16:00:36 <RRSAgent> RRSAgent has joined #ag
16:00:36 <RRSAgent> logging to https://www.w3.org/2021/12/07-ag-irc
16:00:49 <ToddL> ToddL has joined #ag
16:00:55 <GreggVan> GreggVan has joined #ag
16:01:07 <laura> present+ Laura_Carlson
16:01:08 <bbailey> bbailey has joined #ag
16:01:10 <agarrison> present+
16:01:11 <shadi> present+
16:01:14 <GreggVan> present+
16:01:18 <Rachael> present+
16:01:20 <Nicaise> Nicaise has joined #ag
16:01:25 <Nicaise> present+
16:01:26 <Jennie> Chuck: We are working on technical issues and will start the 2nd minute after the hour
16:01:37 <Lauriat> Present+
16:01:41 <Jen_G> Jen_G has joined #ag
16:01:41 <bbailey> present+
16:01:49 <Jen_G> Present+
16:01:52 <mbgower> mbgower has joined #ag
16:01:54 <Raf> Raf has joined #ag
16:01:54 <mbgower> present+
16:02:01 <GN015> GN015 has joined #ag
16:02:10 <kirkwood> present+
16:02:11 <Jennie> Chuck: Thank you for coming. We need a scribe for the 2nd hour.
16:02:17 <Regina> Regina has joined #ag
16:02:20 <Raf> present+
16:02:22 <alastairc> present+
16:02:28 <JakeAbma> JakeAbma has joined #ag
16:02:33 <JakeAbma> present+
16:02:41 <Jennie> mbgower - like this? https://www.w3.org/WAI/GL/wiki/Scribing_Commands_and_Related_Info
16:02:52 <Jennie> Chuck: Anyone want to introduce themselves?
16:02:54 <MelanieP> MelanieP has joined #ag
16:02:55 <ToddL> present+
16:02:59 <MelanieP> present+
16:03:02 <kirkwood> present+
16:03:08 <Jennie> Chuck: Any new topics for future conversations?
16:03:17 <Chuck> zakim, take up item 1
16:03:17 <Zakim> agendum 1 -- Measurement Proposals -- taken up [from Chuck]
16:03:37 <Jennie> Chuck: Jeanne will start first. I did not send the links out to this
16:03:39 <Chuck> https://docs.google.com/document/d/10-1vKOYXBDKLJecfLmtz9tH3T1Y8PRW7pqG6HAcvaHU/
16:03:40 <AWK> AWK has joined #ag
16:03:50 <AWK> +AWK
16:03:58 <Jennie> Chuck: These are scoring proposals we want the Accessibility Guidelines and Silver group to review, consider
16:04:06 <JF> JF has joined #ag
16:04:09 <Jennie> ...Next will be Alastair
16:04:12 <JF> Present+
16:04:17 <thbrunet> present+
16:04:44 <Jennie> Jeanne: This was a consolidation of a number of discussions different groups have had about
16:04:53 <Jennie> ...ways to improve what was in the 1st public working draft.
16:05:07 <Jennie> ...I started with an attempt to make scoring consistent across guidelines
16:05:09 <Wilco> Wilco has joined #ag
16:05:19 <Jennie> Jeanne: many addressed this in content
16:05:32 <Jennie> ...To simplify the scoring to make it easier
16:05:40 <Jennie> ...We will be counting errors
16:05:47 <Jennie> ...And it removes the wholistic tests.
16:05:57 <Jennie> ...It was a really good ideas, but it has raised a lot of questions.
16:06:04 <Jennie> ...I don't think we have a lot of answers for them
16:06:20 <Jennie> ...Personally the idea raised by John Folio is better than the wholistic test idea.
16:06:36 <Jennie> ...I want to propose we remove the holistic tests in Section 4
16:06:57 <Jennie> ...In the testing section we would simplify the types of tests, so it only includes the atomic tests
16:07:02 <Jennie> ...We could rename it test
16:07:15 <Wilco> present+
16:07:15 <MarcJohlic> present+
16:07:28 <Jennie> ...In section 4: atomic could change to unit tests
16:07:37 <Chuck> q?
16:08:04 <Jennie> ...In section 5: testing could be by view instead of
16:08:09 <Jennie> ...We were doing it by element in 5.1
16:08:14 <mbgower> https://www.w3.org/2017/08/telecon-info_ag
16:08:21 <Jennie> ...Instead of by element, recommend chaning it to view
16:08:32 <Jennie> ...It could be used beyond web if appropriate
16:08:48 <Jennie> ...Instead of a different scoring system for each guideline, we would go to a percentage score
16:09:00 <Jennie> ...Guidelines that are true/false could be 100%
16:09:13 <Jennie> ...100% minus failures
16:09:20 <Jason_Rutgers_> Jason_Rutgers_ has joined #ag
16:09:24 <Jennie> ...All guidelines with a failure rate less than 5% would pass
16:09:30 <Jennie> ...No fractional percentages
16:09:43 <Jennie> ...Anything .5 or higher would round up - this responds to one of the github issues
16:09:57 <Jennie> ...Any guidelines with a quality measurement will have specific criteria defining each band
16:10:05 <Jennie> ...So it can be compared with different guidelines
16:10:18 <Jennie> ...I am recommending we remove sections 5.2, 5.3, and 5.4
16:10:26 <Jennie> ...This will simplify a lot of the scoring
16:10:43 <Jennie> ...Critical errors would remain, but they automatically fail when encountered
16:10:53 <Chuck> q+ to ask about percentages and critical errors
16:10:55 <Jennie> ...Testing could stop if encountering a critical error
16:11:11 <jon_avila> jon_avila has joined #ag
16:11:15 <MelanieP> q+
16:11:15 <jon_avila> present+jon_avila
16:11:23 <Jennie> Chuck: If there is a critical failure you can stop testing. But, if one isn't encountered, you are generating a percentage
16:11:32 <Jennie> ...I know that others had expressed concerns about counting the failures
16:11:35 <GreggVan> Q+
16:11:46 <JF> Q+
16:11:51 <Jennie> ...To me it appears that in order to get the percentage, you still are counting the number of failures
16:12:04 <Jennie> ...With a critical failure you can stop - the percentage is not meaningful
16:12:24 <Jennie> ...Is there value in having the percentage if one is encountered, or is it just a zero?
16:12:33 <Jennie> Jeanne: I think that would be up to the tester and what they are reporting
16:12:41 <Chuck> ack Ch
16:12:41 <Zakim> Chuck, you wanted to ask about percentages and critical errors
16:12:42 <Jennie> ...The accessibility consultancy I have worked for will continue to test
16:12:50 <Jennie> ...They want to give a report to the client of all issues
16:12:58 <MelanieP> q-
16:12:59 <Jennie> ...But for conformance reports, they would not need to
16:13:20 <Jennie> ...I will finish the examples, then can go through the questions
16:13:39 <GreggVan> please do go ahead with examples
16:13:44 <Jennie> Jeanne: I know there is a group working on critical errors, and I wrote this back in October
16:13:44 <bbailey> q?
16:13:59 <Jennie> ...since then we have done a lot more work on dividing up the issues
16:14:05 <Jennie> ...I am happy to go with that group's proposal
16:14:25 <Jennie> ...In regards to counting the passes - I am trying to set it up so that is not needed
16:14:37 <Jennie> ...You can have tools that count the instances, but not as the tester
16:14:49 <Jennie> ...Critical error would give a score of zero percent
16:15:01 <Jennie> ...Without one, (reads from the document)
16:15:12 <Jennie> ...This does require tool makers to count the elements
16:15:19 <Jennie> ...Otherwise the test remains the same as today
16:15:22 <Jennie> ...First example
16:15:30 <jon_avila> q?
16:15:43 <Jennie> ...If the view has no critical errors, it would pass
16:15:47 <Wilco> q+
16:15:50 <Jennie> ...Second example
16:15:56 <Jennie> ...(reads from the document)
16:16:06 <Jennie> ...This would fail
16:16:21 <Jennie> ...I would like to point out I did not work with any subgroups about this
16:16:22 <jon_avila> q+
16:16:25 <Jennie> ...These were my examples
16:16:36 <mbgower> q+ to say is there a definition of "functional image"
16:16:44 <Jennie> ...Clear words (reads from document)
16:16:55 <Jennie> ...I made this up, the subgroup may have a very different idea
16:17:05 <Jennie> ...I looked at Silverwriter
16:17:18 <Jennie> ...(reads the calculation from the document)
16:17:35 <Jennie> ...I gave an example in the document
16:17:57 <Jennie> ...I did the same thing with a video without captions and an accessible transcript
16:18:11 <Jennie> ...The 1st example (reads from the document)
16:18:21 <GN015> q+ to say that counting by tool relies on automated tests, while many Accessibility tests are manual
16:18:24 <Jennie> ...The 2nd example (reads from the document)
16:18:30 <Lauriat> qv?
16:18:34 <Chuck> ack Greg
16:18:36 <AWK> q+ to say that the percentage should be whatever it is, and if there is a critical error it fails. Gives better information to teams doing fixes.
16:18:36 <Jennie> GreggVan: Very interesting.
16:18:49 <Jennie> ...View: you said you are going to view vs page
16:19:00 <Jennie> ...WCAG2ICT we tried to make view work, and we never could
16:19:07 <Jennie> ...You have a view, you click on something and it expands
16:19:13 <Jennie> ...1 page could have 10,000 views
16:19:19 <Jennie> ...We were applying it to software
16:19:31 <Jennie> ...That is a great idea, but we couldn't figure out how to make it work
16:19:48 <Jennie> ...Percent score: you are talking about the percent of failures - which could be gamed
16:19:57 <JF> +1 to Gregg
16:19:59 <Jennie> ...You can put a lot of images at the bottom with alt text
16:20:07 <Jennie> ...The content no longer needs alt text
16:20:14 <JF> "...there is a possibility that the system may be gamed in order to make a conformance claim. For example, a site with many additional spacer images with appropriate null alt will more easily meet higher threshold values for the ratings.“(Adobe - https://github.com/w3c/silver/issues/400)
16:20:26 <Jennie> ...When you talk about percent: 95% of the building is accessible. Is that good enough?
16:20:37 <Jennie> ...Or 95% of a cafeteria is accessible -  is that enough?
16:20:45 <Jennie> ...The critical errors might help with this
16:20:53 <Chuck> q?
16:21:12 <Jennie> ...You did talk about a critical error being any functional image, and later you said meaningful images
16:21:23 <Jennie> ...A chart or diagram is not functional, so that is something to think about
16:21:37 <Jennie> ...For the caption example, that would need to be a critical error
16:21:49 <Jennie> Chuck: I am assuming you will mark these questions down for later
16:22:03 <Chuck> ack JF
16:22:15 <Jennie> JF: I have a couple of concerns. Greg called out the counting concern
16:22:33 <Jennie> ...We have heard from multiple people (including a DHS tester and Adobe) that this will be problematic because it can be gamed
16:22:49 <Jennie> ...I was concerned about the tool vendors requirement for counting - this assumes everyone will have a tool
16:22:55 <Jennie> ...Will there be a free and open source tool?
16:23:00 <Jennie> ...We have not created tools before
16:23:10 <Jennie> ...If this is part of a scoring mechanism, we should provide one
16:23:18 <Jennie> ...I am concerned about the functional image testing
16:23:29 <Jennie> ...I can have alt equals button - how useful this is...
16:23:44 <Jennie> ...The whole section around clear words, I have concerns about the list of common words
16:23:47 <Chuck> q?
16:23:55 <Jennie> ...Who creates the list? Impact on internationalization
16:24:00 <Jennie> ...I appreciate the effort here
16:24:08 <Chuck> ack Wil
16:24:09 <Fazio> Fazio has joined #Ag
16:24:11 <Jennie> ...I am concern that lots will prove to be difficult to impelement
16:24:30 <Jennie> Wilco: Tools currently really don't agree on what is an image, what is on the screen, what is off screen...
16:24:47 <Jennie> ...Doing this in a consistent way requires a new category of tools that do not exist today - a big ask
16:24:48 <JF> +1 to Wilco
16:24:48 <Judy> Judy has joined #ag
16:24:52 <Jennie> ...We have nothing like this today
16:24:56 <MelanieP> +1 to Wilco
16:25:01 <Jennie> ...This makes me uncomfortable
16:25:03 <Chuck> ack jon
16:25:17 <Jennie> JonA: In terms of an image - what is an image?
16:25:25 <Jennie> ...An SVG? Icon?
16:25:37 <Jennie> Chuck: Jon is on mute
16:25:46 <Chuck> ack mb
16:25:46 <Zakim> mbgower, you wanted to say is there a definition of "functional image"
16:25:47 <laura> +1 to JF.  Same holds true for quality of not only alt text but quality of captioning.
16:25:50 <Jennie> ...If you have more to say, please put yourself back on cue
16:26:00 <Jennie> mgower: Images are maybe their own outlier
16:26:07 <Jennie> ...Is there a definition of functional image?
16:26:20 <Jennie> ...Also, the automated tool is relatively possible to detect images
16:26:31 <GreggVan> Trace has created or commissioned 3 free tools to allow provisions in WCAG.   Agree with JF but I would not think we should not do the right thing for lack of a free tool.       That doesn't mean we know how to create these tools - but we should think about free versions of tools - before tossing out a good idea.    That said - we should not solve problems with imaginary tools as JA said.
16:26:32 <Jennie> ...understanding when the image serves a purpose is different
16:27:10 <Jennie> ...I think that we can marry what can be automatically tested with consistency
16:27:30 <Chuck> ack GN
16:27:30 <Zakim> GN, you wanted to say that counting by tool relies on automated tests, while many Accessibility tests are manual
16:27:33 <Jennie> ...I determined by doing this...my protocol was this...
16:27:41 <Jennie> GN: I am also struggling with the coutning
16:27:49 <Jennie> ...Counting images with a tool could be possible
16:27:56 <Chuck> q+ to mention JF's efforts on protocols
16:28:01 <Jennie> ...For several requirements we have automatic tests
16:28:02 <jon_avila> Did my conversation come through?
16:28:06 <Francis_Storr> Francis_Storr has joined #ag
16:28:10 <Jennie> ...Color contrast on visual indicators
16:28:12 <Francis_Storr> present+
16:28:15 <Jennie> ...1st one has to judge which lines need contrast
16:28:17 <AWK> @jon - just the first sentence
16:28:20 <Jennie> ...This needs to be decided by a human
16:28:33 <Jennie> ...It has to be counted manually: error prone, huge burden
16:28:42 <Jennie> ...I would rather not count, and just go through the severity
16:28:52 <Chuck> q?
16:28:54 <Jennie> ...blocker? hurdle with a work around? kind of an inconvenience
16:29:01 <Wilco> +1, number of issues + severity
16:29:09 <JF> "...as a certified trusted tester in the DHS Accessibility Test Process for  Web, using the measure of "a percentage of overall instances" to obtain this data will be problematic. It would be burdensome to expect a manual test process to "count" all the instances where a test condition applies.“
16:29:19 <Jennie> Chuck: AWK may be muted
16:29:30 <JF> (Ann Marie Davis - https://github.com/w3c/silver/issues/275)
16:29:40 <Jennie> ...We are meeting regularly
16:29:51 <Jennie> ...reach out to JF and myself regarding the protocols call
16:29:56 <Chuck> q?
16:30:04 <Chuck> ack Ch
16:30:04 <Zakim> Chuck, you wanted to mention JF's efforts on protocols
16:30:14 <jon_avila> We need to have scoring around what is a functionally required or not - as this could move score from 0 to 95.
16:30:31 <Jennie> Chuck: The value in the comments will help the chairs decide how to have a more broad conversation
16:30:37 <Jennie> ...Thank you Jeanne, and all commentors
16:30:41 <JF> "Any single "critical error" results in a score of "very poor" for that outcome and also results in an inability to reach bronze level as an overall score. This seems unreasonable for large-scale websites and applications and negates all of the work that may have been done to meet the outcome."  (IBM - https://github.com/w3c/silver/issues/477)
16:30:47 <Chuck> ack AWK
16:30:47 <Zakim> AWK, you wanted to say that the percentage should be whatever it is, and if there is a critical error it fails. Gives better information to teams doing fixes.
16:30:56 <jeanne> Thanks for the comments.  :)
16:30:59 <jon_avila> People today don't agree on what is a functional image and what would actually cause a critical error or not.
16:31:01 <Jennie> Chuck: We are moving on to Alastair G
16:31:06 <Chuck> https://docs.google.com/document/d/1IVCPcfyfnjm2RJRvgV5pbcQLwzxXKHw66V7R-OW3wiE/edit
16:31:11 <Jennie> ...He has a proposal specific to measuring alt text
16:31:36 <Jennie> Agarrison: I will run through the information
16:31:39 <jon_avila> Counting images is a challenge as we don't have a consistent mechanism to determine what are images - svgs, canvas, role img, img, poster images, video, etc.
16:32:18 <Jennie> Agarrison: (displaying the Measuring Alt Text Quality document)
16:32:27 <Jennie> ...I put together a short paper in response to the call out a few weeks ago
16:32:30 <JF> +1 Jon - on a personal level I think counting is a rabbit hole that we should try to avoid
16:32:35 <Jennie> ...I will read through it
16:32:36 <david-macdonald> david-macdonald has joined #ag
16:32:40 <Jennie> ...(reads through the document)
16:32:42 <david-macdonald> present+
16:33:09 <Jennie> ...We have just heard that color contrast will not be achieved
16:33:29 <Jennie> ...People used to think flying machines would not be possible, but this was achieved in a few years from the statement
16:33:43 <Jennie> ...I always try to keep a question mark as to what capabilities will be available
16:34:00 <Jennie> ...Measuring alt text is not a 1 size fits all approach
16:34:15 <Jennie> ...Images need to be categorized
16:34:25 <Jennie> ...I have split the W3C decision tree
16:34:52 <Jennie> ...Images of text, complex images, informative images, decorative images (and a few others he read)
16:35:08 <Jennie> ...Machine learning - categorization confidence
16:35:36 <Jennie> ...Functional images can have image recognition, optical character recognition (continues reading from document)
16:36:06 <Jennie> ...Informative images showing images: (reads from document)
16:36:22 <Jennie> ...Decorative - I do not have any. Just like the W3C decision tree, this is a catch all mechanism
16:36:26 <Jennie> ...These are just suggestions
16:36:32 <Jennie> ...There will be others used over time
16:36:42 <Jennie> ...The process
16:36:54 <Jennie> ...Needs to be done in a sequence - use a discrimination tree
16:37:01 <Jennie> ...then you stop using the rest of the decision tree
16:37:17 <Jennie> ...Look at each category in order
16:37:22 <Jennie> ...With decorative being looked at last
16:37:45 <Jennie> ...I do suggest that the W3C start to amalgamate more
16:37:53 <Jennie> ...If there are central data sets -
16:37:56 <Francis_Storr> note that the WHATWG HTML spec now also includes a concept of an "ancillary image": https://html.spec.whatwg.org/multipage/images.html#ancillary-images
16:38:00 <Jennie> ...Training against the same dataset
16:38:14 <Jennie> ...Then others may also throw their images into that dataset
16:38:18 <Jennie> ...This could reduce bias
16:38:29 <Jennie> ...Looking at functional images
16:38:37 <Jennie> ...You ask the W3C to collect thousands of images
16:38:50 <Jemma> Jemma has joined #ag
16:39:04 <Jennie> ...You can group these images in a big dataset with labels
16:39:17 <Jennie> ...using a standardized label for all images that represent a printer
16:39:22 <Jemma> present+
16:39:29 <Jennie> ...You do this because if you have a standardized set of labels
16:39:32 <JF> Q+ to ask about context
16:39:46 <Jennie> ...You will be able to easily decide if they have an appropriate alt text
16:40:02 <Jennie> ...Characterization would determine if they are functional and what the associated label should be
16:40:13 <Jennie> ...Informative images showing a single famous object
16:40:18 <Jemma> agenda?
16:40:22 <Jennie> ...You could thousands of images for these
16:40:26 <Jennie> ...like a Google search
16:40:29 <GN015> q+ to say the alt text should resemble the semantics, not the visual, e.g. a search button vs a magnifier icon
16:40:37 <Jennie> ...If using a picture of the Mona Lisa
16:40:53 <Jennie> ...(2nd paragraph)
16:41:27 <Jennie> ...You determine if this is an image of the famous image or not
16:41:37 <Jennie> ...Then you assess if the text alternative is representative
16:41:41 <Jennie> ...Complex images
16:41:48 <Jennie> ...W3C collect thousands of images
16:42:02 <Jennie> ...These are not just for machines, but could train an army of humans to do the same job
16:42:16 <Jennie> ...the categorization would be to determine if it is a chart or graph
16:42:22 <Jennie> ...Is the data correctly represented
16:42:36 <Jennie> ...This is a Draconion move, but it makes it much more automatically testable
16:42:50 <david-macdonald> @shadi Thx
16:42:51 <Jennie> ...If it is linked to a data table, that is half of a good test
16:43:04 <Jennie> ...There is a whole lot of work being done on data extraction from charts
16:43:11 <Jennie> ...In the future you would be able to automate this
16:43:21 <Jennie> ...For images of text, each requires a different mechanism
16:43:36 <Jennie> ...All complex images containing text should already be removed
16:43:57 <Jennie> ...(reads from the doc the W3C decision tree quote)
16:44:08 <Jennie> ...There are two types of text detection
16:44:20 <Jennie> ...Both work in a different way
16:44:36 <Jennie> ...EAST is for text not presented in a typical manner
16:44:59 <Jennie> ...EAST is designed to pull out text from billboards, or text on a book in the image
16:45:11 <Jennie> ...Optical Character Recognition is just for typical text to be extracted
16:45:33 <Jennie> ...If you run both against images you are able to tell the text within the image in both the typical format and the atypical format
16:46:13 <Jennie> ...We are only looking for typical text
16:46:27 <Fazio> Fazio has joined #ag
16:46:32 <Jennie> ...This van has lots of words written all over it
16:46:35 <Jennie> ...This is not a text image
16:46:46 <Jennie> ...Using these two methods we can isolate which to categorize as a text image
16:46:53 <Jennie> ...Informative images of objects
16:47:00 <Jennie> ...W3C should find general objects
16:47:12 <Jennie> ...There are existing datasets for this
16:47:30 <Jennie> ...This area is where a metric called SPICE comes in
16:47:38 <Jennie> ...Tests how good the generation of alt text is
16:47:51 <Jennie> ... I have reformulated this for informative images
16:47:56 <Jennie> ...(reads from the document)
16:48:09 <JF> From the document: "If a context specific alt text description is provided it becomes useless when viewed out of context; however if a non-context specific alt text description is provided it remains equally useful in both contexts." I challenge that assertion
16:48:19 <Jennie> ...Wilco asked me some questions about context and other pieces
16:48:24 <Jennie> ...In terms of context
16:48:37 <Jennie> ...The context the image sits within I think should be not considered
16:48:52 <Jennie> ...Article it sits within is the intended context
16:49:01 <laura> laura has joined #ag
16:49:01 <Jennie> ...When the image is searched is the unintended context
16:49:20 <Jennie> ...Alt text that works in both contexts would be helpful
16:49:26 <Jennie> ...(reads from the document)
16:50:05 <Jennie> ...(reads information related to image of 2 dogs running after a ball)
16:50:23 <Jennie> ...The other question was why not include adjectives or names?
16:50:32 <Jennie> ...They don't form the backbone of the scene
16:50:45 <Jennie> ...The big picture summarization is where? What objects? What are they doing?
16:51:22 <Jennie> ...Looking at the dogs - if you called people on the phone and asked them to draw this photo from the alt text
16:51:23 <Jennie> ...they could
16:51:28 <Jennie> ...They can make inferences
16:51:42 <Jennie> ...If instead the description focuses on names, like Bob and Jack
16:51:57 <Jennie> ...This provides a different image for the person hearing that description
16:52:18 <Jennie> ...More adjectives would provide some aspects of the images, but not others
16:52:43 <Jennie> ...Categorization of such an image: does it contain recognizable objects?
16:52:52 <Jennie> ...You try to analyze the relationship between the objects
16:52:58 <Jennie> ...You then label them
16:53:05 <Jennie> ...example dog 1, dog 2
16:53:12 <Jennie> ...Then you have a ball
16:53:23 <Jennie> ...(reads from document)
16:53:39 <Jennie> ...You can then formulate what you expect your alt text to have
16:53:46 <Jennie> ...(reads from document)
16:53:50 <Jennie> ...You can then measure your alt text
16:54:13 <Jennie> ...Look at the list we generated - we can determine which pieces of the alt text existed in our list
16:54:26 <Jennie> ...It is missing a number of things we said should be included
16:54:35 <Jennie> ...It is missing 4 vital pieces of information
16:54:47 <Jennie> ...So the alt score can be measured
16:55:06 <Jennie> ...Once you have gone through the list, and if you determine it is decorative
16:55:18 <Jennie> ...I have included a revamped version of the W3C's alt decision tree
16:55:26 <Jennie> ...into a simple decision tree format
16:55:31 <Jennie> ...You can run through this manually
16:55:39 <Jennie> ...to assess and measure your alternative text for an image
16:55:46 <Jennie> ...Caveat: this is not the only way to do this
16:55:52 <Chuck> q?
16:56:01 <Chuck> ack JF
16:56:01 <Zakim> JF, you wanted to ask about context
16:56:07 <Jennie> JF: Thank you for this Alastair
16:56:11 <Jennie> ...Lots of good thinking here
16:56:17 <Rachael> q+ to say this is an interesting jumping off point for evaluating quality of some of the more subjective content
16:56:19 <Jennie> ...I agree, machine learning will improve over time
16:56:27 <Jennie> ...I don't think we are near where we need to be today
16:56:32 <Jennie> ...for conformance measuring
16:56:40 <Jennie> ...Under complex images you reference a dataset of charts and graphs
16:56:48 <Jennie> ...Another type would be infographics
16:56:54 <Jennie> ...This is usually a large graphic file
16:57:01 <Jennie> ...This requires a long description
16:57:06 <Jennie> ...I am concerned about this
16:57:12 <Jennie> ...The context piece bothers me
16:57:35 <kirkwood> q+
16:58:02 <Chuck> q+ to change scribes
16:58:02 <Jennie> ...Specificity of age of shoes may be esssential in certain use cases - I don't think articial intelligence will be able to determine when to include this
16:58:29 <Jennie> ...In context the most important elements in other images can be that it is a selfie, not the object the person is in front of
16:58:47 <Jennie> ...Artificial intelligence will get us a long way, but I think that context will be an issue
16:59:05 <Jennie> ...I am also concerned about the datasets - who will compile them? Host them? It is a lot to ask the W3C to do
16:59:18 <Jennie> Agarrison: this is designed as a manual process that is meant to be machine learned
16:59:42 <Jennie> ...I agree that infographics is not included in the complex graphics - we could include that somewhere along the line
16:59:58 <Jennie> ...For the selfie example: this would be under the secondary informative image
17:00:07 <Jennie> ...This would be the focus of the alt text
17:00:15 <Chuck> ack Ch
17:00:15 <Zakim> Chuck, you wanted to change scribes
17:01:07 <Chuck> zakim, pick a victim
17:01:07 <Zakim> Not knowing who is chairing or who scribed recently, I propose !
17:01:08 <Azlan> Azlan has joined #ag
17:01:22 <Chuck> zakim, pick a victim
17:01:22 <Zakim> Not knowing who is chairing or who scribed recently, I propose MarcJohlic
17:01:42 <Regina> I can try never done before
17:01:57 <Regina> scribe Regina
17:01:59 <Chuck> scribe:  Regina
17:02:04 <Jennie> Scribe cheat sheet https://www.w3.org/WAI/GL/wiki/Scribing_Commands_and_Related_Info
17:02:25 <jaunita_george> jaunita_george has joined #ag
17:02:27 <jaunita_george> +1
17:02:30 <Azlan> present+
17:02:34 <jaunita_george> Present+
17:02:44 <bbailey> q+ to ask about text alternatives that serve equivalent purpose versus grading alt text
17:02:45 <Regina> Jonh Eiffell towel primary focus in the image
17:02:50 <GN015> Sorry, I have to drop
17:02:53 <GN015> q-
17:03:07 <Chuck> ack Rach
17:03:07 <Zakim> Rachael, you wanted to say this is an interesting jumping off point for evaluating quality of some of the more subjective content
17:03:08 <Regina> ... it's not the primary meaning in the screen
17:03:20 <mbgower> Great work, Alistair!
17:03:23 <alastairc> s/Jonh Eiffell towel /John: Eiffell tower
17:04:00 <Chuck> q?
17:04:04 <Regina> Rachel sugg content recognizing the value
17:04:08 <Chuck> ack kirkwood
17:04:09 <JF> +1 to Rachel
17:04:10 <Lauriat> +1 to Rachael, very helpful write-up!
17:04:26 <JF> s/+1 to Rachel/+1 to Rachael
17:04:33 <Regina> bbailey great example
17:04:45 <JF> +1 to John K
17:04:47 <Regina> ... that  would be the most important thing
17:04:55 <alastairc> s/Rachel sugg/Rachael: sugg
17:05:08 <Regina> ...context be careful on the idea and subjectivity
17:05:10 <Chuck> q?
17:05:15 <alastairc> s/bbailey great /kirkwood great
17:05:22 <Chuck> ack bbai
17:05:22 <Zakim> bbailey, you wanted to ask about text alternatives that serve equivalent purpose versus grading alt text
17:05:34 <Wilco> q+
17:05:54 <mbgower> q+ to say that an SC on captioning important images may be warranted
17:07:02 <Regina> alaistairc with many images impossible to add information
17:07:08 <Chuck> ack Wilco
17:07:13 <Jemma> this presentation was eye opening for me. thanks for great presentation, Aistair!!
17:07:14 <JF> but... if accuracy of alt text is only 80%, what is the impact of that on persons with COGA issues?
17:07:33 <Regina> Wilco looks solutions in user agents
17:07:49 <Regina> ... potential of some development and testing
17:08:08 <Regina> ...implement new technologies
17:08:35 <Jennie> +1 to JF. I was thinking about how to consider the importance of weighting the scoring to describe relationships of elements rather than just the specific elements.
17:08:46 <Jemma> s/eye opening/giving me alternative perspective to think about alt text I never imanged/
17:08:46 <Regina> alastairc not necessary testing to hundred of images
17:08:48 <Chuck> ack mb
17:08:48 <Zakim> mbgower, you wanted to say that an SC on captioning important images may be warranted
17:09:23 <Regina> mbgower required of captions programatically
17:09:35 <Regina> ...providing ALT text automatically
17:09:42 <Regina> ... provide captions to images
17:09:44 <Chuck> zakim, take up next item
17:09:44 <Zakim> agendum 2 -- Review Writing Testable WCAG 3.0 Outcomes Survey -- taken up [from Chuck]
17:09:59 <Chuck> https://www.w3.org/2002/09/wbs/35422/testable-outcomes-2021-12-2/results
17:10:55 <Regina> chuck describing the agenda and survey results
17:11:17 <Regina> ... Mary longer agree statements regarding editorial changes
17:11:38 <Regina> wilco some putuation suggestions
17:11:45 <Wilco> q+
17:12:08 <Regina> chuck agree in some of the comments
17:12:26 <Regina> ...pages to be tested
17:12:31 <Regina> ... just a sample
17:12:49 <Regina> ...become more complicated
17:13:33 <Chuck> q?
17:13:40 <Regina> chuck reading comments from survey
17:13:42 <Chuck> ack Wilco
17:14:29 <Regina> wilco add a list of standards of terms
17:14:39 <Chuck> proposed RESOLUTION:  AGWG approves amended Writing Testable WCAG 3.0 Outcomes for internal use
17:14:59 <Rachael> +1
17:15:01 <Jemma> I agree with Chuck's approach
17:15:14 <Regina> chuck proposing the resolution
17:15:18 <laura> +1
17:15:19 <Regina> +1
17:15:19 <bbailey> +1
17:15:20 <Wilco> +1
17:15:21 <jeanne> +1
17:15:23 <Nicaise> +1
17:15:23 <ToddL> +1
17:15:25 <jaunita_george> +1
17:15:26 <Francis_Storr> +1
17:15:44 <mbgower> +1
17:16:06 <Regina> not sure where to find it
17:16:15 <Chuck> proposed RESOLUTION:  AGWG approves amended Writing Testable WCAG 3.0 Outcomes for internal use
17:16:35 <Regina> RESOLUTION:  AGWG approves amended Writing Testable WCAG 3.0 Outcomes for internal use
17:16:42 <Chuck> zakim, take up next item
17:16:42 <Zakim> agendum 3 -- ACT New Rules (30 min) https://www.w3.org/2002/09/wbs/35422/act_nov_10/ -- taken up [from Chuck]
17:17:12 <Chuck> https://www.w3.org/2002/09/wbs/35422/act_nov_10/results
17:17:19 <Chuck> TOPIC:  Question 1 - Form field has non-empty accessible name
17:17:26 <Chuck> Chuck has changed the topic to: Question 1 - Form field has non-empty accessible name
17:18:00 <Regina> chuck reading questions Form field has non-empty accessible name
17:18:26 <Regina> ... 7 approved, 2 with comments, 1 not approve
17:18:53 <Regina> ...reading Oliver Comment from documents
17:19:29 <Regina> ... reading Gundula Nieman's comment about not backdoors rules
17:19:53 <Chuck> q?
17:20:12 <Regina> chuck reading Wilco comments
17:20:32 <thbrunet> q+
17:20:43 <Chuck> ack th
17:21:19 <Regina> thbrunet something is disable it will be a fail when shows
17:21:56 <Chuck> proposed RESOLUTION:  Accept publishing "Form field has non-empty accessible name"
17:22:06 <Rachael> +1
17:22:08 <jaunita_george> +1
17:22:09 <Wilco> +1
17:22:10 <Chuck> +1
17:22:10 <thbrunet> +1
17:22:11 <ToddL> +1
17:22:14 <JF> +1
17:22:15 <Azlan> +1
17:22:20 <jeanne> +1
17:22:20 <laura> +1
17:22:20 <JakeAbma> +1
17:22:22 <Regina> +1
17:22:23 <Nicaise> +1
17:22:29 <bbailey> +1
17:22:30 <Raf> +1
17:22:33 <Chuck> proposed RESOLUTION:  Accept publishing "Form field has non-empty accessible name"
17:22:34 <Francis_Storr> +1
17:22:43 <Regina> RESOLUTION:  Accept publishing "Form field has non-empty accessible name"
17:22:44 <mbgower> +1
17:22:57 <Chuck> TOPIC:  Question 2 - Attribute has valid value
17:23:04 <Chuck> Chuck has changed the topic to: Question 2 - Attribute has valid value
17:23:20 <Regina> chuck attribute has valid value
17:23:22 <JF> Q+
17:23:28 <Regina> ...7 people agree
17:23:47 <alastairc> s/chuck attribute has valid value/chuck: attribute has valid value
17:23:51 <Regina> ...Wilco responding to Rochelle
17:24:10 <Wilco> q+
17:24:27 <Regina> ...Oliver said samples will fail
17:24:34 <Chuck> ack JF
17:25:07 <Regina> JF question about autocomplete
17:25:20 <stevelee> stevelee has joined #ag
17:25:23 <Chuck> ack Wil
17:25:33 <Regina> ...modify this rule to include the  attribute
17:25:55 <Regina> wilco I don't think we are going to use it
17:26:09 <Regina> ...About Rochelle comment
17:26:35 <Regina> ...we are not doing it anymore it's not applicable
17:26:48 <jon_avila> q?
17:26:54 <Chuck> proposed RESOLUTION:  Accept "Attribute has valid value"
17:26:56 <Regina> ...rule is the same
17:26:56 <Rachael> Fair enough, as stated, I will not object to the exisiting langauge
17:27:13 <Regina> ...it's not failure
17:27:28 <Wilco> +1
17:27:30 <jaunita_george> +1
17:27:30 <AWK> +1
17:27:30 <thbrunet> +1
17:27:31 <Rachael> +1
17:27:32 <Chuck> +1
17:27:34 <jon_avila> q+
17:27:34 <bbailey> +1
17:27:35 <mbgower> +1
17:27:36 <JakeAbma> +1
17:27:37 <Regina> +1
17:27:40 <laura> +1
17:27:42 <Francis_Storr> +1
17:27:43 <Azlan> +1
17:27:46 <jon_avila> +0
17:27:48 <JF> +1
17:27:54 <Regina> RESOLUTION:  Accept "Attribute has valid value"
17:27:56 <Chuck> q?
17:28:02 <Chuck> ack jon
17:28:05 <kirkwood> +1
17:28:39 <Regina> jon_avila it's not a field requiere use to autocomplete
17:28:59 <JF> Q+
17:28:59 <Wilco> q+
17:29:05 <Chuck> ack JF
17:29:14 <Regina> ...because the whole is predictable
17:29:18 <AWK> +1 to Jon - he's right
17:29:23 <Regina> JF values do somehting
17:29:40 <Regina> ... if you put the autocomplete value it will autofill
17:29:55 <alastairc> q+ Jon's right on this
17:30:01 <alastairc> q+ to say Jon's right on this
17:30:02 <Regina> ...wrong value autofill the wrong information
17:30:12 <Chuck> ack Wilco
17:30:27 <Regina> wilco first assumption
17:30:38 <alastairc> q-
17:30:49 <Regina> ...if autocomplete color
17:31:03 <JF> + to ask if there is a companion rule for misuse?
17:31:16 <JF> Q+
17:31:19 <Chuck> q+ to ask if JA objects to the rule?
17:31:23 <Regina> ...role attribute is used
17:31:23 <Chuck> ack JF
17:31:41 <alastairc> q+ to ask if that can result in false fails?
17:31:50 <alastairc> I don't think that would be a fail
17:32:04 <Regina> JF a lot of rules in place
17:32:09 <Chuck> ack Chuck
17:32:09 <Zakim> Chuck, you wanted to ask if JA objects to the rule?
17:32:11 <Wilco> https://github.com/act-rules/act-rules.github.io/issues
17:32:19 <alastairc> q-
17:32:27 <Regina> chuck asking objecting to the rule?
17:32:36 <AWK> q+ to ask if this rule is fully automated?
17:32:44 <Chuck> proposed RESOLUTION:  Accept "Attribute has valid value"
17:32:52 <Chuck> ack AWK
17:32:52 <Zakim> AWK, you wanted to ask if this rule is fully automated?
17:33:06 <Regina> awk automate rule or manual review?
17:33:16 <Regina> wilco this is fully automated
17:33:31 <Regina> awk about the user
17:33:41 <Chuck> q+
17:33:47 <Regina> wilco this make more difficult automate it
17:34:32 <Regina> awk there is extra cases would represent problems when filling a form
17:34:41 <Regina> ... for autocomplete values
17:34:56 <JF> it's also an author failure
17:35:02 <jon_avila> q+
17:35:02 <Chuck> q?
17:35:05 <Chuck> ack Chuck
17:35:08 <Regina> ...it's a usability issue not a failure but a discovered thing
17:35:16 <Regina> chuck still provide value
17:35:24 <Wilco> q+
17:35:31 <Regina> ... provide meaningful information
17:35:34 <Chuck> ack Jon
17:35:59 <Regina> Jon create a false positive
17:36:06 <alastairc> +1, it could bring up false-fails without scoping to the user-info.
17:36:36 <Chuck> proposed RESOLUTION:  Accept "Attribute has valid value"
17:36:38 <Chuck> q?
17:36:39 <Regina> chuck some hesitation with this rule
17:36:41 <Chuck> ack Wil
17:37:04 <jaunita_george> +1
17:37:11 <Regina> wilco every rule has tested
17:37:25 <Regina> ...bu looking people reporting false positives
17:37:38 <Regina> ...we consider not a big deal
17:37:47 <Regina> ...false positive are reported
17:37:51 <Chuck> q?
17:38:09 <Chuck> proposed RESOLUTION:  Accept "Attribute has valid value"
17:38:16 <Chuck> +0
17:38:19 <Regina> chuck interesting conversation to review
17:38:20 <Wilco> +1
17:38:22 <Rachael> +1 It is informative content so I believe we can update if needed
17:38:22 <jon_avila> +0
17:38:24 <bbailey> +1
17:38:29 <Regina> +
17:38:31 <jaunita_george> +1
17:38:31 <JF> @Wilco - issue #1768
17:38:34 <Chuck> +.5
17:38:36 <jeanne> +1
17:38:36 <AWK> +.5
17:38:38 <alastairc> +1, but would like to know where to report false positive cases...
17:38:40 <JF> +1
17:38:49 <JakeAbma> +.5
17:39:01 <Regina> RESOLUTION:  Accept "Attribute has valid value"
17:39:02 <kirkwood> +.5
17:39:10 <Rachael> q+
17:39:12 <Wilco> @awk, you can report them to ACT-R, or to any of the implementors if you're using any tool / methodology using ACT.
17:39:17 <Chuck> ack Ra
17:39:20 <Chuck> TOPIC:  Question 3 - Element marked as decorative is not exposed
17:39:25 <Wilco> https://github.com/act-rules/act-rules.github.io/issues
17:39:27 <Chuck> Chuck has changed the topic to: Question 3 - Element marked as decorative is not exposed
17:39:48 <Regina> chuck third question
17:40:00 <Regina> ... 6 people approved
17:40:51 <Chuck> q?
17:40:52 <Regina> ... reading comments about image and attributes
17:40:56 <AWK> q+
17:41:16 <Wilco> q+
17:41:17 <Chuck> ack AWK
17:42:09 <Regina> awk concerns about about image presentation
17:42:14 <Chuck> q?
17:42:17 <Chuck> ack wilco
17:42:25 <Regina> wilco you can fail this
17:42:35 <Regina> ...elements can have role attributes
17:42:46 <Regina> ...nad have it not being applied
17:43:13 <Regina> ...there are scenarios where it could be ignored
17:43:24 <Chuck> q?
17:43:36 <thbrunet> q+
17:43:40 <Regina> ... could have role none and be ignored and considerated a failure
17:43:59 <thbrunet> q-
17:44:13 <Regina> ...this rule may fail in some scenarios
17:44:18 <Chuck> proposed RESOLUTION:  Accept amended "Element marked as decorative is not exposed"
17:44:28 <Chuck> +1
17:44:32 <jaunita_george> +1
17:44:35 <Wilco> +1
17:44:38 <bbailey> +1
17:44:39 <Raf> +1
17:44:41 <Regina> +1
17:44:42 <JF> +1
17:44:43 <Azlan> +1
17:44:45 <AWK> +0
17:44:49 <Nicaise> +1
17:44:49 <JakeAbma> +1
17:45:24 <Regina> awk there is not conflict putting in a row
17:45:54 <Regina> chuck Andrew are you objecting?
17:46:17 <Regina> AWK this rule about element decorative in some way but not being exposed,
17:46:23 <alastairc> +1, some of those fails are indicative of some nouanced accessibility tree points, not bothered by the obvious passes.
17:46:41 <Regina> ...things marked decorative are exposed
17:46:55 <thbrunet> img[alt=""] is listed as role 'img' by Chrome. Pass 4 indicates it's 'none'. Is Chrome wrong?
17:46:55 <JF> Q+
17:47:05 <Chuck> ack JF
17:47:08 <Regina> wilco aria label and ALT empty
17:47:34 <thbrunet> q+
17:47:57 <Regina> JF about the relationship between attributes aria label and ALT
17:48:07 <Chuck> ack thbrunet
17:48:27 <Regina> thbrunet in chrome images
17:48:30 <JF> s/JF about the relationship between attributes aria label and ALT/JF: about the relationship between attributes aria label and ALT
17:48:42 <Wilco> https://www.w3.org/TR/html-aria/#el-img-empty-alt
17:48:51 <Regina> wilco it's about html
17:49:09 <Regina> ... how html is done
17:49:23 <Chuck> proposed RESOLUTION:  Accept amended "Element marked as decorative is not exposed"
17:49:34 <Regina> RESOLUTION:  Accept amended "Element marked as decorative is not exposed"
17:49:44 <Chuck> TOPIC:  Question 4 - Letter spacing in style attributes is not important
17:49:50 <Chuck> Chuck has changed the topic to: Question 4 - Letter spacing in style attributes is not important
17:50:08 <Regina> chuck reading from document
17:50:13 <Regina> ...5 people agreed
17:50:20 <Chuck> proposed RESOLUTION:  Accept "Letter spacing in style attributes is not important"
17:50:28 <Wilco> +1
17:50:29 <Chuck> +1
17:50:32 <jeanne> +1
17:50:33 <JF> +1
17:50:34 <Rachael> +1
17:50:37 <JakeAbma> +1
17:50:39 <Regina> +1
17:50:39 <ToddL> +1
17:50:44 <Azlan> +1
17:50:47 <mbgower> +1
17:50:57 <Regina> RESOLUTION:  Accept "Letter spacing in style attributes is not important"
17:51:16 <Chuck> TOPIC:  Question 5 - Word spacing in style attributes is not important
17:51:23 <Chuck> Chuck has changed the topic to: Question 5 - Word spacing in style attributes is not important
17:51:36 <Regina> chuck no objections
17:51:39 <Chuck> proposed RESOLUTION:  Accept "Word spacing in style attributes is not important"
17:51:41 <Chuck> +1
17:51:42 <jaunita_george> +1
17:51:43 <Rachael> +1
17:51:45 <Azlan> +1
17:51:45 <Wilco> +1
17:51:47 <Regina> +1
17:51:49 <JakeAbma> +1
17:51:52 <JF> +1
17:51:57 <alastairc> +!
17:51:58 <jon_avila> +1
17:51:59 <ToddL> +1
17:51:59 <alastairc> +1 even
17:52:07 <Nicaise> +1
17:52:14 <Regina> RESOLUTION:  Accept "Word spacing in style attributes is not important"
17:52:20 <Chuck> zakim, take up next item
17:52:20 <Zakim> agendum 4 -- Focus appearance https://www.w3.org/2002/09/wbs/35422/wcag22-focus-appearance-enhanced2/ -- taken up [from Chuck]
17:52:41 <Chuck> TOPIC:  Question 1 - Focus Appearance complexity #1842
17:52:47 <Chuck> Chuck has changed the topic to: Question 1 - Focus Appearance complexity #1842
17:52:49 <alastairc> q+
17:52:57 <Chuck> ack ala
17:53:23 <Regina> chuck going to questions
17:53:27 <alastairc> q+
17:54:08 <Chuck> ack ala
17:54:44 <Chuck> TOPIC:  Focus appearance & Visual examples #2071
17:55:02 <alastairc> https://www.w3.org/2002/09/wbs/35422/wcag22-focus-appearance-enhanced2/results#xq26
17:55:10 <Chuck> q?
17:56:04 <Chuck> proposed RESOLUTION: Accept amended response to address issue 2071
17:56:32 <jaunita_george> +1
17:56:34 <MelanieP> +1
17:56:35 <Chuck> +1
17:56:36 <Wilco> +1
17:56:39 <Rachael> +1
17:56:42 <ToddL> +1
17:56:51 <Jemma> +1
17:56:54 <JakeAbma> +1
17:57:04 <bbailey> +1
17:57:49 <Regina_> Regina_ has joined #ag
17:58:40 <Regina_> sorry I was kick out of the site
17:59:19 <JF> bye all
17:59:23 <Regina_> thanks!
17:59:33 <Azlan> Bye all
17:59:36 <alastairc> rrsagent, make minutes
17:59:36 <RRSAgent> I have made the request to generate https://www.w3.org/2021/12/07-ag-minutes.html alastairc
17:59:45 <Jemma>  /me thanks everyone for all the great discusison.
17:59:53 <Regina_> rrsagent, make minutes
17:59:53 <RRSAgent> I have made the request to generate https://www.w3.org/2021/12/07-ag-minutes.html Regina_
18:00:06 <ToddL> ToddL has left #ag
18:00:12 <alastairc> rrsagent, make logs public
18:00:25 <alastairc> rrsagent, make minutes
18:00:25 <RRSAgent> I have made the request to generate https://www.w3.org/2021/12/07-ag-minutes.html alastairc