Not sure this always happens in practice. I’ve seen lots of redundant resolvers being implemented for the same type at work because someone didn’t know one already existed.