Taking Notes

How do you capture a note that remains findable, trustworthy, and useful?

A useful note is not a verbatim capture. It is a small knowledge unit with a purpose, provenance, relationships, and a next use. Capture quickly, then graduate deliberately.

Minimum Knowledge Unit

Before a note becomes permanent, give it this shape:

Field	Coaching question
Question	What were you trying to understand or decide?
Claim or observation	What changed in your understanding?
Source	Where did it come from, and when?
Confidence	Is it observed, supported, inferred, disputed, or unknown?
Relationships	What does it depend on, contradict, explain, or enable?
Canonical term	What should retrieve it? Which aliases should point here?
Value	What decision or action could it improve?
Ownership	Is this source storage, agent synthesis, or protected human synthesis?
Review	What event, evidence, or date should reopen it?

Use pen and paper when it helps you reframe instead of transcribe. Use a digital capture when speed, links, search, or provenance matter more. The medium is an instrument; the knowledge unit is the output.

tip

Start with goal or set a question to answer

Systemizing Insights

Organize notes into easily digestible summaries.

Start with the question. Capture only what helps answer it or exposes a better question.

• Key objectives
• State changes and influences
• What core concepts/principles
• Capture unfamiliar concepts/vocabulary (Record time)
• Capture topics where elaboration required
• Write down questions
• Draw connections

Semantic Tree

Build a Semantic Tree by first focusing on understanding the trunk and big branches of knowledge so you have a schema for the leaves/details to hang on.

For example, first skim chapter outlines or watch a video on double speed with loose curiosity. When you find a word or phrase you don't understand, write it down but don't dwell on it.

Once you have skimmed through everything quickly (can be multiple passes), seek to understand those unfamiliar concepts before reading or watching in depth. The Knowledge Schema method shows how to map a new domain against one you already own — accelerating which branches deserve deep attention.

Digitalisation

Digitize paper notes to evolve your digital knowledge schema

• Create a document page(s)
• Quantify gaps or value in spreadsheets
• Extract heuristics and make them memorable
• Create ANKI flashcards referring back to page here

Create digital drawings using standard components to grow and share knowledge.

1. Convert notes into components
2. Connect the dots with comments on relationships
3. Copy components to create different diagrams

Add page for the subject

• Link page to diagram
• Link to related concepts that add context

Do not create a page merely because a note exists. Promote a note only when it has a distinct reader job or improves an existing canonical page. Otherwise, keep it as source material linked to its owner.

Graduate the Note

Use five blocking questions before the note enters a shared knowledge system:

1. Ontology — what is this about? Name the entity, claim, state, or relationship.
2. Epistemology — why trust it? Preserve the source, confidence, freshness, and contradiction.
3. Taxonomy — where does it belong? Choose the owning domain, reader job, artifact type, and lifecycle.
4. Nomenclature — how will it be found? Register one canonical term and useful aliases.
5. Axiology — why retain it? Name the decision, action, or question it improves.

If axiology cannot be answered, the note may be interesting but it is retrieval noise. Keep it in capture until a use appears, or discard it. If epistemology cannot be answered, retain it as an uncertain observation rather than accepted knowledge.

tip

The better your notes the faster you'll learn.

Your Voice

AI assists note-taking at speed. The cost: notes generated by an AI assistant converge toward the average of everything it has read — not your synthesis.

An agent can re-research any topic. It cannot re-think your thought.

The discipline is knowing which notes belong to which category:

Type	Who maintains it	Characteristic
Storage	Agent-maintainable	Could have been written by anyone reading the same sources
Synthesis	Human-owned	Requires your specific prior experience to make sense of

Storage is valuable as foundation. Synthesis is irreplaceable — it is the note only you could write. When the two are mixed without distinction, agents update synthesis as though it were storage. The thinking gets averaged out over time even when each individual session feels productive.

Protect Human Synthesis

When AI contributes to your notes, build an explicit step between “generated” and “permanent”:

1. Review agent-generated summaries
2. Mark notes that accurately represent YOUR thinking as protected — off-limits to future agent updates
3. Leave the rest updatable as new sources arrive

Without this step, the system converges toward the average. With it, two compounding layers develop simultaneously: agent-speed coverage and human-owned conviction — each growing without crowding out the other.

The synthesis layer is where graduation matters most. Cross-domain connections, predictions, hard-won corrections to previous beliefs — these are the notes that cannot be recovered if overwritten. Protect them explicitly.

Proof of Done

• A realistic query retrieves the note through its question, canonical term, or alias.
• A reader can distinguish source material from synthesis and human judgment.
• The note links to its canonical owner and does not compete with it.
• Its confidence and review trigger make staleness visible.
• Its next use is a decision, action, experiment, or teaching move—not “read later.”

Context

• Meta-Learning Pipeline — The 8-step cycle where note-taking sits between Capture and Filter
• Knowledge Schema — How to map a new domain against one you own — determines which branches deserve deep notes
• Research — What research produces that needs capturing and protecting
• Matrix Thinking — Empty cells in a matrix are the research targets notes should fill
• Mission Loop — One gap, one bound, one artifact — note-taking in a single session

Links

• Neuroscience
• Cornell Method

Questions

What is the difference between notes you take and notes you actually use?

• At what point do better notes become a way of feeling productive without doing the work?
• Which note-taking habit increases your learning speed — and which one just increases volume?
• How would your notes look different if their purpose was to be read by you in 5 years?
• Which of your notes could an AI have written — and which could only you have written?
• Where is your synthesis, and is it protected from being updated into the average?