distributed consensus is a huge field. only since bitcoin did byzantine fault tolerant types get big. CRDTs have been used for a long time, skype was the biggest example. bittorrent with it's heterogenous tracker/seed pattern is another.

the property you want for your purposes is consistent and available. partition resistance (big for bitcoin) is not so important. have a look at the pnyxdb consensus for an example. lamport clocks also are useful, i built a simple nostr replication system based on it, just as good as negentropy when the goal is total replication.