If Apple were to ever capture (and keep) more than 10% of the market, you'll see that they too will slave to maintain backwards compatibility. Their niche status affords them ability to make the sorts of changes you want.
I don't understand this intense desire for speedy change -- do we really need to replace the desktop metaphor as quickly as possible? Doesn't that just hurt, rather than help, adoption and understanding of computers? Everytime the GUI changes half the planet has to re-learn how to use their computers. If we had more than 3 primary desktop operating systems, all implementing new ideas, the learning curve for users would be impossibly high.
The fact that one operating system won is good for the industry.
you could have used all those same arguments to claim that guis are bad, let's just stick with ms-dos. back in the eighties, i remember many people making exactly that argument. heck, in the back of my mind, i was sort of thinking that myself! i had a great deal invested in the status quo. i'd spent years learning 8086 assembler and all the tricks necessary to write a dos tsr that wouldn't bring down the operating system. it was an ugly, stupid, convoluted system, but i was a master of it, and so i wanted things to remain the same.
apple is never going to stop throwing out old stuff and replacing it with new stuff, no matter how much of the market they get. not so long as steve jobs is alive, anyway. that's what makes the mac a healthy, thriving ecosystem. once an idea has outlived its usefulness, it's chucked for something better. the fact that the people in charge of windows are not willing to do that means that it is just a matter of time before the entire operating system is obsolete.
i'm not saying there shouldn't be a windows. it caters to a certain crowd that i am not a part of, thankfully, and i think that holds for most of us around here. but the idea that it has to be pretty close to the only operating system is a bad thing. fortunately, microsoft doesn't wield the kind of power that it once did, so there's a chance we'll get to see some real alternatives in the next decade or so.
On the other hand, the number of applications that Mac OS X must run to be accepted can be counted on the collective hands of the OS X engineers. They can almost afford to break whatever they want as long Photoshop runs. Microsoft does not have that luxury.
The command line vs. GUI argument doesn't hold water either. Instead we're comparing the desktop metaphor with some concept that doesn't exist yet. Operating system research is always on going and the really good ideas percolate up into all the current operating systems. You can't just radically change direction -- even Apple knows that. In fact, they've been removing features (like creator codes) that distinguished them from Windows and Linux basically because being compatible with the universe is more important than being innovative.
microsoft has demonstrated time and time again that, with windows, they value backward compatibility ahead of almost everything else. it still bears the mark of design decisions made in the eighties. that makes it pretty difficult to innovate in the windows space, even if microsoft wanted to, and they don't.
apple's macintosh is a viable competitor, but they're after a pretty clearly-defined market as well. they've got millions of customers they can't afford to upset.
three primary desktop operating systems is not enough. it retards the spread of new ideas. we are taking too long to get to whatever comes after the desktop gui metaphor, because any company that created such a thing today doesn't stand much of a chance of being able to sell it.