Or cost... or abundance of existing tooling for... or a ton more factors that are all considered. And quality itself is multidimensional. Something might be good for engineering flexibility but terrible for exposure to end users, or vice versa. And engineers can work around shortcomings much better than end users can.
OSes aren't like Javascript frameworks. There are only so many viable choices for a given domain. It's like if someone chose T-Mobile over AT&T and Verizon and you automatically concluded it's because of quality.
I do not understand the argument. Linux won. It won because it — for a variety of reasons — fits certain domains well.
The OS vendor can change the kernel being used, and those average users won't notice it all.
The only quality is being a free beer OS, that the OEM vendors can leech at will without bothering one second up-streaming their changes.
And when the community started to be too vocal about it, efforts to use MIT/BSD licensed alternatives were set in motion.
In a couple of years those router OSes, smart televisions, Android phones won't be using the Linux kernel any longer.
MIT/BSD licensed stuff is still free software.
I advise you to read about Zephyr OS, FreeRTOS, Azure RTOS, Fuchsia, NuttX,....
A closed-source OS in this space has no real chance, as it has to be adapted by hardware makers. Microsoft may be able to provide the backing for wider ThreadX adoption, but don't hold your breath.
I don't see how the fact that BSD allows proprietary forks implies it is not high-quality software.
Again, where is the source code to Playstation OS 4 or clang bitcode changes for watchOS?
Got to love FOSS.
Try giving an average user Photoshop. See how they weep. These thought experiments are really hard to apply to software on complex domains.
OpenOffice is indeed a terrible example.