Imagine this: you are a supposed “math genius” handpicked for a world-class elite program. Your final exam isn’t differential geometry or algebraic topology—it’s a Beijing high school math paper. And you score 110 out of 150. Not only that, but 16 of your classmates score just as badly. The professor who designed the program explodes in fury.
That’s not a plot from a dark comedy. That’s the reality of the Tsinghua Yau class right now.
For years, the Yau Math Competition—named after Fields Medalist Shing-Tung Yau—was supposed to be China’s ultimate filter for genuine mathematical talent. The program promised to find the next generation of Fields winners. But last week’s results told a different story: the top four individual prizes and all six gold medals went to Peking University. Tsinghua, the host institution, got exactly none.
Everyone is framing this as PKU vs Tsinghua. But the real story is that Yau’s own diagnostic—“overfitting”—applies not just to students but to the entire institutional design.
The Yau class selection process looks rigorous: preliminary screening, basic test, discipline-specific tests (two rounds), interview, psychological evaluation, physical fitness test. But that multi-layered bureaucracy creates a false sense of scientific objectivity. What it actually does is filter for something entirely different from mathematical intuition. It filters for the ability to navigate a complex, gamified obstacle course—a game that connected families have mastered.
The result? Seventeen pre-college students who can’t pass a standard high school exam. They were admitted not because they were geniuses, but because they were well-packaged. Their parents wrote petitions when Yau wanted to expel them, claiming the university broke a “promise” that attending classes guaranteed admission. But the fine print always included a clause: fail the assessment and you’re out. It just had never been enforced before.
Yau built a prestige magnet to attract and cultivate talent. But that very prestige made the program a target for credential-seeking families. The more elite the program became, the more it attracted people who wanted the label without the capability.
This isn’t an isolated scandal. It’s a universal pattern. Any system that converts ability into credentials will eventually be captured by people who optimize for credentials over ability. Whether you’re in tech hiring, startup fundraising, or academic admissions, the Yau-Tsinghua case is a case study in how prestige mechanisms become their own worst enemy.
Even well-intentioned founders cannot prevent institutional capture once the signaling value exceeds the substance. Yau admitted in a 2024 interview that Chinese students are trained to do exercises but rarely ask creative questions. He knows the problem. But he built a selection machine that screens out the question-askers and rewards the problem-survivors. The 17 students don’t represent a bug in the system. They represent the system working exactly as designed.
So what happens now? Yau is reportedly threatening to expel those 17 students. But expelling them doesn’t fix the pipeline. The next batch will be just as well-packaged. The real solution is painful: you have to dismantle the signaling value. Stop making the program a credential. Make it a real test of creative thinking—something that can’t be gamed. But that requires Yau to admit that his own creation has been corrupted, and that the corruption started the moment he made the program prestigious enough for parents to care.
The worst part? Yau can’t win. If he expels the students, he admits his selection failed. If he keeps them, he admits the program is just another elite daycare for the connected. The only way out is to burn the whole thing down and start over with a radically different definition of talent.
Until then, every “math genius” title handed out by this system carries an asterisk. And the rest of us—anyone building a system that rewards credentials over substance—should watch closely. Because this isn’t just about math. It’s about every institution that mistakes prestige for proof.
FAQ
Q: Is the Yau program actually producing bad mathematicians, or is this a one-time scandal?
A: It’s not one-time. The repeated pattern—Peking University dominating Yau’s own competition while Tsinghua lags—suggests the selection pipeline is structurally flawed. The 17 failing students are a symptom, not an anomaly.
Q: What practical lesson can I take from this for my own organization?
A: If you build a prestigious credential, expect people to optimize for the credential, not the skill. Use blind tests, random audits, and reward creative problem-solving over checklist completion. And never assume rigorous-looking processes actually measure what you think they do.
Q: Isn't this just a China-specific problem about guanxi and exam culture?
A: No. The same dynamic happens in Ivy League admissions, Silicon Valley hiring (where 'FAANG resume' is gamed), and startup accelerator selection. Anywhere the signal value of being selected exceeds the underlying ability, capture is inevitable. It’s a human problem, not a Chinese one.