Cassidy James   Mar 21
What do you think is the venn diagram of people who say, “GNOME doesn’t listen to what users actually need? How many users change the default settings?” and “Any metrics or telemetry are inherently evil and must be avoided at all costs!”
jane400 - 628.4 fanMar 22
Show threadCassidy James   

To be clear, I'm very cautious when it comes to metrics or telemetry. But at the same time, I understand that GNOME *literally cannot measure* how many people change any settings, which is always the argument for GNOME doing something differently.

“Everyone installs X extension!” Do they? How can we know that?

“Everyone changes Y setting!” Oh really? How can we validate that?

“Everyone wants GNOME to work like Z!” Oh, I would love to see the user research data!

Mar 21 at 7:50pmWeb
Show threadCassidy James   Mar 21
Without that, we design an opinionated experience, listen to the *aggregate* feedback (not just the loudest people on Reddit or Mastodon), and continuously iterate.
Show threadachillMar 21
@cassidy most people are just reactionaries anyway and not worth listening to
Show threadPaulMar 21

@cassidy It's kind of like voting - don't vote, can't complain about pavements and street lighting.

Providing telemetry and accepting it will be aggregated with others is the easiest (and nicest) way you can help influence an open source project. Even if you do not like a future change, your telemetry has been part of the decision.

Which is a bit awesome really.

Show threadNeal Gompa (ニール・ゴンパ) Mar 21

@cassidy At least with extensions, you *do* somewhat have sizable data points with extensions.gnome.org and extensions distributors ship by default.

Same goes for settings. If distributors are changing them or considering changing them, that's a decent signal for consideration.

As for the last point... 🤷

Show threadEgeeMar 21
@cassidy I think telemetry is super important! But as a privacy-conscious user I need better assurances besides "its totally anonymous - trust us 😉". That's my main reason for turning it off.

On test machines and some homelab stuff I leave it on. Like I have a workstation thing with KDE and its on level 1 or 2 telemetry and that's fine.
Show threadCassidy James   Mar 21
@egee right, the entire stack should be open source and auditable and ideally involve differential privacy/randomized response to guarantee individual anonymity. But it’s hard to propose that when the knee-jerk response from a very loud minority of folks is, “BOO, TELEMETRY EVIL, GNOME EVIL!”