Then I spent about a decade and a half trying to embrace alternatives and I concluded that software that handles general use cases is actually a hard, specialized-skillset problem, and there's a lot of benefit to the full-stack control (and large developer culture, and multi-decade legacy history of successful product) that a long-legacy monoculture provides. Most of the friction is in wanting to use something that isn't in your monoculture, and trying to solve that by building a more general solution really just adds another monoculture to be incompatible with your favorite one (see https://xkcd.com/927/). Virtualization has basically solved that rare problem in my use-case.

To each their own, of course; I got bored with the game of making the computer architecture work over paying someone else to do that and then using the computer to make things.