I may be misunderstanding you, but is this loosely correct? It merges the distributed data like git, except in cases where there's a merge conflict, the newer o

I may be misunderstanding you, but is this loosely correct? It merges the distributed data like git, except in cases where there's a merge conflict, the newer one just wins? It's certainly hard to come up with hueristics for automated merging in all corner cases, and punting and asking the (likely best-informed) user to merge, as git does, seems like the most practical solution. Unfortunately, automated replication doesn't seem to allow that option. Am I really off base here?