The Learning Flywheel: How the Sequence Fits Together
Norudit isn't a pile of study features you choose between. It's one ordered sequence, and the reason I call it a flywheel is that each stage produces exactly what the next one needs, so the whole thing picks up momentum the longer you run it.
Most study tools hand you a toolbox and quietly assume you already know how to use it, which is the core problem with the whole category. This is the opposite of that. A recommended path where the phases are linked to each other rather than just listed next to each other.
The flywheel image is doing real work here. A flywheel is a heavy wheel: awful to get started, but every turn feeds the next one, and once it catches it begins powering itself. You can enter it wherever today's work actually demands, because the sequence is the shape of the machine and not a corridor of locked doors. There's exactly one verification gate in the entire thing and I'll come to it.
The arc in one line
Every phase does one cognitive job, and those jobs run in a specific order:
Encode → Elaborate → Retrieve → Apply. Get it into memory, connect it to what you already know, practise pulling it back out, then use it under pressure.
Here's how that maps onto the phases, and how each one hands off to the next.
Encode: getting it in
Phase 0, the pre-test. Skim the material, then attempt a handful of questions you aren't expected to get right. Guessing and failing before you study measurably improves how well you learn the correct answer afterwards (Kornell, Hays and Bjork, 2009; Richland, Kornell and Kao, 2009), which is the pretesting effect, and there's a whole post on it. It produces two things the rest of the sequence goes on to use: a rough mental scaffold, and a map of your specific gaps.
Phase 1, the overview. A walkthrough that lands the real explanation onto that scaffold, pairing words with visuals, which is remembered better than either on its own (dual coding; Paivio, 1971). It needs Phase 0's scaffold to land on, and what it produces is your first coherent mental model of the topic.
Elaborate: connecting it
Phase 2, Socratic inquiry. Instead of being told things you get asked things, a back-and-forth of why and how-does-this-connect that forces you to tie the new model to what you already have (Pressley et al., 1992), while circling back to the gaps Phase 0 turned up. It needs the explanation from Phase 1, because you can't form an analogy to something nobody has shown you yet, and it produces the analogies and self-references the next phase is about to demand.
Retrieve: pulling it back out
Phase 3, Feynman. Named after Richard Feynman, whose study method survives in one blunt sentence: if you can't explain it simply, you don't understand it. His private notebooks were exactly this exercise, taking a topic he thought he knew, writing it out as though teaching a bright twelve-year-old, and watching for the precise point where the plain words ran out. Here you teach the topic back to something that won't let you hide, which is the one test you can't fake. This is the single verification gate in the app, and you pass it by demonstrating understanding rather than by receiving an explanation. It needs your Phase 2 analogies, it produces a verified model, and passing it unlocks free recall.
Phase 4, free recall. The blank page. Close the material and rebuild the entire thing from nothing, with no prompts, no cues and no multiple choice to lean against. Of every form of retrieval practice this is the most demanding, because the memory gets no scaffolding whatsoever and every fact, connection and example has to come up by its own roots. That effort isn't the unfortunate cost of the exercise, it is the exercise, because effort is the mechanism (Bjork and Bjork, 2011) and whatever you drag out unaided is what the dragging strengthened (Roediger and Karpicke, 2006). It stays locked until the Feynman gate passes, because recalling a wrong model just rehearses the error, so recall only starts paying once there's a correct model there to retrieve. Finishing this first recall is also what builds your review deck, out of the concepts you just pulled up on your own.
Apply: using it under pressure
Phase 5, debate. Argue a position against an opponent, including one you don't hold, and then switch sides and argue the other. Perspective-switching breaks the illusion of one-sided mastery (Galinsky and Moskowitz, 2000), and it needs your verified model and recalled vocabulary or you've got nothing to defend with.
There's a second reason debate earns its slot, and it's emotional rather than cognitive. Arguing a position you've got a stake in raises a moderate, task-intrinsic arousal, the productive kind rather than the panicked kind, and that deepens how firmly the material gets encoded (LaBar and Cabeza, 2006; Pekrun et al., 2002). A dialectic that carries stakes consolidates more than a passive reread of the same points would, precisely because the argument makes the difference between the two positions something you feel rather than something you read.
Phase 6, the exam. A genuine exam-style paper marked to a scheme, spanning the skills the earlier phases built, with error analysis afterwards. It needs everything before it, and it produces the diagnostic that decides what happens next.
Phase 7, the capstone. One open-ended project you complete across three disciplinary roles and then lead into a single synthesis you direct yourself. It's the transfer test an exam can't give you, which is applying the whole topic to a problem it has never once shown you.
The map, at a glance
| Stage | Phase | The move (and its source) | Gated? |
|---|---|---|---|
| Encode | 0 · Pre-test | Metacognitive priming (pretesting effect) | Open |
| Encode | 1 · Overview | Dual coding (Paivio) | Open |
| Elaborate | 2 · Socratic | Elaborative interrogation (Pressley et al.) | Open |
| Retrieve | 3 · Feynman | Self-explanation (Chi et al.) | The gate |
| Retrieve | 4 · Free recall | Retrieval practice (Roediger & Karpicke) | Behind the gate |
| Apply | 5 · Debate | Argumentation + perspective-taking | Open |
| Apply | 6 · Exam | Desirable difficulty + spacing + interleaving (Bjork) | Open |
| Apply | 7 · Capstone | Project-based learning + transfer | Open |
Why it has to be this order
The order isn't arbitrary, because each phase's output is literally the next phase's input.
You can't elaborate on a model you haven't been shown, so Socratic has to follow the overview. You can't teach back a model you haven't integrated, so the Feynman gate has to follow Socratic. And recalling from a blank page only helps once the model in your head is correct, so free recall has to follow the gate and never precede it.
Follow the sequence and every stage arrives pre-loaded with what it needs. Skip around and you may find a stage running on less than it wants, which is a trade you're perfectly entitled to make, and that brings me to the next bit.
Recommended, not enforced
For a while the design was going to force the whole sequence. Locked doors, one phase at a time, no skipping. There's a certain purity to that idea. It's also, speaking as someone who has been a student, terrible UX, and it died the moment I pictured an actual person running into it.
Picture it. You already know a topic and you just want a mock-exam run before Friday's paper. Or you've got one nagging confusion that five minutes of Socratic chat would untangle. And the app says no, go back and do the overview first. You can feel the frustration from here. That student closes the tab, and honestly they're right to.
So nothing is locked. Every phase is open, in any order, on any day. The sequence is a recommendation, specifically the order that makes each phase cheapest and most effective, rather than a cage.
With one exception.
The verification gate
Phase 3 is the only thing in the app that will stop you, and the idea behind it isn't complicated: if you can't explain it simply, you don't know it. You give your explanation, it gets judged properly rather than by vibe, and until it passes, free recall and the review deck stay shut.
Unlike an ordinary locked door, though, this one has something on the other side of it. Pass it and your review deck unlocks, and the deck is actually constructed in that moment, out of the concepts you've just proved you understand. A gate everywhere would be a wall. One verification gate, sitting exactly where the feeling of understanding gets mistaken for understanding, with something real waiting behind it, is a threshold worth making people cross.
Which phases are held loosely
Since nothing's enforced, "optional" really just means how often a phase earns its place.
- Phase 0 is preparation, in the way reading the material beforehand is preparation. Skip it and you go into Phase 1 with a colder start, and that's all you lose.
- Phase 5 needs a topic with two genuinely defensible sides. Settled or procedural topics haven't got them, so it's offered where it actually fits rather than everywhere.
- Phase 7 is the highest-transfer extension, offered at the end of a full run rather than expected of every single topic.
The spine is the overview, Socratic, the gate, free recall and the exam. The rest are extensions, there for when the subject and the moment suit them.
Where the wheel actually closes
Here's what makes it a flywheel rather than a conveyor belt: it doesn't dead-end at the exam.
The exam produces a diagnostic, meaning which concepts are weak and why each specific mistake happened, and that feeds two things. One is a per-concept estimate of what you've genuinely mastered, which is the idea behind Bayesian knowledge tracing (Corbett and Anderson, 1994). The other is a spaced-repetition schedule that pushes your weakest concepts to the front and re-tests them right as you're about to lose them.
Assess, diagnose, targeted review, re-test, and round again. Each turn feeds the next, and because the system remembers you between sessions the wheel keeps its momentum instead of restarting cold every time. The second turn is faster than the first, which is the entire point of building it this way.
Each stage makes the next one cheaper and stronger, and the back end feeds momentum to the front. That compounding, rather than any single feature, is what makes it work.
Concept-varied review
The review loop is there to fix the two ways spaced repetition quietly fails.
Fixed-card tools like Anki proved the underlying science works (Ebbinghaus, 1885; Cepeda et al., 2006), but a card that never changes eventually ends up testing whether you recognise the card rather than whether you know the concept. And a backlog left alone for a week becomes a wall of several hundred due cards that most people bounce straight off.
The deck here answers both. Every time you answer a card it gets rebuilt for the same concept, so concepts return reworded to check the understanding held and facts return as application questions to check you can use them rather than recite them. And a backlog never gets dumped on you, it's sequenced into a finishable first sitting of your weakest concepts with the rest offered in small optional batches. The full story is its own post.
The trio underneath all of it
Now the part I haven't seen anywhere else, and the actual reason the pieces sit in this exact order. Three mechanisms, the Feynman gate, free recall and concept-varied review, are welded together into one test-and-maintain chain.
- If you can't explain it simply, you don't know it. That's the gate, and it's merciless deliberately. No demonstrated explanation, no progress, because everything downstream is assuming a correct model and rehearsing a wrong one only makes it more durable.
- If you can't free-recall it, there's nothing to review. Passing the gate proves understanding, and understanding isn't memory. The blank page is where the memory gets forged, and a review schedule aimed at a memory that never formed is maintenance on an empty room. Which is why the review deck for a topic doesn't exist at all until the gate passes: it gets generated in that moment, from the concepts you just proved.
- If both hold, review starts. Understanding is in place, memory is in place, so concept-varied spaced repetition finally has something real to maintain, and because the surface changes every round what it maintains stays the concept rather than the card.
Understanding, then memory, then maintenance, and each one fairly meaningless without the one before it. This chain is the only thing in the whole app that's forced. Every phase is open to roam through, so go straight to the mock exam if that's what today needs. But this chain is enforced, because it sits on the precise spot where studying quietly fails, where "I've seen it" turns into "I know it" and review becomes ritual instead of retrieval. And it earns its strictness by paying you back, since the deck it unlocks is built out of the understanding you just demonstrated.
The gate, the blank page, and the ever-changing card: that trio is the method's own contribution. Everything else here stands on published science, and the way these three lock together is mine.
A toolbox resets to zero every session. A flywheel keeps spinning.
The sequence is the method, but the academy around it is bigger than the sequence: Lura, the teaching assistant inside every presentation, Norudit Search, the Smart Scheduler with its calendar and Kanban board, Co-Writer with summaries and glossaries, mind-maps, Independent Research and the Mock Generator. The full tour is here.
Everything above is built and running on whatever material you bring it.
Further reading
If you'd rather go to the sources than take my word for any of this, start here.
The two best free places to begin:
- The Learning Scientists, with clear evidence-based guides to retrieval practice, spacing, interleaving, elaboration and dual coding.
- RetrievalPractice.org, Pooja Agarwal's library on the testing effect and how to use it.
The books that shaped the method, walked through in The Bookshelf Behind the Method:
- Brown, Roediger & McDaniel, Make It Stick: The Science of Successful Learning (2014).
- Dunlosky et al., Improving Students' Learning With Effective Learning Techniques (2013).
- Willingham, Why Don't Students Like School? (2009).
- Dehaene, How We Learn (2020).
- Mayer, Multimedia Learning (2009).
The primary papers behind specific phases:
- Roediger & Karpicke (2006), "Test-Enhanced Learning": Phases 4 and 6.
- R. Bjork (1994); Bjork & Bjork (2011), Desirable Difficulties: Phase 6.
- Chi et al. (1989), Self-Explanations: Phase 3.
- Pressley et al. (1992), Elaborative Interrogation: Phase 2.
- Ausubel (1960), Advance Organisers: Phase 0.
- Rohrer & Pashler (2007), Interleaving & Spacing: Phase 6.
- Corbett & Anderson (1994), Bayesian Knowledge Tracing: the mastery model behind the review loop.
Sources
- Bjork, E.L. and Bjork, R.A. (2011) 'Making things hard on yourself, but in a good way: creating desirable difficulties to enhance learning', in Gernsbacher, M.A., Pew, R.W., Hough, L.M. and Pomerantz, J.R. (eds.) Psychology and the Real World: Essays Illustrating Fundamental Contributions to Society. New York: Worth Publishers, pp. 56–64.
- Cepeda, N.J., Pashler, H., Vul, E., Wixted, J.T. and Rohrer, D. (2006) 'Distributed practice in verbal recall tasks: a review and quantitative synthesis', Psychological Bulletin, 132(3), pp. 354–380.
- Corbett, A.T. and Anderson, J.R. (1994) 'Knowledge tracing: modeling the acquisition of procedural knowledge', User Modeling and User-Adapted Interaction, 4(4), pp. 253–278.
- Ebbinghaus, H. (1885) Über das Gedächtnis: Untersuchungen zur experimentellen Psychologie [Memory: A Contribution to Experimental Psychology]. Leipzig: Duncker & Humblot.
- Galinsky, A.D. and Moskowitz, G.B. (2000) 'Perspective-taking: decreasing stereotype expression, stereotype accessibility, and in-group favoritism', Journal of Personality and Social Psychology, 78(4), pp. 708–724.
- Kornell, N., Hays, M.J. and Bjork, R.A. (2009) 'Unsuccessful retrieval attempts enhance subsequent learning', Journal of Experimental Psychology: Learning, Memory, and Cognition, 35(4), pp. 989–998.
- LaBar, K.S. and Cabeza, R. (2006) 'Cognitive neuroscience of emotional memory', Nature Reviews Neuroscience, 7(1), pp. 54–64.
- Paivio, A. (1971) Imagery and Verbal Processes. New York: Holt, Rinehart and Winston.
- Pekrun, R., Goetz, T., Titz, W. and Perry, R.P. (2002) 'Academic emotions in students' self-regulated learning and achievement: a program of qualitative and quantitative research', Educational Psychologist, 37(2), pp. 91–105.
- Pressley, M., Wood, E., Woloshyn, V.E., Martin, V., King, A. and Menke, D. (1992) 'Encouraging mindful use of prior knowledge: attempting to construct explanatory answers facilitates learning', Educational Psychologist, 27(1), pp. 91–109.
- Richland, L.E., Kornell, N. and Kao, L.S. (2009) 'The pretesting effect: do unsuccessful retrieval attempts enhance learning?', Journal of Experimental Psychology: Applied, 15(3), pp. 243–257.
- Roediger, H.L. and Karpicke, J.D. (2006) 'Test-enhanced learning: taking memory tests improves long-term retention', Psychological Science, 17(3), pp. 249–255.
Read more
- Why Norudit is one fixed sequence rather than a pile of features: A Sequence, Not a Toolbox
- The whole method, run yourself with nothing but paper: The Norudit Method, By Hand
- A walkthrough of every tool inside the academy: Inside The Norudit Academy
- Why explaining an idea simply is the one test you can't fake: The Test You Can't Fake