RE: Focus visible

Hi folks,
Just thought I’d offer some considerations on this.

Focus Visible was not moved to AAA as a result of logic errors. A LOT of time was spent getting consistent wording of conditions that could lead to a marriage of decent focus indicator contrast and size. The wording of those conditions was iterated many times, each version applied against dozens of common failures, edge cases, etc., that we had compiled over time to ensure we had an approach that seemed to work well across our samples.

The challenge came when those metrics, which seemed to work most consistently, were applied in real-world testing and assessments. The results suggested that inter-rater reliability was going to be very low in quite common situations.

The three main challenges we identified were as follows: 1)  user agents implementations for common focus indicator techniques – such as dotted and dashed outlines – was so variable that mechanisms for arriving at size/area calculations fell apart for anyone using a non-solid-line indicator 2) where any  focus indicator fell on a variable background – say the item with focus is over an image or a gradient background, both of which happen frequently – the means of calculating contrast became subjective and prone to error based on pixelation, anti-aliasing, etc. 3) Many situations cropped up where other visual elements on a page happened to intersect with the focus indicator, causing reductions in contrast that affected the area/size calculation – this happened much more often than one would think. Such intersections also meant that requireing a ‘solid’ indicator became more problematic.

Change by users to text size, zoom level, breakpoints, etc. can put many of the controls on a page into scenarios where one of the above become more commonplace. We arrived at situations where highly visible focus indicators were failing the criteria.

Please note that the same challenges exist with the wording Guy referenced that existed in 2020. In a nutshell:

  *   A level of contrast for the focus indicator needs to be defined (measured against both other things on the page and against the non-focused state);
  *   A yardstick for assessing the area/size of the focus indicator needs to be defined;
  *   Those 2 considerations both need to exist, and need to be aligned in a manner that results in poor or good designs appropriately failing or passing;
  *   The focus appearance testing results need to be consistent across multiple testers and environments. And yes, testing scale (automation, ease, etc)  also comes into play.

I still believe firmly that something can be created that balances these considerations, but as one of the people who spent 100s of hours working on this, I can attest that it is not easy.

The failure wasn’t a result of lack of effort. We simply ran out of time for WCAG 2.2.

But my sense is a solution is tantalizingly close. If  folks who have been participating in its lament here and elsewhere could bring some new minds and energy to bear on this, I believe a solution could be arrived at. To me, this requirement, alone, would make a 2.3 worthwhile.

Takers?

Mike

From: Guy Hickling <guy.hickling@gmail.com>
Date: Wednesday, May 24, 2023 at 8:45 AM
To: WAI Interest Group discussion list <w3c-wai-ig@w3.org>
Subject: [EXTERNAL] RE: Focus visible
Jan, SC2. 4. 13 is, actually, not the same thing, as it is Level AAA. There were originally two versions of this SC, at AA and AAA. Just keeping the AAA version means all the millions of websites that work to AA level won't look at it, they
ZjQcmQRYFpfptBannerStart
This Message Is From an Untrusted Sender
You have not previously corresponded with this sender.
ZjQcmQRYFpfptBannerEnd
Jan,
SC2.4.13 is, actually, not the same thing, as it is Level AAA. There were originally two versions of this SC, at AA and AAA. Just keeping the AAA version means all the millions of websites that work to AA level won't look at it, they just consider AAA criteria as out of scope. So, of no use to people with vision impairments who want to see all websites, not just the very few who go to AAA!

To clarify, back at the end of 2020, we had 2.4.11 Focus Appearance (Minimum) at AA level, and 2.4.12 Focus Appearance (Enhanced) at AAA. They had almost identical wording, except that the AAA one said the indicator should be 2 pixels thick (as 2.4.13 does today), but the AA version said 1 pixel. (And the AA one had an additional clause as well about contrast with surrounding colours.) So a bit like the colour contrast SCs, with 4.5 to 1 at AA and 7 to 1 at AAA. And that remained the position for a long time. See https://www.w3.org/TR/2021/WD-WCAG22-20210521/#focus-appearance-minimum<https://www.w3.org/TR/2021/WD-WCAG22-20210521/#focus-appearance-minimum>

But then they did a further change in September last year which, the way it was written, did not logically make sense. And now, instead of removing the option A which was causing the logic problem, they have simply deleted the whole thing and left people with low vision no better off than they were in WCAG 2.1 - just an AAA version which will do absolutely nothing for the vast majority of websites! If designers want to use thin dotted outlines, they can, and never mind all the millions of people who struggle to see them!