The Feynman technique, and the version of it that does nothing
The four steps
- Write the concept at the top of a blank page. One concept. Not "thermodynamics", but "why entropy increases".
- Explain it in plain language. Write as if the reader has never met the subject. No jargon, no borrowed phrasing. Full sentences.
- Find where you stalled. The places you went vague, reached for a technical term, or wrote "which basically means" and then stopped. Those are the gaps. Go back to the source for those parts only.
- Simplify and use an analogy. Rewrite the rough parts. If you can map it onto something ordinary without the analogy falling apart, you understand it.
The fake version
Here is the failure mode, and almost everyone starts here. You open your notes, read a paragraph, and write your explanation with the notes visible. It feels fine. You produce a page. You learn close to nothing.
The whole value is in writing from an empty page with the source closed. Familiarity is not knowledge, and reading something you understood once produces a strong feeling of understanding that disappears under exam conditions. Cognitive psychologists call this the fluency illusion, and it is the reason re-reading is such a popular and ineffective way to study.
So: close the book. Write badly. The badness is the diagnostic result, not a failure of the exercise.
A worked example
Take "why does a compiler need a symbol table". First attempt, source closed:
"A compiler reads your code and needs to keep track of the names you used. So it puts them in a table. Then later when it sees the name again it can look it up and know what it is. It stores the type and scope and stuff like that."
Now read it back looking only for weak spots. "And stuff like that" is one. "Know what it is" is another - know what, exactly? And there is a bigger one hiding: nothing here explains why a table rather than just scanning the code again.
That last question is the real gap, and you would never have noticed it by re-reading a chapter. It surfaced because you had to produce a causal explanation and could not.
How you know you are done
You can explain it without the source, in ordinary words, including why it works that way rather than some other way. If your explanation only describes what happens and never why, keep going.
Feynman or Cornell?
These get compared a lot, usually as rivals. They are not doing the same job, and the comparison only makes sense once you see that one runs continuously and the other runs on demand.
| Cornell | Feynman | |
|---|---|---|
| What it is | A capture and review system | A diagnostic for one concept |
| When you run it | Every lecture, all term | When you suspect a gap |
| Input | A lecture or reading | A blank page |
| Output | Notes you keep and re-test | An explanation you usually throw away |
| Tells you | What was covered | What you do not understand |
They also chain together well. Your Cornell cue column is a list of questions you wrote about the material. Any cue you repeatedly fail is a candidate for a Feynman round. That is a much better trigger than picking a topic because it feels hard.
If you have not set that up, the Cornell method is the capture half, and the printable template is on the site if you want to start today.
Doing it on screen
Feynman rounds want a blank page, so paper or a plain text file is genuinely fine. The one place a tool helps is step three, when you go back to the source. Marking the exact lines that exposed a gap, in the margin of the paper or slide deck itself, means next time you know where you fell over rather than re-reading the chapter.
That is the use we built SidePDF for: widening a PDF page so there is room to write "could not explain why - see notes" beside the paragraph that caused it.
Frequently asked questions
What are the four steps of the Feynman technique?+
Pick a concept and write it at the top of a blank page. Explain it in plain language as if to someone who has never met the topic. Notice where you stall or fall back on jargon, and go back to the source for those parts only. Then simplify the explanation again and use an analogy.
Did Richard Feynman actually invent this?+
Not as a named method. He was known for demanding plain explanations and for keeping a notebook of things he did not understand, and the four-step version was assembled later by others from how he worked. The technique is sound, the branding is retrospective.
Why does explaining something out loud help?+
It forces production rather than recognition. Reading a page feels like understanding because the words are familiar. Producing an explanation from an empty page has nowhere to hide, so gaps show up immediately.
Feynman technique or Cornell notes?+
They do different jobs. Cornell is a capture and review system you run continuously through a term. Feynman is a diagnostic you run on one concept when you suspect you do not really understand it. Most people should use both, at different moments.
How long should one round take?+
Fifteen to twenty minutes for a single concept. If it is taking an hour you have picked something too big, like a whole chapter. Narrow it to one idea and run it again.
Try it once, properly
Pick one concept from this week that you think you know. Close everything. Write the explanation from scratch, in plain words, including why it works. Twenty minutes.
Most people are surprised, and not in a flattering way. That surprise is the point.
Mark the gaps where you found them
Widen the margin of any PDF and write beside the paragraph, not on top of it.
See how SidePDF works