The shortcut that shrinks Aadhaar's search 72,000 times, and why it fails

transcript

This trick makes Aadhaar's duplicate search 72,000 times smaller. And it's useless against the one person you're hunting.

The textbook fix is called blocking. Only compare similar people. Same gender. Same age band. Same state. A sound-alike name. About 19,000 candidates, instead of 1 point 4 billion.

But here's the catch. Who chooses those demographics? The person enrolling.

A fraudster wants a second identity. He enrols again. He just gives a new name. A new birth date. A new state. Now he lands in a different block. His first record is never even compared. Blocking only catches careless duplicates.

So Aadhaar searches the full gallery. Indexed, massively parallel. It fuses fingerprints with iris. Independent errors multiply down. Three vendors, three algorithms. Two must agree, or a human decides. At the 2014 peak: over 600 trillion matches a day.

Expensive? Hugely. But it's the only search a fraudster can't dodge. Because the fraudster picks the demographics. Never the fingerprint.

Full breakdown in the lecture.