Something about the whole dreary mudfight of hardcore #Linux and #BSD nerds vs. their own userbases—sadly it would seem that @netbsd also regards itself as at war with ordinary users who are running out of options when it comes to finding and using software free from technofascist corruption—has me wondering if there isn't some novel approach to personal computing that's in desperate need of research and development, based round the idea that fabricating one's own chipsets is not wholly impossible.
Operating system development is such a difficult and fractious thing in part because the task of supporting so many different hardware configurations is extremely difficult and frustrating work. Is there hope, then, in tuning the hardware itself? Folks already pick and choose hardware when assembling a PC with an eye to best OS support but I am vaguely imagining going further, creating a chipset that would need a far less complex OS in order to present a generally useful UX to the end user.
Hardware is like 100x harder to deal with than software and its expensive. You could hire a few people to port linux drivers to genode at a lower price than just getting your first hardware revision out.
Building for a specific device that already exists is significantly easier, but you still end up on someone else's support treadmill.