The new CLI sub-command model helps here I think. Many commands by default now list out their sub-commands when not given one and each sub-command has it's own

The new CLI sub-command model helps here I think. Many commands by default now list out their sub-commands when not given one and each sub-command has it's own help output. This plus command line completion really does a good job at discoverablility IMO. Oh and "memorise a seemingly meaningless sequence of text" makes no sense as all sequences of text are meaningless until we assign them meaning which is what you're doing learning a CLI.