I find this to be an interesting thing about reading the Torvads–Tanenbaum debate, that both sides seemed to assume without much consideration that the best instruction set, and product in general, would always win.

With instruction sets in particular, there is a very large amount of inertia.