Why re-reading feels fast and isn't
Re-reading is fast per pass and near-zero per unit of memory. The reason is fluency: the second time you read a paragraph it goes down easily, and your brain reads that ease as *I know this*. It isn't knowledge, it's familiarity with the surface of the text — which is why people who re-read confidently go blank at a question phrased differently from the book.
Retrieval feels slow because it is effortful, and effort is the mechanism, not the friction. Every time you drag something out of memory without looking, the path to it gets easier to travel next time. Looking at the answer does not do this. This is the whole finding behind active recall and it is the least-disputed result in learning research.
Six methods, ranked by minutes to retention
| Method | Speed to get going | Retention per minute spent |
|---|---|---|
| Self-testing from a blank page | Instant — needs nothing | Very high. The default; everything else is a variation. |
| Flashcards on a real scheduling algorithm | Slow to build, instant thereafter | Highest over weeks, because it stops you reviewing what you know. |
| Memory palace / method of loci | Slow — the palace takes an hour to build | Very high for ordered lists, poor for anything requiring understanding. |
| Mnemonics and acronyms | Fast | High for arbitrary lists, zero for anything you must apply. |
| Explaining out loud from memory | Instant | High, and uniquely good at exposing what you don't know. |
| Re-reading and highlighting | Instant | Near zero. Feels the fastest, is the slowest. |
The forty-minute protocol
If you have one chunk of material and need it in memory by tomorrow, this is the sequence with the best return per minute. It is deliberately uncomfortable.
- 1
Read once, slowly, with a pen (8 minutes)
One pass. Not three. As you go, write a question in the margin for anything you would want to be asked. You are not trying to remember yet, you are building the test.
- 2
Close everything and brain-dump (10 minutes)
Blank page. Write everything you can recall, in any order, badly. This will feel like it is going terribly. The gaps you hit are the entire product of this step — they're a map of what to work on, generated in ten minutes rather than discovered in an exam.
- 3
Check, and mark only the gaps (5 minutes)
Compare against the source. Do not re-read the parts you got right — that's where most people lose the hour. Circle only what was missing or wrong.
- 4
Turn each gap into a question, not a note (7 minutes)
"The three consequences of X are…" beats a paragraph about X. A question can be answered from memory; a paragraph can only be re-read. This is the step that converts today's work into something reusable.
- 5
Second brain-dump, from memory only (7 minutes)
Same blank page, same material. The difference between dump one and dump two is what you learned in the last half hour, and seeing it measured is most of what keeps people doing this.
- 6
Schedule one review for tomorrow and one for day four (3 minutes)
Without spacing, most of this is gone inside a week. Two reviews is the minimum that turns a day's memorisation into a month's.
When mnemonics are worth it — and when they cost you
Mnemonics work by attaching arbitrary information to a structure you already have. They are extremely effective for exactly that: arbitrary, ordered, list-shaped facts — cranial nerves, resistor colour codes, the order of taxonomic ranks.
They are actively harmful when the material has internal logic, because the mnemonic replaces the understanding. A student who has memorised a mnemonic for the steps of glycolysis can list them and cannot tell you why the cell bothers. The exam question is almost always the second kind. Use mnemonics for the arbitrary residue *after* you understand the structure, never instead of it — the memory palace piece goes into where the line falls.
Speed comes from not re-learning things
Here is the arithmetic nobody runs. Memorising 200 facts in a weekend and re-learning them from scratch three weeks later costs you two weekends. Memorising them once and reviewing at widening intervals — day 1, day 3, day 8, day 21 — costs one weekend plus about forty minutes total. The second approach is four times faster over a term and it is the only reason spaced repetition is worth the setup.
- Review before you forget, not after. After means re-learning at full price.
- Let the interval grow. Reviewing a card you know cold is time transferred from a card you don't.
- Never review by reading. A review is a test or it isn't a review.
What to do the night before
If it is genuinely tomorrow, cram — and cram correctly. Do not read. Test, check, test again, on the highest-yield material only, and sleep at least a full sleep cycle before the exam because the last consolidation pass happens there. Cramming versus spacing covers what cramming does and doesn't buy you; the short version is that it gets you through tomorrow and is gone by next month, which is fine if next month doesn't have an exam in it and a disaster if it does.