Of course I'm being somewhat extreme and provocative with my "completely bonkers". However I think it's that 90% incidental complexity you refer to that validates it as a fair description of pure functional programming. And causes my wondering as to why anyone would ever want to do that.
Do you have an example of a case where Functional programming works very well? Something that fits on a page and is understandable by mortals?
No doubt I'm totally missing a point but as far as I can tell it is impossible to compute anything of interest with pure functional programming and immutable data.
My naive mind thinks of it like this:
I can count my sheep in the field by carving successive notches on a stick. Each notch is there forever. If I want to count a new sheep I need to carve a new notch. Anyone with only a partial view of the stick may only see the old notches, unchanged. Or they can see the notches on my old discarded sticks. That is immutable data.
Soon, my farm is so big, I have so many sheep, I run out of room on my stick for new notches, I run out of sticks to use. So I invent a number system that compresses huge counts into little space. I can't use my number system on a stick as I can't remove notches. So I invent the abacus. I now have mutable data.
Some time later the tax man comes by. He asks to see my tally sticks from last year. Oops, now I have a problem! I need records, databases, immutable data structures, extra book keeping work. Incidental complexity.
What am I missing in my shepherds tale?