Nice! I realized talking to nostr:nprofile1qy88wumn8ghj7mn0wvhxcmmv9uq35amnwvaz7tmwdaehgu3ww35x2umpd4jkxct59e5k7tcpz9mhxue69uhkummnw3ezuamfdejj7qgjwaehxw309ahx7um5wf6k2tnrdakj7qg6waehxw309ac8junpd45kgtnxd9shg6npvchxxmmd9uq3wamnwvaz7tmjv4kxz7fwdehhxarj9e3xzmny9uq3gamnwvaz7tmjv4kxz7tpvfkx2tn0wfnj7qgnwaehxw309ac82unsd3jhqct89ejhxtcpz4mhxue69uhhyetvv9ujuerpd46hxtnfduhsqgpxdq27pjfppharynrvhg6h8v2taeya5ssf49zkl9yyu5gxe4qg55zddau4  recently that migrating is a huge part of making outbox work. Planning to add some tools for this to coracle. 

 one thing Nostr can't have without reinventing merkle trees with events (which be awfully inefficient): verifying that migration was pr wasn't complete.  

 Totally. If someone changes their relay list, outbox doesn't know it existed. In a perfect world, NIP-65 would also store timestamps for when a relay was added. When removing the relay, clients user events from that timestamp. Or otherwise, relays are not "removed" from NIP-65 lists, but marked inactive and marked with a timestamp indicating when the user left. 

 pre-coffee notes are imperfect. 

 post-coffee

*clients would sync events from that timestamp to other relays.
*but marked inactive with a timestamp indicating when the user left. 

 Not a bad idea