What this covers

Nassim Nicholas Taleb and Bryan Caplan conduct a wide-ranging critique of formal education, arguing that schools do not teach useful skills or sustain civilization. Instead, they contend that real competence emerges through convex trial-and-error, apprenticeship with practitioners, and learning from those with skin in the game. The conversation moves between philosophy, economics, and practical examples—from Roman cathedral builders to traders working under pressure—to build a case for radical institutional separation between civilizing liberal arts, pursued for their own sake, and practical vocational training.

A central distinction emerges between knowing what will happen (episteme, which is hard) and knowing how to structure your exposure to outcomes (techne, which is often easier). The speakers argue that fields receiving feedback only from peer review—macroeconomics versus reality, academic humanities spinning elaborate theories—become self-validating and disconnected. They invoke research by Alison Wolf and Lant Pritchett showing education correlates with national wealth only when countries are already rich. Real progress comes through optionality: aggressive small failures cost nothing while improvements compound, outpacing experts constrained by theory. They cite historical cases—engineers building robust cathedrals without geometry, the green lumber fallacy of traders succeeding despite incomplete knowledge—and propose restructuring: starting higher education later (after apprenticeship), exposing students to realistic vocations rather than abstract subjects destined for nonuse, and separating the "holy" (poetry, geometry pursued for elevation) from the "profane" (practical skills for income).

Sharpest takeaway

Taleb and Caplan argue that formal education largely fails to transmit useful skills or sustain civilization; real competence comes from trial-and-error, apprenticeship, and learning from practitioners, and society should separate 'civilizing' liberal arts from practical, vocational skill-building.

  • Convex trial-and-error beats top-down theoretical knowledge because errors are cheap and gains compound like options
  • Education works as wealth-transfer/signaling for individuals but is not causative of national wealth (Alison Wolf, Pritchett)
  • Skills must be learned from practitioners with skin in the game, not from communicators or theorists

The argument · threads6 threads · 25 claims
0.85

Practical tacit knowledge and trial-and-error learning outperform theoretical expertise and high intelligence.

5 pointscentrality 5/5
  • Success in a profession often depends on hidden, practice-specific knowledge rather than the naive theoretical knowledge an outsider assumes is essential — illustrated by the trader who made fortunes in green lumber without knowing it meant freshly cut (not green-painted) lumber, while the man who knew all the chemistry lost his money.

    “He knew a lot of stuff, but not the stuff you’re naively inclined to know. That’s the green lumber fallacy.”

  • It is better to be antifragile than smart: a Monte Carlo simulation shows that a population of people who are not very smart but do a lot of trial and error outperforms very smart people (the chemists) who cannot do much trial and error, because the latter only have intelligence enough to recognize marginal daily improvement.

    “You can model it, and I did the Monte Carlo by showing the population of people who actually are stupid but not very stupid. But they do a lot of trial and error.”

  • Trial and error with optionality beats top-down theoretical knowledge: because errors cost nothing (you discard failures) while improvements are kept, a non-expert who aggressively randomizes and ratchets up will outperform experts working from equations — you'd need 1,000 extra IQ points to match someone doing aggressive trial and error.

    “On one hand, you’ve got to ratchet up by convex trial and error because the errors don’t cost you anything. It’s like an option. You have optionality on what to keep or not to keep.”

  • Historically engineers did not connect to mathematics — Roman engineers did not use Greek geometry, and cathedrals were built without clear geometry using rule of thumb, which was actually more robust; formal geometry was only incorporated into engineering education late, and it produces less robust 'ugly corners.'

    “Engineers built cathedrals without clear geometry. It was actually more robust.”

  • Pressure and necessity activate a heightened problem-solving capacity ('necessity is the mother of invention') — like addicts who become extraordinarily resourceful at procuring money for drugs but incapable elsewhere, traders under existential pressure learn deep probability and invert high-dimensional matrices they couldn't otherwise; unlike the addict, skills learned under genuine necessity persist.

    “Discover how smart addicts are at procuring money for their addiction, but they’re incapable of making money outside of that.”

0.79

Formal education serves mainly as a credential signal for employers, not as a vehicle for genuine knowledge transfer.

4 pointscentrality 5/5
  • Liberal arts education originated as what the idle upper class learned to mark themselves as aristocrats (the trivium and quadrivium had nothing practical about them); the desire to imitate aristocrats by acquiring this education — with the illusion it leads to wealth — is the root of the modern conflation of civilizing study with practical training.

    “The liberal education was what people learned in order to become aristocrat and idle upper class.”

  • An education in liberal arts and humanities cannot be what causes or sustains civilization, because virtually no one actually acquires this knowledge — almost no adult can answer even basic questions about it — so schools cannot be 'saving civilization' when they fail to achieve the very knowledge transfer that would supposedly be required.

    “it can’t possibly be the case that an education in liberal arts and humanities is causing civilization or sustaining it because virtually no one acquires this knowledge.”

  • The main purpose of education is to impress employers by getting credentials ('the stamp on your forehead') rather than to learn material — Taleb studied the exact minimum to pass exams (a 'barbell autodidact'), and business school is mostly a way to meet people and get the first job; the signaling value is exhausted after that first job.

    “you go to business school, but you realize that business school is mostly a way to meet people, or to get a job, the first job after business school.”

  • Judging education solely by job performance is a good metric for would-be employees, but the world needs creators (artisans like Bill Gates and Steve Jobs) and menders; these high-impact people are a disproportionate part of the value created and should not be evaluated by the same average variables.

    “It is a good metric for people wanting to become employees, but the world needs people who create things — the artisans — or people who would mend things.”

0.65

Education should separate practical vocational training from civilizing study, keeping them institutionally distinct rather than conflated.

4 pointscentrality 4/5
  • Because we are living longer, front-loading all education early in life makes no sense; people should start undergraduate study later (around age 30) after working in a practical trade, and should learn by practice before theory rather than starting with theory and ending with practice.

    “We’re living longer, so this idea of front-loading education makes no sense.”

  • Countries like Germany and Switzerland succeeded with low levels of formal education and a huge amount of apprenticeship, showing that what works is transmission of technique through doing, not knowledge formatted to be taught in school.

    “These places had a very low level of formal education and a huge amount of apprenticeship, and a huge amount of built-in.”

  • Education should separate the 'holy' (civilizing study like poetry, history, geometry, done for its own sake like going to church) from the 'profane' (practical skills done to make money); just as translating sacred Latin into vernacular degraded religion by making it 'useful,' making poetry or literature practical destroys it — the two should be institutionally segregated in a barbell.

    “We should keep things in the holy instead of the profane.”

  • Rather than exposing students to subjects almost no one uses (poetry, art history, foreign languages), schools should expose them to a diverse menu of realistic, likely-useful options — a few weeks each of plumbing, electricity, customer service — so the curriculum is a 'tasty menu of realistic options' rather than 'pipe dreams.'

    “why not make sure everyone spends a few weeks learning some plumbing, a few weeks learning some electricity, a few weeks going into learning some customer service?”

0.59

Closed-circuit peer-review systems in academia become self-validating and disconnected from reality-based feedback.

3 pointscentrality 4/5
  • Fields where feedback comes from peer review rather than from reality become closed, self-validating systems — a carpenter gets feedback from clients, but a macroeconomist only gets feedback from another macroeconomist like Paul Krugman, allowing such fields to give themselves Nobels while disconnected from reality.

    “the carpenter has feedback from his clients, or her clients, whereas a macroeconomist only has feedback from Paul Krugman, so it’s closer.”

  • In the humanities, the competitive closed-circuit system means you get tenure not by knowing facts but by inventing elaborate theories (e.g., a structural theory of baking beans in Sassanid Persia, or postcolonial/Marxist/gender interpretations), so a generation of scholars is ignorant of basic facts in their own fields.

    “you don’t get tenure from knowing the facts. You get tenure from inventing some full structural theory of baking beans and mint in Sassanid Persia.”

  • You should learn a profession from those who actually do it (a Mafia lord from a Mafia lord, not from Francis Ford Coppola), and you cannot trust two classes of people: educators who are selectively better at explaining than understanding, and science journalists better at communicating than understanding — which produces scientism, the belief that science is settled, since real practicing scientists never rule out skepticism.

    “If you want to learn how to become a Mafia lord, you learn from a Mafia lord. You don’t go learn from Francis Ford Coppola.”

0.57

Education does not cause national wealth and austerity with accountability is more effective than reflexive increased spending.

5 pointscentrality 3/5
  • The objection that vocations can be wiped out (e.g., typewriter repair) doesn't justify teaching known-useless subjects; it is better to train people for things likely to remain useful (plumbing) than for things we know will not be useful (poetry), since whatever the economy looks like in 40 years, poetry will not be a major part of it.

    “it’s better to go and train people for things that seem likely to be useful, like plumbing, than to train them for things we know are not going to be useful, like poetry.”

  • For a developing country like India, it is better to open a thousand schools than one university, with a focus on technical skills (electricity, batteries) rather than producing theoretical graduates like sociologists who add little practical value.

    “Better to open a thousand schools than one university.”

  • In risk and probability, one must distinguish x (knowledge of what will happen, the episteme — e.g., what the stock market will do) from f(x) (how that event affects you, the techne — your payoff/exposure); it is often easier to modify your exposure f(x) (e.g., via an insurance contract with a close-ended clause, a lawyer's job) than to gain better knowledge of x (a statistician's job), and conflating the two is the root error.

    “one of them is knowledge. The other one is how it affects you. It can change you.”

  • Education is antifragile at the individual level — it is a great way to transmit wealth across generations and keep children in the middle class — but at the level of a country it does not produce wealth and may even work in reverse; education is a product of societies that are already rich, not a cause of their wealth (per Alison Wolf and Pritchett's data).

    “it’s a great way to transmit wealth to a generation, because your children are certain to stay in middle class if you educate them... but at the level of a country, it doesn’t seem to work. In fact, it’s the reverse kind of thing. Alison Wolf’s data.”

  • The most effective educational reform is to spend less on education ('educational austerity') — including cutting funding to dysfunctional school systems first and only returning money if they start performing — rather than reflexively spending more, which gets approval without anyone asking what the money will accomplish.

    “when someone says, “We need more money for education,” the reaction is normally, “Yay,” not “What specifically do you plan to spend the money on?””

0.52

Practical ability and common sense are distinct from IQ and best assessed by real-world performance challenges, not standardized tests.

4 pointscentrality 3/5
  • The best test of ability for hiring is not standardized scores (which all candidates already have) but a practical real-world challenge — e.g., 'get me a specific French suit in New York City within an hour' — which reveals who gets confused by the request and who can actually execute under constraint.

    ““Listen, I want a French suit, and you have an hour to get it...” That’s the kind of test we give people, to see if they can go find that suit.”

  • The standard way mathematics is taught actively deters people from liking it; Taleb claims he was lucky to have a poor math teacher as a child so he zoned out, only later becoming self-taught and genuinely loving math, and is self-taught in almost everything he truly enjoys.

    “The teaching, the way that they teach you mathematics, is really to deter you from liking mathematics.”

  • Common sense is a distinct skill from IQ that keeps improving into people's 50s while IQ sometimes declines — demonstrated by survival tests like a car breaking down on a snowy mountain road, where smart 18-year-olds give answers that get them killed (e.g., waving down cars on a winding road) while normal 50-year-olds give answers that ensure survival.

    “common sense is something where people keep improving into their 50s, unlike IQ, where actually, at least sometimes, people are getting worse.”

  • Teachers at parent-teacher nights are 'touchy-feely' people who resist thinking about practical results or whether something was actually learned, focusing instead on whether the student enjoyed the experience — but if you truly know something, a test can be devised to measure it, so the claim that learning can't be measured is an evasion.

    “These are touchy-feely people who do not want to think about anything in terms of practical results.”