@witchescauldron@serapath When I was last playing with it, hypercore was (is?) based on a directed acyclic graph and worked well. It was lacking multi-writer support (without CRDT last I saw) but I believe that was finally addressed in time. The core tech, at least back then, was a genuine attempt to create a replicated datatype and tools based on it without any bloat (or a business model beyond grants). You could likely build on that foundation.