Because of how MVCC is implemented in Postgres. However, it's not terribly hard to add efficient count tracking that works with this implementation of MVCC, usi

Because of how MVCC is implemented in Postgres. However, it's not terribly hard to add efficient count tracking that works with this implementation of MVCC, using row triggers. I use it often in production. This might change in the future with the zHeap storage backend.