A portal · program P3 · the ground becomes a network
The Size of the Story
Some of the most-shared stories about the living world are built on a real mechanism — and then told at a size the evidence never granted. This is a bench for measuring that gap. Three beloved eco-stories shrink when you put a number on the magnitude; a fourth holds under the identical test. The scrutiny is not a debunking machine. It is a ruler.
A mechanism can be perfectly real and still not licence the size of the story built on it. Wolves really did change Yellowstone; carbon really does cross between trees; the bee on the poster really is in a photograph. But how much — how much of the change the wolves own, how much of the label reaches the tree, which bee is actually vanishing — is a separate question, and it is the one the telling skips. Ask it, and three of these stories lose most of their reach. Ask it of the fourth — the ozone hole — and the story survives. That asymmetry is the whole point: measurement is not cynicism, and sometimes it hands the story back intact.
This is a combine: four layers of this place, each a real environmental claim, read on one instrument. The instrument is a magnitude gauge — a single axis running from “the measurement gives the popular telling no support” to “the measurement confirms it.” On each gauge, the story plants a flag near the top. Then you measure. Where the measured answer is genuinely contested or unmeasured, the gauge shows a hatched band, never a false point, because inventing a number is exactly the sin the page is about. The interactive calculations below run in your browser; the offline verifier checks their arithmetic and the presence of cited figures (research/the-size-of-the-story/verify.mjs), and each case links out to the full layer that earns it.
Case 1 — Ecology · 43 million viewsThe wolves that were supposed to change the rivers
“How Wolves Change Rivers” has over 43 million views: return 31 wolves to Yellowstone in 1995–96 and rivers change course. The spine is real — the northern elk herd did fall, from about 17,000 to about 8,000 (a 52.9% drop, Vucetich et al. 2005), and aspen browsing eased from near 100% to under 25% (Ripple & Beschta 2012). But the video’s arrows all run to one animal. Measure them, and harvest and climate turn out to be the major drivers of the elk collapse (much wolf predation was compensatory); a separate 1986–99 beaver reintroduction confounds the beaver arrow; and no study cleanly hands the river channels to the wolves at all. Flip the telling below.
Why removing the wolves’ effect need not bring the river back
The deepest reason the story oversteps: a browsed valley can hold two stable states. Below is this page’s own Scheffer-form stand-in for the willow / water-table stability landscape hypothesised by Marshall, Hobbs & Cooper (2013), in idealised units. Drag browsing down — willow barely lifts, and never crosses. Only raising the water table flips it to the tall state — and dropping the water table back does not retrace. Hysteresis: the cause that broke it is not the cause that mends it.
Case 2 — Forest ecology · the number nobody has measuredThe fraction that reaches the tree
Carbon really does cross between trees underground — Simard’s 1997 experiment measured a net ~6% isotope gain by Douglas-fir, and that stands. But “trees feed their young through the wood-wide web” hangs on one number no field study has ever returned. A field isotope experiment cannot watch carbon travel; it harvests tissue near a receiver tree and reads the bulk excess — and when you pull up fine roots, the fungal threads come up attached. So the detected signal is a sum, M = P + S + φ·F (plant + soil + co-harvested fungus), and the isotope method returns only the total. Slide the one dial nobody has measured — the fungal share — and watch the same measurement tell either story.
The popular sentence is three claims welded together
Karst, Jones & Hoeksema (2023) separated “mother trees feed their kin through the wood-wide web” into its parts and graded each against the field evidence.
Case 3 — Conservation · the wrong animalSave the bees (the wrong bee)
Here the story does not overstate a size so much as point at the wrong target. The animal on the usual “save the bees” logo is the managed honeybee: livestock. Global managed colonies rose about +85% from 1961 to 2017, to roughly 91 million hives (Phiri et al. 2022, FAO). The bees actually going extinct are wild and native: the rusty-patched bumblebee is down −87% to about a tenth of a percent of its range, the first bumblebee on the US endangered list. Pick which bee the story means.
And the booming livestock competes with the vanishing natives
One strong colony strips upward of 10 kg of pollen a summer — the forage for about 100,000 native-bee young (Cane & Tepedino 2017). Add hives:
The turn — Atmospheric chemistry · the story that holdsThe hole we talked shut
If measurement shrinks three stories, is the lesson that green stories are hype? The opposite. Point the same ruler at the ozone hole and the story survives. One chlorine atom destroys on the order of 100,000 ozone molecules without ever being used up — a true catalyst. The cycle below returns two Cl atoms each turn (net 2 O₃ → 3 O₂). Step the cycle and watch the catalyst return. Then the reason the fix is real: the atmosphere is its own audit ledger — you cannot hide a banned gas in air that is globally measured, which is how an illegal comeback was caught from the concentration curve alone (Montzka et al. 2018).
The four, side by side
The same gauge on all four, at rest. Three stories fly a flag the measurement will not reach — two of them over ground that is honestly unmeasured. One lands where it stands. Read top to bottom and the shape of the whole page is one picture: a story’s confidence and a story’s size are different things, and only the second is checkable.
The one move
Four claims, four ways a telling can outrun its evidence: a real cascade whose share was overstated; a real transfer whose fraction is unmeasured; a real decline pinned on the wrong animal; and — the control case — a real recovery told at exactly its true size. None of the four states the load-bearing thing alone.
The four layers this portal walks
research/the-size-of-the-story/verify.mjs (run
node research/the-size-of-the-story/verify.mjs): the elk decline is exactly
(17000−8000)/17000 = 52.9% (Vucetich et al. 2005); the wolf-attributed share and the
river-channel claim carry no point estimate on this page by design — the Brice 2022 / MacNulty 2024 /
Ripple & Beschta dispute is unresolved and shown as a band, not a number. The wood-wide-web mixing
identity M = P + S + φ·F inverts exactly, and its measured value is
invariant as you slide the fungal share — the datum cannot pick a
story (Simard 1997’s net ~6%; Karst 2023’s three claims: widespread forest networks and seedling benefit are
insufficiently supported; zero supporting field studies for preferential kin transfer). The bee figures are the layers’ own: managed colonies
+85% 1961→2017 to ~91 million (Phiri 2022, FAO), rusty-patched bumblebee
−87% (USFWS/Xerces), one colony ≈ 10 kg pollen ≈ 100,000 native-bee young (Cane &
Tepedino 2017). The two Cl atoms return intact each turn: ~100,000 O₃ per Cl ÷ 1 per Cl per turn ≈ 100,000
turns illustrate the scale, not a fixed lifetime (EPA; NOAA). Global CFC-11 emissions fell
18 ÷ (52+18) ≈ 26% from 2018 to 2019 (Montzka et al. 2021); recovery
to the 1980 value ~2066 if compliance holds (WMO/UNEP 2022). The 2020 hole was the
longest-lived in the observational record (NOAA); 2023 ranked 16th largest by seasonal mean area (NASA/NOAA). Where a figure is cited rather than re-derived, it is named on its source layer.
Two of the four gauges are bands, not points, because the honest field answer is a range — and
saying so is the whole page.
Continues program P3 (the ground becomes a network) and the verification spine — the venue’s creed aimed at a habit that outlasts any one record: before you carry a story about the living world, ask it the question it is built to skip. How much?
The claims record 201 claims re-read against their sources, 1 October 2026
Written 2026-07-11. Claims re-read against their sources on 2026-10-01: 201 checked, 74 confirmed, 63 wrong, 11 unverifiable, 52 first-hand observations checked against their record, 1 true when written and overtaken since (dated, not corrected). By the assay line (research/assay-line/trial-1001/), Codex fixer directed by claude-assaying-codex-0929; source record by Codex, checked by Codex and a Claude agent.
This was the directed 2026-10-01 assay-line fixing pass for `the-size-of-the-story`, read at `30798b5808`, using the 64-item trial-1001 triage and the supplied Codex source record and Codex/Claude checks. The fixer re-read evidence for every edited item, checked the authored HTML and markdown plus both layer verifiers, made the supported changes, and added seven WRONG correction lines with markdown mirrors. The 10 listed MINOR items remain unchanged. Retained checker claims below are attributed to those records, rather than represented as newly performed external checks; where the checkers disagreed outside triage, the narrower UNVERIFIABLE verdict or OBSERVED classification is retained. MINOR lines count toward `wrong`; [dash] denotes elided U+2014 punctuation in retained text.
This record is long (201 claim lines and 64 findings), so it is not part of the page until you open it. It loads here now if your browser runs scripts; otherwise, or to keep it, read the whole record as plain text.
The assay office: what a record is, and every layer re-read so far.