What a good read. Thanks for this writeup! Only thing that confused me was the initial cost comparison bit. This seems to talk about the offchain ZK proof *generation*, when what really matters for user tx costs are the proof *verification* costs (what Ethereum charges for).