page displaying fonts

<dj> ... we need to be able to say it in context

<jon_avila> I agree with Gregg - folks will miss the exceptions if they are somewhere else for this to be readable.

<dj> ... they're not going to look way up above in some other thing

<dj> ... i think details about the exception need to be right there

<dj> ... i also think that helps with plain language because all the relevant info is in one place

<dj> ... 2: if we do this, we need to say that this is a policy issue and we don't have authority

<dj> ... i'm concerned about places where things can be called accessible (because the conform to WCAG) despite not meeting the criteria

<dj> alastairc: i'm sensing some resistance

<kirkwood> +1 to Greggs point… not our place to know/determine when something is not accessible

<dj> ... where it destroys the point of what you're trying to do (fonts example), is that still problematic?

<dj> ... compromise could be: could we start with a standard exception where applicable?

<dj> GreggVan: for a long time, movie makers said adding captions destroyed the intent of their captions because of suspense of disbelief and artistic license

<jon_avila> that's the same thing the writers said when folks who are blind wanted text to speech on ebooks.

<Zakim> Chuck, you wanted to echo what I think Alastair proposed

<dj> ... we fought for years to get past that objection

<dj> Chuck: alastairc I thought you said 'let's create standard some language which doesn't apply, and then see at the end if it's applies to enough criteria to be considered universial'

<dj> ... is that accurate?

<Rachael> +1 to creating a text template for exceptions to help with writing.

<kirkwood> “essential” to whom?

<Rachael> a+

<dj> alastairc: I think so. but with addition that language could very across outcomes

<jon_avila> It seems like this is something can be done at the end after looking at the exceptions to determine if it can be done.

<alastairc> In WCAG 2 we have "unless X is essential", how is this different?

<dj> GreggVan: there is some resistance, but this is obviously important to you. why do you feel that way?

<kirkwood> +1

<dj> Rachael: i still think it's worth talking about creating a standard exception for when it applies

<dj> ... that way subgroups can use the already-written exception when the identify something that needs one, instead of writing one on their own

<dj> ... that feels like a separate, second option

<dj> alastairc: i see three options:

<dj> ... 1. universal essential exception

<dj> ... 2. no standard exception; custom exception per outcome

<dj> ... 3. standardized language for people to start with, and can let us no if it doesn't work for their case

<Zakim> alastairc, you wanted to say that we already have that in WCAG 2, how is this different?

<dj> Chuck: GreggVan: what you said about policy changed my mind

<dj> ... alastairc's current proposal isn't counter to that

<jon_avila> It would depend on what the shared language says.

<dj> ... it just gives us a starting point

<dj> alastairc: just to point out, there is a standardized definition of essential in WCAG 2.x

<dj> scotto: i really do appreciate the third option (standardized language that could be applied or adapted)

<dj> ... i get asked all the time what's essential

<jon_avila> The template should also allow for no allowing an essential exception.

<dj> ... i think a standard definition would be really helpful for people

<Zakim> GreggVan, you wanted to say instead of standardized langauage perhaps rules

<dj> GreggVan: I agree with the new approach

<dj> ... one suggestion: we talk about guidelines and recommendations instead of using a template

<dj> ... example: "guidelines for writing exceptions" instead of "exception template"

<Zakim> alastairc, you wanted to ask why the current WCAG 2 exception is split into two?

<dj> ... that would also lead to simpler language because subgroups wouldn't have to shove their language into a template

<GRAHAM> can we add a thing that says "the first rule of exceptions is that you shouldn't use exceptions" like we do with ARIA though! hehe

<dj> alastairc: sounds like there aren't objections to starting off with language on how to approach exceptions

<alastairc> "if removed, would fundamentally change the information or functionality of the content, and information and functionality cannot be achieved in another way that would conform "

<dj> ... why does the current WCAG 2.x exception have two haves to it?

<dj> ... https://www.w3.org/TR/WCAG33/#dfn-essential

<dj> GreggVan: the reason was: a really good test of whether something is essential is whether taking it off removes the purpose of the page

<dj> ... example: taking of font from font page removes purpose of page

<dj> ... un-illuminated illuminated font doesn't remove purpose of page normally, unless it's an example of what an illuminated page looks like

<kirkwood> removes the meaning of the content

<dj> alastairc: thank you

<dj> ... we could come up with something similar and simplier

<dj> ... any other questions or comments on this topic?

<dj> ... (none)

Working Session - Text Alternatives Outcome https://docs.google.com/document/d/1wkQ8ZKiRnNnax9ENaDXfQhl6kh-iYlXx7gMEhKMTQBU/edit

<GRAHAM> document isnt accessible for me, do i need to request access?

<alastairc> Oops, try again, should be viewable

<dj> Rachael: we've been working on outcomes

<Jaunita_George> When is the working session at CSUN?

<Rachael> Writing Testable Outcomes https://docs.google.com/document/d/1sugAtqie_x1XqHDZo1Im7ftDNllWeRV_ty4PULeoTV0/edit#heading=h.xghfays6dhl8

<dj> ... we thought, since we have CSUN next week anyway, why don't we as a group prepare an outcome before hand

<jon_avila> I can access it.

<GRAHAM> all fixed now Alastair, thanks

<dj> ... (please don't edit "Writing Testable WCAG 3.0 Outcomes")

<dj> ... we have a lot of exploratory content

<dj> ... we'd like to develop it further

<dj> ... some steps:

<dj> ... 1. look back at prior work

<dj> ... 2. come up with examples and edge cases

<dj> ... the key is the edge cases

<dj> ... for example: text alternatives edge case might be whether text is decorative

<dj> ... today we're developing test cases and edge cases that can be used to create rough outline of edge outcome (next week)

<Rachael> scratchpad: https://docs.google.com/document/d/1znxw1MZQrWV1pM9G7aQTRRzNkuM3APVdGcdv78WPiso/edit#heading=h.txiccm4cn4nl

<dj> ... chairs are looking at that now and will come back to the group in ~2 weeks

<dj> ... idea that came out of the subgroup: maybe decorative images still need to be presented (user controlled)

<dj> ... - AI alt text needs to be editible

<dj> ... - text alternative needs enough context

* alastairc yep

<dj> ... - image type (photo, icon, etc) indicated

<dj> ... - images have text alternative

<Jennie_Delisi> * Thank you to whoever added the AI editable bullet. That's fantastic.

<dj> ... we're just going to be working on "images have text alternatives"

<jon_avila> What is not-visual mean?

<dj> ... questions?

<alastairc> scribe+ ashleyfirth

<dj> GreggVan: is your intent to add the outcomes at the end of this?

<dj> Rachael: our intent right now is just to work on "images have text alternatives", but i wanted to indicate the other outcomes so they don't get lost

<dj> GreggVan: I meant are we linking to the outcomes from this document, or are we working somewhere else

<dj> ... the document being "writing tests and outcomes"

<dj> Rachael: "writing tests and outcomes" is the instructions

jon_avila: Question: what does "not visual alternatives" mean? Does this include things like sound that are not text?

<dj> gpellegrino: cases where, in books, and image is described by surrounding text, like "in Figure 1, something something something"

<dj> scribe: ashleyfirth

alastairc: Not sure where the non visual part comes from?

jon_avila: A little confused, seems like alternatives we're talking about are not just programmatical? Trying to understand scope

Rachael: A great place to start is to understand the scope of this. Multimedia is already broken out (audio, video)
... I think our examples will determine our scope

alastairc: So are we looking for examples and edge cases?

Rachael: Yes. We could work as a group which will be more inclusive but slower, break out and sort them individually, or break into groups

dj: Prefer if we don't all work on the doc at once, creates confusion

GRAHAM: If I run an AI company and create an image, should I need to provide alt text for the images I create?

<jon_avila> Most AI companies create images based on text.

alastairc: Emojis could also be an edge case

dj: Emojis are characters in unicode, so they're always read

<GreggVan> +1

<GRAHAM> Yes @jon_avila, for clarity I mean if I generate 4 images from a prompt as an AI provider, should I have to describe the contents of the image (is it even possible!) as they could all be slightly different even from the same prompt.

<ashleyfirth_> Duff_Johnson_: Is ASCII art an example of what you're describing?

<ashleyfirth_> jon_avila: Not really. Something like Google Docs where text is drawn on the screen rather than programmatically shown

<rscano> at point 5 typo: ACII ART

<ashleyfirth_> Rachael: How about something that is AI generated, but generated as both image and audio? Does that count as multimedia?

<ashleyfirth_> dj: Are you referring to special characters?

<ashleyfirth_> jon_avila: No. My understanding is how some things are rendered on a screen appears to be text, but is actually being drawn on the screen as an image

<ashleyfirth_> alastairc: Using the doc we're looking at as an example, it's text rendered on a canvas

<dj> special characters (role of text)

<ashleyfirth_> kirkwood: An edge case could be a visual representation between two text fields, such as a red gradient

<ashleyfirth_> GRAHAM: An edge case would be code examples that are shared as images on social media

<ashleyfirth_> alastairc: Importantly, the structure of that text within the image wouldn't be represented by alt text, such as headings

<ashleyfirth_> jon_avila: Maps could be another example

<ashleyfirth_> jon_avila: Anything where the view on the screen changes, such as games or moving a map, the new view would need to be described

<kirkwood> examples may be: family tree, process flow diagram

<ashleyfirth_> Rachael: Did some research a few years ago. Found a big difference between exploratory images and images portraying a set message

<ashleyfirth_> GRAHAM: SVG's that are served via the image tag?

<jon_avila> I was also going to build when you need to explore the image or map ort parts of it.

<ashleyfirth_> scotto: right now, no. if svg is served via an img element it is not expected to be traversed

<ashleyfirth_> GreggVan: Want to make sure we don't suggest anything that assistive technologies won't pay attention to

<ashleyfirth_> GRAHAM: picture element in HTML, you can add multiple images as sources, but can only provide one alt text

<ashleyfirth_> dj: Math can also be served as an image that would hard for alt text

<ashleyfirth_> Rachael: Another edge case: an image where the description is so long that it needs to be in its own file

<ashleyfirth_> alastairc: Are we getting to a point where we could consider this to be a good set of examples

<ashleyfirth_> Rachael: I think so

<kirkwood> dynamic images?

<ashleyfirth_> jon_avila: Sprite images could be another, where you're only showing one part of the larger image

<ashleyfirth_> Rachael: Next step is to draft some assertions that could be used to test each example and edge case

<ashleyfirth_> kirkwood: Dynamic Images or images that change are another example

<ashleyfirth_> Mike5Matrix: We need to define what's an image and what's an animation

<ashleyfirth_> kirkwood: Example I was thinking of is similar to a carousel, which moves when you roll over it on a page

<ashleyfirth_> mbgower: We need to define what's an image and what's an animation

<ashleyfirth_> Rachael: Now we have these edge cases, should we change the use of the term "image"?

<julierawe_> Is the goal to come up with as many edge cases as possible?

<ashleyfirth_> alastairc: How do we define visual content that is mostly static but not a video

<ashleyfirth_> scotto: Variable/dynamic images could be a category/consideration here

<ashleyfirth_> mbgower: We could focus on having "updating imagery" as its own area, separate to general rules on imagery

<ashleyfirth_> mbgower: Also need to add whether imagery is changing on its own, or in response to user interaction

<Zakim> Rachael, you wanted to ask if also generated images?

<ashleyfirth_> Rachael: Should we also in these buckets consider generated content, like AI?

<ashleyfirth_> alastairc: I don't know if that's a different bucket

<ashleyfirth_> dj: AI is set as a the first item in the list

<scotto> don't think it is a bucket, but absolutely agree it should be called out as a requirement that generated imagery would also need to have generated description as to what was created

<ashleyfirth_> mbgower: Has anyone tried distinguishing between important and unimportant images?

<ashleyfirth_> alastairc: We haven't yet. A lot of the types of images we've covered would fall into one of those two categories, but we haven't fully considered that yet

<ashleyfirth_> Rachael:

<kirkwood> sufficient

<Rachael> Rachael read https://docs.google.com/document/d/1wkQ8ZKiRnNnax9ENaDXfQhl6kh-iYlXx7gMEhKMTQBU/edit

<ashleyfirth_> GRAHAM: All sounds good, but need to check the limits of the boundaries of this

<ashleyfirth_> Rachael: The other thing I'm seeing here is that it doesn't cover text not in alt or in the caption

<ashleyfirth_> alastairc: With some tweaking it could. Examples that come to mind are if you've got an image without an alt, and the screen reader will instead read out the file name.

<ashleyfirth_> alastairc: If an image has a good caption but a null alt attribute, nothing is lost

<rscano> +1

<Chuck> +1 excellent point!

<ashleyfirth_> GRAHAM: Thinking about the earlier example of Math in an image from dj, content could be shown correctly in HTML, so should we be covering the most appropriate approach

<Chuck> +1 adding

<ashleyfirth_> alastairc: This is a good point, but takes us quite far into the weeds

<ashleyfirth_> Rachael: We welcome the weeds!

<rscano> and for IA generated, need that in alt text this is noticed for inform user that is a machine-generated content. IMHO

<kirkwood> “no meaning is lost” rather than “nothing is lost”

<rscano> oherwise we need to ask to modify HTML for add an attribute ai-gen="true" :joking

<ashleyfirth_> scotto: Generally this would apply in many instances, so we need to encourage people to avoid this type of content within imagery

<ashleyfirth_> Rachael: Should we add in "should this be an image?" based on different content

<ashleyfirth_> Rachael: Long descriptions are desired in many instances

<Chuck> +1 leave under 3 (not a strong opinion)

<Zakim> alastairc, you wanted to comment on not using an image

<ashleyfirth_> mbgower: I think finding out if an image is important is an easier guide (although still tough)

<ashleyfirth_> alastairc: Now we're putting all image-related criteria in one place, that might help us create broader rules

<ashleyfirth_> Rachael: We could have a method for a content image, and a method for images of text etc.

<ashleyfirth_> Rachael: Or we could create a single method with multiple paths. It's a question of structure

<ashleyfirth_> GRAHAM: What about a broken image? Is that relevant?

<ashleyfirth_> alastairc: We'll add it to the edge cases to explore

<ashleyfirth_> kirkwood: We can't have alt text describing an image that isn't there, as a screen reader user would be receiving different content

<ashleyfirth_> Rachael: Is this a good time to transition out so people can ponder it

<ashleyfirth_> alastairc: Good idea. We'll be carrying on next week with this topic

This is scribe.perl Revision VERSION of 2020-12-31
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/2/2, 3/
Default Present: dj, kirkwood, JakeAbma, Azlan, kevin, Rachael, Chuck, bruce_bailey, Jennie_Delisi, alastairc, wendyreid, mike_beganyi, Francis_Storr, giacomo-petri, JustineP, graham, Makoto, Kimberly, sarahhorton, Gez_Lemon, Detlev, Laura_Carlson, jeanne, Jen_G, scotto, ShawnT, GreggVan, sabidussi, ljoakley, Glenda, mbgower, gpellegrino, jaunita_george, shadi, julierawe, maryjom, jtoles, ashleyfirth, StefanS, ShaneDittmar, Frankie, Wolf, julierawe_, jon_avila, rscano
Present: dj, kirkwood, JakeAbma, Azlan, kevin, Rachael, Chuck, bruce_bailey, Jennie_Delisi, alastairc, wendyreid, mike_beganyi, Francis_Storr, giacomo-petri, JustineP, graham, Makoto, Kimberly, sarahhorton, Gez_Lemon, Detlev, Laura_Carlson, jeanne, Jen_G, scotto, ShawnT, GreggVan, sabidussi, ljoakley, Glenda, mbgower, gpellegrino, jaunita_george, shadi, julierawe, maryjom, jtoles, ashleyfirth, StefanS, ShaneDittmar, Frankie, Wolf, julierawe_, jon_avila, rscano, Frankie Wolf
Found Scribe: ashleyfirth
Inferring ScribeNick: ashleyfirth

WARNING: No meeting chair found!
You should specify the meeting chair like this:
<dbooth> Chair: dbooth


WARNING: No date found!  Assuming today.  (Hint: Specify
the W3C IRC log URL, and the date will be determined from that.)
Or specify the date like this:
<dbooth> Date: 12 Sep 2002

People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.


WARNING: IRC log location not specified!  (You can ignore this 
warning if you do not want the generated minutes to contain 
a link to the original IRC log.)