Zettelkasten Method
The Zettelkasten (“slip box”) is a knowledge management system developed by German sociologist Niklas Luhmann, who used it to write over 70 books and 400 academic papers. Cortex adapts this method for the digital age, adding AI-powered linking and semantic search.
The Three Rules
1. One Idea Per Note (Atomicity)
Each note should contain exactly one idea. If you find yourself writing “and” to connect two thoughts, split them into two notes.
Test: Can you summarize this note in a single sentence? If not, it’s not atomic enough.
2. Write In Your Own Words (Elaboration)
Never copy-paste. Rewriting forces you to process the information, which dramatically improves understanding and retention. This is backed by research on the generation effect — information you generate yourself is remembered better than information you passively read.
3. Declarative Titles (Findability)
Titles should be claims, not topics.
| Bad Title | Good Title |
|---|---|
| ”Spaced repetition" | "Spaced repetition intervals should grow exponentially" |
| "React hooks" | "useEffect cleanup runs before every re-execution, not just on unmount" |
| "Meeting notes" | "Team decided to use SQLite over Postgres for embedded deployment” |
Note Types in Cortex
Cortex extends the classic Zettelkasten with typed notes:
| Type | Purpose | Example |
|---|---|---|
| fact | Objectively verified information | ”SQLite supports up to 281 TB databases” |
| insight | A connection you’ve made | ”Zettelkasten linking is similar to how neurons form synapses” |
| decision | A choice and its rationale | ”Chose Haiku over Sonnet as default model for cost efficiency” |
| experience | Something you observed | ”Users abandoned the app when onboarding took > 30 seconds” |
| belief | A working hypothesis | ”I believe graph-based search outperforms vector-only search” |
| code_finding | Technical discovery | ”FastAPI’s Depends() creates a new instance per request” |
| synthesis | Higher-order note combining others | ”Three patterns emerge from my notes on learning: …” |
Note States
Notes in Cortex have lifecycle states:
- active — Default state. The note is live and searchable.
- contested — Another note or the AI has challenged this note’s claims. Needs review.
- verified — Manually confirmed as accurate.
- archived — Removed from active search but preserved.
The Importance Score
Every note has an importance score from 0.0 to 1.0:
- 0.9–1.0: Critical, foundational knowledge. Should never be forgotten.
- 0.7–0.8: Important findings and decisions.
- 0.5–0.6: Standard notes.
- 0.3–0.4: Minor observations.
- 0.0–0.2: Low-priority, may be pruned over time.
Importance affects:
- Recall ranking — Higher-importance notes surface first in search results
- Decay resistance — Important notes decay slower during pruning
- Flashcard priority — More important notes are reviewed more frequently
Building Your Zettelkasten
Start small. Don’t try to import thousands of notes at once. Instead:
- Read something — an article, a book chapter, a talk
- Write 3–5 atomic notes capturing the key ideas in your own words
- Link them to existing notes (Cortex does this automatically, but review the links)
- Revisit weekly — use
analyzeto find clusters and write synthesis notes
After a month, you’ll have a knowledge graph that genuinely reflects how you think. After a year, it becomes an intellectual partner.