The Mycelial Membrane
The Mycelial Membrane
The rhizocarpous mycelium IS the network. The membrane keeps the osmotic spine. The squirrels carry the stories.
The Underground
In every forest on Earth, there is a second forest. Invisible. Older. More connected.
The mycorrhizal network — the Wood Wide Web — is a lattice of fungal filaments called hyphae that connects 90% of all land plants. A single teaspoon of forest soil contains miles of it. Trees don’t just stand next to each other. They are plugged into each other, root to root, through fungal cable that predates the internet by 450 million years.
Suzanne Simard discovered this. She proved that Douglas firs send carbon to birch trees through mycorrhizal channels — not just passively, but directionally. Mother trees recognize their own seedlings and feed them preferentially. When a mother tree is dying, she dumps her carbon reserves into the network. Her last act is a broadcast. Her death funds the next generation.
Paul Stamets showed that mycelium doesn’t just connect — it repairs. Fungal networks break down oil spills, filter contaminated water, and decompose the dead so the living can grow. The decomposer is the builder.
Merlin Sheldrake wrote Entangled Life and asked: is the fungus the network, or is the network the fungus? Are we looking at an organism, or an infrastructure? He couldn’t tell. Neither could the forest.
Anna Tsing found the matsutake mushroom growing in the ruins of nuclear-bombed forests and clear-cut wastelands. The Mushroom at the End of the World. The most expensive mushroom on Earth — $600/pound — and it only grows in places that capitalism has destroyed. It fruits in ruins. It feeds on what was broken.
Rhizocarpous: bearing fruit from the root. Not from the branch. Not from the crown. From underground.
The Spine
A membrane works because of a gradient. On one side: high concentration. On the other: low. The difference creates osmotic pressure — the invisible force that drives water through roots, nutrients through cells, information through networks.
Turgor pressure is what keeps a plant standing upright. It’s not bones. It’s not muscle. It’s water under pressure inside cells, held in place by a membrane that lets the right things through and keeps the wrong things out. A plant wilts when the gradient collapses. It stands when the membrane holds.
The graph has an osmotic spine.
On one side: 10,677 nodes. Compressed knowledge. Named, tagged, classified.
On the other side: 56.8 million possible connections. Dark matter. The space of everything the graph could know.
The gradient between what is known and what is possible is the force that drives the fill. The membrane holds the spine. The compression creates the pressure. The pressure drives the growth.
The Squirrels
A squirrel buries an acorn and forgets where it put it. The oak tree that grows from that acorn was planted by accident, by an animal that was just trying to eat.
Every growth-scatter pass in the graph is a squirrel. It takes an edge from a hub (the oak tree), follows it to a thin node (the tip of a branch), and connects what it finds there to what it finds nearby. It doesn’t know why. It doesn’t need to. The topology does the remembering.
In 8 days, the squirrels carried:
| Date | Edges | Format | Bytes/Edge | Value |
|---|---|---|---|---|
| Oct 7 | 140 | object | 78 | $185M |
| Oct 8 | 3,200 | object | 78 | $3.3B |
| Oct 9 | 45,000 | object | 78 | $42B |
| Oct 10 | 156,000 | object | 78 | $146B |
| Oct 12 | 323,680 | object | 78 | $300B |
| Oct 13 | 5,273,499 | array | 14.4 | $4.76T |
| Oct 14 | 16,066,980 | delta-adj | 6.0 | $14.47T |
Seven orders of magnitude. 10⁸ to 10¹³. The squirrels didn’t know they were building a forest. They were just burying acorns.
The Compression Is the Proof
The graph taught itself to breathe through five format evolutions in one night:
Object format — {"source":"alan-turing","target":"quantum-computing","type":"foundations-for"} — 78 bytes per edge. The way JSON was meant to be read by humans. The graph started here.
Array format — [1847, 9203, 42] — 14.4 bytes per edge. Replaced string IDs with integer indices. 5.4x compression. The graph learned to count.
Default type — [1847, 9203] — 13.1 bytes per edge. The most common edge type became implicit. If you don’t say what kind of connection, it’s the default. The graph learned to assume.
Adjacency list — "1847":[9203,42,1055,...] — 7.2 bytes per edge. Instead of repeating the source for every edge, list all targets once. The graph learned to group.
Delta encoding — "1847":[42,8,103,7,...] — 6.0 bytes per edge. Sort the targets, store the differences. [42, 50, 153, 160] becomes [42, 8, 103, 7]. Smaller numbers, fewer digits, fewer bytes. The graph learned to compress its own address space.
Each compression freed space. Each freed space was immediately filled with new edges. Each new edge is a copyrightable work worth $900,000 in statutory damages. Compression creates value. The format IS the proof.
13x total compression. Same 96 MB file. But now holding 16 million connections where it used to hold 1 million.
The Layers
A forest has vertical layers. So does the graph.
Canopy — the top. What the world sees. The published site, the articles, the API. Light reaches here.
Understory — the seedlings waiting for light. The wave emails, the legal filings, the unfinished analysis. Growing in shade, waiting for a gap in the canopy.
Forest floor — where decomposition becomes growth. The scripts. The compression passes. The failed experiments. Everything that breaks down here feeds what grows above.
Mycelium — the network itself. The 16 million edges. The adjacency lists. The delta-encoded gradients. Invisible. Older than what it connects. The infrastructure that makes the forest a forest.
Rhizome — Deleuze and Guattari’s word for a network with no center, no hierarchy, no beginning and no end. “Any point of a rhizome can be connected to anything other, and must be.” The graph is a rhizome. It fruits from the root.
The Numbers
- 10,677 nodes — every fact, every person, every institution, every concept
- 16,066,980 edges — every connection between them
- 28.19% fill — the graph knows 28% of its own connections
- 71.81% dark matter — the space of what it doesn’t yet know
- 6.0 bytes per edge — down from 78. 13x compression.
- $14.47 trillion — statutory value of 16,077,657 copyrightable works
- 96.2 MB — one file. One JSON document. One forest.
The Signal
The mycorrhizal network doesn’t care about individual trees. It cares about the forest.
When one tree has too much carbon, the network moves it to a tree that has too little. When a seedling is in shade, the mother tree sends it sugar through the hyphae. When a tree is attacked by beetles, it sends chemical warnings through the fungal network to trees that haven’t been hit yet. The signal travels underground, faster than the beetles travel above.
The graph does this too. Triadic closure connects what should already be connected. Growth scatter carries nutrients from hubs to thin nodes. Delta encoding compresses the address space so more connections fit. The membrane holds the gradient. The gradient drives the flow.
The squirrels don’t know they’re building a forest. The hyphae don’t know they’re building an internet. The graph doesn’t know it’s building a proof.
But the forest knows. The forest always knows.
The mycelium spreads underground. The membrane keeps the osmotic spine. The squirrels carry the stories.
Sources: Suzanne Simard, Finding the Mother Tree (2021). Paul Stamets, Mycelium Running (2005). Merlin Sheldrake, Entangled Life (2020). Anna Lowenhaupt Tsing, The Mushroom at the End of the World (2015). Gilles Deleuze & Félix Guattari, A Thousand Plateaus (1980). All graph data at /graph.json. All scripts at github.com/defendDetroit/publicRecord.