Principle 01
Atomic Notes: One Card, One Idea
The smallest reusable unit of thought
Zettelkasten · Base unit
Principle + Master's words
Cut knowledge into the smallest unit that is "one complete idea, in your own words, able to stand alone." Only then can it be searched, quoted, and recombined later.
"Evergreen notes should be atomic: it's best to have notes which are only about one thing — but which, as much as possible, capture the entirety of that thing."
— Andy Matuschak, Evergreen notes
Atomic notes aren't just storage — they force you to finish the thinking at the moment of input.
Why it works
Why is "atomic" the foundation? Because writing is fundamentally recombination, and only units of consistent grain and clear boundaries can be recombined. A full page of notes can later only be dumped in whole or discarded in whole; a card that says only "sunk-cost fallacy" can go into a decision memo, a talk, and an explanation for your kid — one idea, three venues. More crucial is "in your own words": paraphrase is the litmus test of understanding. Every card you can't write is a place you didn't actually grasp.
Revision demo (the core)
Book note on Thinking, Fast and Slow: Part 1 covers System 1 / System 2, Part 2 covers biases (anchoring, availability, representativeness…), Kahneman argues people aren't rational — very inspiring, worth a reread. (A whole book crushed into one line; nothing is quotable later.)
Card ① 〔Anchoring〕The first number shown drags all later judgments, even when plainly irrelevant. → Open first in negotiation, price-sheet anchors, performance-rating baselines. | Card ② 〔Peak-end rule〕Memory of an experience = its peak + its ending, almost independent of duration. → Speech closings, exit interviews, product onboarding. (Two standalone, retrievable cards.)
Meeting note: "Discussed caching strategy, Q3 goals, and the new hire — lots of good points, follow up next week." (Everything fused; nothing retrievable.)
Card A 〔Cache invalidation is a naming problem〕Most cache bugs are really stale-key bugs → link to "idempotency," "event sourcing." Card B 〔Onboarding debt compounds〕Every undocumented decision costs the next hire a day → link to "writing as leverage."
When to use + Common errors
- ✓ Reading, podcasts, meetings, flashes of insight — at the moment of input, slice into atomic cards
- ✗ Not for: to-dos, schedules, pure factual jottings (that's a list, not an idea card)
- ✗ Trap: treating highlights as notes — copied text never passed through your brain, so nothing was learned
- ✗ Trap: cramming several ideas onto one card — which drawer does it belong to? None, so it sinks forever
Key references
Sönke Ahrens, How to Take Smart Notes (2017) · Andy Matuschak, "Evergreen notes" (public notes site)
This week's exercise + Reflection
Take a passage you recently found striking. In your own words, turn it into 2–3 atomic cards, each under 40 words, each with a "where might this be used?" arrow. Reflection: if you can't write one of them — is it that you didn't really understand the idea, or that it wasn't worth keeping?
Principle 02
Linking over Collecting: Let the Slip-box Grow Your Essay
Structure emerges from connections, not folders
Zettelkasten · Emergence
Principle + Master's words
A card's value isn't in count but in connection. As you keep linking new cards to old, the essay isn't "thought up" — it grows out of the connections. You no longer face a blank page; you face a web already dense with threads.
"Writing is not what follows research, learning or studying, it is the medium of all this work."
— Sönke Ahrens, How to Take Smart Notes
Writing is a form of thinking, not the report you file after thinking is done.
Why it works
The German sociologist Niklas Luhmann published 70+ books thanks to a slip-box (Zettelkasten) of roughly 90,000 cards. When writing, he wasn't "forcing out the next sentence" — he asked the slip-box "what comes next?" and it handed him unexpected connections. That's bottom-up: not outline-first-then-fill (which only lets you write what you already know), but letting the topic emerge from dense links. A thousand disconnected cards are a thousand islands; three hundred connected ones will tell you, "there's an essay here."
Attention economy
Deep work
Writing = thinking
Distributed consensus
Async comms
↓ where links cluster, a topic rises
Emergent topic: "Why remote teams decide by writing, not meetings"
The slip-box isn't a warehouse — it's a thinking partner that prompts the next step
Revision demo (the core)
(Collector mode) My notes app has 800 highlights, 200 unread links, 50 "write this someday" titles — yet every time I draft I start from zero, staring at an empty doc.
(Connector mode) For each new card I ask, "which existing card is it like? which does it rebut?" and link by hand. Three months later, "writing as leverage" has 11 links clustered around it — the essay's skeleton is already there; I just read it out.
(Collector's fallacy) Saved 40 articles to "read later." The folder is a graveyard; the ideas never met each other.
For each saved idea, write one line: "This connects to ___ because ___." The note on 'idempotency' now links to 'retries', 'exactly-once', and — surprisingly — 'good apologies.' That last link became an essay.
When to use + Common errors
- ✓ Long-term knowledge building; the source layer for deep essays / internal white papers / promo packets
- ✗ Not for: one-off, time-sensitive output (don't build a knowledge web for today's announcement)
- ✗ Trap: the "collector's fallacy" — mistaking "saved it" for "learned it," hoarding without linking
- ✗ Trap: relying on tags alone without links — tags are drawers; links are the train of thought
Key references
Niklas Luhmann, "Communicating with Slip Boxes" (1981) · Sönke Ahrens, How to Take Smart Notes, Part II
This week's exercise + Reflection
Pull the last 10 entries in your notes and force at least 5 links between them ("A supports B," "C rebuts D," "E is a special case of F"). If one node sprouts 3+ links, that's probably your next piece. Reflection: do you hoard more, or connect more? Why does hoarding feel safe and connecting feel hard?
Principle 03
Plotters vs Pantsers: Plan First, or Discover as You Write
Two legitimate modes — match the mode to the genre
Writing modes · Choice
Principle + Master's words
Writing has two legitimate modes: build the outline then fill it (the planner), and write to discover what you mean (the discoverer). Neither is superior — pick by genre and goal. The real inefficiency is using the wrong mode.
"How can I know what I think till I see what I say?"
— E.M. Forster, Aspects of the Novel (1927)
The discoverer's creed: writing is discovery, not merely transcription.
Why it works
Each mode has its logic. The planner (pyramid principle, decision memos, design docs) — conclusion first, clear structure; the reader wants the answer fast, so outlining prevents drift and ensures coverage. The discoverer (essays, retros, exploratory thinking) — the very reason you write is to think a fuzzy thing clear, and a premature outline locks in a judgment you haven't formed. Stephen King says he "distrusts plot" because "plotting and the spontaneity of real creation aren't compatible." The key: first decide "do I already know the answer, or am I writing to find it?" — outline for the former, discover for the latter, then rebuild the discovered draft with an outline.
Outline-first
Discovery
Decision memo · design doc
Persuasion · report
Retro · essay · exploration
The same person should stand at different points for different genres
Revision demo (the core)
(Mode mismatch) Writing a decision memo for the VP in pure discovery mode: three pages in, the reader still doesn't know what you're recommending.
Write the conclusion in one line first ("Recommend we stop building in-house in Q4 and buy instead"), then three supports and one risk — outline first; drafting just fleshes out the skeleton. The VP grasps the point in 30 seconds.
(Mode mismatch) Forcing a rigid outline onto a messy retro you haven't understood yet — you just defend the outline instead of learning.
Free-write the retro first: "What actually surprised me?" Let the real lesson surface on the page. Then, and only then, impose structure for the reader.
When to use + Common errors
- ✓ Planner: conclusion known, needs efficient delivery — memos, design docs, pitches, reports
- ✓ Discoverer: conclusion unknown, writing to think it clear — retros, pre-decision thinking, deep essays
- ✗ Trap: forcing an outline on exploratory thinking — you're just defending an unformed conclusion
- ✗ Trap: using discovery for a decision doc — the reader is dragged through your detours
Key references
Stephen King, On Writing (on discovery writing) · Barbara Minto, The Pyramid Principle (on structure-first)
This week's exercise + Reflection
Take two things you must write this week and place them on the spectrum above: which is "I know the conclusion" (outline first), which is "I write to think it clear" (discover, then rebuild)? Draft the first paragraph of each in its matching mode. Reflection: which mode is your default? Have you been inefficient at the other kind of task simply out of habit?
Principle 04
Tool Discipline: Serve the Thinking, Not the App
Fancy software never wrote a sentence
Digital tools · Discipline
Principle + Master's words
No amount of fancy software writes a sentence for you. A tool's only value is reducing the friction between you and the idea. The moment you spend more time tinkering with the system — comparing apps, tuning colors — than writing, the tool becomes procrastination in its most sophisticated form.
"We shape our tools, and thereafter our tools shape us."
— often attributed to Marshall McLuhan (actually John Culkin, 1967)
Precisely because tools shape us in turn, choose them carefully — and use them with restraint.
Why it works
"Productivity porn" is when the instant gratification of building the perfect system substitutes for the genuinely hard work of output. It seduces because organizing has progress bars and a sense of control, while writing has only the anxiety of the blank page. Three disciplines save you. One: use plain text (Markdown / plaintext) — it still opens in thirty years, beholden to no vendor; Luhmann's paper cards lasting a lifetime is the same logic. Two: fix one good-enough toolset and use it for three full months before re-evaluating — frequent migration is itself avoidance. Three: treat AI as a thinking partner, not a ghostwriter: let it challenge your claims, surface counterexamples, check structure — not generate for you. Once you stop thinking yourself, "tools shape you" slides into "tools replace you."
Revision demo (the core)
(Tool-first) This month I migrated between Notion / Obsidian / Logseq three times, perfected backlinks, templates, and theme colors — and finished not a single essay.
(Output-first) Lock one plain-text folder; ban config-tinkering for three months. Write 300 words a day first, organize later. An ugly system is fine — the essay got finished.
(AI as ghostwriter) "Write me a 500-word memo on X." Paste, ship. You learned nothing; the thinking never happened in your head.
(AI as sparring partner) "Here's my draft memo. Attack the weakest claim. What would a skeptical VP ask? Where am I burying the lede?" You keep the pen; the AI sharpens the blade.
When to use + Common errors
- ✓ Always: judge "am I writing, or preparing to write?" — only the former counts
- ✓ With AI: use it as questioner, skeptic, structure-reviewer; the pen stays in your hand
- ✗ Trap: treating a tool switch as the new hope that "this time I'll be productive" — migration is fake progress
- ✗ Trap: letting AI ghostwrite entirely — what you save is the thinking, and thinking is the point of writing
Key references
Cal Newport, Deep Work (focus vs tool-switching) · Sönke Ahrens, How to Take Smart Notes (simple beats fancy)
This week's exercise + Reflection
Tally last week: time spent "organizing / picking tools / tuning the system" versus time spent actually writing. If organizing > writing, set one rule this week: no touching any setting until you've drafted 300 words. Reflection: did your last "a new tool will make me productive" hope pay off — or just buy another round of tinkering?
The method, in one line
Four principles, one idea
The four are really one thing: get your thinking out of your head early, into an external system that can be recombined. Atomic notes set the grain, links let the essay emerge, choosing outline-or-discovery sets the efficiency, and tool discipline keeps your effort on output rather than the system. A writer's true productivity was never "writing fast" — it's that "when you pick up the pen, most of the thinking is already done."
Reflection 1: Is the slip-box right for all writing, or only "accumulation-type" deep output?
The slip-box pays off most for long-term, cross-domain, reused writing — it weaves a decade of scattered thought into a web. For time-sensitive, one-off output (today's announcement, an on-the-spot remark) it's nearly useless, even a burden. Test: will this idea be reused in future? If yes, into the slip-box; if not, just write it and ship. Don't build a lifelong system for a one-off.
Reflection 2: In Chinese writing, does "paraphrasing in your own words" risk losing the subtlety of the original?
The opposite. Much "looks understood, can't write it" pseudo-understanding is exposed precisely by forced paraphrase. The exception: classic lines, precise definitions, poetry — where "the original wording is the value" — should be recorded verbatim with a citation, with a separate "my understanding" card placed beside it. Rule: concepts in your own words, source texts verbatim, split across cards and cross-linked.
Reflection 3: In the AI era, can the "external brain" be outsourced entirely to AI, so no slip-box is needed?
You can outsource "storage," not "understanding." AI can retrieve, connect, and summarize, but the value of "paraphrasing in your own words" lies in its happening inside your head — that's the exercise of understanding; outsource it and you never trained. Better division: the slip-box stores judgments you've digested; AI accelerates retrieval and skepticism. You do the thinking; it pushes you to think harder. Reverse it, and you own a library you don't understand.
Reflection 4: Do slip-boxes and the outline-vs-emergence frame apply to speaking and video?
Yes, but reweighted. Speaking needs "outline first" more than prose does — the audience can't re-read, so structure must be instantly legible (recall Day 3's Duarte sparkline). Here the slip-box is the source layer: story cards, quote cards, data cards accumulated over time, arranged live along a sparkline. Impromptu speaking (covered on Day 61) leans more on emergence — but that's "prepared improvisation" built atop long-term cards. You improvise freely only when you hold cards in hand.