– DRAFT –
Silver Task Force & Community Group

<Lauriat> https://lists.w3.org/Archives/Public/w3c-wai-gl/2022OctDec/0105.html

Preview of next week

Lauriat: AG will meet this coming Tuesday

Lauriat: However, no Silver call next Friday

Review homework: more general outcomes with specific user needs by disability

Lauriat: specificuser needs by disability

<Makoto> https://docs.google.com/document/d/1dOESQPFibABFOwIYnHRGgyggswEKjqHTXmrw4hl1o48/edit#heading=h.dxv4bslptw23

Makoto: may have taken a different approach

Makoto: reviewed main functional needs and user needs docs

Makoto: stuck to 7 types of images enumerated by WAI tutorials

Makoto: wrote outcomes re alt text

Makoto: also tried writing methods

Makoto: notes he does a11y audits for clients

Makoto: have been collating rankings from that work on an annual basis; collating in reports for common issues

Makoto: Suggesting use of alt decision tree

<Makoto> https://docs.google.com/document/d/1dOESQPFibABFOwIYnHRGgyggswEKjqHTXmrw4hl1o48/edit?usp=sharing

Makoto: suggesting outcomes and methods for the 7 image types can be derived from the decision tree

<jeanne> Alt Decision Tree

[people are looking at Makoto's doc]

Chuck: lots to absorb here

Chuck: Very excellent! Need to review to comment, but really looks great

Lauriat: agree; particular articulation of user needs

<Zakim> jeanne, you wanted to say the link to the user needs

jeanne: like the clear tie to user needs in the methods

jeanne: Believe this demonstrates taking functional needs from that articulated doc; then generate userneeds followed by specific methods

Makoto: not sure whether this approach will work for other criteria

<Chuck> +1 on "mind changing" :-)

jeanne: Like how we can go directly to user needs from the functional

Lauriat: this can also inform other parts of our spec; e.g. common images like print

Lauriat: could even support image recognition generated alt for common images

<Chuck> ack

JenStrickland: trying to understand our assignment

Lauriat: yes, believe decision tree can be extended to support this beyond alt

<JenStrickland> Ah, yes! Thank you, Lauriat, the tree is great!

<Lauriat> https://www.w3.org/WAI/tutorials/images/informative/

Lauriat: walks through an example of how that might work ...

Lauriat: even to ACT style decisioning

Lauriat: e.g. two phone numbers one with a fax icon, the other with a phone icon

kirkwood: think it's fantastic; but we may have something not covered

kirkwood: conveying meaning e.g. safety protocols of inserting electrical plug into an outlet in different countries

kirkwood: good description could be a safety issue

<Lauriat> +1, great example of criticality of accuracy

<Zakim> Lauriat, you wanted to explore that with AI

Lauriat: believe there may be ways to support this

Lauriat: one image might be what not to do while another is what to do

Lauriat: user could ask questions to get detail from AI

Lauriat: provides keyword results along with confidence scores

<Zakim> jeanne, you wanted to ask Makoto if he has addressed this

Lauriat: so park with people and buildings in the background would likely start with park, but could be queried "are there people?"

<kirkwood> well done Makoto

jeanne: asking makoto whether has considered where criticality fits?

<Lauriat> Maybe an example of the decision tree that leads to "complex images": https://www.w3.org/WAI/tutorials/images/complex/

Makoto: same image can have different meanings in different contexts

<kirkwood> wether author intent is present or not? is very important

jeanne: think challenge that user need in method may be informative; but criticality may need to be normatively specified

jeanne: or a second outcome

Lauriat: believe criticality already fits in the tree; see above link

<Chuck> +1 to "it fits"

Lauriat: documentation would explain how

kirkwood: would want to be satisfied that author intent is captured in the process

<Lauriat> +1 to that example of maturity of process to reliably achieve the outcome

<Zakim> Lauriat, you wanted to note the decision tree example

SuzanneTaylor: did a similar approach in my homework

<SuzanneTaylor> https://docs.google.com/document/d/1YFLroVspFQSEqUgAcI8qPS2CI6EboZcHA43FQNWpt-8/edit?usp=sharing

SuzanneTaylor: phps the user needs

SuzanneTaylor: guidelines high level

SuzanneTaylor: outcomes focussed on type of audio

<kirkwood> just requested access

<SuzanneTaylor> https://docs.google.com/document/d/1YFLroVspFQSEqUgAcI8qPS2CI6EboZcHA43FQNWpt-8/edit?usp=sharing

SuzanneTaylor: then user needs become very specific and include functional needs

SuzanneTaylor: then the methods become very specific for specific tech

<Chuck> I am denied

Lauriat: thanks Suzanne for walking through her examples; very helpful

jeanne: allows detailed info to be included as informative

SuzanneTaylor: should help innovaters by describing goals

<Lauriat> +1, I really like that start-with-an-ideal approach

SuzanneTaylor: believe we could inspire and direct innovation

Lauriat: starting from that point provides a structured approach to the research that's needed

Lauriat: so that takes us away from what html does today which is very important

<SuzanneTaylor> +1

<jeanne> +1

+1

<Makoto> +1

jeanne: believe we should present both approaches to AGWG

jeanne: notes we eventually come back to an outcome

jeanne: but original task was to get rid of outcomes

jeanne: so AGWG feedback would be helpful for choosing a WCAG3 direction

<Rachael> +1

Lauriat: believe that's a good next step; audio example with decision tree would help with that presentation

jeanne: asks suzanne for example decision tree

<Lauriat> Closest I can find, maybe a good starting point

SuzanneTaylor: yes; for each user need? or to decide which user need?

jeanne: probably for which would be appropriate for which method

jeanne: makoto's allows choosing which method

<Lauriat> Excellent, thank you!

jeanne: believes we're scheduled to present this on Tuesday 6 December; so have a little time

jeanne: requests to move