<Chuck> meeting: AGWG-2021-09-28

<Ben> unfortunately multi-tasking today but will hopefully volunteer to scribe again in the near future

<sarahhorton> I can scribe

<sarahhorton> scribe: sarahhorton

Chuck: Introductions?

JakeAbma: New topic: joining Silver and WCAG, based on different conversations in different groups, idea might be interesting to have something similar to WCAG acceptance criteria

<JakeAbma> https://www.w3.org/WAI/GL/wiki/WCAG_2.2_Success_criterion_acceptance_requirements

JakeAbma: different people have different opinions, regular publishing of information, have conversation on consensus, acceptance criteria for content to be published

Chuck: Having conversation on next week's agenda, Chuck will check with Jake to make sure it's covered
... Will talk about it in more detail
... other topics?
... Been challenging circumstances last 3 weeks, AGWG and Silver struggling
... everybody believes in shared goals to produce quality standard
... Covid makes it difficult to meet up, build relationships, build empathy
... concerns raised by members about behaviors, chairs working to address and resolve
... reaching out, will reach out, to discuss concerns with individuals
... starting point, review code of conduct
... next week talk about processes

CEPC reminder

<Jemma> Thanks so much, Chuck for all the effort the chairs are making.

MichaelC: CEPC reminder, guidance document, interact respectfully together, work together
... everyone has something to learn from it
... go through it, think through ways people receive behavior

<Rachael> https://www.w3.org/Consortium/cepc/

MichaelC: big one, think of others needs from their point of view
... other people don't mean ill, converse be sensitive
... treat each other with respect, diverse perspectives, language differences and their impact

<Jemma> great reminder, Michael!

MichaelC: problematic, bullying, intimidation, talking over, patronizing
... no one has intention to engage in these, people receive differently than intend
... need to be extra careful, understand, hear each other, diverse voices needed
... learn over time, takes practice
... lots of learning in group, continue to learn, less need for escalation
... continue to raise concerns to chairs, staff, helpful

Measuring Success, based on Alt Text

<Chuck> https://docs.google.com/presentation/d/1rWLV8CW4QtoqFzDLLvBj5S8-OL2bLR_Xh4u1mcQ8FX4/edit?usp=sharing

Rachael: Presentation, trying to figure out best way for conformance
... complex, difficult, big problem, after discussion, best way to start from bottom up
... smallest piece, measurement, move up to overarching
... walk through slides, goal not to make decision, point of understanding

<Jemma> distraction: just a quick note that PWE is buidling resources including CEPC. https://github.com/w3c/PWETF

Rachael: recognize, use Silver work, integrate knowledge of AQ, bring to same place and build
... goal, review measurement testing, scoring along with issues

<sajkaj> I believe it's "first public working draft"??

Rachael: talk about alt text, simple, a way of handling conversation
... terminology, need ways to talk about same thing
... measuring/measurement most granular
... testing, scoring, conformance how well score matches
... view, not talking about pages, using view, instance of content
... requirements, 1 multiple ways to measure

<Lauriat> WCAG 3 Requirements direct link: https://www.w3.org/TR/wcag-3.0-requirements/

Rachael: 2 technology neutral

3 motivation, go beyond, 4 regulatory

mbgower: Concerned about definition of view, shouldn't be "or"

Rachael: Agrees with change

GreggVan: If only programmatic, information hidden on page?

MichaelC: Tangent, finish presentation and separate clarification from discussion

<PeterKorn> +1 to collecting these for the end.

<alastairc> We've had that comment from Mike Pluke as well

GreggVan: Tried to switch pages for view previously

<Jemma> +1 to michael's suggestion, Q and A at the end.

Rachael: Please send list of people involved to chairs

<jeanne> +1 to michaels suggestion to only have claridicaation questions now.

Chuck: When question not specific, will note it so it doesn't get lost

GN015: "and" was changed to "or" and not okay, can discuss later

Chuck: Will have discussion outside of call

<bruce_bailey> for later maybe then, with 508 we sometimes use "information and data" instead of "content"

<bruce_bailey> maybe helpful, maybe not

<GreggVan> suggestion to keep the ? mark next to OR on the View item

Rachael: Link available on slides

<Jemma> +1 to section 508 , information and data.

Rachael: from FPWD, terms for today, distinction of atomic (object) and holistic (AT testing, UCD methods)
... subgroup working on scoping, reach out if interested
... functional images and alt text
... unit is images and measurement is percentages
... passes or doesn't number passed total images, results in percentage

<Cyborg> wow side note to say the captioning is doing such a great job with Rachael's voice. As if it was designed for her.

Rachael: manual tests/procedures, FPWD place to start conversation, are we capturing what's needed

<Chuck> yes, captions very helpful.

Rachael: direct port from WCAG 2.x

<Zakim> JF, you wanted to comment on subjectivity

<alastairc> JF - we will be going over the comments next

JF: Already seeing issues that raise concern, got feedback from FPWD comments about counting, level of granularity barrier, working from assumption don't believe
... fraught with subjectivity, concerned might not be right thing

<laura> +1 to JF

<Zakim> Chuck, you wanted to say we will be reviewing the issues that were raised.

JF: reviewing ePub doc, content author tagging images as decorative, subjective, concerned that topic does not have agreement

<Cyborg> wow that someone could get around this by labelling everything as decorative :(

Chuck: Intent to get to point where we can tackle
... laying out starting point
... what was publishing, issues raised

<laura> Don't agree with the 3.0 Decorative Images method.

Rachael: FPWD, starting point, a lot of concerns

<laura> In WCAG 2.x pure decoration is "serving only an aesthetic purpose, providing no information, and having no functionality". I don't know why we would change it.

Rachael: linked to issues that call out concerns, something most of us feel we understand, simple but also hard

<JF> I don't see SVG images on the list of image types

Rachael: complexity, can have conversations
... ratings, aggregating measures, pass/fall, percentage
... ratings, percentages have ratings

<AWK> +AWK

Rachael: critical errors another conversation

Cyborg: Involved with conformance early, then away, fundamentally don't understand how decision made, how poor performance gets rating, accessible or isn't

<alastairc> Topic for later - measures and level of fails allowed

Cyborg: additional things could be addressed if all of 100 was bronze, get to complex quality measures more subjective, how so many failures in bronze, worsening A AA AA problem
... explain how decisions made, thoughts?

<Chuck> I will get to Peter after Jeanne's answer

jeanne: Change is looking at score as aggregate of all guidelines, possible to do poorly, less than perfect on 1 guideline
... because aggregate, possible to have minor errors and pass if overall site is accessible, users don't have any errors that block
... on individual guideline (numbers arbitrary to start conversation), minor errors, shouldn't flunk
... how many images would be okay, part of conversation, make possible for orgs doing good job to pass

<Cyborg> my question - which i think is important - is why not make full accessibility of everything that can be auto-tested a bronze, so that we can use silver and gold increasing qualitative measures and new measures that are a bigger reach.

jeanne: allow some amount of error, tinkering with how much

PeterKorn: When come back, passing automated test, not fully automated but has manual

<Cyborg> (because things are accessible or they are not) - and so fully accessible could really be bronze.

garrison: Critical, non-critical, assess site and ranking 1, 30% issues remain, manually go through 30% to check whether critical

<Cyborg> and then more complex added measures beyond yes this is accessible, but this makes it better than a minimum - could be silver and gold.

<JF> ...and "critical" to whom?

<jeanne2> +1 garrison, that's what we want to look at

garrison: assess entire site, purely human, can't automate

<Cyborg> i understand that this is a global measure, as Jeanne shared, but nonetheless, a higher score per content criteria contributes to a certain colour.

Rachael: Critical errors, separate conversation, log as something to come back to

<Cyborg> i'm quite concerned about a lack of transparency about a score and where the problems are.

Rachael: Makoto's team put together
... thanks for work

<jeanne2> +1 Makoto and his subgroup

Rachael: numbers of arbitrary for now

<Cyborg> gaps are absolutely key to identify. we need to think about gaps. for example, being able to label everything a decorative image is certainly a major gap or design flaw.

Rachael: difficult about counting, three issues on that, links are in slides, slide 9
... concern about need to count, burdensome, manual, percentages

<Cyborg> HMW prevent someone from calling everything a decorative image?

Rachael: images have boundaries, more difficult with other content (text)

<laura> Agree with Cyborg. Make full accessibility of everything that can be auto-tested a bronze, so that we can use silver and gold increasing qualitative measures and new measures.

david-macdonald: counting up all that passes burden on client, not good use of their budget

<Cyborg> +1 to Laura

JF: Missed something on slide, that require manual testing, that's the problem, volume, feedback — less subjectivity, more mechanical
... less ambiguity, more precision, adding to manual testing is no starter

Chuck: Limit to clarifying question, want to get to measuring, people bringing up scope and other concerns, good questions, trying to get to how to count
... focus on clarifying questions on measuring

Cyborg: If 100% images conforming in ways to auto measure, which aspects of conformance could we auto-measure if flipped, lowest score for most easily testing, higher score for thing more difficult to test

garrison: Can test for, get rid of images, alt text, still leaves set that need to be assessed manually
... in machine learning, methods for testing test for correctness
... automate out subjectivity, aim at looking at other metrics for things

<Cyborg> the benefit of giving the lowest score to the auto-testable material, and extra points for things that are harder, we incentivize choosing the fruit that isn't so low-hanging. reach harder, get more points.

garrison: interesting metric through spice

<alastairc> garrison - is it better than the ML generated alt text?

<Wilco_> +1 An area that IMO should be explored

<PeterKorn> +1 on automating more and more over time, as ML/CV supports that. And having as a WCAG 3 goal a mechanism that allows us to more efficiently embrace those.

<Cyborg> this way an auto-testing software could get people to bronze, but they'd have to do more manual work to be recognized for outstanding accessibility.

Detlev: Discussing reliability, looking at all images, for 1 image will be ambiguity re appropriateness, e.g., hamburger menu, may pass, may not pass

<jeanne> garrison, can you forward that email to the Silver email list? public-silver@w3.org

<garrison> Alastairc - SPICE is a metric for assessing alternative text. It is not a generator or alternative text.

Detlev: pervasive, won't get to replicable result
... have to work with uncertainty

<alastairc> garrison - understood, but was wondering if it worked better evaluating than it does for generatingt

<garrison> Alastairc - SPICE is only for evaluation.

Rachael: Question of bronze, silver, gold, defer, have discussed making only auto testing level, suggest putting topic back on queue
... need to decide how to measure before level

<alastairc> garrison - sorry, not being clear. The results of ML generated alt text aren't v. good. Is SPICE better at evaluating alt-text than the others are in generating it? Seems like it would be two sides of the same coin.

Rachael: types of measuring, comments conflict, functional testing (holistic), objective measures, need to be clear about how to get to goal, don't meld
... ambiguity, clarity, flexibility great, inter-rater reliability
... need for simplicity, more complicated, less likely to adopt
... rating scales different for everything in FPWD, if rating should be consistent
... suggest common software bug rating, sample scoring sheet

<garrison> Alastairc - Yes, as it uses human responses from Image datasets (MS COCO, etc) as the reference to test the candidate text against. Which is why it would work fine for testing human created alt text.

Rachael: go read the issues, explore them

<david-macdonald> can someone drop in the url for this doc

Rachael: joint ACT Silver work

jeanne: Started in late April, identify how to bring measure, test closer to WCAG 2 ACT rules, 2 long meetings
... looking at issue with measure, outcomes, raised issues, didn't resolve a lot
... met in small groups, looking at tests, created revised template, AGWG passed CfC, to revise methods, move tests to expectations rather than procedures
... also included simplified tests for beginners
... tests were technical, included get started
... methods closely aligned to ACT rules, now joint group forming as subgroup, Test Reliability Group
... revising outcomes, created new draft, seeing in next 2 weeks, new design for outcome
... improve clarity, reduce ambiguity, more precision in outcomes

<Cyborg> I haven't seen the ACT document but will it be accessible or very technical? Can we put some work into using plain language on that document if needed?

<Cyborg> (so that more can provide feedback on it)

Rachael: 2 aspects to alt text, quantitative, does it have alt text, y/n
... qualitative, is it equivalent
... e.g., hamburger menu

JF: Missing context, how is image used in context

<Chuck> future examples (next slide?) will highlight context

<mbgower> also, it's the name plus the role

<Jennie> +1 to JF. Example: is it in a design document.

<Cyborg> for alt text, it appears that the conformance is based on a decision tree, is that true for all guidance? if so, are the trunks worth more than branches, branches more than offshoots etc?

JF: struggle, alt text always dependent on context

<Cyborg> or are there more points for completing the whole length?

Rachael: Next example, dogs
... quantitative and qualitative point of view

<mbgower> name: Settings, role: button, attribute: aria-haspopup='menu'

Rachael: context, complexity, how used, all delved into it, just a reminder

Cyborg: Trying to go through decision tree, each level gets more granular, points associated with getting perfect at level of decision tree, more points associated with different levels of tree

<JF> alt="our dogs at Wilson Park last Sunday"

<alastairc> I do wonder if we should pass functional images (part of a control) to the equivalent of name/role/value. (But this discussion is still relevant to content images.)

Cyborg: many ways to approach, want full compliance but prioritize different things, adjust priorities, layer, promote ultimate behavior, measure potential for unintended consequences

<jeanne> alastairc, that could be an alternative to critical errors -- an interesting idea to explore

Cyborg: decision tree one potential approach

<bruce_bailey> another alt decision tree:

<bruce_bailey> https://www.w3.org/WAI/tutorials/images/decision-tree/

Rachael: aggregate, how to we thing about image in aggregate within view
... menu with no text, dog photo with good text, aggregate score

<Cyborg> Thank you Bruce! Yes, so if 100% of images passed the first question you get so many points, 100% on next tree question gives you more points, etc.

<Cyborg> what are people's thoughts on that?

Rachael: step by step guide, different possibilities, rating scale, points, completely different, want to start conversation

<mbgower> This is not scalable. You can have a process for creating ALTs, but a testing process that ranks the success is doomed to defeat, IMO

<Cyborg> if the decision tree already exists, can we try it with that?

Rachael: objective, simple, consistent, inter-rater reliability, scoring
... a lot in example

garrison: Sampling, not mentioned, test in different ways if sample or site
... sample might look at all images, if scan site would not go with detail

<jeanne> garrison, Conformance Options subgroup is just starting work on sammpling, if you would like to join?

garrison: where consider sampling, has been discussed?

<Cyborg> if 100% were required to move forward with a certain aspect that was auto-testable or high up on a decision tree, then it becomes like the beeping seatbelt that requires you to put it on or it won't leave you alone. thoughts on this technique?

<Cyborg> +1 to Jennie

<Cyborg> 100% to what Jennie is saying!!!!! i bring this up a lot!

Jennie: Aggregate, considering some people with vision may need alt text, not able to understand purpose of image, consideration, expose alt text to those who need it

<Cyborg> +1 to Jennie

<Cyborg> thank you so much Jennie for bringing that up. i have this conversation so often.

<jeanne> +1 to Jennie, that is included in the user needs, but we have not written browser requirements yet

JF: Not seeing complex images, aria-describedby, prose explanation
... sees textual alternatives, no complex image as part of scoring

<Zakim> sajkaj, you wanted to invite Alastair to join the Conformance Options conversations on sampling--juist getting started

sajkaj: Conformance options taking up sample/reporting, starting with use cases

<jeanne> +1 jf - it is in a different method that isn't in this example.

sajkaj: welcome those interested

<Zakim> Chuck, you wanted to briefly discuss sampling

<ToddLibby> Chuck: How do we measure the lt text for that image.

<ToddLibby> Rachael: Other examples that need to be thought through

<ToddLibby> Rachael: Like to focus on measuring

<ToddLibby> Rachael: We would like to explore different ways of participation. Propose ways to test qualitative content.

<GN015> sorry, I have to step out.

<ToddLibby> Rachael: By October 30th

<Chuck> understood

<ToddLibby> Rachael: Feel free to give measuremtns of how that can be scored

<Zakim> Chuck, you wanted to say I heard one proposal already (decision tree?)

<ToddLibby> Chuck: Looking for ideas making determination of alt text on an image.

<ToddLibby> alistairgarrison: You cannot do it for 3M images. Proponent of using other methods of ML

<ToddLibby> alistairgarrsion: Not a perfect outcome but 80% would be a step forward to automate what we are looking at

<jeanne> +1 to Garrison - 80% accessibility would be a great improvement!

<Zakim> mbgower, you wanted to say that if an image is actionable (i.e., a control) that may be a subset to consider

<ToddLibby> mbgower: Echoes concerns.

<JF> +100 to mbgower

<ToddLibby> mbgower: Identify alts that are actually controls. Need to be actioned.

<alastairc> From Alistair's email: See SPICE - Semantic Propositional Image Caption Evaluation https://panderson.me/images/SPICE.pdf as an example

<ToddLibby> mbgower: Automated tests that flag automated controls

<Zakim> Chuck, you wanted to ask about machine learning

<Cyborg> +1 to Mike for a baseline, to be perfect on those things. but it's important from a cognitive disability perspective that we don't ignore quality of alt text which in some situations can create exclusionary barriers when not done.

<ToddLibby> Chuck: Jazzed about ML

<ToddLibby> Chuck: Another possibility - enough content where we could evaluate content

<Zakim> jeanne, you wanted to talk about the origins of these tests - tight timeline, direct migration from WCAG2.

<ToddLibby> jeanne: Recap origins of this work.

<ToddLibby> jeanne: all day meetings on approach we would use

<ToddLibby> jeanne: Kudos to Makoto and teams work under short dealines

<sajkaj> https://lists.w3.org/Archives/Public/public-wai-announce/2021JulSep/0002.html

<ToddLibby> jeanne: A lot of prep by Makoto's group to migrate from WCAG2 to WCAG3

<ToddLibby> jeanne: really likes ideas and likes to explore all.

<laura> Not a direct migration. As WCAG 2.x "pure decoration" was changed.

<ToddLibby> jeanne: Need for qualitative, need to address problem to address qualitative measurements

<Zakim> alastairc, you wanted to talk about aligning with better outcomes

<Cyborg_> it still makes more sense to me to give baseline points to 100% on quantitative tests and extra points for doing well on qualitative tests.

<ToddLibby> alastairc: Need to combine suggestions around decision tree for base level aspect

<ToddLibby> alastairc: balance that with what will align with a good outcome

<jeanne> +1 Cyborg, that's a good approach. Please write it up.

<ToddLibby> alastairc: ideas floating around and is up for collaboration.

<ToddLibby> alastairc: get ideas down for next month

<ToddLibby> sajkaj: Notes we have upcoming symposium likely to have papers on applications on ML

<ToddLibby> sajkaj: not too late for papers

<Cyborg_> Janina can you please share the link to the invitation you are mentioning?

<sajkaj> If you can find a metric, you begin to move it toward quantative

<ToddLibby> alastairgarrison: There may be other examples of metrics you can use to provide an answer to qualitative testing problems

<Cyborg_> Janina, can I share that call out widely? Or is it for W3C members only?

<ToddLibby> JF: Concern. Measuring quality. What constitutes success or failure.

<alastairc> Noting our current requirement is "a text alternative that serves the equivalent purpose"

<ToddLibby> david-macdonald: when testing, is there real text in there, describes purpose of it?

<Zakim> jeanne, you wanted to talk about rubrics and rating scales

<ToddLibby> jeanne: we don't define at normative level what is alt text.

<Cyborg_> @Janina - can I please share the link you sent me widely? Or is it only for W3C members? can the public submit proposals?

<ToddLibby> jeanne: no repeatability or reliability for testers.

<david-macdonald> we say "Equivalent Purpose" as the qualitative measurement in WCAG 2

<ToddLibby> jeanne: we want to advance, what is good alt text?

<ToddLibby> jeanne: a number of suggestions over the past year.

<Cyborg_> and Jeanne, what are the qualities of AMAZING alt text including for cognitive disabilities?

<ToddLibby> jeanne: idea of doing qualitative measurements are not out of bounds. We should explore.

<ToddLibby> GreggVan: we have to separate this into two parts

<ToddLibby> what do we regulate?

<ToddLibby> GreggVan: What do we regulate?

<ToddLibby> GreggVan: we can give you the way to give you how to judge it

<Cyborg_> regulate should be bronze but then silver is advanced/best practices and gold is modelling reaching forward (look ahead).... is reas

<ToddLibby> GreggVan: That which you can require and strongly push for

<ToddLibby> Wilco_: want to caution that an extent that which we can refine this where it doesn't get a lot of return back

<kirkwood> +1 to Wilco

<ToddLibby> Wilco_: want us to be cognizant we don't need to further refine this one and say WCAG2 did good job at it

<Cyborg_> -1 to Wilco only because there are audiences that have been left out

<kirkwood> very very good point Wilco

<Chuck> Todd: Willing to help

<ToddLibby> Rachael: Link to deck was emailed out and in these notes for meeting

<alastairc> Wilco - true, but we'd still need to convert it to match the framework.

<ToddLibby> Rachael: Really great discussion today, thank you.

<KarenHerr> scribe: KarenHerr

<Wilco_> Alastair - not if the new framework allows for what we have in 2.x

<Zakim> Chuck, you wanted to change scribe and topic

WCAG 2.2 Focus visible (Q1-3) https://www.w3.org/2002/09/wbs/35422/wcag22-focus-appearance-enhanced2/

<Cyborg_> thanks everyone.

<garrison> Bye

Question 1 - Adobe Comment #1887

<mbgower> These responses seem to exceed a response to the survey question (and I think are addressed in later survey questions)

Chuck: David Macdonald would like it to be less confusing

<kirkwood> agree with David this is too complicated to understand.

david-macdonald: comments on wcag being less complicated

<GreggVan> One way to make that simpler is to just break the sentence out into bullets that can be understood separately

AWK: pr in here doesn't make changes to normative text.
... understanding documents have become longer and more difficult to comprehend.
... suggest here to make it easier to read. two versions - changes order so minimum area is first
... The other is The focus area indicator.
... change to say indicator area is either....longer but clearer sub-bullet.
... figures out if area is right size and if area has the right ratio and adjacent contrast. still too long

alastairc: have put both versions on screen
... if a focus indicator is solid, this doesn't matter.
... if not solid becomes important as to what guideline is about

<mbgower> yep, had the same concern, Alastair.

Wilco_: largely what i was pointing out too. i feel this can be fully automated. i spent some time and am quite far. There will be tools.

<Zakim> Rachael, you wanted to say that I support removing the words Outline and Shape but not the reordering

Rachael: i wanted to suggest that removing bolded text at beginning makes it more readable
... i spend more time trying to resolve first word with text after it

<Zakim> Chuck, you wanted to ask wilco about progress

<Wilco_> AWK - you trust the HTML validator to tell you if your HTML is valid, no?

<AWK> @wilco - I trust it to verify it, but I go into the process with a clear idea of the direction

Chuck: one thing oracle is lookinig for - tell us how a tool does it. we will build our own tool. which of these is an improved explanation

david-macdonald: interesting point on automation. i share both opinions. contrast ratio worked when steve faulkner came up with a contrast ratio calculator

<Zakim> mbgower, you wanted to say one option is to remove the "shape" sub-bullet entirely

mbgower: think the shape is contributing a lot of mental overhaed. in there for designs that don't make perimeter requirement.
... consider pulling out shape requirement. biggest sticking point in terms of readability

<kirkwood> +1 to removing shape, per m gower

<Zakim> bruce_bailey, you wanted to ask if it grammatical to say an area is "thinner"

AWK: contrast area and minimum area

bruce_bailey: like edit just made. words shifting around. focus on minimum area of focus. not sensible to say area is thinner. needs more words

<AWK> "and no part of the focus indicator is thinner than 2px"

Chuck: i was about to do a poll. but i hear we need more tweaking on second part.

<AWK> "and no part of the focus indicator is less than 2px in width"

GreggVan: large doesn't necessarily mean area

<Chuck> proposed RESOLUTION: Accept amended PR 2000 to address issue #1887

<david-macdonald> +1

<jaunita_george_> +1

Chuck: and amendment is the text we have been working on

<Zakim> mbgower, you wanted to say I have another approach

mbgower: might work better

<mbgower> When user interface components receive keyboard focus, the item with focus is not entirely hidden by author-created content, and the focus indicator meets the following: Minimum area: The focus indicator has an area that is either… Contrast: The minimum area has a contrast… Adjacent contrast:

mbgower: suggest text read...

<Zakim> Wilco_, you wanted to ask for time

Wilco_: ask for time to think about this. i don't feel entirely comfortable voting after reading it once

Chuck: alastair has been one primarily working on this. Mike has a revision that could be reviewed

<Zakim> alastairc, you wanted to answer david

alastairc: hidden aspect - treating separately from
... item with focus not being entirely hidden
... pass if typing down page if footer partially covered item with focus

<Zakim> bruce_bailey, you wanted to ask about thin line between two thick bars

<Zakim> GreggVan, you wanted to ask does "not entirely hidden" mean it is entirely visible or that at least 1 pixel is visible

alastairc: the latter - could have a pixel showing

<mbgower> please, let's not go down this road again!

GreggVan: so focus indicator could be entirely obsured except for 1 px?

alastairc: it would be the element itself, rather than the focus indicator

GreggVan: would like explanation

Wilco_: would also like explanation

Chuck: people would like opportunity to review changes. not enough time to advance

<alastairc> https://www.w3.org/WAI/WCAG22/Understanding/focus-appearance-minimum.html#unobscured

Chuck: would like conversation to take place offline. lots of history and time spent on this.

alastairc: if questions, read understanding document in link above

Chuck: we will adjourn meeting. thank you for attending.

<ShawnT> thank you

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/liekly/likely/
Default Present: sajkaj, Ben, alastairc, Laura_Carlson, ToddLibby, Azlan, JakeAbma, sarahhorton, Raf, Jennie, Francis_Storr, MelanieP, SuzanneTaylor, garrison, MarcJohlic, JF, PeterKorn, GreggVan, Lauriat, julierawe, Makoto, Nicaise, jaunita_george_, Jen_G, StefanS, mgarrish, Detlev, Jemma, kirkwood, Rachael, stevelee, mbgower, jeanne, Cyborg, Wilco_, jenniferS_, david-macdonald, AWK, Katie_Haritos-Shea, KarenHerr, Cyborg_, ShawnT, bruce_bailey
Present: sajkaj, Ben, alastairc, Laura_Carlson, ToddLibby, Azlan, JakeAbma, sarahhorton, Raf, Jennie, Francis_Storr, MelanieP, SuzanneTaylor, garrison, MarcJohlic, JF, PeterKorn, GreggVan, Lauriat, julierawe, Makoto, Nicaise, jaunita_george_, Jen_G, StefanS, mgarrish, Detlev, Jemma, kirkwood, Rachael, stevelee, mbgower, jeanne, Cyborg, Wilco_, jenniferS_, david-macdonald, AWK, Katie_Haritos-Shea, KarenHerr, Cyborg_, ShawnT, bruce_bailey
Regrets: Breixo Pastoriza, Rain Michaels
Found Scribe: sarahhorton
Inferring ScribeNick: sarahhorton
Found Scribe: KarenHerr
Inferring ScribeNick: KarenHerr
Scribes: sarahhorton, KarenHerr
ScribeNicks: sarahhorton, KarenHerr

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: 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.)