Compromising a $6B big data project through poor data quality: the Aadhaar case study
summary
Claim: country‑wide offline enrollment creates a window for mass fraud. Enrollment clients are Java SE apps whose patched jars remove biometric and GPS checks so offline zip files can be forged and uploaded. Field evidence shows an operator account making 30,000 fraudulent enrollments, use of crack software sold for ~₹5,000, and auditors/whistleblowers flagging 47.82 crores missing documents. The system’s eventual consistency and distributed CSC/60,000 agents enable scale and concealment.
AI-written summary · about
chapters (22)
- 0:00Intro
- 0:23About myself
- 1:08All Weaknesses are exploited
- 2:45You control what you write
- 6:28Human Checks
- 7:53Software Checks
- 9:36Back End Checks
- 10:21The default fingerprint
- 12:30Eventual Consistency
- 13:08The missing documents
- 14:31The Whistleblower
- 15:33Enrollment Client
- 18:15Youtube Support Channel
- 21:51The Assembly Factory
- 23:00The Human and the Database
- 23:10There is just the Database
- 23:34Unviable Business Model
- 25:49Crime and Fortune
- 26:56Scale Comprehension is HARD
- 27:52Data Acquisition Lessons
- 29:10Control is Intractable
- 29:47The $6B Lesson
related talks
44mKeep it Clean: Why Bad Data Ruins Projects and How to Fix it
18mData and the City
49mKeep it Clean: Why Bad Data Ruins Projects and How to Fix it
54mLifecycle of a Phone Fraudster: Exposing Fraud Activity
39mImproving delivery of safe water to African communities with
38mIs It A Bug Or Is It A Story?
25mKeynote: Extending from Open to Usable: A Commerce Data Conundrum
33mGovernment Big Data for All