The thing is, there are a large number of different reasons to store an SSN as a long int or a string depending on how it is used with the rest of the data. For a phone number, there can be a valid reason to store the area code separately to speed up data queries that narrow down by area code instead of all in one field and peeling it apart. There are also reasons to have additional, seemingly redundant, columns that can be used for optimizing searches or simplifying how queries are written.

A common one is that using 1 and 0 instead of Y an N is often faster for massively large dataset optimization, but isn’t as easily human readable.

There are complex reasons for choosing different approaches in a database, and the most important thing is generally consistency within the database. His point is meaningless without context beyond consistency, and the different government systems will have had different priorities, not to mention trying to update all of the databases to make them consistent is a MASSIVE fucking undertaking. And the systems can stay the way they are as long as they have APIs or other methods of transferring data that ARE normalized and consistent.

I have personally been working with reporting data to federal systems for 15 years as a semi knowledgeable technical person. This is what I do for a job. What he is saying is pointlessly small trivia used to justify tearing things down instead of improving them.