Colorblindness modes do not adjust label colors to be differentiable

[Phoenix616]

Select Topic Area

Bug

Body

If you have some form of colorblindness then certain label colors aren't differentiable from each other even with the special colorblind themes from the feature preview, even though they are different colors for non-colorlind people. One would expect the label colors to be adjusted in the same way that the rest of the colors of the theme get adjusted.

[ghostinhershell]

Hi @Phoenix616, thanks so much for reaching out with this feedback!

Can you give me more information about this issue so that I can send this feedback over to our Accessibility team? For example, what theme are you currently using? Do you have any example screenshots you can send over to demonstrate this issue? Any additional info would be helpful.

[Myndex]

Hi @ghostinhershell

I won't speak for Phoenix, but I've posted about this in the past, though that was a part of a very long thread. Here's a short recap:

There is a lot of misinformation regarding the actual user needs of those with color vision deficiency. In particular, there are a number of "simulators" out in the wild that are simply bunk, and will lead designers to incorrect conclusions.

Especially the achromatopsia sims.

The simulators that are accurate and based on clinical testing and peer review use the Brettel/Viénot Model. Dalton Lens is one, we have one at myndex.com/CVD/, you can process a local image or screenshot and it shows all the key CVD types.

The Github CVD schemes

I'm not sure what the intent was, but most of what I see makes no sense to me—the CVD schemes all look the same as the "standard" colors, making them of little utility. At the same time, luminance contrast is suffering throughout the site.

Are we there yet?

In other words, these schemes intended for the color insensitive don't actually do much of anything as far as I can tell, though to be sure I haven't tested out every page and every combination.

But as an example, in the examples below, the dark mode that is indicated for protanopia is actually worse for protanopia than the standard setting. deuteranopia and protanopia have unique needs, especially for dark modes. Protanopia really should not have any red against black, due to protan red insensitivity.

What puzzles me is GitHub is not doing all that much with color, anyway. It would be helpful to first get the visual hierarchy straightened out, and useable contrast for the dark modes, all of which are too low in contrast IMO except the high contrast one which is barely useable, at least for me and my low vision issues.

Remember that WCAG 2.x contrast math does not function for dark mode—I'd be happy to show you why on a zoom call. And if you are concerned regarding conformance, WCAG 2 provides that if the user has selectable choices, only your default layout needs to "pass" WCAG 2, all the other layouts can use a method appropriate for the various goals.

Because of the problems with WCAG 2 dark mode, I suggest using a perceptually uniform method. APCA is one, and the draft APC Readability Criterion provides for more granular guidance.

Thank you for reading,

Andy

Andrew Somers
Director of Research
Inclusive Reading Technologies, Inc.
W3C WAI AGWG Invited Expert, WCAG 3

Any opinions stated are mine and do not necessarily reflect the views of the W3C/WAI.