Address Fields vs Address Book Fields in a customer search

I am trying to export customer addresses for cleanup because someone loaded zip codes and dropped all the leading zeros during the import.

In trying to write the saved search I am finding more address lines than what appear in the customer record. I am seeing 2 results for each address that is displayed. They have different Address INIDs if I add the join INID vs the primary address INID value.

The address 1 is also different so I am using a formula to restrict the results.

What I don't understand is why/how there are hidden or older versions of that address record able to be queried. Do I need to worry about them? Do I fix those as well? The odd thing is that one of the 4 lines (in this customer example) has a 5-digit zip, but the phantom line doesn't so it also shows up in my search to be fixed, but visually it does not need to be fixed.