My theory is that old systems were too limited for modern BS like this. BIOS had to fit in a couple hundred kB. There was no room for nonsense like UEFI has.

My theory is that old systems were too limited for modern BS like this. BIOS had to fit in a couple hundred kB. There was no room for nonsense like UEFI has. Other than POST, it booted pretty quickly because even a slow CPU can do anything that credibly fits in 200kB pretty quickly. (O(n^2) algorithms were pretty unlikely at boot time.) Having three or four different control panels each controlling part of, say, the network configuration takes developer time, compile time, and distribution space. When everything needs to fit on one CD or, better yet, a few floppies and the companies were new and very small and scrappy by modern standards, this would be utterly impractical. There was one control panel because it was cheaper and less code! Want one developer and zero professional designers to make a dialog box? You use one well-designed widget set (from the OS vendor!) and you just do it. Everything is easy to use because the resources needed to make pretty-but-undiscoverable UIs weren’t available. (We see this today writ small in the embedded space. For example, Secure Boot is an abomination that, as implemented in essentially all cases, requires truly bizarre CPU resources and complexity. So embedded systems use vastly simpler chain-of-trust systems.)