The commands for git are an example of that -> they don't make sense. They are very poorly designed and counterintuitive.
Linus I think has this belief that 'technology & complexity = power' when really, it's 'products that do useful things' that matter.
Making seamless experiences matters. Consumers really don't care how many techies think something is cool, they don't care about drivers, or open-source, they just want it to work.
Linux could absolutely provide the foundation for the desktop, but whatever is on top of it will have to be equally innovative and provide a lot of value, so much so that I'd argue that the high level OS layer will be 'the hard part' not the kernel.
Here's an example: "system, orangeFS, bringing the number of supported file systems to 35. (Remind me, how many file systems do Mac OS X and Windows support?)"
Really? Who cares. You only need to support ONE, the right one, and support it well. Maybe there is an advantage in supporting a few others, but not really. This is a perfect example of 'engineering' that doesn't necessarily ad up to any end-user value.
"Pretty is not my primary thing." and here it is. Linux conflates 'aesthetics' with 'design'. A very common failure. They are not the same thing, not remotely.