flip RFP's prefers-color-scheme to dark
RFP reduces this binary metric to useless (in our Tor Browser set of users) by always returning light
. We can achieve the same FPing protection by always returning dark
.
This is, IMO, not technically an accessibility issue, as the CSS standard is arbitrary, not universal - i.e only a few websites (but arguably large popular websites) implement it. That said, I am not an accessibility expert, or knowledgeable about or experience light hurting eyes and creating migraines etc. I will say I've never heard of anyone claiming dark sites did the same (but of course the default is light and we enforce light)
In ESR115 as a major milestone, we could change test always returning dark
. My logic for this is
- entropy is not affected
- accessibility may be helped
- I strongly believe accessibility re colors is best served under existing/upcoming standards that are universal (which we could preset/harden)
- there are degrees of usefulness, and accessibility advocates indicate that this helps (maybe they're lying just to advocate their perference, but I'm inclined to agree that it can't hurt and would likely help)
We currently get RFP users (and tom will confirm), who complain about the same few RFP items: it's always timezone, prefers-color-scheme, and now timing (60FPS). It is my belief that no matter what we do, people will complain, but by returning dark
, user's complaints are no longer anywhere near the validity of e.g. saying it causes migraines - in fact users who complain they get dark themed sites are just aesthetics (unless someone can prove dark themes are an accessibility problem)
In other words - flipping to dark cannot hurt fingerprinting, and can/would help usability
Class, discuss! cc @donuts