I sympathize too. Things being out of control is not easy, but it's also not a priori wrong. Maybe gardening is a good metaphor. You attend to what you can, but a good garden isn't defined by "use cases".

With the lists stuff, you did exactly the right thing, which was observe what existed in the wild (and should have been documented in NIPs as far as that goes), then invent your own solution when you hit a wall. The interesting question is what do you think about the recent NIP 51 re-write? It deprecates 30001 entirely, replacing several ad-hoc solutions with new, different kinds. This was done with the consensus of an overwhelming number of developers, but was top-down (i.e., spec first, then implementation).

Just watched your talk. I love the project! I've also thought long about the WoT problem. Two things to look into — are you aware of NIP 89? It solves the client cataloging/recommendation problem in a WoT-compatible way. And are you familiar with NIP 32? I wrote that one, the goal was basically "decentralized lists" — basically, you apply an adjective to any target (event/pubkey/link), and then anyone can filter those adjectives based on WoT or some other criteria. Seems less cumbersome than merging recursive lists, but I'm not closely familiar with the work you've done.