Silver Conformance Options Subgroup

11 Feb 2021


Bryan, Jeanne, Jeanne Spellman, Jemma, JF, John_Northup, KimD, PeterKorn, sajkaj, sarahhorton, Spellman
Azlan, Bruce, Wilco

Meeting minutes

Agenda Review & Administrative Items

Janina: Some inconsistencies between google doc and working draft.

Revisiting Principle #6 re Critical Failures

Janina: Revisiting principle 6; what is a critical error.

Jeanne: Rachael and I do not disagree on critical errors; it's the final resolution of the critical error. Concern that critical error for flashing (zero) may be averaged out by higher scores in other areas and thus overlooked.

<PeterKorn> https://w3c.github.io/silver/guidelines/#conformance-levels-0

<PeterKorn> For content that conforms to the bronze level: The total score and score within each of the functional categories MUST be at least 3.5; and Views and processes MUST NOT have critical errors.

Peter: Doesn't see how the calculation method could result in the outcome that Jeanne describes.

Jeanne & Peter: Discussion on calculations and possible outcomes.

Rachael: Low scoring items that are not critical do balance out, but the purpose of the critcial error is to be a hard stop. Would not agree with a model that passed critical errors.

John F: Ensure we are not focusing on flashing; other errors are critical.

Janina: ... but flashing is life-threatening...

John F: Other things can be life threatening too depending on content and context.

Peter: There has been a discussion about the difference between content itself being life-threatening, and the context being life-threatening.

Peter: Marvel film trailer may trigger a seizure. Does it make a difference if the content is third-party? Do we treat this content differently if there is a warning?

Janina: Proposal to look at WCAG 2; possibility to programmatically correct problems; e.g., user agent detecting and stopping flashing.

<Zakim> sajkaj, you wanted to reconsider programatic solutions to flash, audio on load, etc

Brian: Challenging/impossible to catch every use case; need to keep it simple to drive adoption.

Jemma: If blocking a video is preferable to warning, how to score that?

Janina: Don't know yet.

Peter: Focus on creating use cases; then will move toward solutions.

Jeanne: Use cases valuable at this stage.

Peter: More to discuss about bugs?

Peter: Do we have an opinion on what should happen when a critical error is found?

John F: Do you mean time frame for remediation? A score is at a certain point in time...

John F: Discussion of lag time between scoring and remediation - if no followup evaluation, then old score is still the latest.

Janina: Scheduling by priority...

Jeanne: Met yesterday with DHS Trusted Tester, discussed measuring a11y errors compared to other bugs. They were cautiously enthusiastic.

Jeanne: Goal is a11y bugs treated same as other bugs. Just as critical as security or usability bugs.

Jeanne: They are enthusiastic about the point system and critical errors because it helps with VPAT

Bryan: Likes prioritizing by severity.

Sarah: Are we talking about bugs as equivalent to critical errors, or bugs with different degrees of severity?

Janina: Trying to specify that a11y bugs are as important as other bugs; not de-prioritized.

<Zakim> JF, you wanted to note that a11y, security and privacy bugs are *MORE* critical than other software bugs, due to legal obligations.

<JF> Precious metals like gold and silver are measured using the “troy” system where there are 12 ounces in a pound. If you normalize everything to mass a pound of feathers is 454 grams and a pound of gold is 373 grams, therefore the feathers are technically heavier.

John F: Bugs related to security, privacy, a11y... not all bugs are created equal.

John F: Legal obligation puts a11y ahead of other bugs.

John N: Likes the idea of commingling a11y and other types of bugs, so that a11y isn't diminished.

Peter: So much content is programmatically-generated--a bug in the script that generates content over several pages, which may become critical.

Janina: The challenge is finding those instances.

Sarah: Perhaps not refer to "critical bugs" due to confusion with existing term "critical errors".

Peter: Discussing equivalence across diff types of bugs (security vs a11y)...

Jemma: Seeking clarity around critical errors

Sarah: Not quite far enough along with Errors work to definitively answer.

Bryan: Is the word "critical" a stumbling block, even though the general principle is appropriate...

Sarah: Principle is sound. Use case introduces the issue of time... should that factor into the conformance model...

John F: Just because something is critical, the fix may not take immediate effect. Score could change moment by moment as content changes...

<JF> Time issue *IS* real, but shold it have an impact on "Score"?

<Jemma> I think that Peter is touching the issue of impact scope of "critical error"

Peter: Acknowleging the time issue, while we also try to consider a11y in the context of other bugs...

Sarah: Much about culture/practice/procedures that should weigh into conformance - beyond just products...

<PeterKorn> bye bye

Minutes manually created (not a transcript), formatted by scribe.perl version 127 (Wed Dec 30 17:39:58 2020 UTC).


Maybe present: Brian, Janina, Peter, Rachael, Sarah