Re:The New ID Theft: Millions of Credit Applicants Who Don’t...
by jstarfish on Thursday March 08, 2018 @03:27PM
(#16547359)
You're overstating the role of the CRA. They don't provide facts, they provide intelligence and let the recipient draw their own conclusions.
FWIW there is a boolean "deceased" flag on credit reports (internally). I don't know by what means or how often it gets updated.
They get a lot of denormalized data. Just because an SSN:name pair presented to them isn't legit doesn't mean there isn't a real person associated with it.
There's a limit to the accuracy of credit reporting given the nature of their inputs. 123-45-6789 for Janice Smith, 123-45-6789 for Jane Smith and 123-46-6789 for Janice Deer are all the same person, but there's value in maintaining even the inconsistencies-- when an inquiry comes in at 123-46-6789 for Tran Nguyen, it's obvious something is up with one of the two parties, but without further investigation we don't know if it's a typo or fraud. Then the manual review process starts.
Entity resolution is a very complex problem to do correctly at scale. Phone numbers and addresses are shared across entities over time and can be typoed. Social security numbers can be typed incorrectly, or the name (and nowadays, gender!) associated with them changes freely. For all we know Tran and Jane are the same person; there are circumstances that have led to such oddness. Foreign passports or IDs may not match up to anything seen before, and they can't exactly hit a Romanian government API to validate numbers presented, yet they can't deny the existence of an applicant based on metrics they don't know about-- they can only share what they know.
FWIW there is a boolean "deceased" flag on credit reports (internally). I don't know by what means or how often it gets updated.
They get a lot of denormalized data. Just because an SSN:name pair presented to them isn't legit doesn't mean there isn't a real person associated with it.
There's a limit to the accuracy of credit reporting given the nature of their inputs. 123-45-6789 for Janice Smith, 123-45-6789 for Jane Smith and 123-46-6789 for Janice Deer are all the same person, but there's value in maintaining even the inconsistencies-- when an inquiry comes in at 123-46-6789 for Tran Nguyen, it's obvious something is up with one of the two parties, but without further investigation we don't know if it's a typo or fraud. Then the manual review process starts.
Entity resolution is a very complex problem to do correctly at scale. Phone numbers and addresses are shared across entities over time and can be typoed. Social security numbers can be typed incorrectly, or the name (and nowadays, gender!) associated with them changes freely. For all we know Tran and Jane are the same person; there are circumstances that have led to such oddness. Foreign passports or IDs may not match up to anything seen before, and they can't exactly hit a Romanian government API to validate numbers presented, yet they can't deny the existence of an applicant based on metrics they don't know about-- they can only share what they know.