In his People's Daily essay on basic research, Shi Yigong names supporting "non-consensus" topics โ€” ้žๅ…ฑ่ฏ†่ฏพ้ข˜, research directions the mainstream hasn't yet validated โ€” as a deliberate value at Westlake University, alongside ้•ฟๆœŸไธปไน‰ (long-termism) and resisting the pull to chase whatever research area is currently trending. This is a specific, structural choice, not just a nice sentiment: it shows up in how the university evaluates its own scientists.

What "long-cycle evaluation" actually replaces

Most research institutions, and most schools, evaluate people on relatively short cycles against relatively legible metrics โ€” this term's grades, this year's output, this quarter's targets. Westlake's assistant professors, by contrast, are evaluated over long periods, with a single standard: has this person made, or are they on track to make, a breakthrough that is genuinely irreplaceable โ€” described in the reporting on Shi Yigong's essay as one that, worldwide, could only have come from them.

That's a fundamentally different bet than "did you hit this year's numbers." A short-cycle, metrics-driven system rewards ideas that can prove themselves quickly, which structurally disadvantages exactly the kind of idea that takes years to either pan out or fail โ€” and non-consensus ideas, almost by definition, take longer to prove than ideas everyone already agrees with.

Why "non-consensus" doesn't mean "probably wrong"

It's worth being precise here, because this argument is easy to caricature as "back any weird idea and you'll eventually strike gold." That's not the claim. A non-consensus idea is one the field hasn't yet validated โ€” which includes genuinely wrong ideas, but also includes ideas that are simply ahead of what the field currently has the tools or framework to recognise as correct. The entire history of science includes ideas that were dismissed for years before being vindicated, and separately, plenty of ideas that were dismissed and stayed wrong. The uncomfortable truth Shi Yigong's argument leans into is that you can't tell the two apart quickly โ€” which is precisely why a system that only funds fast-provable ideas will systematically miss some of the real ones.

What this means for how a family responds to a child being wrong

This has a direct, practical translation to how a household handles a child's unconventional idea, guess, or approach to a problem. The instinctive move โ€” understandably, given time pressure and exam schedules โ€” is to correct a wrong answer quickly and move on to the next question. That's often the right call for straightforward factual errors. But it's worth distinguishing between a simple mistake and a genuinely unconventional approach that happens to be wrong this time: an unusual method for a maths problem, an atypical interpretation of a text, a hypothesis in science that doesn't immediately check out.

Shi Yigong's long-termism argument suggests that if every one of those gets corrected and closed down at the same speed as a simple factual slip, a child never gets the extended experience of sitting with an unconventional idea long enough to actually find out whether it had something to it โ€” which is the exact experience non-consensus research requires, just at a much smaller scale.

Where this connects to how a tutor should actually respond to wrong answers

This is part of the argument behind a hint ladder that probes and questions rather than immediately supplying the correct method โ€” see why tutors don't give answers. A student's unconventional (and currently wrong) approach, met with "here's the correct way to do it," gets closed down before the student has had any chance to test it further. Met with a question โ€” "what makes you think that would work here?" โ€” the student gets a bit more room to either discover the flaw themselves or, occasionally, to find that the unconventional approach actually holds up in a way a faster correction would have missed.

The trade-off Westlake itself accepts

It's worth being honest that this approach has a real cost, which Shi Yigong's own institution accepts deliberately: long-cycle evaluation and tolerance for non-consensus ideas means slower, less predictable measurement of progress, and inevitably some ideas that were given years of support and simply didn't pan out. The bet is that this cost is worth paying because it's the only structure that doesn't systematically exclude the ideas that take longest to prove โ€” which, per his argument, is disproportionately where the real breakthroughs live.

FAQ

What is a 'non-consensus' research topic?

A research direction the mainstream scientific community hasn't yet validated โ€” which can include genuinely wrong ideas, but also ideas ahead of what the field currently recognises. Shi Yigong argues supporting some of this kind of work is necessary for real breakthroughs.

How does Westlake University evaluate its scientists differently?

It uses long-period evaluation for assistant professors rather than short-cycle output metrics, with the single standard being whether they've made or will make a breakthrough described as genuinely irreplaceable โ€” one that, worldwide, could only have come from them.

What does this have to do with a KS3 child's schoolwork?

It's a useful frame for how a family responds when a child's idea or approach turns out to be wrong. If every wrong attempt is corrected instantly and moved past, a child never gets to hold an unconventional idea long enough to find out whether it was onto something.

Source

Adapted from ๆ–ฝไธ€ๅ…ฌ (Shi Yigong), "ใ€Š'ไปŽ0ๅˆฐ1'็š„็ช็ ด้ ไป€ไนˆใ€‹" ("What Does a '0 to 1' Breakthrough Depend On?"), People's Daily, 16 May 2026.

Duke Harewood ยท aitutors.me ยท Updated 14 Jul 2026.