Not everybody is good at every task. When he made his website in the mid 1990s for his first business after college where he had business listings and such (I made a similar website from 2007-2014 for my area), he most likely used a flat-file database and not a relational database because web tools were not very sophisticated at that time.
It wasn’t until the late 1990s that you started to see more powerful tools for website databases, around the time they were working on Paypal.
“de-duping” is something you might do on a flat-file database. He is likely seeing the same social security number showing up with variations of names
So in his mind, there should be only one social security number per name – AND THERE IS, kind of.
People make spelling errors, there are married and divorced names, etc. so really there are a LOT of different versions of names that can resolve to a single social security number.
If you look at the correct table, you’ll see a “master list” of ss numbers where each of them are uniquely associated with whatever names are associated with that number. Every version of a name has to be captured that’s in use. That probably shows a ss number in a row with every variation of the name that’s in use.
BUT you might see another view which combines that table with others, then you might see
ss numb1 Name one
ss numb1 Variation of name one
ss numb1 Unmarried name
etc.
and it will look like the same number is used for different people, but it’s all the same person.
If you just picked one name and “deduped” by deleting the others, you’ve just deleted information belonging to that person, all because it is a comprehensive database that’s catching every version of a person’s name in use.