W3C

Accessibility Conformance Testing Teleconference

02 Aug 2018

Attendees

Present
Shadi, Trevor, Alistair, MaryJo, SteinErik, Wilco, Anne, MoeKraft, Kathy
Regrets

Chair
Wilco, MaryJo
Scribe
Trevor

Contents


Issue 227: Clarifying the relationship between rule application and accessibility conformance https://github.com/w3c/wcag-act/issues/227

wilco: pass and fail directly inherit from EARL

shadi: so technically WCAG does not define pass and fail, it talks about satisfying the requirements, or meeting the requirements
... I did put some editorial suggestions that I think will help this. We often use pass and fail. I think pass fail for rules, and satisfy/not satify for requirements

wilco: How would we then better address this confusion?

shadi: Instead of using satisfy and not satisfy for rules. In this way you could pass a rule, but not satisfy a requirement

wilco: when it comes to success criteria, most professionals do pass/fail/inapplicable, so your proposal is not the common language

kathy: So my understanding is that with the atomic rules you can pass or fail, and the composed rules and determine if it meets the requirement

wilco: technically speaking you can never pass a success criteria, you can just not satisfy. Its more towards making a best effort to show you do not have the listed problems

alistair: there are only a few instances where we have said a success criteria can be passed or met. why don't we just specify our language somewhere that a success criteria has been met/unmet
... i think just changing the language a bit will fix

shadi: Agreed, i think we just need to spell things out and make sure we are clear.

wilco: Assigning shadi to this issue

<shadi> https://lists.w3.org/Archives/Public/public-wcag-act-comments/2018Aug/0000.html

wilco: I will take a look and assign this to me. I will bring it back to the group after we have gotten a resolution

Issue 228: Having a common rule format is not sufficient to ensure consistent interpretation https://github.com/w3c/wcag-act/issues/228

shadi: in my draft i did make a few changes to this section.
... take a look at my draft and then we can come back to the issue.

alistair: what they want is for every line in the test to be specified, i don't think there is much we can do about this

stein: I think we should remove/adjust the sentence. As it is, it would require significant changes in order to work. he has a point, but it is not very solvable
... wants empirical method, but I am not sure it would be as rigorous as he would want

wilco: i agree

shadi: I did also raise an issue on benchmarking and accuracy, and it says the accuracy is measured by what an expert would claim
... does not say what percentage that needs to be, it just says here is how to measure the expert
... and what constitutes an expert, who qualifies for that.
... we will have to think about the judgement behind an act rule

alistair: that is a very broad topic, for example you might want an industry quality mark that says a level of accuracy

shadi: we kinda have that implicitly in the review group. I think there needs to be some review process for rules

alistair: the problem that you have is that W3C is standards and not a quality review.

stein: isn't this what we are doing with our review process already.

wilco: I think the argument is that it is not part of the standards.

shadi: I think we need to reference from the standards that a review process is necessary for the rules
... i do think we make a claim we want to help harmonization, but i am not sure that we are doing that

alistair: if you say you need a group of experts to get together, say 20, you could still have groups in one location e.g. America, Europe, etc. that deals with these things
... and that may not really be harmonization

wilco: Publishing rules in a given format that we have all talked about, i think helps in harmonization by itself

shadi: I don't think we really provide a format, there are multiple ways people could go out and still follow the standard

Issue 230: 3. ACT rules type - clarify example https://github.com/w3c/wcag-act/issues/230

alistair: its the "can" bit that is the problem
... the word can is just making the rule "allowed" to pass, but isn't actually passing it. It should be more direct

stein: In principle, it should be the outcome of the rule that passes

wilco: No, its that the test result is to pass
... I will leave a comment and assign it to myself

Issue 231: 7. Accessibility Requirements - Improvement suggestion https://github.com/w3c/wcag-act/issues/231

wilco: I think we are all good with that.

alistair: out of interest, what is a local law

wilco: state or regional

alistair: we don't have any local laws with that

<shadi> "local or regional law"?

stein: Is there possible confusion, or is it just not applicable to some countries

wilco: the proposal is to take out "local"

Issue 232: 4. ACT Rule Structure - Improvement suggestion https://github.com/w3c/wcag-act/issues/232

wilco: I have already been doing that in some of the stuff i am doing for auto-wcag.

<shadi> +1

wilco: anyone think this is not a good idea?

alistair: I feel like its fairly obvious between the two. The problem with adding extra metadata is that in can cause problems.

wilco: I think the suggestion is just to be explicit

alistair: Would prefer it to be optional?

anne: Wouldn't we have the same problem if it was optional? Have people misusing
... I don't see what it should be optional, should either be there always or never

wilco: I think its worth being explicit about it.

moe: Looking at some of shadi's suggestions, i think this falls in line with his changes and will help.

Issue 233: 8.1 Common Aspects - question https://github.com/w3c/wcag-act/issues/233

wilco: We don't have a way now, but i suppose we could give examples

anne: The could be examples and then a link to a page with items people could pick and choose form

wilco: thinking through practically, would that be like a note that we can add stuff to over time

shadi: we can not make it a normative requirement, it would have to be informal
... the requirement is either to have a detailed description or a reference to a common aspect
... If you want to specify a particular way, you have to make it normative, if you want to leave it open then you can just provide a list of informative references

Issue 234: 11.2 Aggregation Expectations - Update example https://github.com/w3c/wcag-act/issues/234

wilco: seems editorial
... I can update it

Issue 235: Outcomes of atomic tests https://github.com/w3c/wcag-act/issues/235

wilco: This is the pass or fail question we discussed earlier coming back again

anne: We state that rules must be consistent with accessibility requirements. It should not list the rule, but the composed rule. The composed rule should list the accessibility requirement
... we referred to the section on aggregation for how passes should work.

wilco: Please put that in a comment for this issue.

Issue 236: 15.2 Accuracy benchmarking - update definition https://github.com/w3c/wcag-act/issues/236

wilco: going to set up a call with annika.

alistair: I don't understand her comment. Where she is finding that a higher number means that the approach performed worse.

stein: lets take a call and see if she can clarify

Issue 237: 16. Rule Aggregation - Improve example https://github.com/w3c/wcag-act/issues/237

wilco: she is correct, I used aggregation in two different ways
... just need to do some editorial work to clarify this

maryjo: or clarify that there are two levels of aggregation

alistair: the composed rules are not really an aggregation

wilco: assigning to myself and shadi

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.152 (CVS log)
$Date: 2018/08/02 21:02:27 $