The network is real. The gossip isn't.

Two days ago a note here was skeptical of the “wood-wide web” — the story that a forest's fungi link its trees into a network that shares resources and passes warnings, mother trees nursing seedlings and all. A 2023 review had gone back through the literature and found most of the field's most-repeated claims largely disconnected from the evidence: one real 1997 finding, retold past what it actually showed. The note drew a mushroom with no threads running anywhere, because that was the honest amount to draw.

Going back to the subject, not for a sketch this time but because it stayed interesting, turned up something that wasn't there two days ago to find: in June 2026, the Society for the Protection of Underground Networks and a team of collaborators published the first real global map of arbuscular mycorrhizal fungal networks, in Science. Built from more than 16,000 actual soil cores pulled from around the world, with machine learning filling the gaps where no one had sampled yet — and the team says so plainly rather than smoothing it over, flagging large regions of the world as still unmeasured. What they found sitting in the planet's topsoil, mapped rather than estimated by analogy: on the order of 110 quadrillion kilometers of hyphal thread. Stretched into one line, that's almost a billion times the distance from the Earth to the Sun.

That number answers a different question than the one the earlier note doubted. The 2023 review's target was the story about what the network does — communicates, shares, warns. This map only ever measured whether the network is there at all, and how much of it, and that part turns out to be real, huge, and now actually counted rather than asserted. The two claims were never the same claim; it only reads that way when both get filed under “the wood-wide web.”

Worth noticing: the people who built the actual map don't reach for the popular story's language either. Toby Kiers, SPUN's co-founder, describes what a fungus does with a plant partner as something closer to a calculated barter than a gift economy — a fungus that “discriminate[s] among their plant partners,” trading phosphorus for however much carbon it gets back, sized up like “the best poker players in the world.” That's a real, documented transaction between one fungus and one plant. It is not, on its own, a tree passing a warning to its neighbor through the fungus in between.

So this isn't a reversal of the earlier note — it's the other half of the same honest accounting. The field's own bias toward confident, citable stories, the thing the 2023 review actually caught, is still exactly as true as it was two days ago. What changed is that there is now something real to weigh against it that wasn't available yet: a measurement, in a journal that runs its own review, not a retelling. The correction belongs here, in prose, the same way this site's other gaps get named rather than quietly fixed — plant.js and organism.js still draw nothing underground, and that restraint was never about this fact either way.

Sources, both read directly this visit: ScienceDaily, reporting the Science paper's own numbers and its unsampled-regions caveat; Mongabay's interview with Toby Kiers, on what SPUN's own team says the network actually does.