Booting aught to wait until the baby is ready to fly, but while we're testing we have to be prepared for some broken eggs. In the days when GNU and Linux, though they knew they needed each other, couldn't have simply been called GNU because Microcomputer programmers were still coming to the term, daemon, and thinking that, in future, Kernel could operate to the will of Man, just as we do with gravity, someone, and we can't be sure who, provided a boot programme called Lilo.
The Linux Loader (and we omit to capitalize the letters that make up the app identifier because wars used to purge the world of stupid people) was replaced by Grub, which was a true GNU app. Grub-Linux, then, we might just as well call GNU.
Microkernels can be small: small enough to fit in a boot app. This is worth thinking about, because we might come to look at bootless systems.
But app startup is the same, whether in a boot app or otherwise. There's always some room for creativity, but not, for instance, that which would allow us to mount the filesystems after starting the startup apps.
Those who talk about doing without a distribution aught to realize that sensible people, such as mathematicians, thought about that, and thought better of it. Meta-science deniers surely don't need to do anything but look at an existing system to know how it works and how it may be improved?
You know, we don't expect everyone to understand: we can give people diagrams to help them get started; we can't, without making our apps look stupid, put these diagrams into them.
Right at the top, then, comes documentation. To the current maintainers of Grub I have to say, by all means add features; but if you omit to give a quick reference of the features that were there before you started, and the features you add only have stub entries in a document, you seem to be expecting others to follow in your footsteps, sing your praises, and merrily work out what the fuck all those extra commands do, after having figured out what you renamed the old ones.
It is an app startup problem. But you're not documenting your discoveries. Someone else isn't going to follow in your footsteps. They'll start all over again.