Nothing personal. I just wanted to set the record straight, keeping in mind your comments about reading what is not there and stretching the meaning of words beyond their author's intent.
hey ... you missed an opportunity to explain how the old-n-good MailAddressSecondLine issue works for you. Are we in agreement there are attributes that do not call for any relationship and nobody cares if they carry or not a value?
If nobody cares then by the YAGNI principle the attribute should be left out of the database altogether.
I think the problem you are referring to is that it can be a mistake to assume certain properties about addresses which do not hold true in reality. That doesn't have very much to do with a discussion about normalization. It just means that as part of the design process you should determine what dependencies your data is actually going to satisfy.