What’s Gone Wrong in Schooling?

Written by Mason Hartman

Today, as at many points in the past, humanity faces significant challenges beyond its current capacity to resolve. As we always must, we entrust our future to the generations that follow us. We try to prepare them for the work. We offer what we know, and we hope to provide the resources and encouragement necessary for the discovery of what we don’t. We speak urgently about fostering a lifelong love of learning and “twenty-first century skills.”

Laudable goals, certainly! But we’ve failed to see them through, and — quite concerningly — very few of us have even noticed the failure.

It’s time we take a sober account of the situation we’ve created, and begin to look for the signposts that might guide us back.

How to train a human parrot

A parrot trained to answer a prompt —“What’s your name?” “Polly!” — has not learned any facts about the world. She does not know what a name is or that she has one. She’s only learned that she can earn a reward by producing one series of noises in response to another.

Human beings can do this, too: if you “teach” a child that engines work “because of combustion,” she may later repeat that:

“What makes an engine work?”


But if you were to ask her what combustion is — whether that substance is combustible, how she might tell whether something is combusting or might combust — she won’t be able to answer these questions. She does not understand how an engine works by virtue of learning to say the word “combustion” when an adult asks.

Yet this is how much of our toolkit for evaluating learning in children works — standardized tests most obviously, but also much of the rest.

For the test, the child learns to force many of these associations into a cue-recall format in his mind: “engine: combustion,” and so on. For the essay question, the child learns to store a little meta-data regarding which of these associations occurred in roughly the same cluster, and to semi-faithfully reconstruct source texts or lecture notes from memory. For the term paper, the child learns to find extant work on the relevant topic, smash excerpts together, and then fiddle with sentence structure and word choice until the swagger of the resulting Frankenstein’s monster doesn’t too closely resemble that of any one of its parts.

A child subjected to these protocols will become a complex parrot.

Worse, because we can’t admit to ourselves that we’re in the business of training human parrots, we don’t offer children the tools that might at least allow them to succeed as parrots. We evaluate and reward rote memorization, but we don’t teach the proven mnemonic strategies; we reward shallowly-disguised plagiarism, but we don’t provide guidance for rephrasing others’ work.

The result is that some children learn to “play the game” (with varying degrees of self-awareness), and the rest are made to feel stupid — and to be deemed stupid by their peers. This is a self-perception that can persist for a lifetime.

What is it to know something?

A person who understands how an engine works will be able to make some predictions about why one might break and how it could be repaired, what properties could be used to identify a fuel, and the circumstances under which operating one might become dangerous. A person who only knows that a thing called combustion makes an engine work will not be able to do any of these things.

How would a person come to actually understand combustion? Roughly the same way that humanity did: through a curiosity-driven, outcome-directed process of investigation and experimentation. The advantage that language gives us isn’t the ability to transfer facts into an empty-bucket mind, it’s the ability to offer simulated observations and experimental results. A good teacher will not only illustrate the process with a degree of clarity and detail that invites good questions, but also stand up to the resulting interrogation. She’ll also recognize that this simulated combustion is less information-dense than real combustion, and where possible, she’ll arrange for her student to witness the real thing.

Most importantly, she’ll recognize that learning is not and can never be something done to a person, that the real substance of teaching is done by the learner himself. The teacher can only provide the game: the environment, the system, some tools. The student must pick up a tool or ask the questions that will allow him to experiment with or simulate the subject matter. Whether he asks out loud or not is irrelevant; all learning is generative, and the learner is the generator. To build out his model, he must be able to notice where he’s confused — where a placeholder like “combustion” is obscuring unexplained magic.

All learning is ultimately play.

How did we get here?

Many have argued that the archetypal modern school — with bells, lectures, whiteboards, and frantic note-taking (or doodling, or snoozing) — is a factory. As the story goes, the Prussian model for schooling — which the United States adopted roughly in concert with its determination that the whole thing be compulsory — was designed to produce a homogenous human product: compliant industrial or agricultural workers, with a few basic competencies relevant to rote work.

The “factory model of education” makes little sense as a frame for understanding what’s occurring in schools, because schools are not efficiently producing anything. The median student in his final year is not prepared for farm or industrial work, nor is he prepared for the service-oriented or creative careers he’s more likely to pursue. He is largely innumerate, unable to understand complex texts, and lacks job skills or experiences.

If more than a decade of schooling hasn’t equipped him with literacy, numerical competency, reasoning skills or the ability to do any job, what has it done, and why? This question is actually several:

What did the designers intend when they created the system? What do the people currently responsible for executing the day-to-day activities of a school believe they’re doing?

Perhaps most crucially: how is this system adaptive? In other words: how does it serve the real (stated or unstated) incentives of the people who are necessary for it to persist?

In the case of modern schooling, the answer to the first question appears somewhat damning: in the United States, the adoption of the Prussian model for compulsory education was in large part an indoctrination project — explicitly so, by many accounts. It was a useful early tool for the faithful replication and spread of Protestant Christianity, which as a matter of doctrine placed the responsibility for interpretation of the scriptures (and thus salvation) on the individual. Soon after, the same memorization-based, one-to-many model became a tool of the state.

As Edward Ross, one of the founders of American Sociology, put it:

“Copy the child will, and the advantage of giving him his teacher instead of his father to imitate is that the former is a picked person, while the latter is not.”

The teacher’s job, Ross further elucidated:

“…to collect little plastic lumps of human dough from private households and shape them on the social kneading-board.”

The modern school, though in many ways unchanged, represents a lattice of incentives: teachers and administrative staff rely on it to provide a living and (ideally) some sense of personal meaning; parents rely on it in order to free up their weekdays for work, and children are told that it’s a necessary stepping stone toward any sort future they’d reasonably want.

Interestingly, the most ardent stakeholder is also the most distant: school staff could find other work, and most children would rather be elsewhere. It’s the parents who are in a bind: with the social value of full-time parenting by choice at an all-time low and most households far from poised to lose an income-earner, homeschooling has been ferreted away to the realm of religious zealots and the eccentric rich. Whatever other functions the school serves, first and foremost it gives legs to an economy that can accommodate child-bearing but must relegate child-rearing to off hours only.

What do the teachers, administrators, and support staff believe they’re doing? Generally something like “teaching critical thinking,” which is in most cases the educator’s equivalent to “combustion.” As the economist Bryan Caplan details in his explosive 2018 book,  teachers rarely provide a concrete description of any aspect of “critical thinking,” and insofar as they can, the evidence we have does not demonstrate that schooling imparts it. We also know that students retain very little of the object-level information they’re taught, generally reverting to failing grades on the same tests they’d previously passed weeks or months after the corresponding subject matter was covered.

The modern school is adaptive. It’s just not adaptive for any reasons related to learning.

The power of play

Zoom out a little bit, and you might notice something that gives away the game: infants, toddlers, and older children are incredible learning machines. A child will learn her first language (or languages, if available) purely by osmosis, by surrounding herself with the noises of family life, mimicking what she hears and seeing how others respond to her. If she has the opportunity, she’ll easily learn to walk and run, use electronics, sing, and play musical instruments.

She’s capable of learning all of this with no explicit instruction; she only needs to observe others and experiment through play.

As observant educators have long been telling us and researchers are beginning to reiterate, play is the work of children — and it’s critical work. Virtually all young mammals do it, expending valuable energy on apparently frivolous romps. Why? Because that’s how they learn.

Ultimately, all play is experimentation: a goal is determined, the conditions and constraints are evaluated, and a strategy is tested. This is true when you’re playing Monopoly, but it’s also true when you’re testing out jokes with a new friend or climbing a tree. Sometimes you play the wrong card, say the wrong thing, grab the wrong branch. The nature of your successes and failures provides the scaffolding for your continuously evolving model of the world.

The vogue in education is to hamper this process in the name of greater efficiency. If we already know the answers, why not just tell them to the children? A child can be made to sit down and copy what you tell him, after all. He can’t be made to play any particular game in earnest.

This is the logic that’s led us to accept the transformation of curious children into parrots: the fear that natural learning is too slow, too reliant on the whims of children and too prone to permit gaps. But beneath its peeling edges, an uncomfortable truth can’t be fully obscured: that the system we’ve created is labor-efficient childcare, not education.

In order to rescue education, we must rediscover play, and we must learn to trust the natural inclination of children to guide their own learning — as natural scientists.

About the Write of Passage Fellowship

This article was written by Mason Hartman as part of the Write of Passage Fellowship. It’s a short version of a 10,000-word essay I’ll be writing on this topic in the next few months.

We’d love to hear from you. If you have feedback, you can find the author on Twitter.

Here is the list of essays we’re writing as a group:

Click here to follow the Fellowship by email.