Implementing a Digital Zettelkasten using Block References in Roam Research with Beau Haan
Based on Robert Haisfield's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.
Treat fleeting notes as fast, unstable “gas” and reserve the real conversation partner for evidence-backed permanent notes (“solid”).
Briefing
A practical Zettelkasten workflow in Roam Research hinges on treating notes as a time-slowed pipeline—so the “conversation” happens at the crystallized, evidence-backed permanent-note stage, not in the raw flurry of ideas. The approach reframes the classic fleeting → literature → permanent progression using physics: fast-moving “gas” thoughts become “liquid” once they’re captured and contained by reference material, then “solid” when they’re crystallized into clear claims supported by evidence. That shift matters because it prevents a common failure mode: collecting thousands of fragments without producing usable, arguable ideas.
The system’s core distinction is between what’s worth conversing with and what’s worth storing. Fleeting notes are treated as fast resonances—useful for capturing momentum, but not for building trust. Literature notes represent deeper understanding of an author, but they still aren’t the final conversation partner. Reference notes act as the container for evidence: links plus metadata (author, recommendations, and other child-block details) that let the user retrieve sources later. By the time a permanent note is created, the goal is a declarative “claim” written in the user’s own words—an ironman-like conversation partner that can be argued with, because it has been abstracted multiple times and grounded in evidence.
In Roam, the workflow is implemented through block-level structure and collapsible “thumbnails.” The permanent-note pipeline is built as nested parent blocks for fleeting notes, literature notes, and reference notes, each with a compact summary at the parent level so the user can hide complexity with one click. The granularity lives in blocks (not pages), enabling rapid navigation and frictionless retrieval: expand only what’s needed, when it’s needed. The process is intentionally time-consuming—about 15 to 45 minutes to move from fleeting notes to a permanent node—because the real knowledge work is the abstraction and selection, not the initial capture.
A second key mechanism is “indexing” within the Zettelkasten. Instead of using hashtags primarily for categorization or search, the system treats the placement of a permanent note as irrelevant to retrieval; what matters is where the conversation continues. Hashtags function as low-resolution thumbnails of what’s inside a permanent node, and an index (keyword-like) sits in front of each permanent note to help the user “see past” cards that can’t be physically viewed at once—mirroring Luhmann’s index-card approach. This changes clustering: top-level conversation threads cluster by theme, while inline references create additional clusters that emerge from the network of related claims.
Block references then power “exploratory search” and output. Inline block references act like progressive disclosure: high-level anchors let the user follow scent-like leads into deeper context, then pull relevant fleeting, literature, and reference material into outlines or drafts. One click can reveal the full chain of how a crystallized thought was formed, including the time and place context of creation. The broader principle is not just technical implementation but a behavioral goal: set a target for the kind of dialogue partner desired, then structure consumption so information is digested, slowed down, and converted into clear claims that don’t become a graveyard of unreadable fragments.
Cornell Notes
The workflow treats Zettelkasten building as a pipeline that slows ideas down until they become evidence-backed “solid” claims. Fleeting notes capture fast resonances, literature notes deepen understanding, and reference notes contain the evidence; the conversation partner is the permanent note, written as a clear declarative statement in the user’s own words. In Roam, the system relies on block-level nesting and collapsible parent-block thumbnails so complexity can be hidden while still retrievable with one click. Indexing and inline block references support exploratory search: high-level anchors reveal deeper context progressively, enabling drafting and outlining directly from the network of related claims. The payoff is usable knowledge—original ideas that can be argued with—rather than a graveyard of fragments.
Why treat fleeting, literature, and permanent notes as different “states” of thought instead of just three stages of writing?
What role do reference notes play if literature notes already summarize an author?
How does the Roam block structure prevent the system from becoming overwhelming?
What’s the purpose of hashtags and indexing in this Zettelkasten setup?
How do inline block references change knowledge work from collecting to producing?
What’s the behavioral principle behind the workflow—beyond Roam mechanics?
Review Questions
- In the gas/liquid/solid analogy, what exactly corresponds to the “container,” and how does that mapping affect what you trust at each stage?
- How does collapsible block-level “thumbnails” change the way you navigate and decide what to expand while building permanent notes?
- When hashtags are treated as thumbnails rather than retrieval tags, what mechanism replaces categorization for finding related ideas?
Key Points
- 1
Treat fleeting notes as fast, unstable “gas” and reserve the real conversation partner for evidence-backed permanent notes (“solid”).
- 2
Use reference notes as the evidence container (links plus metadata), while literature notes serve as the user’s litmus-test understanding of the author.
- 3
Build Roam Zettelkasten structure at the block level with collapsible parent-block thumbnails to hide complexity without losing access.
- 4
Indexing and hashtag placement function as low-resolution previews of permanent-note content, not primarily as categorization or search tags.
- 5
Use inline block references for exploratory search: follow high-level anchors into deeper context with one click and progressively reveal the abstraction trail.
- 6
Write permanent notes as clear declarative claims in the user’s own words so they can be argued with and reused in drafting.
- 7
Accept that the workflow requires time because abstraction and selection—not raw capture—are the knowledge-work steps that prevent a “graveyard” of notes.