[esarbe]

Gnome does regular UX testing[0].

There's no 'feedback' channel (other than, you know, the regular feedback channels of matrix, gitlab. They exist.) because that tends to skew the data because of self-selection.

A/B testing is only possible when you can do randomized trials. I do think that randomly switching between competing implementations of a feature could be somewhat confusing for desktop users and create a big kerfuffle. It's easy with constantly-changing websites - people expect a website to change. On the desktop, change is seldomly appreciated.

[0] https://blogs.gnome.org/shell-dev/2021/02/15/shell-ux-change...