It minimizes data storage duplication, but not reading. The language does not have a simple way to receive unique records across N-1 joins.