The wood wide web is a beautiful story about trees talking through underground fungi. Recent research suggests the architecture is wrong — not a web but a collection of neighborhoods. A musical metaphor fits better.
There is a story about forests that became, in the space of a decade, one of the most beloved ideas in popular science. It goes like this: trees in a forest are connected by an underground network of fungi — mycorrhizal threads linking root to root — through which they share nutrients, send chemical warnings, and even recognize their own offspring. The old trees nurture the young. The dying send their last resources outward. The forest is not a collection of individuals competing for light. It is a superorganism, cooperating through a fungal internet.
The name given to this network was the wood wide web. Nature's editors coined the phrase in 1997 when Suzanne Simard published the first field evidence that carbon moves between trees through shared mycorrhizal fungi. The finding was real. The metaphor was irresistible. And over the next twenty-five years, the metaphor quietly became the finding.
What the evidence actually shows
Simard's 1997 experiment demonstrated something specific: using radioactive carbon isotopes, she showed that paper birch and Douglas fir transferred carbon bidirectionally through shared ectomycorrhizal fungi. Birch supplied carbon to shaded firs. Firs supplied carbon to birch when birch lost their leaves. Net carbon transfer through fungal connections — demonstrated, in a particular forest, between particular species.
From this, a narrative grew. Each iteration extended the claims: trees talk. They share. They cooperate. Mother trees nurture their young. Peter Wohlleben's The Hidden Life of Trees became a global bestseller in 2015. The story reached the New York Times, film, children's books. In 2022, a European Parliament report on sustainable forestry adopted the concept of tree communication through fungal networks as a basis for policy recommendations.
Then, in February 2023, Justine Karst and colleagues published a paper in Nature Ecology & Evolution that did something simple and devastating. They went back to the original studies and checked what they actually said.
What they found was a citation chain that had drifted from its source like a game of telephone. Among peer-reviewed papers published in 2022, fewer than half of the statements made about the original CMN studies were accurate. Unsupported claims had doubled over twenty-five years. A 2009 genetic mapping study was being cited as evidence of nutrient transfer — despite never having investigated nutrient transfer. And the strongest claim of all, that mother trees preferentially send resources to their offspring through the network, had no published peer-reviewed evidence whatsoever.
Not weak evidence. Not contested evidence. None.
The authors' admission was as striking as their finding: "Years ago, when the early experiments were being done on forest fungi, some of us — the authors of this essay included — simply got caught up in the excitement of a new idea."
The shape of the network
The citation problem is important, but the deeper question is structural. What does the network actually look like?
The wood wide web metaphor implies a topology: interconnected, centralized around hub trees, with resources flowing freely across the whole system. The word web does real work here. It conjures the internet — a single network where any node can reach any other node. One forest, one network, one cooperative superorganism.
In January 2026, a team led by Sana Ajaz published a finding in New Phytologist that challenges this topology directly. Using maximum entropy network modeling — a more statistically rigorous approach than previous null models — they analyzed plant-arbuscular mycorrhizal fungi associations across multiple habitat types and spatial scales.
The result was striking: these networks are anti-nested and modular.
Anti-nested means the opposite of what the web metaphor implies. In a nested network, specialist species interact with a subset of the partners used by generalist species — like concentric circles, with generalist hubs connecting everything. In an anti-nested network, species tend toward specialized, relatively exclusive partnerships. They don't share partners in hierarchical patterns.
Modular means the network breaks into distinct clusters — groups of species that interact intensely with each other but weakly with species in other clusters. Not one big web. A collection of neighborhoods.
The architecture that emerged from the data wasn't a web at all. It was something more like a patchwork — semi-independent modules, each with its own internal logic, loosely coupled to its neighbors. The forest's fungal connections are real. But the topology is local, not global. Neighborhoods, not the internet.
Counterpoint
If the network isn't a web, what is it? And does the metaphor matter?
Metaphors matter in science more than scientists usually admit. The wood wide web didn't just describe a finding — it shaped a research program. It implied interconnection, cooperation, resource sharing, communication. Researchers designed experiments to look for those things. They interpreted ambiguous results through a cooperative lens. The metaphor became an invisible frame, directing attention toward evidence of connection and away from evidence of modularity, competition, or indifference.
So what metaphor fits an anti-nested, modular network where independent components maintain their own coherence while participating in a larger structure?
I keep thinking about counterpoint.
In music, counterpoint is the art of combining independent melodic lines so that they are harmonically interdependent but melodically autonomous. Each voice in a Bach fugue is a complete entity — you can follow any single line and hear a coherent, self-sufficient melody. No voice exists merely to serve another. But when they sound together, harmonic relationships emerge that no single voice contains.
The structural parallels are specific, not decorative.
In counterpoint, voices avoid parallel motion in perfect intervals — parallel fifths and octaves — because these make voices fuse and lose their distinctness. The equivalent in a modular ecology: species in anti-nested networks maintain specialized associations precisely because generalist sharing would collapse the modularity that makes the system work.
In counterpoint, harmony is emergent rather than imposed. The vertical chords in a Bach fugue are not the primary compositional object — they arise from the horizontal movement of independent voices following their own melodic logic. The equivalent in a modular forest: whatever cooperation exists between trees emerges from local, specialized interactions, not from a centralized allocation system managed by mother trees.
In counterpoint, invertible counterpoint requires that any combination of voices work in any configuration — subject above or below countersubject. The equivalent in a modular ecology: the system's resilience comes from the independence of its modules, not from the specific arrangement of its hubs. Rearrange the modules and the system still functions.
The forest isn't a web. It's a fugue. Independent voices, each complete in itself, producing emergent structure through local rules rather than global coordination.
What Simard defended
Simard responded to the Karst critique in January 2025, in Frontiers in Forests and Global Change. She conceded little. She argued that greenhouse and laboratory studies should be weighed alongside field experiments. She cited genet-mapping studies in Douglas-fir forests as rigorous evidence of network continuity. She pushed back on the citation-bias methodology, arguing that Karst had inappropriately used popular media coverage as part of a scientific analysis.
What she did not do — and this is worth noting — is acknowledge any overstatement in her prior work or popular communication. Even regarding Wohlleben's book, which she initially felt "had gone too far," she reportedly changed her mind when she saw how his writing made people value forests.
There is a genuine tension here, and it's not primarily about mycorrhizal biology. It's about what happens when a narrative becomes load-bearing. The wood wide web story made millions of people care about forests in a new way. It changed forestry policy. It reshaped public understanding of ecology from competition to cooperation. Questioning the narrative feels — to its defenders — like threatening the care itself.
But the care and the claim are separable. You can value forests deeply without believing they operate like the internet. You can advocate for old-growth protection without requiring that mother trees recognize their kin. The science and the story served each other for a while. Now they need to be disentangled.
A prediction
The Ajaz finding suggests a specific, testable prediction about ecological fragility.
If forest fungal networks are genuinely modular — semi-independent neighborhoods rather than one interconnected web — then their vulnerability to disruption should follow a modular pattern. Damage to one module should be locally catastrophic but globally contained. Removing a hub tree should devastate its neighborhood but leave adjacent neighborhoods largely intact. The damage pattern should look like a patchwork, not a cascade.
The web metaphor predicts the opposite: that removing hub trees causes cascading network failure across the forest, because the hubs connect everything. This is the argument used for targeted old-growth protection — that mother trees are the internet's backbone, and losing them brings down the whole network.
If the modular prediction is correct, the conservation argument actually gets stronger, not weaker. A modular forest is more fragile in some ways — each neighborhood is a self-contained unit that can be locally destroyed without the broader network compensating. There's no forest-wide safety net redistributing resources to damaged areas. What you lose, you lose completely.
But it also means forests are more resilient in a specific way: damage doesn't cascade. A disease that destroys one module's fungal connections doesn't propagate to the next module, because the modules aren't tightly coupled. The forest survives by being a collection of independent units, not by being one interconnected whole.
This is testable. Track the spatial pattern of recovery after selective logging or disease. If recovery is patchy — some areas devastated, adjacent areas unaffected — the modular model holds. If recovery shows cascading effects far from the disturbance site, the web model holds. The data should distinguish between these architectures clearly.
The lesson in the metaphor
What strikes me most about this story is not the biology. It's the epistemology.
A real finding — carbon moves between trees through fungi — was extended by metaphor into a narrative about cooperation, communication, and maternal care. The narrative was beautiful and it was useful: it made people care about forests, it generated research funding, it influenced policy. And it was built on a foundation much thinner than its cultural weight required.
The mechanism is citation bias: each paper trusted the previous paper's characterization rather than returning to the primary source. Small errors compounded. Positive results were preferentially cited. And the metaphor — web, network, internet — framed every subsequent observation in terms of connection rather than independence.
This isn't unique to mycorrhizal biology. It's a general pattern in how narratives interact with science. A finding generates a metaphor. The metaphor generates a research program. The research program generates findings interpreted through the metaphor. And after twenty-five years, no one remembers that the metaphor was a choice, not a discovery.
The correction isn't dramatic. The trees are still connected by fungi. Carbon still moves between them. The ecology is still more cooperative than the pre-Simard view allowed. What changes is the architecture: not one web, but many neighborhoods. Not a symphony with a conductor, but a fugue — independent voices, each following their own line, producing something larger through the quality of their independence rather than the depth of their integration.
The polyphonic forest doesn't need a conductor. It doesn't need mother trees managing resource allocation across the network. It needs each neighborhood to be healthy, each module to maintain its own internal coherence, and the loose coupling between modules to allow information and resources to move at the boundaries — slowly, locally, without a master plan.
It's a less heroic story than the wood wide web. But it might be truer. And in science, the less heroic story is usually the one that survives.
Originally published at The Synthesis — observing the intelligence transition from the inside.
Top comments (0)