You’re building a fraud detection system. You need to figure out if two people are related. You look at their 18-digit Chinese ID numbers and notice the 15th and 16th digits match. “Aha,” you think, “those are the local police station codes. They’re from the same family!”
It sounds brilliant. It’s also completely, catastrophically wrong.
The most dangerous lies aren’t the ones told by malicious hackers; they’re the ones printed by official media and coded into our AI.
Recently, a risk control product manager clashed with a data analyst over this exact scenario. The analyst wanted to use the 15th and 16th digits as a hardcoded link between individuals. The analyst’s source? “Common knowledge.” Even major news outlets have proudly declared that the 15th and 16th digits of a Chinese ID represent the local police station code.
It’s a myth. And if you build a system on it, your logic will collapse.
Let’s use first principles and look at the actual national standard (GB11643-1999). The rules are clear:
Digits 1-6: Address (Province, City, County)
Digits 7-14: Birth date
Digits 15-17: Sequential code
Digit 18: Checksum
Notice what’s missing? The word “police station.” The 15th, 16th, and 17th digits form a single, three-digit sequential number.
But let’s give the media the benefit of the doubt. Let’s do the math.
Assume the 15th and 16th digits are fixed police station codes. That leaves exactly ONE digit—the 17th—for the individual sequence. But the 17th digit also has a job: it denotes gender (odd for male, even for female).
If the 17th digit is the only variable, a police station has exactly five odd numbers (1, 3, 5, 7, 9) for males and five even numbers (0, 2, 4, 6, 8) for females.
If a system breaks the moment reality touches it, the system was built on a lie.
If the myth were true, a police station could only register five baby boys and five baby girls per day. On the eleventh baby boy, the universe would implode. The system would crash. In any major city, this mathematical ceiling is shattered every single morning.
The myth persists because of a cognitive glitch. Police stations do hand out the numbers, so people assume the numbers represent the station. But “who distributed the number” and “what the number means” are two entirely different things. Police stations manage number segments; they don’t embed their own codes into your lifelong ID. If they did, every time a station was renamed or merged, your ID would become invalid.
This isn’t just a quirky trivia fact. It’s a warning shot.
We are entering an era where AI can spit out a 10-page report in seconds. But AI is trained on the internet, and the internet is a landfill of confidently stated falsehoods like the “police station code.” If you ask a generative AI how to link family members using ID numbers, it might hand you the exact broken logic the analyst tried to use.
In the age of AI, the ability to get an answer is worthless. The ability to recognize a bullshit answer is the only thing keeping you employed.
Stop trusting second-hand interpretations. Stop accepting “authoritative” sources without running them through the meat grinder of basic logic. The future belongs to the clear-headed thinkers who can look at an AI’s output, look at a media headline, and say, “The math doesn’t work.”
Verify everything.
FAQ
Q: But what if the national standard changed to include police station codes?
A: It didn't. And mathematically, it can't. The standard (GB11643-1999) explicitly defines digits 15-17 as a sequential code. The mathematical constraints of a 2-digit station code + 1-digit gender sequence make it physically impossible to support daily birth rates in any major city.
Q: How does this actually affect my system design?
A: If you're building validation rules or fraud detection logic, do not use the 15th and 16th digits as geographic or relational anchors. Treat the 15th-17th digits as a single, opaque sequential block, or you will generate false positives and miss real connections.
Q: Isn't AI smarter than relying on old media articles?
A: No. AI models are trained on the same polluted data corpus that contains these media errors. If the loudest sources on the internet say something false, the AI will confidently regurgitate that falsehood. AI amplifies existing bias; it doesn't automatically fix bad logic.