Learn anything in three moves

The chat you can’t remember

Think of the last thing you asked a chatbot to explain. Maybe it was why prices keep going up. The answer was clear, well organised, even a little elegant. You read it, nodded, and closed the tab.

Now try to say what it told you. Most people can recall that they asked, and roughly how the answer felt, and very little of what it said. The one sentence you didn’t quite understand stayed not quite understood. And nothing about the experience warned you, because reading a clear explanation feels almost exactly like understanding one.

That feeling has a name. In 2002 Leonid Rozenblit and Frank Keil asked people to rate how well they understood everyday things — a zip, a flush toilet, a cylinder lock — then to write out, step by step, how each one works, then to rate themselves again. The second ratings fell sharply. They called it the illusion of explanatory depth: until you try to produce an explanation, you can’t see the holes in it.

This article is about a way of learning that is built around that finding. It has three moves — understand, connect, teach it back — and each one exists to catch a different way that understanding quietly fails. We’ll walk through an afternoon on a real question, why do prices rise?, chosen on purpose to show that the method isn’t only for programmers.

Move one: understand it

Start with a good explanation. It can come from a textbook, a teacher, a well-made video or an AI, but it should have a recognisable shape: the idea in one sentence, how it works, an example, and where people usually go wrong. That last part matters more than it looks. A list of common confusions tells you what to check in your own head.

For prices, the one-sentence version might be: prices rise across the board when spending grows faster than the economy can produce, when the cost of making things goes up, or when people expect prices to rise and act on it. The example might be the energy shock of 2022, where the cost of fuel and power pushed up the price of almost everything made or moved with them. And a common confusion: one product getting expensive isn’t inflation. Inflation is the general level of prices rising, which is why it is measured with a basket of many goods.

Before you read, ask yourself one question: what do I already know about this? The educational psychologist David Ausubel put it bluntly in 1968 — the most important single factor influencing learning is what the learner already knows. Even one line of what you know gives the new explanation something to attach to.

Then do two small things that make a large difference.

Ask one question about the part you don’t follow. Not “explain it again” but something specific: why would people expecting inflation make it happen? Asking why and how something is true — psychologists call it elaborative interrogation — is one of the few study techniques that held up well in a large review of the evidence, well ahead of rereading and highlighting (Dunlosky et al., 2013).

Put the idea in a sentence of your own. Not a copy of the summary, but your phrasing, with the note out of sight if you can manage it. In a classic study, eighth-graders who were prompted to explain each line of a text about the circulatory system to themselves ended up with correct mental models far more often than students who read the same text twice (Chi et al., 1994). Your sentence will probably be clumsy. That’s the point. The clumsiness is where the understanding is still being built.

What move one leaves behind: a clear explanation, one question you asked and had answered, and one sentence in your own words.

Move two: connect it

Every explanation mentions other ideas. Our one about prices mentions spending, production, costs, expectations and, if it is any good, interest rates and wages. Most people read past these. They are the most useful thing on the page.

Pick the one that seems to carry the most weight — here, probably expectations — and follow it. Get it explained the same way: one sentence, how it works, an example, the usual confusion. Expectations matter because a business that expects its costs to rise next year raises its prices this year, and a worker who expects prices to rise asks for a bigger raise. Each of those decisions is reasonable on its own, and together they can keep inflation going after the original cause has faded.

Now the important part. Don’t just file the new idea next to the old one. Say, in a sentence, how the two connect. Expectations turn a one-off price rise into a lasting one, because they change what businesses charge and what workers ask for before the costs have actually arrived. That sentence is the link, and it is worth more than the two explanations on either side of it.

Why insist on the sentence? Because recalling a connection is a different act from looking at one. In a study published in Science, students who practised retrieving what they had read learned more than students who drew detailed concept maps of the same material — even when the final test asked them to draw a concept map (Karpicke & Blunt, 2011). Stating the relation from memory is what makes it yours.

Do this two or three times — interest rates back to spending, wages to expectations. You don’t need to follow every thread; write the ones you leave as a short list of what to look at next.

What move two leaves behind: two or three more explanations, each joined to the first by a sentence of yours, and a short list of threads not yet pulled.

Move three: teach it back

When you think you’ve got it, close everything and explain the whole idea to someone who doesn’t know it. A friend is ideal; a rubber duck, or a page addressed to an imaginary twelve-year-old, will do.

Teaching does two things at once. The first is diagnostic: it exposes the gaps that reading hid, which is exactly what Rozenblit and Keil found. The second is that it strengthens what you know. Students who studied expecting to teach recalled more, and recalled it in better order, than students who expected a test (Nestojko et al., 2014). And students who went on to actually teach — not just prepare — still did better a week later (Fiorella & Mayer, 2013).

The listener matters. A good one asks the questions a newcomer would: why?, can you give me an example?, what does that word mean?, what would happen if…? They don’t correct you and they don’t supply the answer. Your job is to keep going until you hit a question you can’t answer.

You will hit one. Maybe it is if higher interest rates slow spending, why don’t central banks just raise them as soon as prices rise? — and you realise you don’t know what higher rates cost. That isn’t failure. That is the most valuable sentence of the afternoon, and it should be written down as a question, attached to the idea it belongs to.

What move three leaves behind: your explanation, in your words, and a handful of honest questions about where it broke.

What you have at the end

After an afternoon you have something that a chat transcript never gives you: a small set of linked notes about one subject, a few of them in your own words, the connections between them stated as sentences, and a list of questions that mark the exact edge of what you understand.

The questions are the part people underrate: each one sends you back to move one, on a smaller and sharper topic. Understanding a subject is rarely one pass. It is several turns of this loop, each starting from the gaps the last one found.

Doing it on paper

None of this needs software. A notebook and a pen will do:

1. Write the big question at the top of a page. Under it, one line on what you already know.
2. For each concept you explore, write its one-sentence version, an example and its usual confusion, then one sentence in your own words.
3. Between concepts, write the connecting sentence: A matters to B because…
4. Keep a list in the margin of concepts mentioned but not yet explored.
5. Close the notebook. Explain the whole thing aloud. Write each place you stalled as a question, on the page it belongs to.

The bookkeeping — a consistent shape, an up-to-date list of unexplored concepts, each question in the right place — is the tedious part, and the part that teaches you least.

How Graphite does this

Graphite is a notebook built around these three moves. Ask it to explain anything and the Guide writes the explanation as a note, labelled as its own. Ask the Guide beside any note — or select a line, choose Ask Graphite and type your question — and it answers from your notes, then asks you one question back. Link related proposes links, each with its reason, and adds none you don’t accept. When you’re ready, close the note and teach the idea to Graphite playing a curious student; it marks your explanation against your own notes and puts each gap under What’s missing. Then the Guide can arrange a first draft from your own sentences, to publish as a page you own. Everything is a plain markdown file you keep. You can try the notebook in your browser.