> When Ubuntu needs to experiment with something new, they don't need to design a "new innovation cycle" or some fancy new "sandbox process". They just go and do it - like any company using open-source does. Zero "bikeshedding" getting in the way. Then, if and only if the experiment was successful, they can consider submitting parts of it upstream and let the bikeshedding start. But now the proposition has been battle-hardened first.

Yeah…but how often have they ended up in a dead end with "nothing" to show for it? Mir? Upstart? How long until snap is abandoned for flatpak too? How many such experiments have made it upstream? I find a lot of the issue is that there's an attempt to keep some key parts of things "locked up" (either with a CLA and/or actual proprietary bits). I can't imagine Fedora going along with something like that. Though I admit I've been *far* outside the loop on project-wide efforts for the past…multiple years beyond LWN coverage despite being a user and a packager.

Personally, my latest gripe with Ubuntu has been their abysmal communication about releasing fixes for the last splurge of Linux kernel vulnerabilities. That seems awfully…core to being a distro to me.

> I realize this downstream approach does not apply well to infrastructure changes like Forgejo but it does seem to apply to pretty much everything else. Even with Forgejo: has IBM already been using it in anger internally?

AFAIK, there was a *lot* of community discussion around it (some of it even covered here). Of course, this is perhaps an instance of *Fedora* forging (heh) ahead on a custom solution (Pagure) instead of finding something out there. But then again, the forge "scene" was *very* different back when it was being selected too (pre-2016)… Though Pagure seems about 4+ years older than that, so perhaps just a "wrong horse" problem instead.