Comment by pwg

And now you've just discovered why relational databases have a "null" concept that is separate from the "empty string" concept.

In a relational table, all rows in the table have a "middleName" column. So you can't "omit" the column entirely for some rows like you can in a doc/KV store.

"NULL" in a relational DB is to store, in a table where every record has a "middleName" column, the concept of "missing key" from a key/value store type layout.

NULL is the relational DB's way of representing: "a record without a middleName key" exists here.