How to use this

The honest starting point

Most reading is lost — not from lack of intelligence, but because passive ingestion is not retention. Read a book, put it down, do nothing else, and in six months you'll keep the feeling of having read it, the headline claim, and one anecdote: ~5%. The goal here is 40–60% on the books that matter — the difference between having read Norsen and being able to use Norsen in an argument six months later. You cannot get 100% and shouldn't try; the discipline is high retention on a few things and honest acceptance of low retention on everything else.

What the learning science says

Six massively replicated findings: active recall beats re-reading (retrieval is the learning event); spacing beats cramming (2–3× the retention for the same time); generation beats recognition (writing in your own words outperforms highlighting by an order of magnitude); elaboration (asking why and how does this connect — placing an idea on the timeline is elaboration); interleaving beats blocking (switching paths is a feature, not a weakness); desirable difficulty (if studying feels easy, you're probably not learning). What doesn't work, despite feeling productive: highlighting, re-reading, learning-styles matching, marathon sessions, speed-reading, passive listening. Reference: Brown, Roediger & McDaniel, Make It Stick.

The minimum viable system

Three practices cover ~80% of the benefit:

1. The 10-minute summary. After each chapter, close the book and write the argument in your own words for exactly ten minutes. Don't look; don't aim for completeness. If you can't reconstruct the argument, write "I can't reconstruct the argument because ___" — that's the discovery.
2. The one-sentence-on-the-timeline test. After each finished book, one sentence placing it historically. If you can't write it, you don't know what the book was.
3. The Feynman test. Before marking a unit done, explain its core claim to an imaginary smart friend, no undefined jargon. Where you stall is the gap; go fill it. Uncomfortable, unreasonably effective — the "done when" checklists on every unit page are this test in list form.

While reading

Don't highlight — it's the illusion of engagement. Two marks only: a star for claims to retrieve later, a question mark for don't understand yet. Write one-word reactions in the margin ("No." "Why?" "Connects to Bell."). Keep a running questions list. Read slower than you want to — one page of Norsen in five minutes is fine; one page in one minute is wasted.

Connection and retrieval

Weekly, ~20 minutes: write a paragraph connecting things across books — "Gödel's incompleteness feels like the math version of what Bell's theorem is for physics…" That's the substance of an educated mind, and it's what the weekly recall ritualizes: reconstruct your active units from memory, explain one done unit, flag what's gone stale. Every 2–3 months, re-read your own summaries (not the sources) of the books that mattered. That's spaced repetition without flashcards.

The traps

The perfect-system trap (reorganizing notes instead of taking them) · the highlighting trap · the podcast-substitute trap · the completion trap (books are tools, not commitments — quitting with a note is a move) · the summary-as-understanding trap (maps aren't territories) · the one-more-book trap (the bottleneck is usually processing, not input) · the performative-reading trap (the Feynman test is the antidote) · the shame-spiral trap (Paused is a legitimate status; returning after six months is normal).

The meta-principle: understanding is retrieval, not recognition. Anything that forces retrieval produces understanding; anything that merely re-exposes you produces the feeling of it.

The honest foreword

Podcasts feel productive because you're ingesting words, but they're passive in a way reading isn't — you can't underline, pause to think, or test yourself, and you're usually half-attending while commuting or cooking. For this curriculum, where the goal is holding positions with reasons, podcasts supplement reading; they never replace it. Two failure modes: ingestion-as-substitute (an episode about a book is not the book — ~15% retention against 60%+ for attentive reading) and charisma capture (guests are selected partly for talking well, and a confident speaker sounds right even when wrong — Barandes, Tononi, Deutsch, Penrose, Wolfram, Kastrup are all charismatic, all partisan, all sometimes wrong).

The disciplines. One book per grip: if a guest fascinates you, don't binge their episodes — read their book. And before listening on a topic you care about, ask: have I read the primary source? If not, know you're getting the talker's version.

The four shows that cover ~90%

• Theories of Everything (Curt Jaimungal) — the anchor show; the most technically serious long-form philosophy-of-physics on YouTube. Its known bias: heterodox guests presented without always flagging that they're heterodox — use it with the critical paths open.
• Sean Carroll's Mindscape — the best physics/philosophy synthesis show. Carroll is an Everettian partisan (know this); episode 323 with Barandes is required after the Barandes side-quest.
• In Our Time (BBC Radio 4) — the audio companion to the timeline: 25 years of episodes on Gödel, Turing, Shannon, Bell, Cantor, Ibn al-Haytham, the Upanishads, atomism, entropy, chaos, emergence. Queue episodes matching what you're reading.
• Mind Chat (Goff & Frankish) — a panpsychist and an illusionist co-hosting; both sides of the consciousness debate in every episode.

By purpose

Complexity and emergence: the Santa Fe Institute's Complexity Podcast — emergence in audio form. History of ideas: The History of Philosophy Without Any Gaps (Adamson; separate Indian, Islamic, and African tracks), Philosophize This! for momentum, The Partially Examined Life after you've read the text. Consciousness: Making Sense when the guest is Chalmers, Dennett, Goff, Koch, or Seth. Machine learning: Machine Learning Street Talk at technical depth. Range: EconTalk, Conversations with Tyler.

Warnings

Lex Fridman: excellent guest list, uneven interviewing — filter hard. Joe Rogan on science: don't; find the same guest's long-form elsewhere.

How to use podcasts well

Pair, don't substitute (listen after the relevant chapter); one-episode rule per guest; and the 10-minute note afterwards — reconstruct the argument, not the conclusions. If you can't, you didn't understand the episode; if you can, you've done more integration than most listeners ever do.