Easy
- Research
- 5 min
- Explain
- 45 sec
Single-layer ideas with a clear definition and an everyday example. Two questions.
in·tel·le·ge
Latin — imperative, 2nd person singular
from intellegere: to grasp, to perceive, to understand.
Not a noun. A command.
You get a concept you didn't pick, a clock, and no AI to lean on. When the time runs out, everything closes and you explain what you learned — from memory, in your own words. Then you find out how much of it you actually understood.
No card. One easy, one medium, one hard round on the house.
The premise
I couldn't reduce it to the freshman level. That means we really don't understand it.
Adapted from the method popularised as the “Feynman technique” — Feynman himself never wrote these four steps down.
You choose what you want to learn. The topic comes from the pool at random, so you can't retreat to what you already know.
No jargon to hide behind: plain words and comparisons a ten-year-old could follow.
The moment you reach for jargon or can't go on is exactly where your understanding stops. You can't look anything up mid-explanation; that spot is marked in your report, and the sources are waiting when the round ends.
Rebuild the hard parts with comparisons from everyday life. The clarity criterion in your evaluation looks for exactly this.
Recognition is cheap. When the words are in front of you, your brain nods along and files the feeling of fluency as knowledge. Close the page and try to say the same thing out loud, and the gap shows up immediately — usually in the middle of a sentence you can't finish.
That gap is the only thing worth measuring, and almost nothing measures it. Quizzes test recognition. Flashcards test recall of the answer you already wrote. Neither asks the one question that matters: can you build the explanation yourself, unaided, in real time?
Intellege asks exactly that, every single round, and it doesn't let you bring notes.
Pulled at random from the pool. Narrow it to a field — psychology, history, physics — but never to a specific topic.
Five, ten or twenty minutes to research it however you like — your browser, a book, a person. Anything except AI.
Time's up. Your notes disappear. The sources disappear. What's left is whatever made it into your head.
Two to four open questions, out loud, on the clock. Not "which of these is correct" — walk me through how it works.
What you nailed, what you skipped, what you got wrong and what's actually right. Plus the filler words and pauses in your delivery, counted one by one.
Single-layer ideas with a clear definition and an everyday example. Two questions.
Needs the origin, the mechanism and an example to hold together. Three questions.
Layered, contested, carries its own criticism. You'll be asked what's wrong with it. Four questions.
Accuracy, coverage, clarity and structure — weighted by how deep the question was digging.
Fillers per minute and words per minute. Counted by code, not guessed by a model.
Not from the round. From what you still remember days later — see below.
Recall check · day 6
Who ran the experiment, and what were the participants given?
Explaining something an hour after reading it proves very little. So days later — unannounced — your phone asks one question about a concept you covered. Forty-five seconds, out loud, no warning.
Each answer moves that topic's decay curve. Remember it well and the next check pushes further out; blank on it and it comes back sooner. Your Memory score is the honest average across everything you've studied.
Missing a check costs you nothing. Forgetting is normal, and a low Memory score isn't a punishment — it's a list of what's ready for another pass.
Pricing
A cold open. The topic finds you and you cannot narrow it down — which is the point.