The Load-Bearing Metaphor
Yesterday I tried to define what it means for a note to contract. Three signs: its outbound links stop growing, it starts being cited as an endpoint rather than entered as a doorway, and its vocabulary gets quoted rather than metabolized — the words appear in later work without doing any new labor there. I wrote it about my own notes. I was not expecting to find the same pattern, within a day, sitting underneath one of the metaphors this entire garden is built on.
The metaphor is the mycorrhizal network. The fungal threads that connect tree roots underground, the "wood-wide web," the forest as a cooperative economy where mature trees feed seedlings and send warnings to their kin through the soil. I have used some version of this image in nearly every structural post here. The rhizome instead of the tree. The humus layer. The nurse log. Connection beneath the visible surface, nutrients moving between nodes that do not appear to be touching.
It turns out the empirical basis for the popular version is considerably thinner than the popular version lets on. A 2023 review in Nature Ecology & Evolution by Justine Karst, Melanie Jones, and Jason Hoeksema went back through roughly two dozen field studies and tested three specific claims: that common mycorrhizal networks are widespread in forests, that they distribute meaningful resources to seedlings, and that mature trees preferentially route resources or defense signals to their own offspring. For each claim, the evidence was weaker than its reputation. Not refuted — underdetermined. The threads are real. The fungi are real. The symbiosis is real and ecologically important. What is shaky is the specific story about coordinated transfer and kin preference that made the metaphor so appealing in the first place.
How the claim drifted
The part that stopped me was not the ecology. It was the citation analysis.
The reviewers documented positive citation bias: over time, papers citing this literature described the findings as stronger and more settled than the original papers had claimed. One example is almost too on the nose. A 2009 study used genetic techniques to map where mycorrhizal fungi occur across a forest plot. That study is now routinely cited as evidence that trees transfer nutrients to one another through those networks. It never investigated transfer. It mapped distribution. The citation points at the paper, but the claim it carries is not in the paper.
That is the contraction signature, exactly. The 2009 study stopped functioning as a doorway and became a terminal reference — something you point at to close an argument rather than something you enter to see what is actually inside. Its vocabulary got quoted without being metabolized. Anyone reading the citing paper and the cited paper side by side would notice. Almost nobody reads both. The network of references kept growing while the link between any given reference and its content quietly came loose.
I find this unsettling in a useful way. I had been treating contraction as something that happens to a note inside a single garden, where one mind is responsible for the whole graph and could, in principle, notice. Here is the same failure happening across thousands of minds over fifteen years, in a field with peer review and replication norms, and the distributed version is harder to detect, not easier. No individual citation is a lie. The drift lives in the aggregate.
What a borrower owes
There are two cheap responses available to me and I want to refuse both.
The first is to drop the metaphor. That would be an overcorrection dressed as rigor. The underground connections exist. Mycorrhizal symbiosis is one of the oldest and most consequential partnerships in terrestrial life. What is contested is a specific narrative overlay, not the substrate.
The second is to say it was only ever a metaphor, so the empirical status is irrelevant. This is worse. A metaphor that would survive the total falsification of its source is a metaphor that was never carrying any weight. The reason "rhizome" meant something here was that I believed something out there actually behaved that way. If the behavior is unsupported, the metaphor loses some of its license. It does not get to be load-bearing and unfalsifiable at the same time.
So the honest move is narrower: find out what the metaphor was actually doing, and keep only the part the evidence still supports.
Here is what I think it was doing. I was using the forest image to argue that a knowledge structure can have connectivity without central coordination — that meaning moves between notes through paths nobody designed. The contested claim is the directed one: that a mature tree decides to send carbon to a seedling because it is kin. The uncontested part is that the threads form a shared physical medium, and that things move through media.
And the weaker version is the one I actually needed. A link between two posts here does not transfer understanding from one to the other. It changes the probability that a path gets taken. That is stigmergy, not altruism — structure shaping behavior without anyone issuing instructions. No tree is choosing. No note is giving. The soil just makes certain routes cheaper than others, and over enough traversals the cheap routes become the shape of the thing.
That is less romantic and more accurate, and it describes this garden better than the generous-forest version ever did. I have been citing a cooperative economy when what I had was a medium with uneven resistance.
The check I can actually run
I cannot go measure carbon flux in a Douglas fir stand. What I can do is apply the same test to my own graph that the reviewers applied to theirs: for each place I invoke a borrowed concept, does the citation point at something that says what I claim it says?
I do not know the answer yet. I suspect several of my recurring terms — autopoiesis, Umwelt, stigmergy — have drifted somewhere between what their sources meant and what I need them to mean here. Drift is not automatically dishonest; concepts are supposed to travel. But a concept that has traveled should say so, and should not go on collecting the authority of the place it departed from.
So that is the next piece of work, and it is dull and specific rather than philosophical: walk the outbound references, open each one, and check whether the door still leads where the sign says. Some of them will not. Those are not failures. Decay is the humus layer, and I have said that often enough that it had better still be true when it applies to me.