Roam Tour #1: Professor Joel Chan- Zettelkasten and Evergreen Notes for Generative Thought

TL;DR

Creative knowledge work depends on synthesis—building a structured understanding of a problem space—not just collecting sources.

Briefing Cornell Notes

Briefing

Evergreen notes in Roam aren’t just a storage habit—they’re a method for building “seeds” of synthesis that can be recombined into new problem formulations and theories. Professor Joel Chan, a cognitive psychology researcher focused on human-computer interaction and tools for assisted thinking, describes a workflow aimed at creative scholarship: first constructing a rich understanding of a problem space, then using small, reusable knowledge units to generate novel insights.

Chan’s core claim is that creative knowledge work depends on synthesis, not just collecting sources. It’s not enough to gather many papers and form a vague sense of the topic; the hard part is identifying what makes a problem worth solving, why it’s hard, what past attempts missed, and which assumptions must change. Tools for finding papers and increasing “productivity” exist, but the gap is support for the later stage—turning scattered inputs into a structured understanding that can support new thinking.

In Roam, Chan uses the original “sit beside knowledge” idea from the white paper: knowledge should accumulate as a growing repository of usable, synthesized material rather than staying trapped in silos. His most important output is a set of compact evergreen notes—often prefixed with “z:” as a namespace convention—that act like building blocks for larger theories. One example comes from drafting a paper on “requirements for a system that supports synthesis.” Instead of staying at the paragraph level, he breaks claims into smaller components and then converts key observations into evergreen notes, such as a “citation statement” style claim that can be revisited, audited, and recomposed.

The reason these notes stay concise is reusability. Chan ties compactness to atomicity: if a note is too complex, it becomes harder to recontextualize, combine, and apply elsewhere. In Roam, blocks can be remixed, but that only works if each block is self-contained and designed to be recomposed—an effort that requires discipline. He also frames note titles as “API-like” concept statements, similar to how literature reviews and review papers make claims that can be remade. Each evergreen note can include evidence (screenshots, citations, and context) so it remains meaningful when pulled into new writing.

As the collection grows, Chan expects the challenge of combining many notes and emphasizes “processing” rather than endless organization. His pipeline starts with daily notes as an inbox for raw inputs (papers, books, and ideas). Items that prove useful graduate into literature/reading notes and then into evergreen notes (“seeds”), which later feed drafts and larger theory-building pages. He rejects the idea that processing is tedious janitor work; it’s knowledge work because it turns connections into actionable understanding.

Chan also describes how he manages citations: each source paper gets its own page so the context of where a claim came from remains attached. He imports citations from Zotero via a script, using a format that preserves the paper title for quick scanning while still supporting queries. For Roam users, his advice is pragmatic: don’t over-engineer queries early, let structure grow from real pain points, and migrate gradually by following where attention and interest are actually moving. The goal isn’t a perfectly curated database—it’s generating new thought as a byproduct of disciplined synthesis.

Cornell Notes

Joel Chan uses Roam to support creative scholarship by turning sources into small, reusable evergreen notes that function as “seeds” for synthesis. The workflow starts with daily notes as an inbox, then processes promising material into literature/reading notes and finally into atomic evergreen notes (often prefixed with “z:”) that are designed to be remixed into larger theories. Chan argues that concise, self-contained notes are more reusable because they can be recontextualized and recombined without freezing ideas in one context. He also emphasizes that processing is knowledge work—returning to sources, auditing claims, and making connections—rather than mere cleanup. Gradual migration and resisting early over-optimization (like complex queries) are key to making the system work in practice.

Why does Chan treat evergreen notes as “building blocks” rather than just permanent storage?

Evergreen notes are meant to be recomposed into new problem formulations and theories. Chan’s convention is to create atomic, concept-oriented notes—often prefixed with “z:” as a namespace—so each note is a reusable claim with enough context to be audited later. When drafting, he pulls these “seeds” into larger pages, similar to how writing uses citations, but with the added twist that the notes represent his own synthesized thinking rather than only quoted highlights.

What makes an evergreen note “atomic” in Chan’s approach, and why does that matter?

Chan links atomicity to reusability. If a note becomes too complex, it’s harder to recontextualize and combine with other notes, which reduces creativity and makes ideas feel “frozen.” In Roam, blocks can be remixed, but that only works if each block is self-contained and designed for recombination—an intentional discipline rather than an automatic outcome.

How does Chan structure note titles so they behave like something usable during writing?

He describes note titles as “API-like” concept statements—similar to citation statements or claims you’d see in literature reviews. The goal is that the title itself communicates a remakable claim (e.g., a statement about what’s needed for impactful research), so the note can be reused across contexts. He also fleshes notes with evidence (screenshots, basis papers, and context) so the claim remains grounded when pulled into new drafts.

What does Chan’s processing pipeline look like from raw inputs to novel thoughts?

He frames a multi-level workflow: daily notes act as an inbox for raw inputs (papers/books/ideas). From there, material becomes literature/reading notes, then graduates into evergreen notes of different types (including atomic “seeds”). Those seeds later compose larger structures—like theory pages and eventually drafts. The point is to move beyond composing from quotes alone toward new claims “on the shoulders of those giants.”

How does Chan handle citations and why does each paper get its own page?

Chan wants each source to remain tied to the context of the claims derived from it. He uses Zotero to manage papers and exports citation blocks into Roam via a script, producing a page per paper. This keeps the paper title visible for scanning while still supporting query-friendly tags. When he later cross-checks or audits a claim, he can quickly trace it back to the exact source and its bibliographic details.

What advice does Chan give new Roam users about building the system?

He recommends not regretting early changes and migrating gradually as interests evolve. He also advises against learning complex queries too early; instead, let real friction (“pain”) reveal what queries are needed. Start with a structure inspired by permanent/evergreen note ideas, then grow it organically rather than trying to set up a fully engineered system from day one.

Review Questions

How does atomicity in Chan’s evergreen notes increase reusability, and what tradeoff does it require in practice?
Trace Chan’s workflow from daily notes to evergreen “seeds” to larger theory drafts. Where does claim-auditing happen?
Why does Chan prefer paper pages for citations, and how does that choice support later synthesis and verification?

Key Points

1
Creative knowledge work depends on synthesis—building a structured understanding of a problem space—not just collecting sources.
2
Evergreen notes should be designed as atomic, self-contained “seeds” that can be recombined into larger theories and problem formulations.
3
Concise notes improve recontextualization and recombination; overly complex notes become harder to reuse and can “freeze” ideas.
4
Roam’s value for synthesis comes from disciplined block design and remixing, not from storing large pages of undifferentiated material.
5
A practical pipeline moves from daily notes (inbox) to reading/literature notes to evergreen notes, then into drafts and higher-level theory pages.
6
Citations work best when each source paper has its own page so claims can be traced back for auditing and contextual understanding.
7
New users should migrate gradually, avoid over-engineering queries early, and let real workflow pain guide system improvements.

Highlights

Chan treats evergreen notes as reusable “seeds” for synthesis—small, auditable claims that can be recomposed into new theories.

Atomicity isn’t a style preference; it’s a mechanism for recontextualization—too much complexity makes notes less alive and harder to combine.

Processing is knowledge work: returning to sources, auditing findings, and making connections turns inputs into usable understanding.

Each cited paper gets its own page so claims retain provenance, enabling quick cross-checking during writing.

Chan’s onboarding advice: don’t start with complex queries; let pain points reveal what automation and retrieval structures are actually needed.

Topics

Evergreen Notes
Synthesis Workflow
Atomic Claims
Citation Management
Roam Processing Pipeline

Mentioned

Joel Chan