For solving all things complex, there must be a plan.
Never is a long time and grand plans need to be executed for their real flaws to appear. The kind of flaws that are never what we thought in advance, usually much worse.
We do need to think about it until the end of humanity. We've build countless societies/civilizations and non of them survived the test of time. It's our ultimate puzzle.
There is probably [say] someone at MS who knows how an OS should work but replacing parts in a running machine isn't easy. Burning everything to the ground isn't ideal either but it does make building more attractive.
Now that I think about it though, it is more to do with inflexibility of the plan, rather than having a plan itself. If you are working off of a ideological commitment, rather than setting an end goal with a fuzzy time frame and a loose path to get there, then that's when you land in trouble.