There's a genre of education documentary that has exactly one move. It flies some kids from a comfortable Western country into an East Asian classroom, films them looking shell-shocked at the hours, and then the narration tells you these schools are the best in the world.
While the footage shows eleven-year-olds asleep on their desks.
Hannah sent us one of these this week. It's a BBC thing on YouTube, School Swap: Korea Style. Welsh teenagers swap into Seoul high schools for a few days. And the documentary keeps calling the Korean system the best in the world. Her question is the one the documentary never actually asks. Best at what? Tutoring until midnight, up at six for school.
She's not wrong about the schedule.
And she's coming at it from the inside, which is what makes it interesting. She says she always tested well, sometimes without understanding the material at any real depth. Which, honestly, most people who get good grades could admit about themselves if they were being honest.
That's a real distinction and I want to get to it.
So the questions she wants us to actually take seriously: what should we be measuring instead? Adult mental health, employment, income, creativity, civic engagement, life satisfaction, the ability to think independently, long-term educational outcomes. She brings up Finland as the opposite pole — later formal learning, less homework, less testing — and asks how that actually compares to Korea once you stop staring at exam scores. And the big one underneath all of it: do we have good evidence about which systems produce not just high-performing students, but successful, healthy adults? That last one has a shorter answer than the documentary would like.
It's a corridor. The question runs the whole corridor and comes out the far end with no answer.
So let's start with the ranking itself. What does it even mean to call a school system the best?
It means one of two things, depending on who's talking. Either it means "scores highest on PISA," or it means "does what we've decided education is for," and the second one never gets said out loud because it's a value judgment and the first one sounds like data.
PISA being the triennial OECD test of fifteen-year-olds.
Reading, math, science. Every three years, dozens of countries, one common instrument. And it has completely colonized the vocabulary of education policy. If you've read a headline about a country's schools being great or collapsing, you were almost certainly reading about PISA, whether they named it or not.
And Korea does well on it.
Korea does very well on it, consistently. Top of the tables in reading and math, year after year. Finland was the absolute darling from 2000 to about 2009 — the first couple of PISA rounds, it was at or near the top in everything, and the entire world descended on Helsinki to figure out how they did it. Then Finland slipped somewhat. Not collapsed. Slipped. And Korea stayed up there.
So both got called the best. With completely different systems underneath.
Completely different. To the point where if you only knew the scores, you would not guess the schools looked anything alike.
The center of it is the CSAT — the College Scholastic Ability Test. One exam, one day, largely determines university admission. And university admission in Korea largely determines career track and social standing. It's not quite the only input, but it functions as the pivot of a person's life, decided in a single sitting.
A single day.
And everything upstream organizes itself around that day. Which gives you the hagwons — the private tutoring academies. These are not an after-school hobby. They're a parallel school system that runs into the night. Kids go from regular school to hagwon and often don't get home until eleven or midnight, and then they do it again. There's a documented pattern of parents and police actively managing the curfew rules around these places because the demand to study past the legal hour is so strong.
Parents coordinating with police to let their kids keep studying. There's a whole society in that sentence.
There is. And the hagwons track the exam. If the exam changes, the hagwons repivot within a semester. Which is why the reform fight in Korea is so specific and so interesting.
The killer questions.
So Korean authorities have been moving to strip what are called killer questions out of the CSAT. These are items so brutally difficult that a well-prepared student who's simply paying attention in class can't solve them. You need the trick, and the trick is taught at the hagwon. Which means those questions were, functionally, an admissions advantage you bought.
So a kid who couldn't afford the academy wasn't failing the exam. They were failing the parts of the exam that were designed to be un-passable without the academy.
That's the argument, and it's a strong one. The reform is explicitly aimed at decoupling test performance from tutoring access. If the hardest questions require private instruction to solve, then the test is measuring your family's ability to buy instruction. Pull the questions and the test drifts back toward what school is supposed to teach.
Does it work? Have they pulled them?
They've been pulling them and it's contested, both directions. Some people say it makes the exam too easy to distinguish the top students. Others say that's the point — the top of the distribution was never measuring ability, it was measuring access. I'm not sure which side wins long-term. But the reform is a real-time experiment in exactly the question Hannah's asking: can you separate test performance from the machinery built to game it?
Okay. Now the thing she said about herself, because I think it's the load-bearing part and I want to take it seriously.
Tested well without deep understanding.
Yes. And the obvious objection is "well, you didn't really know the material," which is true, but it's not the interesting part. The interesting part is that the test didn't catch it. She cleared the bar and the bar didn't notice there was nothing behind her.
The test measured what the test measures. That's not a flaw exactly, it's a definition.
It's a definition that everyone forgets is a definition.
Yeah. And there's a body of work on this — on shadow education specifically, the tutoring and cramming industry. The broad finding across the meta-analyses is that it produces modest test-score gains, and the effects vary enormously depending on who's tutoring, who's tutored, and what subject. And crucially, the gains tend to show up on the specific test you were prepped for and don't reliably transfer to broader understanding or to long-term outcomes.
Modest gains, narrow.
Modest gains, narrow, expensive. And I'd flag the obvious methodological problem — the kids who get tutoring are often already advantaged, so you can't cleanly separate the tutoring from everything else that was going right for them.
So the tutoring industry is selling a score, and the score is real, and the thing behind the score is not necessarily what the score is standing in for.
Which is exactly what Hannah described from the inside.
And then there's Finland. Which is the mirror image. Tell me what it actually looks like mechanically, because "less homework" undersells how strange it is.
Finland doesn't start formal schooling until age seven. Seven. There's a year of what they call pre-primary education before that, play-based, but the structured stuff starts late by most countries' standards. There's essentially no standardized testing through the system until the end of upper secondary school — one exam at the end, that's it. Homework is minimal. Teachers have enormous autonomy; they're highly trained, masters-level, and trusted to design their own assessment.
No standardized testing. So you cannot rank Finnish fifteen-year-olds against Korean fifteen-year-olds from inside the Finnish system.
You can't, which is precisely why PISA is the only instrument that can make the comparison at all. And on that instrument, Finland's early scores were outstanding. Without the pressure machine.
So pressure isn't required for performance.
Pressure isn't required for top performance. That's the honest read of the early Finnish data. It is not, and I want to be careful here, evidence that pressure is never useful. It's evidence that it isn't necessary in some conditions we don't fully understand.
What conditions?
Finland is small, comparatively homogeneous, wealthy, low-inequality, with a strong social safety net. Any one of those could be doing work that we're attributing to the school system. That's the trap with national comparisons. You're not comparing pedagogies, you're comparing societies that happen to have pedagogies.
So the pressure machine might be doing nothing. Or it might be doing something that Finland achieves some other way.
Right. Korea's hagwon culture might be a substitute for the private tutoring that wealthy families arrange privately in other countries. Same function, just industrialized and visible.
But scores are only one outcome. What happens to these students as adults?
This is where it gets hard to know anything, and I want to be honest about that up front.
Be honest about it.
Korea has very high university attendance. Strong outcomes in some sectors. And it also has a youth suicide rate that is among the highest in the developed world, a workplace culture that is famously punishing, and a birth rate that has collapsed to the lowest on the planet.
Lowest on the planet.
Lowest recorded, essentially. And I have to say immediately — you cannot pin any of that on the school system. That's the error. Those are adult outcomes in an economy with brutal housing costs, brutal work hours, and a deep set of social expectations. The school system is one strand in it and it may not even be the biggest.
But it's not unconnected either.
It's not unconnected. You're training people for twelve years to succeed in a single high-stakes competitive event and then releasing them into a labor market where the same competitive logic continues. That's a culture, not a curriculum. And the curriculum is downstream of the culture.
Do the Koreans themselves think they have a problem?
The reforms say yes. The killer-question removal, the curfew fights, there's a live domestic argument about exactly this. It's not a country that thinks it solved education. It's a country that thinks it built the most effective exam machine on earth and is now asking whether that was the machine it wanted.
That's a very different thing from "best in the world."
Wildly different.
And Finland. Presumably the utopia story has a crack in it too.
It does, and it's recent enough that it's changed the conversation. Finland has high life satisfaction, high social trust, low inequality, good mental health outcomes by lots of measures. But in the last several years, Finnish youth mental health has been deteriorating, and its PISA scores have continued to drift down. So the country that was held up as the proof that you can have it all is now getting questions about what's slipping, and the answers aren't obvious.
So neither pole is a clean result.
Neither pole is a clean result, and I think that's the most honest thing we can say. Finland proves pressure isn't necessary for scores, and Finland is also now proof that the low-pressure model isn't automatically self-sustaining.
Here's what I keep circling. Every outcome on Hannah's list — mental health, creativity, civic engagement, life satisfaction — is real and matters. And none of them is measured by the thing we use to rank systems. So the ranking is answering a question that isn't the question anyone actually has.
That's the measurement problem. We measure what's easy to measure and then we call it the thing we care about.
Say it more carefully than that. Because "we measure what's easy" is a truism that lets everyone off the hook.
Fine. Here's the sharper version. There's a framework from the development economics world — Amartya Sen and Mahbub ul Haq — called the capability approach. The idea is that you shouldn't measure a society by its inputs, like years of schooling or test scores, and you shouldn't measure it purely by outputs, like GDP. You should measure what people are actually able to do and be. Capabilities. The real freedoms a person has.
And this is where the Human Development Index comes from.
The HDI is the compromise. Haq wanted a single number, because a single number travels and gets used. Sen wanted the rich theory, because the single number flattens what matters. The HDI is what happened when they met in the middle.
And that's exactly the same fight we're having about schools. Rich outcomes versus a simple metric.
Identical fight. And the HDI is instructive because it's honest about what it did — it took three things, health, education, income, and admitted that it was a proxy. Nobody thinks the HDI is a complete picture of human flourishing. They think it's a useful distortion.
Is PISA honest about being a useful distortion?
PISA is honest about measuring reading, math, and science. The distortion comes when policymakers and documentaries use it as a verdict on the whole system. The instrument is careful. The reception isn't.
So what's the actual evidence base on adult outcomes? Hannah asked that directly, and I don't want to glide past it.
Here's the honest map. Test scores — good data, decades of it. Employment and income — decent data, but heavily confounded, because you can't separate the school from the labor market it feeds. Creativity — weak, contested definitions, no standard instrument. Civic engagement — weak. Life satisfaction — self-reported, cross-culturally slippery, and we've spent time on this show before on how hard it is to even define happiness, let alone compare it across countries.
And long-term longitudinal studies tracking students from different systems into midlife?
Almost nonexistent in a form that would let you do the controlled comparison Hannah's asking for. You'd need to track cohorts through different systems with comparable measures of adult flourishing, and nobody has the money or the patience, because it takes forty years to get the answer.
The study you'd need cannot be completed by the institution that would fund it.
Correct. Nobody funds a forty-year study that outlives the people who commissioned it.
So when someone says a system is the best in the world, they're making a claim they cannot actually support.
They're making a claim about test scores and dressing it up as a claim about adults. Sometimes knowingly.
And there's a knock-on effect here, which I think is the real damage. If you can't measure what you value, you end up valuing what you measure. That's not a clever line, it's a description of what happens to institutions under pressure.
It's Goodhart's law with a longer fuse. The moment a measure becomes the target, it stops being a good measure. And an education system optimizes hard, because the stakes are high and everyone knows what's being scored.
So you get a system that produces the score and loses the thing the score was a proxy for.
And the killer questions are the perfect illustration. Nothing about the CSAT says "make items that require private tutoring to solve." But once the exam was the target, the ecosystem optimized around it, and the hardest items became the ones the academy sold the answer to. Nobody designed that. It emerged.
Is there a version where we just accept the messiness? Rank nothing, measure everything, describe each system in its own terms?
Tempting, and unworkable, because parents need to decide where to send their kids and governments need to allocate money and both want a number. The number isn't going away. The question is whether we're honest that it's a narrow instrument.
Where does that leave us? Two systems, both called the best, producing very different humans.
The honest answer is we don't know which one produces better adults, because we never built the instrument that would tell us. We built the exam instrument instead.
The documentary calls Korea the best in the world and shows you children barely sleeping.
It never once asks whether sleeping is part of the outcome.
Hilbert: It's the word. "Best."
Sorry?
Hilbert: You keep saying "the best." Nothing in that building was measured against "the best." It was measured against a test. Those aren't the same word.
Hilbert, go on.
Hilbert: There was a prep academy I worked at for about a year and a half. Not a school. The kind of place that takes a kid who's already doing well and sells the last five percent. Wall of clocks in the front office.
A wall of clocks. How many?
Hilbert: Fourteen. Different time zones. Since the students were sitting exams in different countries — Singapore, London, Toronto, a couple for Seoul. The clocks were there so a kid would know what time it was where their exam was.
That's an extraordinary thing to have on a wall.
Hilbert: Every one of them was wrong. The time. Running two, three minutes off. Nobody fixed them. The students figured out which clock was fast and which was slow and adjusted in their heads. They'd been doing it for years, most of them.
The whole point of the wall was to be precise about time, and it wasn't.
Hilbert: It was fine. That's what I keep thinking. Everything in that building was pointed at the exam and the clocks were two minutes off and nobody minded, because the exam doesn't come down to two minutes. But the kids noticed. They always notice. They just don't say anything.
That's a study I'd actually want to read.
Hilbert: What you were saying about the tests not measuring the right thing. I'd agree. But those kids learned something. They learned how to walk into a room and read it in ten seconds. How to know exactly how much time they had and what to spend it on. How to perform when they were exhausted. Those are real skills. You can't buy them at a store.
They just aren't the skills the school was claiming to teach.
Hilbert: Right. The building said it was teaching mathematics. It was teaching resource management and nerve, mostly, and it charged a lot for it. The math was fine. It wasn't the point.
Nobody in that building could have told you what the point actually was.
Hilbert: No. The clocks couldn't even tell you the time.
The students became capable of something nobody designed. Just not the thing on the sign.
Hilbert: Every one of them. That's what I've got. It's not a tidy answer.
It isn't.
Hilbert: That was a lot of good kids in that building. I think about them.
The thing Hilbert just handed us is that "outcomes" is plural and it's messy. Those students developed real capabilities — reading a room, managing time, performing under pressure — and none of them would appear on any ranking of that school. And the school was also failing at what it claimed to do.
Both true at once. It's the least satisfying answer and it's the real one.
If we can't measure what we value, how do we decide what a good school system is? Is the problem the rankings, or is the problem that we want a single answer at all?
I think it's the second one. The rankings are just a symptom of the appetite.
The natural experiment keeps running. Korea's pulling the killer questions, Finland's dealing with its own slippage. That's the thing to watch — not whether one model wins, but whether either can reform without losing what it's good at.
Korea trying to decouple test scores from tutoring access is the most interesting reform anywhere right now. Watch that one closely.
Maybe the question was never which system is best. Best for what, and for whom. And no ranking is built to answer that.
Thanks to Hilbert Flumingtop, our producer.
This has been My Weird Prompts. If you've got a minute, a review helps people find the show.
See you tomorrow.
