> Last Writer Wins is not the most compelling use case for CRDT because if those fruits were denormalized into Postgres rows, you'd already have the behavior de

> Last Writer Wins is not the most compelling use case for CRDT because if those fruits were denormalized into Postgres rows, you'd already have the behavior described. I'm not sure from the article, but I'm guessing the difference here is that it will apply the updates in the correct order (using last writer wins) regardless of when the yjs doc hits the server. I guess the idea here is you can get decent offline-ish support without directly storing timestamps, version numbers, soft-deletes etc and associated logic.