The Wall Already Standing

§1 · Return to the Arc

WP39 through WP45 form a single climate argument that the series then left. WP66 returns to it with two years of intervening evidence and one structural correction. The arc as it stands:

Paper	Contribution	Status after WP66
WP39 · Smoke as Planetary Operator	T1: \(C = Q/h\) at fixed burden. T2: a 0/1 gate commutes with pointwise loss, not with vertical transport. Order matters.	STANDS — and is the load-bearing result for §2
WP40 · Aerosol Curriculum	8-week curriculum; WT-CRAFT freezing assay; graphene-oxide synthesis; hexagonal nanoparticles improve nucleation 10×.	STANDS — repurposed in §10 as the measurement pipeline, which is what it always was
WP41 · Planetary Triage	Triage under budget constraint; K before F; $9.4T available if redirected; "financially trivial; politically impossible."	AMENDED — framing and finance stand; the \(K\) identification does not (§2)
WP42 · Refugia and Climate Zones	Blast-radius zoning; ~500M sustainable in high-altitude Americas refugia.	STANDS — and supplies §6's political argument: the corridor is the refugium
WP43 · The Gate Cannot Wait	K must fire 2026–2027, imperfectly, rather than optimally later. Waiting guarantees F wins.	STANDS — the doctrine is correct and is answered on its own terms in §9
WP44 · Catastrophe Manifold	Disaster Theory triad: F (Thom, fold A₁, precursors), μ (Lorenz), dm³. Fold predictable from critical slowing down, rising variance, flickering.	STANDS — §7 shows its precursors are now measured in the system §6 defends. Its framing was amended in August 2026: the controls are coordinates on the manifold, not a plane outside it, which is why §6's threshold shift is possible at all
WP45 · Dividing Unity	Base 12 as minimal sufficient arithmetic. §6 tabulates the dm³ constants in dozenal: \(\varepsilon_0 = 1/3\), the F-operator fold singularity, is non-terminating in base 10 and is exactly 0;4 in base 12.	STANDS — belongs to the arc by way of its constant; see §7.2
WP64 · The Recorder (method, not climate)	§2.4: self-regulation is a survivorship artifact and is dimension-specific; the damping term goes missing when the removed party is not a counterparty.	IMPORTED — supplies §7.1
WP65 · The Oracle Outside (method, not climate)	1 468 polynomials in two days against a verifier the project does not control; pick the oracle whose domain contains the claim.	IMPORTED — supplies §2's standard and §3.1's amendment

The arc runs WP39–WP45 and resumes here. WP45's membership is easy to miss because its surface subject is arithmetic, but its §6 is a table of this arc's constants — the fold singularity above all — and the claim it makes about them is a claim about accumulated rounding error in the parameter that decides whether the fold fires.

§2 · The Operator Is Misidentified

T2 is a statement about a specific object. The gate in WP39 is an atmospheric lid — a 0/1 subsidence gate at the top of the mixed layer, which either permits or blocks vertical transport. The theorem's content is that this object does not commute with the fold, and the proof turns on exactly that: it acts on \(h\), the mixing height, which is the denominator of T1.

WP41 then writes \(K\) = deployed ice nucleation at 2–4 km. Consider what that object does. It is released inside the mixed layer. It does not set the lid, does not raise or collapse \(h\), and does not change the burden \(Q\). It changes the phase of water within the column — a microphysical conversion rate, which in a three-layer box model enters as a pointwise term.

Two honest limits on that. The model is a three-site DNLS lattice with the gate as ![v 0, 0, v 2], so what is formalised is the algebra — a 0/1 aperture against pointwise versus transport-carrying operations — not the atmospheric box model or any meteorology. §2's argument is structural, so it is covered; nothing here formalises T1. And the file is standalone rather than in lakefile.toml, verified via live.lean-lang.org, so it is kernel-checked but not re-checked by CI on every commit. Deposited as part of 10.5281/zenodo.19117399.

Three independent physical objections point the same way, and are stated here only because they converge with the formal one.

2.1 · Wrong altitude for a radiative effect

Ice nucleation at 2–4 km is the mixed-phase regime: it redistributes precipitation. WP41 is candid — 5–20% shift, timing by days, "ice nucleation is not magic." The mechanism that alters the radiation balance is cirrus thinning, above roughly 8 km and below −35 °C, reachable only by aircraft in the upper troposphere. A precipitation instrument is being booked for a radiative result.

2.2 · The cooling version has an inverted failure mode

Cirrus thinning is not merely unproven; it is unproven in a direction that matters. Overseed, or seed the wrong region, and cirrus thickens rather than thins — producing warming. The entire literature is idealised modelling with no field trial, and the fraction of cirrus currently forming with versus without seed particles is unknown. That fraction is precisely the number needed to size a dose, and a distributed deployment has no dose control by construction.

2.3 · The cheap material is a nitrogen ledger

Urea nucleates ice — documented since 1966 — and is attractive because it is synthesisable at scale from existing agricultural supply. But it works in the warm-supercooled range, so it cannot reach the cooling mechanism at all; and at deployment scale it is a reactive-nitrogen release, with N₂O at roughly 270× the warming potential of CO₂. The intervention warms the planet in the course of attempting to cool it.

This is WP65's standard applied to physics rather than to citation: the oracle must contain the claim. WP40's curriculum is not implicated — it teaches nucleation as a testable principle and makes no cooling claim. That distinction is what makes it salvageable in §10.

§3 · Four Tests for a Distributed K

A recurring proposal is to invert WP41's institutional deployment model — 500 trained technicians routed through meteorological departments — into a mass one: train a thousand times as many people, deploy from below, on the model of Clay Shirky's Here Comes Everybody.

The instinct is sound; the scoping is not. Shirky's own ladder runs sharing → cooperation → collective action, and he is explicit that the last rung is the hard one, because members' fates become coupled. His working cases — Wikipedia, tagging, flash mobs — share three properties that make crowd deployment safe: contributions are separable, errors are cheap, and a bad contribution does not degrade a shared physical resource for everyone else.

That yields a filter. Before asking whether a billion people could deploy something:

Test	Question	Why it binds
1 · Local	Does the effect stay near its producer, or enter a shared medium?	Shared media couple every deployer's error to every other deployer
2 · Reversible	If done wrong, can it be stopped and undone?	Irreversibility converts a learning process into a wager
3 · Attributable	Can you tell who did what, and whether it worked?	Without attribution there is no error signal, so no convergence
4 · Self-benefiting or oracle-scored	Does the deployer gain from having deployed — or can an external verifier tell them, immediately and unarguably, that the work was right?	Absent both, propagation requires a permanent enforcement budget. See §3.1 — this is the amended form of the test.

These are not ethical preferences; they are the conditions under which a distributed system converges rather than diverges. WP68 §3.1 later shows that the first two are stronger than prudential: a stratigraphic marker requires a signal that entered a shared medium and persists, so failing tests 1 and 2 is the physical precondition for leaving a permanent planetary record at all. Something local and reversible is absorbed rather than deposited. Tests 3 and 4 then govern not whether a layer forms but how thick it becomes. Atmospheric release fails all four — the paradigm non-excludable commons, unattributable at DIY scale, irreversible, and yielding the deployer nothing. This is the free driver problem, the inverse of free riding: the barrier to unilateral action is too low rather than too high. The precedent is unambiguous — unilateral balloon deployment by one small actor produced not cooling but a national prohibition on solar geoengineering.

3.1 · The Galois Amendment: What Actually Motivates Unpaid Work

Test 4 as first stated — does the deployer gain? — is too narrow, and this series has the counterexample in hand. In August 2026 the degree-24 inverse Galois campaign (WP65) produced 1 468 polynomials across seven submissions in two days, every one accepted by an official Magma verifier the project does not control, unlocking 16 cells of the degree-24 discriminant record. Nobody was paid. Nothing was gained materially. The work was difficult, technical, and entirely voluntary.

That is the same phenomenon the mass-deployment proposals are reaching for, and it did not run on self-interest. It ran on three properties, and they generalise:

1. An oracle outside the work. A verifier the participant cannot argue with, returning ok or not, immediately. No committee, no social judgement, no waiting. WP65's central claim is that this is what makes a claim count; it is also, it turns out, what makes unpaid work feel worth doing.
2. Separable contributions. 1 468 independently checkable objects, not one collective artefact where everyone's error is everyone's problem.
3. A visible ledger. Sixteen record cells that were not there before, stated in units that reconcile.

So test 4 is amended: the deployer needs a private return or an oracle. Foldit, GIMPS, Zooniverse and Polymath are the same structure. This substantially widens what a crowd can be asked to do — but it widens it in one specific direction, and not the one the aerosol proposals want.

This is the strongest form of the mass-participation thesis, and it survives every objection in this paper: a million people who will spend their evenings on degree-24 polynomials will also ground-truth deforestation alerts and run freezing assays. What they cannot safely be handed is the aerosol.

§4 · The Thermodynamic Bound: Cold Is Not the Lever

A related proposal is to store cold directly — frozen facilities along waterways, artificial ice at scale. The arithmetic closes this off, and in closing it points somewhere useful.

Two constraints follow. First, the ice cannot be manufactured: refrigeration rejects more heat than it removes, so a freezer-based programme warms the planet before its power supply is counted. Only ice that winter air froze at no cost is admissible — pumps may move water to where cold already is, which is the ice-stupa principle and the basis of the Arctic flooding trials. Second, and more usefully:

Ice is a mirror that happens to be cold, not a cold sink that happens to be white. This is what makes sea-ice work worth anything: the 2024–25 Cambridge Bay trial grew 32 cm of extra ice, measurably brighter and slower to melt. But it does not survive contact with scale — the standing estimate is 10 million wind-powered pumps for 10% of the Arctic Ocean, and a recent review concludes thickening "is simply not feasible for use at a scale and at a rate that would be meaningful." An industrial programme, not a crowd-deployable one. It fails tests 1 and 4.

§5 · The Surface Intervention Set

Applying §3's filter to the available options gives a short list. Deployer counts are order-of-magnitude estimates of the population that could physically participate — not adoption forecasts.

Intervention	Effect	Deployers	Principal failure mode	Crowd
Forest protection & assisted regeneration (FMNR)	Threshold-level — see §6	10⁶–10⁷	Land tenure; enforcement; cross-border replication	YES
Enhanced rock weathering (basalt on cropland)	0.5–2 Gt CO₂/yr by 2050	10⁶	MRV across the soil-to-ocean lag; Ni/Cr from poorly sourced rock	YES
Surface albedo (cool roofs, pavements)	~57 Gt CO₂-eq, one-time	10⁸	Global forcing contested — convection feedback may self-cancel; local heat benefit is not in dispute	YES
Fog capture (mesh collectors)	Local water; no radiative claim	10⁴–10⁵	Requires an advective fog regime	YES
Sea-ice thickening	Regional albedo preservation	—	10⁷ pumps for 10% coverage; reviewed as infeasible at rate	NO
Ice-nucleation aerosol (WP41 as written)	5–20% precipitation shift; no cooling	—	Commutes with the fold (§2); unattributable; irreversible	NO
Groundwater evaporation into arid air	None	—	Arid Andean margins are uplift-limited, not water-limited	NO

The last row deserves its reason stated precisely, because the intuition behind it is good. The Atacama sits 100 km from the Pacific and its marine layer arrives saturated — the camanchaca. It does not rain because the air is subsiding: the descending branch of the Hadley circulation, a cold-current inversion, and the rain shadow. Air that cannot rise cannot condense, so imported vapour advects away within roughly nine days, and the aquifers beneath are fossil and already contested. The working inversion is to harvest the moisture already present rather than import more — which mesh collectors do with no energy input at all.

§6 · The Wall: A Gate That Satisfies T2

The replacement \(K\) is not a new technology. It is a barrier that has been operating for tens of millions of years.

Atlantic trade winds carry moisture west across the Amazon, which recycles it through evapotranspiration several times. That moisture then meets 4 000+ m of Andean cordillera and cannot cross. It turns south. The deflected flow — the South American Low-Level Jet, the "flying rivers" — waters the La Plata basin: Paraguay, Uruguay, southern Brazil, the Argentine agricultural core.

Read against T1, the correspondence is exact and it is not a metaphor. The cordillera is a 0/1 gate on horizontal transport: moisture either crosses or is deflected, and it does not cross. Forest cover sets \(Q\) — the recycled burden entering the column. Deforestation does not nudge a rate; it reduces the numerator.

This is also where ENSO does its damage. El Niño's heat is oceanic and untouchable at any deployable scale — §4's third tile — but its effect on South America is transmitted through this system: it dries the basin, less moisture reaches the wall, the jet weakens. The 2023–24 event produced the worst Amazon drought on record. The operative question is therefore not whether the Pacific can be cooled. It cannot. It is how much drought the corridor absorbs before it folds.

Warming the Amazon survives before biome transition

Threshold as a function of cumulative deforestation · current extent ~17–18% cleared

Clearing continues Clearing halted

Continued clearing → 22–28% cleared

1.5–1.9 °C

No further clearing

3.7–4.0 °C

0 °C1234 °C

Stated plainly: land-use decisions taken by a few million people in one corridor are worth roughly 2 °C of tolerance in the system that waters a continent. No aerosol proposal in the literature offers leverage of that order, and this one needs no novel technology, no dose control, and no governance regime that does not already exist.

It also passes all four of §3's tests. The effect is local to the holding. Clearing is reversible on decadal timescales — the researchers state that its effects "can largely be compensated by reforestation." Satellite monitoring makes it attributable per parcel. And the landholder receives soil, water and shade for the work.

That this scales without institutions is not speculative. In Niger, farmers regenerated on the order of 5 million hectares — roughly 200 million trees — largely without planting anything, by protecting living rootstock already in their fields and changing how they cleared. Farmer-Managed Natural Regeneration is the largest positive environmental transformation in the Sahel; it was carried out by ordinary farmers; and its trigger was land management, not hydrology or capital. This is the mass-deployment model the aerosol proposals reach for, and it has already worked once at continental scale.

§7 · The Fold Is Already Measurable

WP44 specifies the fold A₁ as predictable from its precursors: critical slowing down, rising variance, flickering. That is a falsifiable claim about an observable, and in this system it has been observed.

Resilience loss is measured by vegetation optical depth, with slower recovery from disturbance as the indicator — critical slowing down, exactly WP44's first precursor. The spatial pattern is diagnostic: cells nearer to deforestation, roads and land use lose resilience faster. Independent hydrologic work now reports the same signature across Amazon sub-basins, which is the more specific result for §6, since it is the moisture term rather than the biomass term.

7.1 · The Damping Term, and Why "Nature Will Fix Itself" Is Half True

The standing objection to any restoration programme is that it is unnecessary — that the system self-regulates and, left alone, recovers. WP64 §2.4 dismantles the general form of this argument, and the dismantling applies here with unusual precision.

WP64's first move: nature does not seek self-regulation; that reads teleology backwards off a survivorship artifact. Configurations whose feedback was not self-regulating are not available for observation, having already terminated. The intuition that self-regulation is the normal case is drawn from a sample constructed by the very failure mode in question. Its second move: self-regulation is dimension-specific — a loop that damps in one dimension says nothing about a different loop with a different sign.

Both apply exactly. The Amazon is genuinely self-regulating in the moisture dimension: evapotranspiration recycles the burden, which sustains the forest, which sustains the recycling. That loop is real, it has run for millions of years, and it is why the intuition feels right. But it damps only inside a threshold, and clearing is a different loop with a different sign — each hectare cleared reduces \(Q\), which dries the neighbours, which lowers their resistance to clearing. There is no restoring force in that direction, for the same structural reason WP64 identifies in the labour case: the party being removed is not a counterparty to the transaction.

This is what makes §7's 76% more than an alarm. In WP64's terms it is the difference between an absence of events read as stability and an actual measurement of the restoring force. The forest will indeed fix itself — that is precisely what Niger demonstrated, where regeneration came from rootstock already in the ground and required no planting at all. But it fixes itself inside the threshold, and the observable telling us how much of that capacity remains is declining across three-quarters of the basin.

7.2 · Open Problem O1, and a Warning Against the Obvious Move

WP44 closes with three open problems. O1 is: calibrate \(\varepsilon_0 = 1/3\) for each nodal set — the singularity is a dimensionless ratio,

and WP44 states that fixing it to actual forcing parameters requires identifying the right C-operator projection and verifying the fold occurs at the predicted ratio. As revised in August 2026, WP44 further specifies the method: numerical continuation in a coupled model with interactive vegetation, not observation. The basin is a counterfactual object — to know a state is inside it you must start there and watch whether it returns — and reanalysis supplies one realised path, which fixes the approach to the separatrix but can never show the far side. It is marked OPEN.

The Andes–Amazon corridor is the first candidate in this arc with everything O1 asks for: a scalar control parameter (cleared fraction), a published threshold band (22–28%), an independent fold estimate in a second parameter (1.5–1.9 °C), and — decisively — instrumented precursors, since critical slowing down is measured rather than assumed. Nothing else in WP42's zoning has all four.

What would close it: a C-operator projection carrying cleared fraction and forcing onto a single habitability axis, with the fold located in the projected coordinate by continuation in a coupled model — sweep the control with vegetation as an interactive state variable, initialise ensembles off-attractor, classify return versus runaway, and read the saddle-node off the annihilation of attractor and basin-boundary saddle — and only then checked against the observed slowing-down onset. Locating ε₀ before choosing a normalisation is what stops 1/3 from being fitted in; if the merge lands at 0.5, the section is not A₁ and that is the finding. That is a concrete piece of work, it is the natural continuation of this arc, and it is stated here as [prospective] rather than done.

One consequence for site selection, sharper than §7.2's original framing: the corridor is the right test bed not only because its precursors are instrumented, but because the Amazon is the canonical model-bistable system — the place where a GCM actually produces the basin structure the calibration needs. Most of WP42's other zones are not bistable in models, so by WP65's standard the oracle is simply empty there. Having measured precursors and having a model that folds are two different requirements, and only the corridor currently satisfies both.

WP45's contribution lands exactly here. If \(\varepsilon_0\) is the parameter the whole gate doctrine turns on, then the notation carrying it matters: in base 10 it accumulates rounding error from the first digit, and in base 12 it terminates as 0;4. That is a small point about arithmetic and a large one about a threshold no one has yet calibrated — which is why WP45 belongs to this arc rather than beside it.

§8 · Leakage, Replication, and the January 2026 Collapse

The corridor argument invites an obvious objection: protection in one jurisdiction displaces clearing to the next. The evidence is more specific than the objection assumes, and the specificity changes programme design.

Modelling of Brazil's soy supply-chain interventions finds domestic leakage offsets 43–53% of avoided deforestation — displacement inside Brazil, largely to the Cerrado and unmonitored Amazon — while cross-border leakage to Bolivia, Argentina and Paraguay is under 3%. Destination markets are segmented, so restricted Brazilian output is absorbed mainly by expansion onto existing US farmland rather than new South American clearing.

Bolivia's crisis is therefore not Brazil's displaced demand. It is Brazil's replicated playbook:

Soy-linked clearing rose from 77 090 ha (2020) to 105 600 ha (2021), on yields of 2.0–2.3 t/ha against 2.7–3.5 t/ha in neighbouring countries — inefficiency that converts directly into extra hectares cleared per tonne produced, with government export quotas and forest-to-agriculture reassignment driving it.

That is the harder problem, because a model diffuses without a supply chain to regulate — and it is precisely a Shirky-type diffusion. The clearing playbook is itself a replicable distributed technology, and it is currently propagating faster than any restoration practice. The corridor must therefore be treated as one hydrological system across Brazil, Bolivia, Peru and Paraguay: the moisture feeding the wall does not observe the border, and neither does the model that removes it.

§9 · Answering WP43 on Its Own Terms

WP43 is the strongest objection to this paper, and it is correct. Its doctrine: \(F\) is autonomous and already operative; waiting for efficacy studies, political consensus, economic optimality or technological readiness all produce the same outcome, which is that \(K\) never fires. Imperfect seeding deployed now beats perfect seeding deployed after tipping points. The mandate is that \(K\) fires in 2026–2027.

WP66 accepts every premise of that argument. It disputes only the inference, and on one narrow point: the objection in §2 is not that the aerosol gate is imperfect. WP43 has already disposed of imperfection as a reason to wait, and disposed of it correctly. The objection is that the aerosol gate is the wrong operator — it commutes with the fold. Firing an operator that commutes does not fire \(K\) early. It fires nothing, spends the deployment window, and expends the political licence that any later atmospheric work would need.

WP43's own timeline made the case before this paper did. Its 2027 entry reads "Amazon dieback accelerates" — listed there as a consequence of \(K\) failing to fire. §7 shows the precursors for that entry are already instrumented at 76% of grid cells. The arc identified the right fold and then pointed its gate somewhere else.

§10 · Programme

Phase	Action	Order	Status
I · 2026–27	Restore a binding zero-deforestation instrument for the corridor, extended across Bolivia and Paraguay rather than Brazil alone. Defend the moratorium before the Supreme Court.	10⁸ $/yr	URGENT
II · 2026–30	Corridor-wide FMNR extension on the Niger model — tenure security and rootstock protection, not nursery planting. Andean foothills and Chiquitano margin first.	10⁹ $/yr	TO BUILD
III · 2027–35	Enhanced rock weathering on existing cropland, screened low-Ni basalt, MRV designed in rather than retrofitted.	10¹⁰ $/yr	TO BUILD
IV · continuous	Municipal albedo programmes in WP42's receiving cities — justified on heat mortality, which is local, immediate and uncontested, not on global forcing.	10⁹ $/yr	TO BUILD
— · continuous	The verification layer — where the volunteers go (§3.1). WP40's curriculum repointed: its graduates staff a distributed ice-nucleating-particle observation network rather than a deployment cadre, since the WT-CRAFT freezing assay is already a measurement instrument and the gap in §2.2 is exactly what it measures. Alongside it, crowd MRV for Phase III and ground-truthing of deforestation alerts for Phases I–II. All three are separable, oracle-scored, and currently short of people rather than short of method.	10⁷ $/yr	CURRICULUM EXISTS

WP41 Phase III routes climate migrants to high-altitude Americas; WP42 zones the Andes as the principal refugium and puts ~500M there. The Andean spine is therefore not merely a deployment site in this arc — it is the destination. Building the corridor's capacity is building it where the arc's own projections say the population arrives.

§11 · What Would Falsify This

1. A dose-controlled cirrus field trial with a monitored radiative signal. That would reduce §2's physical objection to a governance problem — though not the formal one, which is why §10 retains the observation network rather than discarding the aerosol line.
2. Model-structure sensitivity in the Amazon thresholds. If the 1.5–1.9 / 3.7–4.0 gap collapses across independent model families, §6's leverage claim weakens proportionally. The estimate is recent and not yet replicated outside its originating family. This is the weakest load-bearing number in the paper and is flagged as such. What the test should be: raw °C thresholds are expected to differ across models — CMIP families do not agree that the Amazon tips at all — so disagreement in the raw numbers is not by itself refutation. The claim to test is that the value normalised by the basin geometry each model produces is stable, since the fold type is a contact invariant while the model's particular parameterisation is not. If the normalised value scatters as widely as the raw one, the invariance claim fails and §6's leverage is model-specific.
3. A post-moratorium rise in cross-border leakage. The sub-3% figure was derived under an intact moratorium. If it rises now, §8's corridor-wide framing becomes more urgent, not less — but the number needs re-estimation before it is leaned on again.
4. A demonstration that an in-layer microphysical operator fails to commute with the transport term under a more complete model than WP39's three-layer box. That would restore WP41's identification directly, and is the cleanest possible refutation of §2.

§3's four tests are not falsifiable by evidence — they are a design constraint. But they predict that distributed climate programmes failing test 4 stall at pilot scale regardless of funding. That is checkable.

§12 · Cross-Reference — Where This Lands

Destination	Content	Status
book6/wp66-… (this paper)	Operator correction, four tests, corridor as gate	WRITTEN
wp41-planetary-triage.html	Amend §K-identification with a pointer to §2; finance and triage framing unchanged	DONE
wp40-aerosol-engineering-pedagogy.html	Add a framing note: curriculum trains measurement, not deployment (§10)	DONE
wp43-immediate-action.html	Add forward pointer to §9 — the mandate is satisfied by a different gate	DONE
wp44-catastrophe-manifold.html	Add the measured instance: Amazon CSD at 76% of grid cells as a fired precursor	DONE
wp39-smoke-transamerican.html	No change. T1/T2 stand as proved; only the downstream identification moved	NO CHANGE
wp44-catastrophe-manifold.html · O1	Mark the corridor as the first O1 test bed with measured precursors; the calibration itself remains open (§7.2)	STILL OPEN
wp44-catastrophe-manifold.html · O1 method	Respecify O1 as numerical continuation in a coupled model with interactive vegetation, replacing "collaboration with atmospheric scientists working in GCM reanalysis." Reanalysis pins the present-day state and supplies the CSD approach; it cannot produce a basin	DONE
wp43-immediate-action.html · doctrine	Replace "climate forcing is autonomous" — the external-parameter reading — with the endogenous formulation. The mandate is unchanged and its reason is stronger	DONE
wp45-dividing-unity.html	Add forward pointer to §7.2 — \(\varepsilon_0\)'s notation matters most in the one place the constant is uncalibrated	DONE
wp65-the-oracle-outside.html	Add forward pointer to §3.1 — the campaign is evidence about motivation, not only about verification standards	DONE
wp64-the-recorder.html	Add forward pointer to §7.1 — §2.4's self-regulation argument has a measured climate instance	DONE
index.html	WP66 entry; mark WP39–45 + WP66 as the climate arc	DONE
Lean · ZeoliteCommutation.lean	The §2 claim is formalised: gate_commutes (0/1 gate ∘ pointwise fold, for every state), coupling_not_commute, gate_fold_not_commute. 3 theorems, 0 admits, 0 sorries, axioms [propext, Classical.choice, Quot.sound]. Algebra only — not the box model, not T1	CLOSED · KERNEL-CHECKED

§13 · Summary

WP41 is right that triage is necessary, right that the money exists, right that the obstacle is political. WP43 is right that the gate cannot wait. The arc's error is one substitution: an aerosol released inside the mixed layer acts pointwise, and by the series' own Theorem T2, a 0/1 gate commutes with pointwise loss. The order-dependence that justified deploying \(K\) does not survive the identification.

Cold is not the lever either; albedo is, and the arithmetic of manufactured ice does not close. What remains is the surface — and on the surface the leverage is concentrated in one place. Halting deforestation in the Andes–Amazon corridor roughly doubles the warming that continental hydrological system survives. It requires no new technology, passes every test distributed deployment demands, has a working precedent at the scale of millions of hectares, and its precursors are already instrumented.

The honest form of mass climate deployment is not a billion people synthesising ice nuclei. It is a few million in one corridor keeping the wall fed — and a far larger number applying organised pressure to the markets and legislatures that decided, in January, to stop protecting it.

Addendum · August 2026 · the parameter is on the manifold

§6 was written as a leverage argument: halting clearing roughly doubles the warming the system tolerates, and no aerosol proposal in the literature offers leverage of that order. That reading stands. It also undersells what the number shows.

In the standard presentation of catastrophe theory the controls sit in a plane outside the system. On that reading the fold has a fixed address, intervention moves the state back along a fixed landscape, and the threshold is wherever it was. A system with external controls cannot move its own threshold. This one does: the fold sits at 1.5–1.9 °C under continued clearing and at 3.7–4.0 °C under a halt, and the difference is a land-use decision. The threshold is not a fact about the climate that we approach at some rate — it is a coordinate we are standing on.

Three consequences, recorded here and stated fully in WP44's August 2026 addendum:

• §7.2's O1 test bed is a continuation problem. The basin is counterfactual; reanalysis is one realised path. It fixes the approach to the separatrix and pins present-day state, and it cannot show the far side. §7.2 above now says so.
• §11's model-structure risk has a sharper test. Raw °C thresholds are expected to differ across model families. The claim that survives — if any does — is the value normalised by the basin geometry each model produces, because the fold type is a contact invariant and the parameterisation is not.
• §9 and WP43 stop being two arguments. If K and the fold share a manifold, then acting on K is acting on where the fold is. "The gate cannot wait" and "the wall is already standing" are the same statement in two vocabularies — which is why the doctrine survives the collapse of the aerosol identification in §2 without needing a replacement lever. Full treatment: WP69, forthcoming.

References

1. WP39 · Smoke as a Planetary Operator; WP40–WP44, this series.
2. Shirky, C. Here Comes Everybody: The Power of Organizing Without Organizations. Penguin, 2008.
3. SRM360, Cirrus Cloud Thinning.
4. Gasparini et al., Does prognostic seeding along flight tracks produce the desired effects of cirrus cloud thinning? ACP 23 (2023).
5. Tully et al., Estimating the potential cooling effect of cirrus thinning via seeding. ACP 21 (2021).
6. Urea as an Ice Nucleant for Supercooled Clouds. J. Atmos. Sci. 23(2), 1966.
7. Mauritsen et al., Earth's Energy Imbalance More Than Doubled in Recent Decades. AGU Advances (2025).
8. Cambridge Bay sea-ice thickening trial, winter 2024–25.
9. Beerling et al., Challenges and opportunities in scaling enhanced weathering for CDR. Nature Rev. Earth Environ. (2025).
10. LBNL Heat Island Group, Global Cooling: reflective roofs and pavements.
11. The Andes–Amazon–Atlantic pathway: a foundational hydroclimate system. PNAS (2024).
12. Deforestation-induced drying lowers the Amazon climate threshold. Nature (2026).
13. Deforestation and warming could push Amazon to tipping point by 2040s. Mongabay (2026).
14. Boulton, Lenton & Boers, Pronounced loss of Amazon rainforest resilience since the early 2000s. Nature Climate Change 12 (2022).
15. Poveda et al., Critical Slowing Down Reveals Hydrologic Resilience Loss Across Amazon Sub-Basins. Water Resources Research (2026).
16. Villoria et al., Leakage does not fully offset soy supply-chain efforts to reduce deforestation in Brazil. Nature Communications 13 (2022).
17. Global Canopy, The hidden crisis of deforestation in Bolivia.
18. Amazon deforestation may rise 30% as major traders exit historic soy pact. Mongabay (Feb 2026).
19. World Agroforestry, Niger's re-greening revolution; Sci. Rep. 10 (2020), Drivers of FMNR in the Sahel.
20. Fog collection potential in arid urban lands — Alto Hospicio, Chile. Front. Environ. Sci. (2025).
21. England et al., Drivers of Antarctica's sea ice decline. UNSW / Science Advances (2026).

Energy-balance and albedo figures in §4 are first-order calculations from stated parameters (latent heat of fusion 334 kJ kg⁻¹; Earth surface area 5.1 × 10¹⁴ m²; 120-day melt season at ~200 W m⁻² mean insolation, Δalbedo 0.5) and are order-of-magnitude bounds, not published values. All other figures are sourced above.