OK, hi all. Very sorry that it took me so long to get back to this breakout topic. This was a lot of comments to read through, and honestly I’m not sure that I learned very much from the last 200 or so. It looks like we’ve reached a point where a few people are repeating the same or similar points again and again. The Fedora community has long known this doesn’t work well (or win arguments!) on mailing lists. I don’t want to suggest we stop sharing our opinions here, but if we did so less repetitively, this discussion sure would be a lot easier to read and more people would feel comfortable participating.
Anyway, news: we are willing to try an Accept/Reject prompt instead of a toggle. I’m quite nervous that the consent rate might not be satisfactory, but we can give it a try and find out.
We actually should be able to get a decent enough estimate of the consent rate based on dnf countme data. I was wrong about not being able to do that. So if you don’t consent, we shouldn’t need to record that via the telemetry system after all.
I’m tentatively planning to use the “dark pattern” of making the Accept button suggested (blue) and the Reject button not suggested (normal gray color) so that accepting is presented as the choice Fedora suggests that you make, which should improve the consent rate and increase the chance that the Accept/Reject prompt is successful. I know several of you have said you won’t be happy with anything other than a neutral prompt, but at least there should be very low chance anyone would click through without noticing that they’re consenting to data collection, and it would undoubtedly be an opt-in. We don’t have any mockups to share quite yet, but we won’t design it to be confusing or trick users into clicking Accept.
My intention is to link to a wiki page with detailed information on what metrics will be collected (which would open in Firefox). Most users admittedly won’t click the link, but the information will be easily accessible for those who wish to see it. I really do not want to show that much information in the OS itself because (a) it will change relatively frequently, but UI cannot easily be changed (requiring e.g. localization), and (b) it will make the UI much more complex, but GNOME is designed to be clean and simple.
I will need a few more days to prepare additional revisions to the change proposal. Anyway, hopefully now we can at least we can move on from arguing about opt-in vs. opt-out, and debate more exciting topics instead, like, um, button chromatism.