What this is (and is not)
- A critique of the architecture, incentives, and assumptions of mass schooling.
- Not a critique of teachers. Many are doing extraordinary work inside constraints they did not choose.
The Classroom Was Built for a World That Is Leaving
The origin story
The modern classroom was designed in the 19th century.
Prussia created the model: age-segregated cohorts, a single teacher transmitting information to rows of passive recipients, standardized content, standardized assessment, a bell that tells you when thinking time is over.
It was explicitly modeled on the factory floor.
The mismatch
For 150 years, that was fine. The economy needed factory-floor humans.
That match is breaking. The economy now needs people who can:
- Think across domains
- Generate original ideas
- Navigate ambiguity
- Learn continuously
- Work in self-directed ways
- Collaborate without a manager in the room
Memorization Was Never Learning. We Just Confused the Proxy for the Thing.
What happened
We needed a way to measure learning at scale. Direct observation of understanding is expensive and slow. Tests are cheap and fast. So we optimized for what we could measure: recall, reproduction, the ability to produce a correct answer under timed pressure.
And then, because that is what got measured, that is what got taught.
What real learning is
Real learning is the ability to transfer:
- Apply understanding across contexts
- Recognize patterns that connect new problems to old ones
- Explain why an answer is correct
- Predict what changes when conditions change
Transfer is what compounds. Transfer is what makes someone genuinely educated rather than just well-tested.
The Degree Is a Credential. Credentials Are Signals. Signals Can Be Replaced.
The degree was never proof of capability
It was a filter — a social contract between employers who could not afford to evaluate everyone and universities that agreed to pre-sort the population. That trust is eroding.
Why it is eroding
- The cost of a degree has outpaced inflation for four decades.
- The debt burden is structurally irrational relative to the income premium.
- The premium is compressing as the signal becomes noisy.
- Skills-based hiring is growing. Portfolio hiring is growing. Apprenticeship models are returning.
What replaces it
Distributed, verifiable, public proof of capability: GitHub. Published work. A course you built that people completed. A company you started. A community you ran. A body of work anyone can inspect.
Learning Has a Pace Problem. Institutions Have a Speed Problem.
The half-life of a skill is shrinking.
In 1990, you could learn something in school, apply it for twenty years, and still be employable. The world moved slowly enough that institutional education could stay close enough to relevant. The gap is now measured in months, not years.
By the time a university identifies a skill shortage, designs a curriculum, hires faculty, runs a pilot cohort, and scales a program, the landscape has shifted.
The most important learning is now happening outside institutions: in communities, on YouTube, in cohort programs that spin up in weeks, in Discord servers where practitioners share what is working before textbooks exist.
Cohort Learning Is Not a UX Improvement. It Is a Different Theory of How Humans Learn.
The self-paced course problem
Self-paced online courses have a dirty secret: completion rates hover between 3% and 15%.
This is not a laziness problem. It is a motivation architecture problem.
The human mechanism
Humans are social learners. We learn in context:
- Watching others
- Being watched
- Feeling social pressure to show up
- Having our ideas challenged in real time
Cohort-based learning reintroduces the social layer — not as a wrapper around content, but as the core mechanism. Content is abundant and cheap. Structured social context around content is scarce and valuable.
Learning In Public Is the Most Undervalued Career Strategy of This Era
A person who learns in public for two years accumulates more opportunities than a person who learns privately for ten. Not because they learned more — because their learning is visible.
What visible learning does:
- Compounding distribution: public learning creates signals that reach collaborators, employers, investors, customers, and mentors.
- Forces clarity: you cannot explain what you do not understand.
- Builds proof of trajectory: how you think, how fast you grow, how you approach problems.
- Creates community: people learning the same things find you.
Teachers Are Not the Problem. The System They Were Put Inside Is.
The single most important variable in educational outcomes is teacher quality. Not curriculum. Not class size. Not technology. Not budget. Teacher quality.
And yet: we pay teachers poorly. We give them little autonomy. We bury them in admin work. We measure them on compliance metrics. We pack classrooms so full that individual attention becomes structurally impossible.
The teachers transforming lives are almost universally doing it by breaking the rules — ignoring the curriculum long enough to connect with a student, staying after hours that are not compensated, caring more about the person in front of them than the test score. They succeed despite the system, not because of it.
Curiosity Is the Only Renewable Resource. We Are Teaching People to Deplete It.
Every child is curious. Curiosity is not a personality trait. It is the default state of human cognition until something suppresses it.
And then school happens.
By the time many people finish twelve years of formal education, curiosity has been replaced by compliance. Not because teachers wanted to destroy curiosity, but because the system rewards the right answer and penalizes the interesting question.
AI Is the Most Significant Pedagogical Event Since the Printing Press. We Are Responding to It Like It Is a Cheating Problem.
The printing press changed who could read, what counted as an authority, how knowledge was organized, and who got to be educated at all. AI is that kind of event for education.
Not because it can write essays — because it changes the economics of personalized instruction. Bloom's 2 Sigma Problem has been known since 1984: one-on-one tutoring produces dramatically better outcomes than classroom instruction. We could not solve it because one-on-one tutoring at scale was economically impossible. It is not impossible anymore.
The wrong question vs the right question
- Wrong question: How do we prevent cheating?
- Right question: If AI can handle information transmission better than lecture, what should teachers be doing instead?
The answer: the things only humans can do — mentorship, challenge, context, community, and the relational infrastructure that transforms information into wisdom.
The Goal Was Never Knowledge. It Was Always Agency.
The point of education was never the accumulation of facts.
It was the development of people who can think, decide, act, and build. Agency: the capacity to shape your life and contribute something original. Everything else — knowledge, skills, credentials — is supposed to be in service of that.
If the answer to "what kind of person are we trying to produce?" is a capable, curious, self-directed human who can contribute something original to the world, almost everything about how we currently run education needs to change.
A final word
Education is not a system in need of optimization. It is a system in need of a different question.
The current system asks: how do we process the most students through the most content, measured by the most scalable assessment, at the lowest cost per student?
The right question is: what kind of person does the world need now, and what conditions actually produce that person?