Comment by bak3y

Social Security receives payments as well as makes them. SSNs are keys for both.

The “super old person” SSN numbers are in the DB mostly because non-citizens are using them to pay into the system. If you delete those numbers, the next payroll run will inject them right back in.

And you would remove important accounting metadata for each payment. Metadata that is consumed by the systems that prevent fraudulent payments from going out.

The only way to stop the fake/bad SSNs is to go into the field and address each instance with employers. This is time-consuming and expensive, which is why no one has done it much.

The reason given that the SSA does not clean up the data is it would cost too much for little to no administrative benefit. They also don't want to add new inaccurate data to the system.

The no administrative benefit bit checks out with napkin math. Of the 18.9 million entries for people age 100 or older they are paying out benefits to 44,000. The total number of people in the US age 100 or older is around 90k to 100k, depending on time period for comparison.

There's an Inspector General audit report in a nearby comment for source.

> Presence in the DB allows for downstream fraud, even by accident.

That's like saying null columns in a particular database table must be filled in (or have the row entirely erased) because someone, somewhere, somehow, might infer the wrong thing about them, if they completely ignore all the other tables and business rules.

___

"Hello, I am Oldy McOldperson. Give me money."

"...Sorry sir, but that person would be almost 150 years old now, and that's well past our Impossibly Old threshold of 115 years. Furthermore, one our other databases says that person was reported as missing 90 years ago."

"But Oldy's--I mean, my precise confirmed date of death is still blank, therefore I'm alive, so give me money!"

"Sir, only a complete moron would believe that's how it works."