I don't think FOSS is even necessary for what I'm talking about (although it seems to be sufficient). Non-networked binary distribution of yore was also (mostly) aligned with the interests of its users.
It's much more related to who is paying the bills. Technical concerns do make some difference, but are eclipsed by the money question.
Our current problem is that people mostly don't pay for software anymore. So we are left with FOSS and anti-user software.
I've long pondered if an economy around Free software could be better implemented. We take it as a forgone conclusion that wide source code availability precludes charging per-instance, but really that's just an assumption.
Of course the idea is likely moot in the light of network effects, which seem to be a major underlying factor of the current harsh dichotomy.
(b) can be done with a proposition in the State of California and a few million spent on a very convincing ad campaign ("Vote yes on Proposition 42 to take back total control of your computing devices! Don't let advertisers and data leeches control and dictate your life from behind the curtain!"). California because it's home to a big chunk of Big Tech, and because it's a market too big to simply ignore and not sell into.
The flip side is all of this device churn is all the actually devices left in the wake, available for cheap/free. If software is done right, even last-decade specs are perfectly capable devices good for another decade. Which is what we need, because convincing someone to pave over their current device or spend a lot on a new "weird" one isn't going to happen. Tablets got a foothold by being a new type of device, rather than an immediate replacement.