Introduction
Study & Recall
Explaining something from memory is the fastest way to find out you can’t.
Here’s the short answer before the article properly starts. To use the Feynman Technique for studying: pick one focused topic, close the source, explain it in plain language from memory, mark every point where you went vague or hand-wavy, go back to the source and fix those specific bits, then explain it again — and come back to it days later.
That’s it. What it is not is the “pretend you’re teaching a child” theatre it usually gets sold as. The imaginary child is a prompt, not the method. The method is a retrieval-and-checking loop: produce an explanation without help, find where it breaks, repair it against a reliable source, prove the repair held.
If you’re working through an online course, a professional qualification or a technical cert, that difference matters. Simplifying language is easy. Producing a correct explanation with the tab closed is the bit that tells you whether you actually learned anything.
What the Feynman Technique is, in one paragraph
The Feynman Technique is a study method named after physicist Richard Feynman, who was known for explaining hard ideas in unusually plain terms. The technique takes that habit and turns it into a study loop: you choose a concept, write or say an explanation of it in ordinary language without looking anything up, notice where your explanation goes fuzzy or collapses into jargon, then go back to the source material to fix exactly those points and explain it again. The value isn’t the simple wording. It’s that attempting a full explanation from memory makes gaps visible that rereading quietly hides.
1. Choose one focused topic
Not “networking”. Not “module 3”. One concept you could name in a short phrase: how a DNS lookup resolves a name, what a foreign key constraint enforces, why an index speeds up a query, what the difference is between authentication and authorisation.
Broad topics produce broad explanations, and broad explanations hide gaps beautifully. A tight topic gives you nowhere to waffle.
2. Close the source
Notes shut. Course video paused and off-screen. No documentation open in a second tab “just in case”. If the source is visible, you’re paraphrasing, not retrieving, and the whole exercise loses its diagnostic value.
3. Explain it in plain language from memory
Write it out, say it aloud, or record a voice note — whichever you’ll actually do. Aim for language a competent person outside your subject could follow. Include what the thing is, what problem it solves, how it works step by step, and one concrete example.
Keep going even when it gets uncomfortable. The uncomfortable bit is the data you came for.
4. Mark the exact gaps and hand-waving
Go back through what you produced and highlight every spot where you did one of these:
Be specific. “I don’t really get caching” is not a gap you can study. “I can’t explain what decides how long a cached record is kept” is.
5. Check the original source and correct errors
Now open the material. Two jobs here, and people usually only do the first: fill in what was missing, and check whether what you confidently wrote was actually correct.
Confident-and-wrong is the most expensive category in any study session, because nothing about it feels like a gap. The only way it surfaces is by producing the explanation and then checking it against something reliable.
6. Explain again, then revisit later
Close the source and re-explain — the whole thing, not just the repaired fragments. Then put a return date on it. A topic you fixed on Tuesday and never touched again is a topic you fixed on Tuesday.
A real Feynman technique example: “What does DNS do?”
Theory’s cheap, so here’s a worked one. Say you’re studying for an IT cert and the lesson covered DNS. You close the material and explain it from memory.
The first attempt (be honest — this is what it looks like)
“DNS is like the phone book of the internet. You type a website name and DNS turns it into an IP address so your computer can find the server. It looks it up somewhere and sends the address back.”
Which sounds fine. It’s the answer most people would nod along to. It’s also mostly a slogan.
Checking the source and correcting it
Back to the material, and several things get fixed. DNS isn’t one phone book — it’s a distributed, hierarchical system. Your device usually asks a resolver (often run by your ISP or a public provider), and the resolver does the work of finding an answer, consulting authoritative name servers for the relevant part of the domain hierarchy when it doesn’t already have one.
DNS also doesn’t only return IP addresses. It returns records of various types — A and AAAA records for addresses, MX for mail routing, TXT for arbitrary text data used by things like domain verification, CNAME for aliases. “Name to IP” is one common case, not the whole job.
And the path a lookup takes varies. If the resolver already holds a valid cached answer, it returns it immediately and nothing else is contacted. If not, it may need several queries before it has an authoritative answer. Records carry a time-to-live that controls how long they can be cached, which is why DNS changes don’t take effect everywhere at once.
The second explanation
“DNS is the system that turns human-readable names like example.com into the records computers need — most commonly the IP address of a server, but also mail records, text records and aliases. It’s distributed rather than one central list: your device asks a resolver, and if that resolver doesn’t already have a valid cached answer, it queries authoritative name servers responsible for that part of the domain hierarchy until it gets one. The answer comes back with a time-to-live, so it can be cached for a while, which is why most lookups are near-instant and why a DNS change can take time to be seen everywhere.”
Same topic. Same person. Twenty minutes apart. The second version isn’t better because the words are fancier — it’s better because producing the first one showed exactly which four things needed checking.
Note what the first attempt was worth. Getting it thin and slightly wrong was the useful part. That’s the same principle behind Fail First : find out what you can’t do before you spend an hour studying things you already knew.
Why the Feynman Technique can work
Two things are going on, and they’re worth separating.
First, explaining from memory is retrieval practice . You’re pulling the material out of your head rather than letting the page hand it to you, and the act of retrieving strengthens later recall. Roediger and Karpicke’s well-known 2006 experiments with prose passages found that repeated studying produced better performance on an immediate test five minutes later, while students who had been tested on the material retained more on delayed tests two days and a week later. Rereading flatters your short-term performance; retrieval holds up over time.
Second, producing an explanation makes your knowledge visible and checkable . A vague understanding can survive indefinitely while you nod at notes. It cannot survive being written down in full sentences with a worked example attached. This is closely related to self-explanation, and to research on learning by teaching — Fiorella and Mayer (2013) looked at teaching expectancy and actually teaching, finding that the act of explaining to others, not merely preparing with the expectation of teaching, was associated with better understanding.
Roughly: retrieval does the strengthening, explanation does the diagnosing. You need both, which is why the checking step isn’t optional.
An honest note on the evidence
Worth being straight about this, because a lot of study advice isn’t.
There is no body of experimental research proving that the branded, named “Feynman Technique” — as a fixed four- or six-step sequence — is uniquely effective or better than alternatives. What the evidence supports is the ingredients : retrieval practice, self-explanation, and learning by teaching all have real research behind them.
So the sensible claim is this. The Feynman Technique is a practical way of packaging several things that are well supported individually. Whether it works for you depends heavily on how you run it. Explain from memory, check against a reliable source, and come back later, and you’re doing things with genuine support. Paraphrase your notes with the notes open and you’ve done none of them, regardless of what you call it.
Copying the notes into simpler words with the source open
The single most common failure. It feels productive, produces a tidy document, and tests nothing. If the source was visible, you translated — you didn’t retrieve.
Choosing a topic that’s far too broad
“Explain cloud computing” gives you enough room to sound competent for four paragraphs without ever landing on something checkable. Narrow it until vagueness has nowhere to hide.
Making the explanation simple but false
Simplification has a failure mode: you smooth an idea into something clean and wrong. “DNS is the internet’s phone book” is exactly that. Always check the simplified version against the source, not just the missing bits.
Stopping the moment you find a gap
Finding the gap is the halfway point, not the finish line. “Ah, I don’t really know that bit” followed by closing the laptop is a diagnosis with no treatment.
Letting AI write the explanation for you
Asking a chatbot to explain the topic simply produces a perfectly good explanation that tests nothing about your memory. Use AI afterwards if you like — to generate questions, or as one source to check against — but the first explanation has to come out of your head or the whole thing is pointless.
Never checking the source
An unchecked explanation just confirms what you already believed, confidently. The check is where wrong-but-fluent gets caught.
Never retesting later
One good explanation on a Tuesday is one good explanation on a Tuesday. Without a return visit you have no evidence it survived the week.
Using it for an online course, cert or professional qualification
If you’re self-studying something with an assessment at the end, a few adjustments make this a lot more useful.
Run it at the end of each lesson, not the end of the module. One explanation per concept, while the material is fresh enough that the gaps are about understanding rather than total forgetting.
Match the explanation to the assessment format. If you’ll be answering scenario questions, explain the concept and then explain when you’d use it and when you wouldn’t. If it’s practical — code, labs, configuration — your explanation should include the steps you’d actually take, not just the definition.
Keep a running gap list. Every marked gap goes on one list for the course. That list is your revision plan, built from evidence rather than vibes, and it’s far better than rereading chapters in order the week before the exam.
Explain the boundaries too. For technical material, “what does this not do” and “when is this the wrong tool” separate people who’ve memorised a definition from people who can use it.
How it fits with active recall and spaced repetition
These aren’t competing techniques, and treating them as a menu to choose from is a mistake.
The Feynman Technique is a form of active recall — specifically the self-explanation kind, where instead of answering short questions you produce a complete account of the topic. It’s one of the more demanding forms, because you can’t half-answer your way through it. If you want other retrieval formats to rotate through, the active recall study methods guide covers several.
Spaced repetition decides when you come back. Feynman decides what you do when you get there. A review that says “explain this topic from memory, then check it” is a far better use of a scheduled slot than opening the notes again — which is the trap covered in rereading vs active recall .
Combined, it looks like this: explain from memory today, check and fix, then re-explain on Day 3, Day 7 and Day 21 using the spaced repetition intervals. Each return visit is a fresh explanation, not a reread. Gaps that keep coming back get more attention; topics that come out clean twice in a row can wait longer.
The five-minute version for low-energy days
Some evenings you have no intention of writing a full explanation, and pretending otherwise just means doing nothing. So here’s the cut-down version.
Five minutes, one gap found, one gap fixed. That is meaningfully more than a tired hour of scrolling through notes, and it’s repeatable on days when the good version isn’t going to happen.
The short version
If you skip everything else:
Choose one focused topic → close the source → explain it in plain language from memory → mark the exact gaps and hand-waving → check the source and correct what’s missing and what’s wrong → explain again, and revisit it later.
The simple language isn’t the point. The point is that you can’t produce a complete explanation of something you only half-understand, and finding out where it breaks is worth more than another pass through the notes. Feynman’s reputation for clear explanation wasn’t a communication trick — it came from refusing to accept a fuzzy understanding of his own. That’s the bit worth copying.
Sources and further reading
• University of York — the Feynman Technique study guide — a university study-skills walkthrough of the steps and how students are advised to apply them.
• Roediger & Karpicke (2006), Test-Enhanced Learning — the prose-passage experiments comparing repeated study with testing at five minutes, two days and one week.
• Fiorella & Mayer (2013), The relative benefits of learning by teaching and teaching expectancy — on whether the benefit comes from expecting to teach or from actually explaining.
Keep learning
• 7 Active Recall Study Methods → Other retrieval formats to rotate through when explaining gets stale.
• How to Do Spaced Repetition → Decide when to come back and re-explain, instead of reviewing at random.
• Rereading vs Active Recall → Why familiar notes feel like knowledge right up until you have to produce it.
• Why The Stack Works → How Fail First, retrieval, explanation and retesting fit into one loop.
• More study guides → Browse practical guides to study methods, recall and finding knowledge gaps.
Structure the explanation attempt
Study & Recall includes a Feynman — Explain It Tool that walks you through choosing a topic, explaining it from memory, marking the gaps and writing the improved version. To be clear about what it does: it structures the attempt and keeps a record of it. It doesn’t understand your topic, it can’t tell you whether your explanation is correct, and it doesn’t replace checking against the source. That part is still yours.