Silver Task Force & Community Group

29 May 2020


jeanne, Rachael, sajkaj, KimD, OmarBonilla, CharlesHall, MichaelC, Chuck, Lauriat, bruce_bailey, JF, AngelaAccessForAll
Shawn, jeanne


<jeanne> queue:

<scribe> scribe: Chuck

Jeanne: subgroup checkins
... Any volunteers to go first?

Chris: We met Thursday (bruce, andy, me). Andy will work on math ml transition from current formulas and placing into markup for what we have currently.
... We'll meet again in 2 weeks from yesterday.

Jeanne: Does Andy have contacts of everyone who can help?

Chris: yes. I'll reach out specifically to him and confirm.

Jeanne: No reason to force him to ask for help if he doesn't need it, but we have people Janina works with that volunteered to help if needed.
... I sent contact info months ago. Text me if that info has been lost.

Bruce: he didn't balk about converting. That wasn't an issue.
... Averty (???) from original research.

Jeanne: Did he have any thoughts about scoring and the chart you made Bruce? Instead of pass/fail, any conditional scoring?

Bruce: We did talk about the chart. Looked at it before. Didn't care for how improvised it was, and plans on taking another look.
... We think we are good for any conformance model. It has a flexible metric. On the other hand you are still trying to meet a minimal number.
... Higher numbers are better.

Jeanne: See what he says about it.
... We'll know more on Tuesday. Jake has a proposal for conformance that weaves in most of the pieces of the different proposals, and it's pretty consistent. I'm excited.
... Jake will present on Tuesday. You may want to have a discussion with him to see how the chart fits in with his proposal.

Bruce: Looking forward to it.

Jeanne: anything else?

Chris: That's it, including what Bruce mentioned.

Jeanne: For your info, it would be good to circle back and look at the other tabs that aren't completed yet. We need to have all the tabs filled out.
... And what I pasted into the tabs, you may want to move somewhere else. I'm not attached to that content. I picked and chose what went where.
... Is Jan here for clear language?
... We had a clear language meeting scheduled this week, but some sick and busy. We'll have a report on tuesday.
... Makoto is not here, 2am his time.
... Anything from functional needs, Michael?

MC: I gathered troops to meet today, but didn't manage. I requested times to meet next week.
... I hope we can have a preliminary meeting next week to lay foundation on how to proceed. Hope to have more substance next week.

Jeanne: Others outside of AGWG?

MC: No, just the 3 that have volunteered.

Jeanne: Are you planning to bring in others, or are you just the coordinator?

MC: We haven't talked yet, I imagine we would welcome additional perspectives and workers. I myself help to coordinate with other activities.
... Don't know if that would turn into one big group or other smaller groups.

Jeanne: OK, keep me posted so I know the plans.

Charles Hall: I have continued to expand the document with the needs that are identified and resources we identified, so that they are all in one place.

Charles Hall: I re-organized the doc to clearly identify the various sources.

Jeanne: Thank you!

<CharlesHall> https://docs.google.com/document/d/1eJkgXqbh7dx3uD6XAy8XAANmwfbbVZ5GKb_gbsUdkVs/edit?usp=sharing

Jeanne: Can you drop a link to the current doc?
... That's everyone here.
... I mentioned to Chris and Bruce, I met with Jake this morning. He has a really substantive proposal that he's testing. Detlev will work on testing, as will I.
... He will be prepared to show it on Tuesday.
... Anybody in subgroups who want more time to ask questions or ... Bruce, any progress in audio description and xr? I know you didn't meet this week.
... Anything I missed?

Bruce: no, nothing else.

Jeanne: Hopefully we'll have something on Tuesday.

Janina: I think we have two weeks in a row of holidays on different continents.

Shawn: I thought that was this past week.

Janina: I think Josh has a holiday.

Shawn: Is there another one?

Janina: There's another one.

Jeanne: Common this time of year.
... Everybody wants summer holidays.
... I'll review email and see.
... Anybody in subgroups have any issues that they have encountered, any comments about the process so far?

continuing on changes to the writing process document

Jeanne: I'd like to continue on the process of making some updates to the content writing document.

<jeanne> Template https://docs.google.com/document/d/1Smly4XDxfzfXHa7AoUxoLXLy_3PdOXMkh0ZwtgksSPk/edit#heading=h.8j6pwbsnl608

Jeanne: We have two links, the template (first link). Easier to understand.
... And the process doc. Which is all the detailed instructions, maybe too many instructions.

<jeanne> Directions https://docs.google.com/document/d/1gfYAiV2Z-FA_kEHYlLV32J8ClNEGPxRgSIohu3gUHEA/

Jeanne: Last time we talked we left off at tags for information architecture.
... I see that Chris put in the links to the EO work that we discussed.
... Which I'll gladly put in the doc.
... What's the 2nd link Chris?

<jeanne> https://www.imsglobal.org/accessibility/afav3p0pd//AfA3p0_DRDinfoModel_v1p0pd.html#toc-5

Jeanne: That's cool. Here's a link for those not looking at doc. It appears to be a list of every acronym and definition.
... For access for all. This is an interesting doc.

Chris: JF and Jan were talking to that in a prior call. I was able to pull and drop into the doc.

Jeanne: Thanks!
... Is there anything in particular that you wanted to draw our attention to?

Charles Hall: That resource was shared also in the context of the functional needs discussion. I pulled in the 2.1 ... section

Jeanne: Excellent!
... This is what I miss, the literature review part of silver research years ago. Never really happened. This is when having that ability to pick things up would have been handy.
... Thank you, I appreciate everyone putting them in!
... Impressive docs.
... Developing tests....
... The instructions start with "known solutions", for WCAG techniques, ACT or easy-checks.
... Are there other things we should list?
... Objections to any of these?
... This is a big item. I'd like to have a conversation. Is this the right place to start?

<ChrisLoiselle> Yes

Chuck: What's easy checks?

<ChrisLoiselle> Chuck: Haven't seen easy checks before, what is that exactly?

<ChrisLoiselle> https://www.w3.org/WAI/test-evaluate/preliminary/

Jeanne: This is an eo document that was published 2-3 years ago. Which has tests for beginners. People just starting accessibility testing.
... Great to have fast people!
... In the directions I had a link to easy-checks directly.
... It identifies the basic things to test for.
... If you look down the page, page title, alt text, headings, color text, etc...
... Gives you step by step directions.
... Web oriented, limited in that way, but a good tool. I think for anyone that is working on a guideline that would be looking for a place to start, that would be good.

JF: Nowhere in any of this test process or how to write/develop tests is anything that addresses partial conformance. My example had times where headings were present but hierarchy was wrong. I'm not seeing for that partial right.
... Prior was pass/fail. I'm not seeing anything that accounts for the middle.

Jeanne: That's in the next section for new tests. I want to circle back, and get your expertise.

JF: Part 4 says new methods. I don't see "write new tests".

Jeanne: It's in part 3. #6.

JF: #5?
... it only goes up to 5 in the version I'm looking at.

<ChrisLoiselle> JF: New tests for Silver Name of new test and describe how to perform the test on https://docs.google.com/document/d/1Smly4XDxfzfXHa7AoUxoLXLy_3PdOXMkh0ZwtgksSPk/edit#heading=h.i7znkzomudrc

JF: I'm looking at the google doc...

Jeanne: I did some work and changed some number.


<CharlesHall> in that doc it is 5c: Are there potential solutions that could be tested if a more flexible test than true/false were used? What new tests should be considered for this? These tests are:

Jeanne: You are right John, it's #5. I need to match template.

JF: Even where it says "tips for writing tests", nowhere is there anything that suggests a partial compliance store?

Jeanne: You are right. Where should it go.

JF: Tips for writing tests. I'd note that this is a potential outcome. It will be increasingly expected. The problem we are trying to solve is WCAG 2.x conformance.

<CharlesHall> it should be included in the Tips prior to the Process

JF: Reality is that nobody is perfect, but many are close, and we are trying to measure that "close". We need to think about measuring conformance on that scale. Beyond pass/fail.

<Lauriat> +1 to having a not on this included in the tips for writing tests

JF: The other thing is, I don't see how tests get linked to scoring.

Shawn: It's in the template, but instructions don't match that yet. I see a placeholder section in the template, but it's not in the same section in the instructions that you and I were both reading through.

JF: There's another gap.

Jeanne: JF, any recommendations, in tips for writing tests. Principal that ACT follows is that they make tests more granular. I started to write that, but this is not something I have a lot of expertise in.
... First bullet, the more granular the test, the easier to write.

JF: That's one way of... I'll use the test for heading. Are headings present? Is hierarchy preserved? I'm not testing for "did you use properly"? It breaks into two parts.
... That's a granular test. In that scenario, both are answered by t/f. When we start taking about other tests, there's going to be... Charles Hall I've never seen a cognitive test presented.
... That complex walk through is broken down into a series of atomic tests that resolve to true/false. I don't see how that is explained here.

Jeanne: Where we are trying to go... btw that is helpful. If you want to put in notes in the doc that would be helpful. What we are trying to do is help the subgroups take the step from user needs to functional outcomes to writing tests.
... The first time we went through this process, they struggled. We didn't have functional outcomes. People couldn't go from user needs to tests.
... We want to make it easier to do.

JF: We don't have functional outcomes yet. I am hoping we could look at en301 and others to write about functional outcomes. I don't know what it is based on disability type.

Jeanne: Let's switch back to template for a sec, it's shorter.
... We start by defining user needs. That's the list of disabilities and the barriers they encounter. In content creation process. Should come from EN301549.

<Lauriat> Template link: https://docs.google.com/document/d/1Smly4XDxfzfXHa7AoUxoLXLy_3PdOXMkh0ZwtgksSPk/edit

Jeanne: Also WCAG understanding, and other resources, and list user needs. At the time we didn't have a subgroup working on functional needs.
... but functional outcomes are different than the needs. More granulary by guideline. We write more specific things related to the guidelines. for headings... it needs to do this this and this.

JF: Let's use that. One of the things heading needs to do is convey hierarchy. Critical for cognitive issues. For non-sighted users, important, but I would argue more important for cognitive.
... for most people with cognitive, unless using AT, they don't have that for them. There's different severity and importance levels depending on the disability in play.
... My peers who are blind roll with the poor hierarchy. It's important, but for them its nice to have. For those users who are dependent on the structure, failing to get the hierarchy is more important.
... How do we test and measure through those measures?

<CharlesHall> related resource for Cognitive Walkthrough (using the gov.uk personas): Sarah Pulis and Andrew Arch on the 13 Letters Podcast: https://www.bemyeyes.com/podcasts/accessibility-down-under

Jeanne: I added another functional out come, we can work on the wording. Quoting you... "conveys a sense of hiearchy". Is that ok for an example for now?

JF: It's as good as anything else. The way I envision it is it's a matrix. 7-9 columns with disabilities that have been defined. If we map existing SC against those category types, we can think about the types of tests necessary.
... For deaf, zero impact. We need to recognize that as part of testing and functional outcomes. Different requirements will have different severities. It's not a single test. It's a collection test. I suggest we need a deeper thought on what's required.
... What is the impact on those different user groups, and what do we need to test for. That summarizes the outcome for that disability type.

Jeanne: Do you think we haven't captured what we need to do to make that transition?

JF: I don't think so. I haven't seen the 7-9 groups, we haven't confirmed or revisited COGA, which we've said multiple times needs fleshed out.
... If that comes from COGA, AG, or this tf, that's an important step we haven't taken.

Jeanne: That's the subgroup with Michael Cooper.

JF: We know the requirements, what does that mean based on disability type what I'm testing for, and why. I also assert if we get that matrix fleshed out, it will help with our scoring discussion.

Jeanne: The way we have been handling it, because we didn't have an easy answer, they are going to be working on that. It needs to be coordinated with other parts of WAI. That work is going.
... In the meantime we have used EN301549 as a placeholder, knowing it will need to be expanded.
... Based on mandate 376.

JF: For every test, for every functional outcome, I envision a big matrix table on the top that has all the tests that list the tests and types. Some will have more tests than others.
... Even if the test is repeated because it's satisfying another type, then that's a more important test because it helps multiple groups. Has more value.
... Even if they are duplicates, that's ok. Let's start thinking in terms of functional outcome and disability type, and classify the tests that we need.

Jeanne: For people working in the sub groups, especially if you have struggled to write the tests, do you think a table or matrix would be more helpful to you in doing it?

Chuck: Maybe pilot it.

<ChrisLoiselle> Chuck: Have one or two groups try it out and report on experiences.

Jeanne: Do you think you can set up in this doc, or in another doc? Or just describe and have me do it?

JF: I'm happy to try my hand at a first draft and help to visualize and spark the conversation.

<Lauriat> Thanks, JF!

Jeanne: Great! We have struggled with it.
... Makoto's group spent a lot of time looking at existing techniques and relating to modern usage.
... But when you say "write new tests", we have not had a lot of success. I think whatever is going to help people make that jump from "this is what the user needs, these are the outcomes, how do we test"? I'm very willing to try.

JF: I'll try and work on it this week.

Jeanne: Anyone else who worked on the earlier round of tests? Earlier round of guidelines? Comments on writing new tests?

JF: I'm concerned as well when tests have compound steps. For example ... textual alternative... the test is mirroring the process that is used in accessible name calculation.
... It almost follows that recursive algorithm. How do we express those kinds of tests today?

Shawn: Which test?

Jeanne: We can use act tests as they are.

JF: Probably. Illustrative example. Where... you are right in the ACT process, it's recursive. They are scriptable. When we are writing other tests, as tips and tricks for writing tests, use a recursive notation method.
... In that case we can point to the test for 1.1.1 if they want to see an example.
... Another q.... <massive echo>
... In that case, everywhere we go, the #1 rule of aria is use native element instead of aria. We have a conflict between best practices vs naming calculation algorithm works. When writing technique, do we write from technical perspective? Or best practices?
... Valid questions, I don't have the answer. But we should be standardized on one or the other. And that would be in the tips for writing tests.

Jeanne: for the tests, that would start with writing the technical "what is" and what can be tested. And in best practices that would be in method.

JF: Aren't we focusing on functional out comes?

Jeanne: That was part 2 and we are now on part 3.
... Starting at the very technical, testing for the accessible name. When we write the method, which is the wrapper that the test is a part of, that's where we would say "the best practice is use the symantic element first".

JF: We are going to have a recursive test. The best practice of those 3 is to use the alt attribute. If you use aria, the functional outcome for the user is the same.
... End user doesn't care which technique is used. But we are saying "use" alt first. Now my question becomes, in a recursive test where any of the 3 tests works, but one is preferred, does one get a better score?

Shawn: +1 to decouple tests from technical implementations. Working backwards, no you would not get more points for using native symantics, tests should validate outcome.

JF: Is that a consensus position that techniques don't have impact on score?

Shawn: Not sure we can say as a blanket rule, but hopefully we can make it a blanket rule.

JF: Working hypothesis.

Shawn: It should have a partial score to reflect that (headings).
... For your example on images, or headings, should look at accessibility tree, rather than raw dom structure. I appreciate you making that point.
... If that ever changes, then our test is invalid. If we focus on functional outcome, then it doesn't matter if there are new ways of getting there.

JF: I'm fine if that's the policy, but it negates good or best practice.
... Not seeing a way to reward people who use best practice. If it is the best, should be rewarded.

Shawn: We wouldn't say here are 12 ways to add alt text, we would say here's the preferred way.

JF: Shawn, seems we want to show the 12 ways if that's the ways there are.

Shawn: I disagree, that will always be in flux.

Rachael: I don't think we need to push best practice. The benefit is inherent. Best way to ensure across platforms. I don't think we need to add more score.
... If it doesn't have the widest applicability and generating the right outcome, why is it the best practice?

JF: If you used the best practice, you get the best outcome.
... We used bronze, silver gold, do we acknowledge use of best practice?

Rachael: You may need to change it if tech stacks chane.

Shawn: If something provides the best experience it should get the best score.

<Zakim> jeanne, you wanted to say that the partial score would be by the functional Outcomes

Jeanne: What I've been hearing Jake talk about is that it you wouldn't have to say "this is the best practice", but you would say "what is the practice that meets all the functional outcomes", and you can do partial scoring on functional outcomes.

<KimD> +1 to Jeanne - "best practice" means different things in different companies.

Jeanne: ...doesn't convey a visual hierarchy, doesn't get a full score. It doesn't meet all the functional outcomes.
... Jake is also building in the adjectival, and others. It's slick. Come on tuesday!

JF: i gotta go.

Shawn: thanks everyone.

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version (CVS log)
$Date: 2020/05/29 18:59:05 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision of Date 
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/recersive/recursive/
Succeeded: s/JF/Shawn/
Present: jeanne Rachael sajkaj KimD OmarBonilla CharlesHall MichaelC Chuck Lauriat bruce_bailey JF AngelaAccessForAll
Found Scribe: Chuck
Inferring ScribeNick: Chuck

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.)

[End of scribe.perl diagnostic output]