<scribe> Scribe:ChrisLoiselle

Sub-groups: please review ported content in The Draft

<Lauriat> Draft: https://w3c.github.io/silver/guidelines/

<Joshue_108> sortofpresent+

Jeanne: For templates in Google docs, the templates changed. We did move material into different places in FPWD .

<Andy> Andrew Somers present+

<jeanne> https://w3c.github.io/silver/guidelines/#text-alternatives

Jeanne: There have been some style changes as well.
... We've been sorting details of ideas. For text alternatives, a number of methods were written for one outcome.

If rating was in method, we would have broken accessibility testing tools. This flagged where we were putting the rating.

<Andy> And here I thought twists were a glazed donut...

<CharlesHall> Twisty? I refer to these as details and summary

One rating for all of the methods was written by Makoto's group. Now these live in the methods, critical failures and rating for text alternative available.

<CharlesHall> The WAI site uses details/summary

MichaelC: The element uses is the details / summary . Open to improving accessibility of the expand collapse element with headings, aria-label , etc.

Jeanne: Feedback definitely welcome on it.

Wilco: I like the way this is going. There are a lot of details to be worked out , but going in right direction.
... Methods are included. Would that make it harder to add methods to normative text?

Jeanne: Do we move list of methods back to how to?

MichaelC: do we want methods at the outcome level?

Jeanne: Yes, we do.

Janina: I'm concerned on decisions for normative / non-normative . I think we can ask it , but saying anything more than that may be an issue.

Jeanne: Should less things be normative?

<CharlesHall> a WAI page that uses details/summary does so with H3s in summary, and contains an expand and collapse all: https://www.w3.org/WAI/fundamentals/accessibility-principles/

Janina: Yes. When someone challenges a claim, it needs to point through scoring approach...normative vs. informative is based on what our models are.

Jeanne: I think we have addressed the rubric. The rating system allows people to pass at less than 100 percent. We have substantially conforms, but haven't written it out. We can review that further.

Rachael: I agree methods shouldn't be normative, but it should stay where it is within the hierarchy.

Jeanne: We will probably have to link out to a list of methods.

Rachael: The link should be from here to the methods.

Wilco: Why is critical errors exclusive to process?

Jeanne: It is different for every outcome.
... talks to XR and emerging technology. Those have no critical errors.

I.e. clear words, critical errors is more of a cumulative score. Flashing is an interference. Another example is alt text on an element that prohibits someone from completing a task.

Wilco: Icons and lack of alt text a critical error if it is not in a process?

Jeanne: We are trying to give some criticality measures to images and alt text. If an image is not important, the company should be able to pass . If it is something blocking , then the company should not pass.

Rachael: Not interference vs. failure techniques. I.e. stops task completion or a break in task.

<Lauriat> +1, that.

The concept is that it stops someone from finishing what they are doing.

<MichaelC> https://www.w3.org/TR/WCAG22/#cc5

<Zakim> MichaelC, you wanted to say I think critical errors correspond to non-interference in WCAG 2

MichaelC: Non interference link provided by Michael C.

Wilco: If it includes not on the web, it is far broader and not sure how we could test that.

<Fazio> smart homes

I.e. the fire alarm example. I get a notification that I should get out of my house...

Wilco: I'll provide another example.

<Sheri_B-H> loudspeaker doesn't work for my deaf daughter either

DavidF: I signed up for disaster notifications . I get an alert if air pollution is bad etc. The police usually would come down street to tell you to evacuate . It is relevant for accessibility community to have alternative notifications to alert them and have a text alternative to this issue.

Jeanne: If the notification doesn't have alt text, then it would be a critical failure.

Rachael: A conformance claim would talk to what tasks you are testing.

<Andy> As an examples, considering visual contrast, the "task" is reading, so a thought on critical failures is along these lines (I think)

<Andy> Hard or critical visual contrast failures that get an automatic 0:

<Andy> -- Content that is intended to be read has a font color is specified, but not a background color or vice versa.

<Andy> -- Lockout the user’s ability to zoom readable text content larger.

<Andy> -- Lockout the user’s ability to use “reader mode” for long-form readable content (articles)

<Andy> -- -- -- Does not apply to lock outs related to paywalls or content protection.

<Andy> -- Content that is intended to be read is using a font-size smaller than 12px.

<Andy> -- Content that is intended to be read is using a color against the background of less than APCA 40%.

<Andy> -- Text that is intended to be read overlays other text or content such as an image with substantial interfering content (i.e. text over a “busy" image).

<Andy> Hard or critical visual contrast failures that get an automatic 0:

<Andy> -- Content that is intended to be read has a font color is specified, but not a background color or vice versa.

<Andy> -- Lockout the user’s ability to zoom readable text content larger.

<Andy> -- Lockout the user’s ability to use “reader mode” for long-form readable content (articles)

<Andy> -- -- -- Does not apply to lock outs related to paywalls or content protection.

<Andy> -- Content that is intended to be read is using a font-size smaller than 12px.

<Andy> -- Content that is intended to be read is using a color against the background of less than APCA 40%.

<Andy> -- Text that is intended to be read overlays other text or content such as an image with substantial interfering content (i.e. text over a “busy" image).

<Andy> Hard or critical visual contrast failures that get an automatic 0:

<Andy> -- Content that is intended to be read has a font color is specified, but not a background color or vice versa.

<Andy> -- Lockout the user’s ability to zoom readable text content larger.

<Andy> -- Lockout the user’s ability to use “reader mode” for long-form readable content (articles)

<Andy> -- -- -- Does not apply to lock outs related to paywalls or content protection.

<Andy> -- Content that is intended to be read is using a font-size smaller than 12px.

<Andy> -- Content that is intended to be read is using a color against the background of less than APCA 40%.

<Andy> -- Text that is intended to be read overlays other text or content such as an image with substantial interfering content (i.e. text over a “busy" image).

JohnK: Critical path discussion , one issue is the three door scenario. You are choosing a direction, woman's room, men's room and other. The one door that is not labeled, is it part of the critical path?

<Andy> Hard or critical visual contrast failures that get an automatic 0:

<Andy> -- Content that is intended to be read has a font color is specified, but not a background color or vice versa.

<Andy> -- Lockout the user’s ability to zoom readable text content larger.

<Andy> -- Lockout the user’s ability to use “reader mode” for long-form readable content (articles)

<Andy> -- -- -- Does not apply to lock outs related to paywalls or content protection.

<Andy> -- Content that is intended to be read is using a font-size smaller than 12px.

<Andy> -- Content that is intended to be read is using a color against the background of less than APCA 40%.

<Andy> -- Text that is intended to be read overlays other text or content such as an image with substantial interfering content (i.e. text over a “busy" image).

<Andy> Hard or critical visual contrast failures that get an automatic 0:

<Andy> -- Content that is intended to be read has a font color is specified, but not a background color or vice versa.

Wilco: Critical failure the same as critical error?

<Andy> -- Lockout the user’s ability to zoom readable text content larger.

<Andy> -- Lockout the user’s ability to use “reader mode” for long-form readable content (articles)

<Andy> -- -- -- Does not apply to lock outs related to paywalls or content protection.

<Andy> -- Content that is intended to be read is using a font-size smaller than 12px.

Jeanne: We are changing to critical error.

<Andy> -- Content that is intended to be read is using a color against the background of less than APCA 40%.

<Andy> -- Text that is intended to be read overlays other text or content such as an image with substantial interfering content (i.e. text over a “busy" image).

<Andy> Hard or critical visual contrast failures that get an automatic 0:

<Andy> -- Content that is intended to be read has a font color is specified, but not a background color or vice versa.

<Andy> -- Lockout the user’s ability to zoom readable text content larger.

<Andy> -- Lockout the user’s ability to use “reader mode” for long-form readable content (articles)

<Andy> -- -- -- Does not apply to lock outs related to paywalls or content protection.

<Andy> -- Content that is intended to be read is using a font-size smaller than 12px.

<Andy> -- Content that is intended to be read is using a color against the background of less than APCA 40%.

<Andy> -- Text that is intended to be read overlays other text or content such as an image with substantial interfering content (i.e. text over a “busy" image).Hard or critical visual contrast failures that get an automatic 0:

<Andy> -- Content that is intended to be read has a font color is specified, but not a background color or vice versa.

<Andy> -- Lockout the user’s ability to zoom readable text content larger.

<Andy> -- Lockout the user’s ability to use “reader mode” for long-form readable content (articles)

<Andy> -- -- -- Does not apply to lock outs related to paywalls or content protection.

<Andy> -- Content that is intended to be read is using a font-size smaller than 12px.

<Andy> -- Content that is intended to be read is using a color against the background of less than APCA 40%.

<Andy> -- Text that is intended to be read overlays other text or content such as an image with substantial interfering content (i.e. text over a “busy" image).Hard or critical visual contrast failures that get an automatic 0:

<Andy> -- Content that is intended to be read has a font color is specified, but not a background color or vice versa.

<Andy> -- Lockout the user’s ability to zoom readable text content larger.

<Andy> -- Lockout the user’s ability to use “reader mode” for long-form readable content (articles)

<Andy> -- -- -- Does not apply to lock outs related to paywalls or content protection.

<Wilco> https://w3c.github.io/silver/guidelines/#dfn-critical-error

<Andy> -- Content that is intended to be read is using a font-size smaller than 12px.

<Andy> -- Content that is intended to be read is using a color against the background of less than APCA 40%.

<Andy> -- Text that is intended to be read overlays other text or content such as an image with substantial interfering content (i.e. text over a “busy" image).

<Fazio> +1

<Andy> YIKES

<Andy> I'm sorry my user agent pasted that multiple times

<Andy> Sorry about that!

<Andy> SO sorry ignore what I pasted, it got pasted multi times

Jeanne: Made a note, we need to change critical failure to critical error in alternative text

Wilco: Rating at the outcome, the rating is not going to be a decimal score, correct? Or do you apply them per method? Question on scope.

Jeanne: Once you hit outcome level, it is 0 - 4. It leaves the percentage at that point.
... Talks to sensitivity of testing. If we had only a percentage outcome to final score, it was overly sensitive. Difference between 86 and 83 had too much weight.

We want to reflect how good their accessibility is, rather than just a score.

We want to take it in bands, once you hit that band, the percentage no longer matters and the rating moves forward for final score.

Wilco: I'm not sure that answered the question.
... Input , output. The input to the rating is a percentage. The output is a rating from 0 - 4. The methods , all of them , are averaged and goes into the rating?

Jeanne: It goes by guideline.
... For text alternatives, yes they are averaged.
... From sub groups, are there anything you want to discuss?

MichaelCrabb: I've added in some detail.

Walk through the scoring and reminder to fill out the scoring survey

<Rachael> https://drive.google.com/file/d/1cJDPHgKoO4D3eY5fdquv-8oQMqjzu1z3/view?usp=sharing

Rachael: downloads to excel to review formulas
... We have our guidelines, outcomes, type of test, view 1, 2 , 3. # of failures in a path, total percentage passed , total errors in path, then test approach.

For text alternatives for images of text, 90 percent had alternative text. For functional images, 100 percent. For clear words, we score a 1. Scoring captions, it is a rating scale. We score that a 2

Once we score that, the outcomes tab in spreadsheet shows aggregate and suggested rating .

<Zakim> jeanne, you wanted to say error in text alternatives separate methods

Based on that, the final score is given looking at functional categories, above a 3.0 to get Bronze level.

Jeanne: Talks to merging of methods into outcome. 4 methods. One outcome. One score.

<Wilco> +1 to Jeanne's concern

Rachael: Question on methods vs. outcomes and merging and scoring.

Sarah: The way I understand is that the score is zero if a critical error.

Jeanne: I agree.

Sarah: I fail that it isn't addressed correctly then.
... I think from a testing perspective, if I have a critical failure, why would I continue working through this sheet?

Andy: One of things on the scoring model was content intended to be read , navigational element vs. not intended to be read, i.e. copyright. If testing tools can't make that decision, that we need to group them together . There is a compelling reason from a design standpoint.

I.e. terms of service , small text vs. the actual desired outcome of using the product.

Rachael: I agree on both points from Jeanne and Andy.

<Chuck_> chris: My comment is based on conversation from Sarah and critical failures and level, bronze or inaccessible.

<Chuck_> chris: What we are calling it, "critical failure" and having it a zero on fpwd, and how that would relate...

<Chuck_> chris: Someone should go through and test all the critical failures. But if you find one and stop, you still have zero. How that factors in is my main concern.

Rachael: Critical errors were introduced for severity purposes.

<Andy> Would it be useful to have a "get out of jail free" class... a class that can be added to a tag to dismiss it from automated tool testing, OR testing at a particular level/

<Andy> <div class="non-content"....

It is a judgement call, if it is only one error , then it also needs to be fixed.

Wilco: By having two methods for alternative text, you are introducing multiple ways of scoring. I'm worrying about opening up for abuse based on ways of measuring.

Jeanne: I feel we address Sarah's comments, when a tester hit a critical failure, they can stop. I feel Rachael shows in the spreadsheet is valid.

The intention is to showcase with the spreadsheet how it could work.

<Jan> request of Wilco - can you expand on how you see critical errors being abused?

<Fazio> frequency is critical for cognitive - mental fatigue

<Fazio> strongly disagree

Sarah: This is helpful. I like the critical errors aspect. I think we are struggling on audit , accounting and the scoring of the outcome. Frequency and severity are separate measures. It is headed in a great direction. I think more is needed.

Is this intended for an audit template or a score card so to speak.

Wilco: On text alternative , there is an equal weighting of methods.

<Rachael> +1 to wilco's concern

Wilco, could you place that in text within IRC , couldn't type all that, sorry!

<Fazio> IMPORTANT: frequency is critical for cognitive

<Wilco> +1 to splitting

Jeanne: Wilco raised a good point on scoring and whether things are watered down too much. This is a good example to split decorative images to own outcome.

<Rachael> In the cognitive methods, frequency can be a critical error

DavidF: Frequency is critical for cognitive. It is a show stopper for cognitive.

<JustineP> +1 to David's point

Jeanne: Clear language is an example of how we are handling that and we can point to as we migrate.

Jeanne, I need to drop, can you send out the minutes?

<st> +1 to Dave's point for cognitive load in education as well

<Andy> Cheers

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)

Default Present: jeanne, Crispy, ToddLibby, JakeAbma, mikecrabb, Fazio, JustineP, sajkaj, Lauriat, MichaelC, CharlesHall, sarahhorton, Sheri_B-H, Rachael, Francis_Storr, Joshue_108, Makoto, shari, kirkwood, Jan, bruce_bailey, SuzanneTaylor, ChrisLoiselle, in, hte, next, bit
Present: jeanne Crispy ToddLibby JakeAbma mikecrabb Fazio JustineP sajkaj Lauriat MichaelC CharlesHall sarahhorton Sheri_B-H Rachael Francis_Storr Joshue_108 Makoto shari kirkwood Jan bruce_bailey SuzanneTaylor ChrisLoiselle
Found Scribe: ChrisLoiselle
Inferring ScribeNick: ChrisLoiselle

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.)