how to learn 200-300 pages of dense material in ~6/8 hours and retain 80-90% of it forever

there is a way to learn 200-300 pages of dense material in ~6/8 hours and retain 80-90% of it forever. (without ever rereading that material) it doesn't have an official name. it's not taught in schools. and it feels completely wrong while you're doing it. — your brain is not a storage system. it's a retrieval system. everything you were taught about studying optimizes for input. none of it trains working retrieval. this is why it doesn't work. Dunlosky et al. (2013) — 10 techniques, hundreds of studies, 55 pages. highlighting: LOW utility. rereading: LOW utility. retrieval practice: HIGH utility. this has been available for 12 years. 76% of students still reread as their primary strategy (Kornell & Bjork, 2007). the problem isn't effort. the wrong strategies feel like learning. the right ones feel like struggling. — why rereading specifically fails: Koriat & Bjork (2005) — fluency illusion. rereading creates familiarity. familiarity gets misread as knowing. predicted recall after rereading: 75.7% actual recall: 60.3% you stop studying because of a false signal. — what actually works: Roediger & Karpicke (2006) — close the book, write everything you remember, no looking. restudy group forgot 52% in one week. recall group forgot 13%. effect size: d = 0.83. Rowland (2014) — 159 effect sizes: Hedges' g = 0.50 for testing over restudying. Adesope et al. (2017) — 217 studies: g = 0.61 overall, g = 0.67 in classrooms. — but retrieval alone isn't the complete picture. your brain doesn't store isolated facts. it stores connected patterns. information with no connections gets pruned automatically. Bransford & Johnson (1972) showed this — same passage, schema activated vs not: nearly double the recall. context before showing text: 6.10 comprehension 8.00 ideas recalled context after showing text: 3.30 comprehension 3.60 ideas recalled no context: 2.30 comprehension 3.60 ideas recalled this is why building a map from scratch matters, not just recalling. Nesbit & Adesope (2006) — 55 studies, 5,818 participants: ◈ maps constructed from thinking: g = 0.82 integration ◈ maps merely studied: g = 0.37 integration same format. twice the effect when you build it yourself. —  and wrong maps are not a failure. (doing it wrong) Kapur (2014) — productive failure. students who attempted problems before instruction — even failing completely — showed deeper understanding than students taught directly. Sinha & Kapur (2021) — 53 studies, 12,000+ participants: g = 0.87 at high implementation fidelity. Kornell, Hays & Bjork (2009): completely unsuccessful retrieval attempts still enhanced subsequent learning vs reading the answer directly. the wrong attempt is the mechanism, not the obstacle. — Blunt & Karpicke (2014) brought it together: concept mapping with text visible: significantly worse. concept mapping from thinking: as effective as free recall. same maps. one condition required retrieval/thinking. one didn't. retrieval/thinking is the active ingredient. the map adds schema construction on top. — the full picture by mechanism: retrieval practice: g = 0.50–0.61 self-constructed maps from memory: g = 0.72–0.82 generation effect: d = 0.40 productive failure: g = 0.36–0.87 vs what most people do: highlighting: no significant comprehension benefit rereading: negligible retention, inflated confidence verbatim notes: shallower processing, worse outcomes ----------------------- how to actually apply this (TL;DR). the technique is called blind schematization. here's the full protocol. — the underlying mechanic first on why this works. your brain runs a constant filter. any new information that arrives in isolation gets binned immediately. not because you're bad at this — because your brain is doing its job (unless you take hdac inhibitors ). the only way to make something stick is to show your brain it's connected to something else. importance is never intrinsic — it's always relational. a piece of information becomes important to your brain when it has consequences, influences other things, connects to a pattern. no connections = no memory. no exceptions (literally). your job when learning is not to collect information (we talked about this). it's to find the schema — the underlying structure that makes it all make sense. once you have the schema, you can navigate the whole topic. without it, you're just accumulating disconnected facts that will be gone in 48 hours. (if you’re lucky) — the protocol. step 1 — extract 15-20 keywords before you touch the material in detail (book, video, etc): pull out 15-20 keywords that represent the main ideas of what you're learning. you can skim the material quickly to get them. you can generate them from memory if you've got knowledge. you can ask an LLM to give you the keyword list for the topic. that last one is fine — and actually good (use claude projects for this). the keyword generation step does almost nothing for your learning. cut it to seconds so you can spend your time on what actually matters. what you now have: a list of disconnected pieces. this is the puzzle dumped out of the box. — step 2 — pick one keyword and ask why it's important look at your list. pick whichever word feels most central. doesn't matter which. (as we said this ideally should be wrong) now ask: why is this important? importance is only ever defined by relationships. something is important because it influences X, causes Y, is required for Z. if you can't answer why it's important, you don't have the right surrounding keywords yet. add them. (just by thinking the importance of it we are already doing more than rereading) — step 3 — start drawing connections. be wrong. take that keyword and the ones connected to it. draw them on paper with arrows. (this ideally needs an iPad or Whiteboard, infinite canvas basically) rules: ◈ keywords only — no full sentences, no bullet points ◈ arrows show direction and relationship (but don’t add things to the arrows) ◈ it doesn't have to be right ◈ it doesn't have to look good this is not about accuracy. this is about getting the puzzle pieces on the table so you can see them. if you try to make it correct from the start you will freeze. you won't know where to begin. the perfectionism kills the processing. a wrong schema that exists is more useful than no schema (we saw this on nesbit 2006). you can build on wrong but you can't build on nothing. — step 4 — keep adding and connecting add keywords. draw connections. don't overthink it. repeat. your map will grow. it will get messier. arrows will cross. groupings will feel off. that's what it is about. keep going. the mental effort you feel during this — the back and forth, the slight frustration, the active searching — that is your brain learning. (even without even knowing the correct yet) that is not a sign you're doing it wrong, you’re doing it right. if it feels easy and passive, you are doing it wrong. — step 5 — when it gets overwhelming, stop and rebuild at some point the map will feel like too much. you'll look at it and struggle to follow it. redraw the map — but cleaner, organized and making more sense to anyone: ◈ regroup: find keywords that belong to the same cluster and group them ◈ rearrange: move things so the arrows don't cross as much ◈ reconnect: remove connections that don't make sense anymore, add ones that do this step — the rebuilding — is where the majority of learning actually happens. not the reading. not the initial mapping. the reorganizing. because reorganizing forces your brain to re-evaluate every relationship. every decision you make while cleaning the map to make easy and understandable is your brain strengthening that connection. (and it feels good) — step 6 — now watch/read the content and repeat the cycle consume a bit more → map it → simplify when it gets messy → repeat. within one session on a complex topic you might rebuild your map 7 or 8 times. (and with 5 minutes differences you might see another way to make it even easier and understandable, this is learning) every rebuild makes it cleaner, denser, more accurate. at the end of this process you should be able to look at your map and feel: this makes sense. i could reproduce most of this from memory. the relationships feel intuitive. when it reaches that point — you understand it. not "i've read it." understood. — what most people get wrong. most people over-consume and under-digest. they read 40 pages then try to review. they watch a 2-hour lecture and take notes the whole way through. they highlight paragraphs and re-read them. none of that is digestion. it's all consumption. the ratio should look more like this: 45 minutes of active mapping and rebuilding for every 15 minutes of consuming new information. if you're spending most of your learning time passively reading, you're overfeeding a system that can't absorb what you're giving it. one more thing. (tips and must-knows) your notes are not reference material. your notes are thinking tools. they exist to help your brain form connections — not to capture the information for later. (and if you studied you know how scary is too look at old notes and say wtf is this, nobody looks nor wants to look at old notes, this is not storage) this means: ◈ no full sentences ◈ no transcription, no copy paste ◈ the uglier and messier the better, as long as you can follow it ◈ word count going up = processing going down. always. (this has been shown on Mueller & Oppenheimer 2014) it also means you don't need to make them pretty. pretty notes are a sign you optimized for the wrong thing. — for genuinely isolated facts that won't connect to anything else: flashcards. spaced repetition. that's it. but most things you think are discrete isolated facts are only isolated because you haven't built enough of the schema yet. as the map grows, the connections appear, and the easier it becomes. — the full system in one line (this is not enough, read the article!): consume less → connect more → rebuild until it's clear → then consume and repeat. so don't waste your nootropics, and learn how to learn first (all of them are useless without this) credits go to keyloz one of the best threads I have read definitely worth the time reading it