The Commonplace Garden

A repository for collected material and personal elaboration, held in the same space.

Vault structure

The Commonplace Garden lives in a folder on your file system. It has two subfolders, and only two.

repository/ holds all living thought. Notes at any stage of development: a single collected quote, a quote with reaction, an elaboration in progress, a finished synthesis. No obligation of completeness or coherence. Notes take whatever form and length they take — one line, three pages, a quote followed by two paragraphs of reaction. They are not broken into atomic units. A note that naturally contains several related thoughts stays as one note.

archive/ holds finished products. A published essay, a delivered chapter, a post that went out, a manual in its final version. Closed material that no longer changes. When something in the repository reaches a definitive form that will not be touched again, it moves to archive.

Everything you are actively working with or might return to lives in repository. Everything that has reached a definitive form and will not change lives in archive. When in doubt, leave it in repository.

Minimal metadata

Each note in the repository carries a frontmatter block at the top of the file. The block contains one required field and one optional field.

---
type: graft
source: "Ursula K. Le Guin, The Left Hand of Darkness"
---

type describes the nature of the note at the time of writing. Four values are available, drawn from a botanical metaphor:

seed — collected material with no added elaboration. A quote, a fact, a reading note, an observation. The material belongs to someone else, or is raw experience with no commentary yet.
graft — someone else's material alongside your own annotations, reactions, or commentary. The note holds both the external material and your thought about it, in the same space.
sprout — your own elaboration in progress, whether it grew from external material or from nothing. Something you are thinking through, not yet finished.
fruit — a mature, autonomous synthesis that stands on its own. It may have grown from seeds and grafts, or it may have appeared more fully formed. It does not reference external material as its primary content.

These types are not a mandatory progression. A seed can remain a seed forever. A sprout can emerge with no seed behind it. A fruit can appear without passing through the earlier stages. This is not a workflow; it is a description of what a note actually is at the moment you write it. The type in the frontmatter reflects the current state, not a destination.

source indicates provenance, when it exists: an author, a book, an article, a conversation, a direct experience. The field is omitted entirely when the note is your own elaboration with no specific external origin.

Here is how a note of each type might look in practice — four different examples to make the distinctions concrete.

A seed:

---
type: seed
source: "Walter Benjamin, The Arcades Project"
---

"Not to find one's way in a city may well be uninteresting and banal.
It requires ignorance — nothing more. But to lose oneself in a city —
as one loses oneself in a forest — that calls for quite a different schooling."

(Project N, on the method of this work — passage on the collector's posture)

A graft:

---
type: graft
source: "Walter Benjamin, The Arcades Project"
---

"Not to find one's way in a city may well be uninteresting and banal.
It requires ignorance — nothing more. But to lose oneself in a city —
as one loses oneself in a forest — that calls for quite a different schooling."

Benjamin's distinction is between orientation (functional, navigational) and
genuine lostness (perceptive, receptive). I think this maps onto reading: you
can navigate a book — following the argument, marking the conclusion — or you
can lose yourself in it, where the argument stops being the point and the texture
becomes the thing. The second kind of reading produces different notes: looser,
more associative, less organized.

A sprout:

---
type: sprout
source: "Prompted by Benjamin"
---

There might be two modes of reading that call for different note-taking.

Navigational reading: you are trying to extract an argument, a fact, a method.
Notes should be precise and organized. This is the reading that serves research,
citation, knowledge-building.

Receptive reading: you are not trying to extract anything, but something is
happening to you as you read. Notes here should be loose — reactions,
associations, images, half-questions. Trying to make them precise kills the
thing you are trying to preserve.

(Unresolved: is the distinction in the text or in the reader's stance?
Can the same book be read both ways? Almost certainly yes.)

A fruit:

---
type: fruit
---

# Reading as navigation vs reading as lostness

All reading involves extraction, but what you extract depends on how you arrive.

When you read to build knowledge — to master an argument, follow a method,
gather evidence — the text is a map, and efficiency is a virtue. You move
through it, take what you need, and leave.

When you read to be affected — to find language for something half-felt,
to encounter a mind unlike your own, to slow down inside a problem — the
text is a place, not a path. What you find there cannot be predicted in advance,
and the notes you take will be strange: reactions, images, questions that do
not yet know what they are questions about.

These are not two kinds of books. They are two stances a reader can take toward
any book. And they call for different instruments.

Titles

The title of each note is the primary tool for finding it again. It should describe what the note is actually about, in your own natural language, specifically enough that scanning a list of titles is sufficient to recognize the content without opening the file.

The practical test: if you see the title in a list three months from now, with no other context, can you tell what is in the note? If yes, the title is good. If no, it needs more specificity.

A few comparisons, from more opaque to more useful:

Opaque	More specific
`Benjamin note`	`Benjamin on navigational vs receptive reading`
`Writing thoughts`	`On the difference between note-taking as extraction and as reception`
`Meeting L.`	`Conversation with L. on restructuring chapter 5's argument`
`Ritual ideas`	`Rituals of attention as alternatives to productivity management`

The opaque titles were probably adequate at the moment of writing — you knew what they referred to. Six months later, they will not be. The more specific titles cost a few extra seconds to compose but save those seconds at every future encounter with the note.

Fruit notes can carry stronger, more general titles — not because they are more important, but because their content is more autonomous and self-contained. A fruit titled "The ethics of ambiguity as a design principle" stands out from a seed titled "Note on de Beauvoir, The Ethics of Ambiguity, chapter 2" without any filter or tag.

Links

Links between notes are spontaneous and never mandatory. When you are writing a note and a connection to another note comes to mind, you place the link. When nothing comes to mind, you do not. No debt, no incompleteness.

The resulting graph of connections is a byproduct, not a project. It grows passively over time. Some notes will accumulate many links because the material they contain is frequently relevant; others will have zero links and remain perfectly valid notes. You do not curate the graph, complete it, or use it as the primary way of navigating the vault.

Occasionally, when you browse the graph, you may notice unexpected clusters — several notes connected around a theme you had not consciously recognized as a theme. This is the graph's occasional value. But it is not something you manage toward. It appears, if it appears, as a consequence of writing.

Index pages

When you realize, while working, that you have several notes on the same theme, you can create a note that gathers them: an index page. It might be called "Ideas on attention" or "Notes on narrative structure" — a note whose primary content is links to other notes on that theme.

Index pages are reactive, not proactive. They emerge from need — specifically, from the need to see related notes side by side — not from an obligation to catalog every theme in the vault. When the need arises, you create the index. When it does not arise, you do not.

An index page lives in the repository like any other note. If it is never updated after the first draft, there is no debt. It remains a pointer to the notes it was created to gather, and that is enough.

This is secondary writing — writing about existing notes — and it counts as work. It is not maintenance. Creating an index page for notes on a theme you are actively developing is part of developing that theme.

The fruit as a springboard

Fruit notes — mature, autonomous syntheses that stand on their own — naturally function as starting points for future work. They contain a complete enough thought that they can be picked up at any point and developed further, or used as a foundation for something new.

They do not require a different location. Fruits stay in the repository alongside seeds and grafts. Their distinctiveness is in the title and in the type field, not in their physical location. A fruit titled "Secular ethics of doubt" stands out from a seed titled "Note on MacIntyre, After Virtue, ch. 3" without any filter — because of how it is named, not where it is stored.

Fruits also tend to attract links naturally. Because they are complete thoughts with strong, specific titles, other notes find them when you need to reference an idea they contain. Over time, well-titled fruits become the high-density nodes of the vault — the notes that other notes point toward when the theme is relevant.

Discoverability

Five channels, in order of immediacy:

Full-text search — for when you know what you are looking for. Works best with specific terms: a name, a phrase, a keyword you remember using.
Scanning titles — for rediscovering what you had forgotten you had. Works best when titles are specific enough to be recognizable at a glance.
Index pages — for seeing existing notes on a theme side by side, when you need to gather material on a specific subject.
Passive graph — for unintentional connections, consulted occasionally and without expectations. Best used when you have no specific goal and are curious what is connected to what.
Random note — for surfacing material buried by accumulation.

The first four channels share a structural limitation: they all require, to some degree, knowing or remembering what you are looking for. After months or years of deposit, good material becomes invisible simply because you do not remember having written it, or cannot find the right words to search for it.

The random note solves this at the lowest possible cost. In Obsidian, the "Random note" core plugin (included, no installation needed) opens a random file from the vault with one button. When you open the vault with no specific purpose, pressing that button two or three times brings forgotten fragments back to the surface. This is not a periodic review; it is the equivalent of flipping a physical notebook to a random page. Sometimes the note will be irrelevant. Sometimes it will trigger a spontaneous link or a new index page. The cost of the button press is near zero.

This does not make the entire repository visible. After years of accumulation, some material will remain buried regardless. But the alternatives — systematic tags, periodic reviews, exhaustive indexes — carry a cost that exceeds their benefit for most users. The Commonplace Garden accepts this as a known tradeoff, and the random note is the lowest-cost mitigation available.

What this method does not do

It does not require maintaining the system as an activity separate from writing. There are no tidying sessions, no cataloging rituals, no upkeep.
It does not impose atomization of notes. A note that contains several related thoughts stays as one note.
It does not rigidly distinguish between collected material and your own thought. The graft type formalizes the continuum: external material and your reaction to it belong in the same note.
It does not use thematic tags, dataview queries, dashboards, or any infrastructure that presents the complexity of the vault as an object to be managed.
It does not treat the link graph as a product to be built, completed, or maintained.

Operating principle

The repository is a notebook, not a database. You write in it. You do not periodically review it, catalog it, or maintain it as a system separate from writing. You return to it when you need something for work in progress. If a note is never retrieved, that is fine.

This method is not without structure. It has an opportunistic structure: the structure is not built beforehand and is not maintained as a dedicated activity. It emerges from the gestures of writing and working. A link appears when it comes to mind. An index page appears when it is needed. A fruit becomes a high-density node because of how it is written, not because of how it is classified.

Organization is a byproduct of use, not a prerequisite.