I agree, but alternatively, we could study git and the lessons learnt there and design a distributable ACID database with version controlled data, extensible user collections and the performance of a more traditional database.