Temporarily wrapping up color contrast discussion

Hello all,

We will get into deep discussions of color contrast in the near future but lets move that to subgroups and Github after we recharter.  Please feel free to continue polite email discussions with individuals if both parties are interested, but some of our members find long email discussions like this difficult to track and overwhelming.

At this time, the most important next step for WCAG 3 is gathering research on contrast and readability. I've started an issue at https://github.com/w3c/wcag3/issues/10  to collect that research and included a link to Andrew's list as the first comment.  If you have additional research, please add links to the research in that issue.

Kind regards,

Rachael
________________________________
From: Andrew Somers <andy@generaltitles.com>
Sent: Friday, August 4, 2023 4:49 AM
To: Gregg Vanderheiden <gregg@vanderheiden.us>
Cc: Jonathan Avila <jon.avila@levelaccess.com>; w3c-waI-gl@w3. org <w3c-wai-gl@w3.org>
Subject: Re: British Tourist Authority, England’s national tourist board contrast discussion


CAUTION: This email message has been received from an external source. Please use caution when opening attachments, or clicking on links.

On Aug 3, 2023, at 10:39 PM, Gregg Vanderheiden <gregg@vanderheiden.us<mailto:gregg@vanderheiden.us>> wrote:

Please do not try to create success criteria (or their equivalent in 3.)    by example — or borderline cases

We haven’t,  not in the slightest.



With all of the contrast measures you will find color combinations that pass that are horrendous.

We have not been focused on edge cases at all. We have ample documentation on the basis, as linked in previous posts.



And for almost every WCAG rule (including 1.1)  — I have found examples of pages that — if they follow the SC strictly are LESS accessible ——  or rather there are more accessible ways to do something that do not conform to a success criterion.

In the case of WCAG 2 contrast, the problem is not edge cases. Independent studies that I’ve linked to earlier have taken thousands of random color pairs and found that with WCAG 2 contrast math, ~47% of passes were too dark to read for all users and should have been rejected, and paradoxically, some 22% of rejected colors should have passed, and were clearly better for color vision deficiency.

I and others have written about this extensively, and not focused on edge cases.



I will also remind us all — that for some people with reading disabilities — high contrast is the opposite of what they want.   And the text that meets WCAG is less readable to them than low contrast (which is more readable by others).

Yes, user personalization is a key user need that is under represented today, though that is changing.

But also there are great misunderstandings here, particularly regarding “high”, “too high” and “low” contrasts, and these are polarity sensitive issues and (often) related to adaptation state, glare from chromatic aberration (red/blue issues), etc. it is indeed a very complicated space.

These factors are already considered and discussed in the documentation, or under evaluation.


 So the fact that something is more or less readable - or certain color combinations that pass are not signs that something is bad — or everything we propose will be bad.

This isn’t about certain colors, or edge cases, this is about across-the-board inaccuracies that amount to errors of as much as 250%, for all users.


Let’s see some research -  that has been conducted on a number of people with low vision.  And a study with   color blindness   (and studies with both)     and see which measures are better overall.  And the colors should be randomly selected — and not include something right on the border  (though if a sample is big enough they can be used as a covariate).

Let’s see some empirical research actually conducted on WCAG 2.0

I’ve listed in previous posts the various documentation the science that supports the current work. I have a standing offer with you for a zoom call to go over everything.

Much of this is is already well understood through the readability research of Lovie-Kitchin et alia, Legge, Arditi (among my favorites in the field).

As I mentioned previous, this work divides into:

  *   A perceptually uniform contrast prediction model (complete, some new features to come)
  *   Aligning that with the existing science in readability (complete)
  *   Applying this to guideline development (a version is complete, but development continues)
  *   Wide evaluation studies for the system (public beta conducted for 2.5 years, some specific studies pending funding)

Lets work off of
- something that is based on studies of readability of text by people with low vision - and dyslexia - and colorblindness

100% of APCA Readability Criterion (i.e. the guidelines) is based/traces to the wealth of peer reviewed science in readability.


- lets not do this based on anecdotes or people with no visual/reading disabilities.

APCA is not based on anecdotes at all, and I’ve offered to go over the background and science on a zoom call with you. Such a statement is notwithstanding.

- and not on anecdotes or edge case -  but on the whole of the colorspace — or random sampling of it

I really don’t know how/where this idea that APCA is somehow based on edge cases came from, though I do know that there are a number of false and derogatory rumors floating around, the motivation of is not clear???? APCA is not based on edge cases at all, and there is no basis for such a statement.

Thank you for reading

Andy




Thanks


gregg

------------------------------
Gregg Vanderheiden
gregg@vanderheiden.us<mailto:gregg@vanderheiden.us>



On Aug 3, 2023, at 6:05 PM, Andrew Somers <Andy@GeneralTitles.com<mailto:Andy@GeneralTitles.com>> wrote:

Thank you Jon, I very much value your feedback on this issue.

I’m going to follow up with additional questions when I get back to the studio.

Andy

Sent from Andy’s iPhone

On Aug 3, 2023, at 17:10, Jon Avila <jon.avila@levelaccess.com<mailto:jon.avila@levelaccess.com>> wrote:


The WCAG contrast failures found in the cited Tourism document such as white on red and white on gray and light blue on white are difficult for me to read.  The white on red is also almost painful for me to read.

I am thankful for a current WCAG 2.0 contrast model that has correctly flagged these combinations of colors that provide problematic contrast.  While the model is not perfect most of the time for me it seems to flag things that are difficult for me to read.  I agree that we need to evolve the model or start over with changes in technology – but simply removing something that does have benefit with a void is not something that would be helpful to me.

Best Regards,

Jonathan

From: Andrew Somers <andy@generaltitles.com<mailto:andy@generaltitles.com>>
Sent: Thursday, August 3, 2023 6:04 PM
To: Patrick H. Lauke <redux@splintered.co.uk<mailto:redux@splintered.co.uk>>
Cc: w3c-wai-gl@w3.org<mailto:w3c-wai-gl@w3.org>
Subject: Re: CFC - Publish WCAG2ICT First Public Working Draft


On Aug 3, 2023, at 11:26 AM, Patrick H. Lauke <redux@splintered.co.uk<mailto:redux@splintered.co.uk>> wrote:

"Continued promotion" or "dealing with the reality that WCAG has been either enshrined into, or referenced by, legislation around the world right at this point, so we're having to make do with the less-than-ideal situation out here in the real world"...


Hi Patrick.

Right, 1.4.3/1.4.11 specifically are not things that should be incorporated into statute law as they stand. That they have become such is setting actual accessibility backwards. While not good as voluntary guidelines, as elevated into “law” becames bad law. The 508 has reasonable exceptions, unfortunately the EU does not, and it needs to.

Something you may find amusing from a WCAG trash-panda perspective, is this guidebook on accessibility<https://www.visitbritain.org/sites/default/files/vb-corporate/visitengland_national_accessible_scheme_serviced_standards.pdf>  This booklet on accessibility from the British Tourist Authority, England’s national tourist board, This accessibility guide was created in 2011:

https://www.visitbritain.org/sites/default/files/vb-corporate/visitengland_national_accessible_scheme_serviced_standards.pdf



The part that is amusing is not only that it fails WCAG 2.0 in areas, but how those fails are properly passes with a perceptually accurate model, such as APCA (the fact it’s a pdf notwithstanding).

Here’s an example from the inside front cover, the text is white on red—for color deficient vision this is ideal—yet WCAG 2 insists that black on red is better.

For comparison I created a matching version with black text—Black against red is notably worse, especially for common color vision deficiencies. WCAG 2 contrast presents conditions that are harmful to readability, particularly for those with color insensitive vision. This has led to a massive misunderstanding in the accessibility community as a result, with promotion of the mistaken belief that WCAG 2.x contrast is “doing something special” for CVD. It isn’t, at least not in a good way.

This is not an isolated case, and it is one worth noting.


<image001.png>






P
--
Patrick H. Lauke

https://www.splintered.co.uk/ | https://github.com/patrickhlauke

https://flickr.com/photos/redux/ | https://www.deviantart.com/redux

https://mastodon.social/@patrick_h_lauke | skype: patrick_h_lauke