So, in short, I think I agree with the op:

> I don't see anything that indicates how spacetime solves the core problem CRDB does, which is guaranteeing that single node failures can be tolerated with zero data loss or loss of availability