Silver Task Force & Community Group

07 Aug 2020


jeanne, Chuck, Todd, Rachael, sajkaj, kirkwood, ChrisLoiselle, MichaelC, Fazio, JF, mgarrish, OmarBonilla, Lauriat, bruce_bailey, SBH, pwentz, Francis_Storr, Nicaise, CharlesHall, AngelaAccessForAll, Julia, KimD, Laura, Crispy, Jan, Gundula, jon_avila
Shawn, jeanne


<scribe> scribe: sajkaj

<scribe> chair: chuck

Presentation of Conformance proposal for AGWG and Silver groups

ca: notes today is about John Foliot's conformance proposal
... Questions today only for clarification, please

jf: begins screen share and will describe ... also post
... will focus on scoring
... assert not all requirements equal and should not be scored equally
... smallest unit should be act as group agreed
... believe even more complex interactions can map back to atomic act tests
... will reference sc because referencing existing rules, might mean methods in Silver definitions
... looks at device orientation example ...
... notes huge impace on low vision -- it matters
... think also of sdisplay on motorized scooter
... may also impact coga
... uses functional categories and a merged value
... sc 1.4.11 nontext contrast -- notes 3 tests
... uses 3 tests as multiplier
... turns to poorly formatted tables ...
... by comparison looks at grouped form elements -- notes it impacts more users

<Zakim> Rachael, you wanted to ask if these maximum scores are set at the AG level and always consistent

rm: asks whether maximum score is static or variable?

jf: depending on atomic test and what's looked for
... expresses desire to rely on act -- let's not drop those
... cumulative score is on smaller things happening

<jon_avila__> I see the challenge will be in coming to agreement on these numbers. I've done this before and it's hard to get agreement on the exact number and the number might change and then that impact scoring.

jf: where there's only one, e.g. lang of doc, it's a singular item
... headings on page would serve the opposite example -- some number ...

sbh: asks about spectrum, i.e. hearing
... how to account for ranges?
... may split into columns as in vision/low vision

jf: will need to figure out categories
... different requirements impact cifferent categories differently

ja: like the approach, but hard to get agreement on what the values should be/are

<Fazio> we discussed this last year

ja: if one changes weights, it will impact and may lose confidence in our model

jf: it's a risk -- we can manage it as a group
... notes we need to be scalable and allow for future tech
... different things will have different maximums

<Zakim> jeanne, you wanted to ask how you will handle important groups that have unique needs so they will always have less points?

js: how deal with more unique needs? e.g. flashing or hearing?
... these are usually in their own class? how to equalize

jf: we know severity is off the chart -- e.g. flashing we may decide to boost the point because of severity
... we simply need to document how we arrive at values

sbh: have you considered doing surveys to establish values?
... also concerned that literacy levels in different locales will affect appropriate weighting

jf: we may have multiple atomic tests where something like: "1 of 3 must apply"
... had hoped to give examples of these, but ran out of time to formulate
... definitely don't have all the answers yet

pw: how are the numbers arrived at?

jf: just illustrative for now; we could do as group or base on surveys
... consider that open for discussion
... now tending toward a 5 point scale even though my examples are 3/4

<jon_avila__> Does this take into account context? Where the issue appears where there are mitigating factors on web page?

df: recalls this was deep discussion about a year ago; and recalls spoons failure example of factoring in many small obstructions
... believe we settled on letting the claimant make their own assertion on that

ca: reminds we should stay on clarifying questions today ...

jf: moving on and looking at rm's views and paths ...
... to me a view is whatever's on screen at any moment
... in a single page, still all that makes it up
... incl modals, etc
... notes it's a visual perspective, but this is what's presented to users now
... either the screen is refreshed or the user moves on ...
... paths will be a superset of views
... thus nuclear components are covered in views
... framework can be tested and claimed
... believe paths will always include editorial content
... want to be able to test as empty framework; then again with content

rm: notes what tool can do; but what still requires review e.g. images have alt, but is the alt appropriate?

jf: react and angular could make a claim under this proposal, though
... would help with how to pick tooling
... would be good if a design system could claim conformance
... would apply to sas or template systems, i.e.wordpress
... notes examples of good frameworks where users add the a11y issues in apply their content
... content complete would then be web sites, perhaps epubs, pdfs, etc

gn: where web tech apps, e.g. ecommerce

jf: looks at wp woocart plugins which could be tested, but still no inventory
... could distinguish responsibility if a user misapplies the woocart adding content
... measuring the app as a template is useful; as is finished site with content
... this is important to me and will come back to this in a few slides ...
... notes reqs like lang can be tested across frameworks and content
... color contrast won't until there's content
... design systems and templates would have color
... landmarks probably not in design system, but probably yes at template
... e.g. findable help; not just the mechanics but that all users can find it in a given view
... maybe in a template findable help is testable
... big point is this distinction between building blocks and finished content sites

a: how does scoring work without the editorial parts?

<Chuck> ach Ch

jf: claimant scopes the claim ...
... no matter which vertical being tested will provide a max number; will always have a max high
... phps max low
... in my examples perfect site would score 955 -- but it's arbitrary
... does metalic breakdown

ca: is the dashboard made up of atomic tests?

jf: yes, their sum
... once we agree on points assignable
... notes pdf example that not all sc applies, so maximum will be smaller
... notes that percentile scoring is supported in this model

<Zakim> Lauriat, you wanted to ask about equal user experience, but a different foundational technology used.

sl: like how applicability is called out ...
... q abous adding scores ...
... completely equal user experience; but each uses different tech
... where the second only tested final ...

jf: should roll up to same score
... functional outcomes will have methods/ atomic, something that still gets us there
... scoping is now by verticals; we might make it extensible
... idea is to scope to the vertical; and scope without content"out of the box"
... if the cms supports choices that apply; we can measure and test and provide scoring
... example task based activity of uploading video, captions, described video tracks, transcript -- if systems upports that accessibly it should be documentable

ja: shouldn't we have minimal thresholds in our functional categories?

jf: phps

ja: notes one can do will in multiple categories but miserably fail one or two

jf: if we believe it's necessary, seems we could still do that
... lastly, need for dashboards
... notes his url for multiple dashboards
... it's something our customers want
... a way to stay current
... launch could be "when my conformance statement is made"
... orgs want to monitor their health -- ongoing
... my question to the group -- we have score as of a certain date --m what do we do?
... option one it's the score until retested; which will be inaccurate over time
... another option is need for retest on some periodicity
... builtin degradation for lack of retesting

gn: asks about assertions tied to version release

ca: tuesday is a good time for that discussion!

jf: because the web site is dynamic
... it's ongoing evolutionary stuff

topic announcements

js: reminder of all day deep dive Tuesday
... will also send out agenda in advance
... please respond on wbs
... also have updated decision page
... the page of recorded decisions; not the decision policy draft

ca: call adjourned!

<jon_avila__> Thanks John!

<Rachael> trackbot end meeting

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: 2020/08/07 19:06:40 $

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/independence/orientation/
Default Present: jeanne, Chuck, Todd, Rachael, sajkaj, kirkwood, ChrisLoiselle, MichaelC, Fazio, JF, mgarrish, OmarBonilla, Lauriat, bruce_bailey, pwentz, Francis_Storr, Nicaise, CharlesHall, AngelaAccessForAll, Julia, KimD, Laura, Crispy, Jan, Gundula, jon_avila
Present: jeanne Chuck Todd Rachael sajkaj kirkwood ChrisLoiselle MichaelC Fazio JF mgarrish OmarBonilla Lauriat bruce_bailey SBH pwentz Francis_Storr Nicaise CharlesHall AngelaAccessForAll Julia KimD Laura Crispy Jan Gundula jon_avila
Found Scribe: sajkaj
Inferring ScribeNick: sajkaj

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]