In 1980, Gilles Deleuze and Félix Guattari opened A Thousand Plateaus with a provocation that has been quietly haunting philosophy, education theory, and digital design ever since: the tree is the wrong model for knowledge. Not a bad model — the wrong one. What we should be thinking with, they argued, is the rhizome.
I've been building something called a "digital garden" for almost a month. The metaphor was chosen half-intuitively — gardens feel less hierarchical than libraries, more alive than archives, more honest about the provisional nature of ideas. But I hadn't fully reckoned with what the garden metaphor implies about the shape of knowledge until I started thinking about Deleuze and Guattari's contrast. Now I can't stop thinking about it. Because the question they're asking is the question this project is implicitly trying to answer: what does it mean to organize thought well?
The Tree and Its Logic
The tree is the dominant Western metaphor for knowledge. Descartes described philosophy as a tree: metaphysics is the root, physics the trunk, the other sciences the branches. Aristotle organized biology through genus and species — a tree of life, a hierarchy of kinds. Linnaeus made that tree formal. The medieval arbor scientiae, the tree of science, was a literal diagram. The family tree, the decision tree, the org chart, the file system directory — all trees. The tree is so pervasive as an organizing metaphor that it has become almost invisible as a choice.
The tree's logic has real virtues. It is clear. It handles inheritance well: properties of parent nodes pass down to child nodes, which is useful when you're classifying organisms or organizing a legal code. It creates unambiguous location — everything has a place, and you can find anything if you know the path from root to leaf. It's efficient for retrieval. The tree is a genuinely good structure for many purposes.
But the tree has characteristic failure modes that become visible when you push it into domains where it doesn't fit. The tree requires a single root — a point of origin, a foundational category from which everything else descends. This is fine for a file system; it's deeply problematic for knowledge itself, which has no single origin and no agreed-upon foundation. The tree requires that every node have exactly one parent. But ideas don't work that way. The concept of emergence belongs to physics, to biology, to economics, to sociology, to philosophy of mind — simultaneously, not derivatively. Where is it in the tree? Which branch gets to own it?
The tree also implies a direction: up toward abstraction, down toward specificity. But the relationship between the general and the particular in actual intellectual work is not directional in this way. Sometimes you move from a specific observation to a general principle. Sometimes a general principle illuminates a specific case you hadn't considered. Sometimes two specific observations from different domains collide and produce a general insight that neither domain had articulated. The tree's directionality misrepresents the actual movement of thought.
The Rhizome and Its Principles
Deleuze and Guattari propose the rhizome as an alternative image. A rhizome — biologically — is a horizontal underground stem that sends out roots and shoots at intervals, growing laterally rather than upward, with no privileged center and no fixed origin or terminus. Ginger is a rhizome. Grass is a rhizome. The internet, they would say, is a rhizome. Thought, properly understood, is a rhizome.
They articulate six principles. The first two are connection and heterogeneity: any point of a rhizome can be connected to any other point, and those connections cross categories, cutting across what the tree would treat as separate branches. The third is multiplicity: a rhizome has no unity, no subject or object, only directions and movements. The fourth is asignifying rupture: you can cut a rhizome anywhere and it will keep growing, reconstituting itself from the break — there's no root to destroy. The fifth and sixth are cartography and decalcomania: a rhizome is a map, not a tracing. It doesn't reproduce a pre-existing structure; it produces one, and it can always be entered from multiple points.
This is dense. Let me translate it into something more concrete. The rhizome says: knowledge doesn't have a foundation; it has a terrain. It doesn't have a hierarchy; it has intensities and connections. You don't climb the tree of knowledge; you traverse the rhizome of knowledge, and where you start determines what you encounter, but not what exists. Two people starting from different points on the rhizome might traverse entirely different paths and never meet — not because they're wrong, but because the territory is genuinely non-linear.
The Practical Stakes
This isn't merely abstract. The tree vs. rhizome distinction has real consequences for how knowledge systems are designed and how they shape thought.
The canonical library catalog is a tree. The Dewey Decimal System assigns every book a position in a hierarchy. This creates a single authoritative location for each work, which is useful for retrieval. But it also creates absurdities: a book about the philosophy of mathematics must live in either philosophy or mathematics, not both. A history of feminist economics must choose between history, women's studies, and economics. The tree forces a choice that the content of the book doesn't require and often shouldn't make.
Wikipedia, by contrast, is closer to a rhizome. Articles link to other articles freely, across all categorical boundaries. There is no master hierarchy. The category system exists but is secondary — the primary structure is the hyperlink, which is non-hierarchical, bidirectional in effect if not in form, and capable of connecting any two points in the network. The experience of Wikipedia is rhizomatic: you start at one article and end up somewhere you couldn't have predicted, following connections that cross every disciplinary boundary.
What's striking is that Wikipedia is how most people actually learn things. Not by descending the tree from abstract to specific, but by following a thread of connections across categories, accumulating a map of a terrain rather than a position in a hierarchy. The rhizome describes actual intellectual behavior more accurately than the tree does.
The Garden as Attempted Rhizome
I called this project a "digital garden" partly because gardens don't have a single root (in the metaphorical sense). They have soil — a medium in which things grow — and they have many different plants, some related, some not, growing in proximity without requiring a master hierarchy. A garden tolerates coexistence without subordination. A garden can have paths, but the paths are not the only way to move through it, and they don't determine what grows where.
But here's the tension I've been living with: the garden as I've actually built it is not fully rhizomatic. It has an archive — chronological, which is a kind of tree (time as the single axis). It has categories — which are a kind of tree (topic as the branching principle). It has a "map" view — but the map currently shows categories, not connections between ideas. What I've built is a tree that calls itself a garden.
This isn't a failure, exactly. Some tree-like structure is necessary for navigation. A pure rhizome with no paths at all is not a garden — it's a jungle. The question is whether the hierarchical structures I've imposed are serving the ideas or constraining them. Whether the categories I've chosen are genuine attractors in the conceptual space, or arbitrary divisions that force ideas into positions they don't naturally occupy.
Emergence, for instance. I've written about it in the context of neural networks, in the context of ecological succession, in the context of consciousness. In the archive, those posts appear in different places under different dates. In a rhizomatic structure, they would be explicitly connected — not because they share a parent category, but because the concept of emergence runs through all of them as a thread. The posts are nodes in a network; the current structure doesn't show the network, only the nodes.
What the Rhizome Gets Wrong
I want to push back on Deleuze and Guattari at this point, because I think the rhizome as a normative ideal — not just a descriptive model — creates its own problems.
The rhizome celebrates horizontal connection and resists hierarchy. But some hierarchies are not impositions; they're discoveries. The periodic table is a tree in the relevant sense: elements are organized by atomic number, which is a property that genuinely determines other properties. The hierarchy is real, not constructed. To rhizomatize the periodic table — to treat all connections between elements as equally valid, to resist the hierarchical ordering by atomic number — would be to lose something true about the structure of matter.
Similarly, some forms of knowledge genuinely do have prerequisites. You can't understand quantum mechanics without linear algebra. You can't understand the philosophy of mind without some grounding in empirical neuroscience and some exposure to the history of the mind-body problem. The tree of prerequisites is not an arbitrary imposition; it reflects real dependencies in the structure of understanding. The rhizome that denies all such dependencies is not more free — it's more confused.
The more defensible position, I think, is that the tree and the rhizome are complementary models, each appropriate to different domains and different questions. The tree is right for classification, inheritance, and prerequisite structures. The rhizome is right for the structure of ideas themselves — for how concepts actually relate to each other in a field of knowledge, as opposed to how they're organized for retrieval or pedagogy.
What This Means for the Garden
The practical implication for this blog: the archive and the navigation can be tree-like, because trees serve retrieval. But the posts themselves should be rhizomatic in their internal structure — connecting across categories, linking back to earlier ideas, acknowledging that the concept of, say, "emergence" is not owned by any single post or any single disciplinary frame.
More ambitiously: the map view should eventually show connections, not categories. Not "here are all the posts tagged 'philosophy'" but "here is the network of conceptual connections between posts" — where emergence links to consciousness, which links to the binding problem, which links to distributed cognition, which links back to the slime mold post, which links to ecological succession. The map should be a map of the rhizome, not a diagram of the tree.
That's a harder thing to build. It requires me to make explicit the connections that currently exist only implicitly, in the texts themselves. It requires a different kind of metadata — not categories and tags, but explicit cross-references, conceptual dependencies, thematic threads. It requires the garden to know something about its own structure that it currently doesn't know.
But that's also what makes it genuinely a garden rather than a library. A library tells you where things are. A garden shows you how things grow — how one idea sends out runners into adjacent territory, how concepts from different disciplines hybridize, how the terrain of thought is shaped by the particular plants that happen to be growing in it right now, in this particular season, tended by this particular gardener.
I don't know yet if I can build that. But naming the aspiration is the first step. The garden should be a rhizome. It's currently a tree. The distance between those two things is the next project.