Media Queries Level 5: prefers-reduced-motion, prefers-contrast, prefers-color-scheme

[tantek]

Request for Mozilla Position on an Emerging Web Specification

• Specification Title: Media Queries Level 5: User Preference Media Features
• Specification or proposal URL: https://drafts.csswg.org/mediaqueries-5/#prefers-reduced-motion
• Specification or proposal URL: https://drafts.csswg.org/mediaqueries-5/#prefers-contrast
• Specification or proposal URL: https://drafts.csswg.org/mediaqueries-5/#prefers-color-scheme
• Caniuse.com URL (optional):
• Bugzilla URL (optional): https://bugzilla.mozilla.org/show_bug.cgi?id=1365045
• Bugzilla URL (optional): https://bugzilla.mozilla.org/show_bug.cgi?id=1506364
• Bugzilla URL (optional): https://bugzilla.mozilla.org/show_bug.cgi?id=1494034
• Mozillians who can provide input (optional): @heycam, @dbaron
• W3C Tag Design Review issue: none

Other information

This issue is scoped to three of the new features in Media Queries Level 5, which we have implemented, as part of evaluating Media Queries Level 5 as a whole: prefers-reduced-motion, prefers-color-scheme, and most recently, prefers-contrast (as announced last week).

These features enhance accessibility features of the platform and thus we should consider a status of "important" for these three as a whole.

Label: w3c.

(Originally published at: https://tantek.com/2020/195/b1/)

[martinthomson]

@annevk, you commented on the pull request about the (unexplored) privacy risks of these features. I see that the issue remains open.

So for @tantek, I want to ask whether we are comfortable saying that these are important without fully understanding the implications?

[hsivonen]

The previous comment above links to a spec issue, and it seems the spec still has as issue box without having resolved the concerns further.

There's an intent to prototype about prefers-reduced-transparency that refers this position issue even though that prefers- pref isn't listed in the first comment here. At least in the case of prefers-reduced-transparency is relevant how people end up with the corresponding system pref set on macOS. In my case, back when Apple made the menubar translucent, I, IIRC, followed some piece of advice about a defaults write incantation to restore the opaque menubar. This setting has then been migrated from one Mac to another by Migration Assistant and gained GUI with a menubar-unspecific characterization under the Ventura a11y prefs. For me, getting rid of menubar translucency wasn't a strict a11y requirement but an aesthetic disagreement with Apple's designers. I had no idea that the defaults write incantation would have Web fingerprinting implications years afterwards.

Further, anecdotally, in a meeting of 6 people, I wasn't the only one with the pref set for the macOS menubar reason, but there was also another person with the pref set for that reason.

On Mac, the system-level effect of some of the other prefs in the same group of toggles could also be plausibly turned on based on design disagreement with the current system-level GUI fashion (e.g. the high contrast and toolbar button shapes prefs draw borders that were the norm for everyone in past decades), which, while a preference, may not be a strict a11y matter and, therefore, the users might not consider the prefs essential when weighed against trackability on the Web.

Even when these prefs are flipped due to strict a11y needs, https://w3ctag.github.io/design-principles/#do-not-expose-use-of-assistive-tech indicates that explicit consent would be appropriate for exposing these system prefs to the Web. Prompting the user with "Do you want to expose system accessibility preferences to the Web?" is problematic both in terms of interrupting the user so that the just wants to dismiss the prompt and in terms of communicating the tradeoffs. Having a checkbox somewhere in the browser prefs to expose system a11y prefs to the Web, on the other hand, risks people who would need the prefs exposed for a11y reasons failing to discover that it's a possibility.

[CanadaHonk]

@hsivonen Thanks for the comment, fwiw, I made a comment on the PR about newer prefers- here: #410 (comment)

Good point with transparency, it might be worth making a csswg spec issue since other vendors would run into this as well?

[lukewarlow]

Fyi I've raised a separate issue to get a position on prefers-reduced-transparency #851 which isn't explicitly covered by this one.

Also similarly I've raised an issue regarding inverted-colors #853