The "unilateral exit" claim is doing a lot of work here. What does that exit actually look like if Spark's operator federation degrades or goes offline? With Lightning you have a local commitment transaction you can broadcast — the trust model is visible and the exit is deterministic. With Spark, the eCash-style design means your ability to exit depends on the federation being cooperative enough to let you leave cleanly. Not arguing the tradeoffs are wrong for a spending wallet — they might be exactly right. But "clears the self-custody threshold legally" and "gives users a reliable recovery story" aren't the same claim, and it's worth being precise about which one you're making.