Silver Conformance Options Subgroup

01 Jul 2021


Azlan, Jeanne, PeterKorn, Rachael, sajkaj, Wilco_
Bruce, John_Foliot, Kim_Dirks, Todd_Libby

Meeting minutes

Agenda Review & Administrative Items

sajkaj: we promised what we might say re 3rd party content - how do we present this in the outcomes?

Possible not for the August timeframe but hopefully December

Will put forward definition of first and second party.

PeterKorn: Propose where Accessibility is not fully solvable in these guidelines we consider whether we believe that a site that has done everything possible but has not resolved all issues, is this enough for bronze or perhaps an alternative lower level. We can come back to this.

jeanne: Request to clarify what is second party?

<PeterKorn> +1 to avoid legal terms

sajkaj: The user is second party

jeanne: Avoid use of legal terms and clarify definitions.

3rd Party in Visual Contrast Outcomes https://w3c.github.io/silver/guidelines/#visual-contrast-of-text

sajkaj: the environment the user uses is controlled by the hosting platform so visual contrast should be the same as in general. More problematic in author arranged content and media.

PeterKorn: we need to be clear on the scope of visual contrast in txt - in images or only in text content. Won't apply to movies as text is in the image of the movie. Embedded content that does its own rendering may not be so controllable by the host.

<jeanne> +1 to embedded code

<Zakim> jeanne, you wanted to say that media uploads with low contrast should still be continued - graphs and charts from 3rd party, but that's not part of visual contrast yet.

jeanne: like the iframe comment from PeterKorn. Visual contrast is strictly text against a background. Text in images and movies is not currently included.

Do you envision a third party component for every guideline? Or just where we know there is an impact?

PeterKorn: For these three try putting it in the guideline because the scoring may differ

sajkaj: we could look to success criteria from 2.2?

Wilco_: there are examples of user content where font, colour and background are user controlled.

PeterKorn: this is an interesting scenario. One mitigation is the tools to make something inaccessible may have a user setting to allow overriding.

PeterKorn: part of a site doing everythign it can is to not allow this and should therefore get a lower score

<jeanne> Peter: proposes an approach where the scoring is tied to the quality of the accessibility. The third party proposal could augment that scoring that for 3rd party content where the quality is worse will impact the scoring..

we can imagine this is a guideline. We might have a different quality scale for machine generated where the highes bar is the same as human generated but we are prepared that if using machine generation if they are getting 1 word wrong out of 20, whatever our quality rating for machine may not have as high a requirement as for human generated.

<LuisG> thanks!

Rachael: the need for simplicity. Scoring is complex. We should keep this in mind. Like what PeterKorn suggests but would like to ask is there a way to make this simpler?

<Zakim> jeanne, you wanted to say that I think we have two approaches: Where there is a separate task to improve the 3rd party and where there isn't.

jeanne: we have two categories of guidelines. E.g. visual contrast if it's embedded content there's not anything the author can do to improve it, but with captions there is. They can list where captions are available. We can give them things to do. If there's something we want them to do we can weave into the guidelines.

Wilco_: not sure there is "nothing" that can be done. There may be practical limitations. Things we can't do just yet is the nuance.

PeterKorn: don't like the idea that a document that is not easily changeable today. Would love to see if we could we are going to be updating techinques periodically for machine remediated content and that will become the bar as the technology improves.

<Zakim> jeanne, you wanted to talk about sorting guidelines on the basis of practical present day solutions and how to handle those which don't have it.

<Rachael> I think going through the exercise is useful because then we can extrapolate up as needed

jeanne: that was in agreement with the proposal. The techniques are not normative and can be updated when we want. Things we can do something about can be in and can flow up to the scoring. Catch all for what we practically cannot do today. As things become more possible we put those into the methods and that can flow into the scoring.

PeterKorn: Wonder about taking jeanne's suggestion and applying to alt text. Add on to rating 1 for third party images you have done all possible to ensure you are prompting fro.. and maybe rating 2, you have followed the below for machine generated.

For rating 2 in 2022 there is no machine generated we will update.

jeanne: in the methods we can put that kind of techniques. Wouldn't worry about the ratings and scoring just yet.

PeterKorn: this isn't about exempting. If you've done all you can you get a minimum score. How do we convey that without an example?

jeanne: we can do that with aternative text

what can be done today? Turn that over to Wilco_ 's group

Wilco_: by doing this on methods, you change the requirements and that should be a normative change

sajkaj: looking at current draft for alternative text - saw accounting of how prevalent your alt text was. Could not see what percentage was deemed appropriate.

<Zakim> sajkaj, you wanted to ask whether we have "appropriate" definition?

PeterKorn: Sounds and updated guideline is not possible for December

Wilco_: Not working on guideline, but the methods

jeanne: Could we make a third party outcome? PUtting exceptions around so it only applies to third party

sajkaj: should be in the definitions

Wilco_: curious what people think makes third party that it should be exempt

jeanne: it's not exempt - its additional things that must be done

PeterKorn: when you have exhausted all your efforts, here is how you can still include and get a minimum score

sajkaj: Alistair has interesting examples in the WBS last AGWG meeting

PeterKorn: scoring is changing so trying to do this in scoring would mean throwing work away. Methods isn't headed in this direction. WOuld be great if we could have a top level treatment rather than making scoring more complex

jeanne: would like proposals. What practical solutions can be done today that we can recommend without jumping to the rating scale?

Focus on what should people do?

sajkaj: Alt text - they should prompt for it. Give pointer how to write good alt text. Highlight where it hasn't been done. Machine checking for poor/incorrect alt text

jeanne: See steps to conform as the generic catch all. Where we have them list here is what the author must do.

Make the proposal - this is what you have to do

PeterKorn: We can do that - not exhaustively but we can for a few.

jeanne: I'd put it above steps to conform, in the guideline changes

Minutes manually created (not a transcript), formatted by scribe.perl version 136 (Thu May 27 13:50:24 2021 UTC).


Succeeded: s/I'm in I think. No one else is here, so I might be on the wrong call//