[mediaqueries-4] Add `prefers-handedness` media query for left/right-handed users

[slafleche]

I propose adding a new media query to detect user handedness preference, similar to existing prefers-* queries like prefers-reduced-motion and prefers-color-scheme.

Proposed syntax:

@media (prefers-handedness: left) {
  /* Styles for left-handed users */
}

@media (prefers-handedness: right) {
  /* Styles for right-handed users */
}

Use case:
Handedness significantly affects UI interaction, particularly on mobile devices where touch targets and navigation controls should ideally be positioned for comfortable thumb reach. Approximately 10% of the population is left-handed, yet most mobile UIs default to right-hand optimization.

Practical applications:

• Positioning floating action buttons and navigation controls on the appropriate side
• Mirroring toolbar layouts for thumb accessibility
• Adapting swipe gesture directions
• Optimizing one-handed mobile interface layouts

Current situation:
Developers currently have no way to detect handedness preferences without building custom in-app settings. This forces users to adapt to right-handed defaults or requires each app to implement individual solutions.

Precedent:
This follows the established pattern of existing accessibility and preference media queries (prefers-reduced-motion, prefers-contrast, prefers-color-scheme, etc.) that expose OS-level user preferences to web content.

OS integration:
Many mobile operating systems already have handedness settings (particularly for keyboard placement). This media query would expose that existing preference to web applications.

This seems like a straightforward accessibility improvement that would benefit a significant portion of users with minimal implementation burden.

[emilio]

This seems like a straightforward accessibility improvement that would benefit a significant portion of users with minimal implementation burden.

Main concern here would be how much the accessibility improvement is vs the extra fingerprinting vector that we expose, fwiw.

[linc]

Main concern here would be how much the accessibility improvement is vs the extra fingerprinting vector that we expose, fwiw.

It's a reasonable point, but I'm skeptical it's any worse than light/dark mode in isolation, and I feel more concerned about "slippery slope" arguments against accessibility attributes than in favor of them.

[emilio]

Sure sure, not particularly against this feature tbh, just pointing out that it's never so straight-forward as it seems :)