I had similar frustrations with a well-known deep-learning framework.
I'd get well-vetted PR's for the build system, CI system, etc. committed. Only to have them later undone, without discussion, by people in the inner circle.
It doesn't take many experiences like that to lose all motivation.
I think some open-source projects deceive themselves (and therefore, outsiders) about their contribution processes.
Nothing stifles progress and contribution more than having a maintainer take your effort and implement it themselves or carelessly merge your work in without review only to have it reverted for not contributing code correctly. For a lot of FOSS the review process is that the maintainer does it themselves when they feel like it.
And you just know the discussion by that inner circle that could have happened during code review is happening in some private email or chat app somewhere away from public eyes.
I'd get well-vetted PR's for the build system, CI system, etc. committed. Only to have them later undone, without discussion, by people in the inner circle.
It doesn't take many experiences like that to lose all motivation.
I think some open-source projects deceive themselves (and therefore, outsiders) about their contribution processes.