This is the chapter where the argument becomes geography, and where it very nearly becomes the worst thing in the book. Both facts are in the arithmetic, so the arithmetic goes first.
Three Colours
Working Paper 42 sorts the world into three conditions rather than a gradient, which is the right shape for a folded system. Green: warming under 2 °C, precipitation within ten per cent of historical, water abundant, sea level under 0.3 m. Orange: two to four degrees, precipitation swinging twenty per cent, water marginal — habitable, but only with intervention. Red: over four degrees, precipitation down thirty per cent or worse, water exhausted, sea level over a metre. Habitability breaks. Not degrades.
The colours are not predictions about particular places at particular dates. They are a classification of conditions, and the useful question is which places pass into which class, and when. [MODEL]
What the Refugia Actually Hold
| Zone | Geography | 20-year capacity | 200-year | Binding constraint |
|---|---|---|---|---|
| Andes Altiplano | Peru, Bolivia, southern Colombia; 2,500–3,500 m | 5–8M | 5–10M | Water: glacier → aquifer; limited crop range; altitude |
| Mountain West, US | Colorado, Utah, Wyoming, Montana, New Mexico; 1,500–3,000 m | 2–5M | 2–4M | Snowmelt → aquifer decline; low agricultural density; fragmented land title |
| Pacific Northwest | Seattle, Portland, Vancouver — marginal | 3M, under stress | 1–1.5M | Summer water; sea level; requires dikes and imported food |
Total green and marginal capacity at the two-century horizon: 12 to 18 million people.
Against a present population above eight billion. That is one refuge place for every four to seven hundred people alive.
The Number Forbids the Obvious Reading
Print that ratio next to the word refugia and the romance dies immediately, which is why it is printed here rather than in a footnote.
Refugia are not a plan for humanity. They cannot be. At one in five hundred, a refuge strategy offered as the response is not a response at all — it is door C from The Accounting, relocation of whoever can pay, wearing the vocabulary of planetary care. That door is the one history usually takes, and a book that argued for it while sounding like this one would be doing real harm.
The refugia are the residual, not the strategy. They are what the arithmetic leaves after mitigation has been pursued as hard as it can be, and their smallness is the argument for mitigation rather than a consolation for its failure.
The partial-relocation figure in WP41 — two to five hundred million people, $20–100 trillion, resource-feasible if subsidies are redirected — is a different and larger operation than these three zones, and it depends on receiving capacity being built rather than found. The table above is what exists. Everything beyond it has to be made, starting with water.
The Refugia Are Not Empty
Everything above is written as though the Altiplano were a shelf with room on it. It is not. It is inhabited, and has been for thousands of years, by communities whose claim to it long predates the arithmetic in this chapter and does not weaken because somebody in New Jersey ran a carrying capacity model.
WP42 raises this in a line and it deserves considerably more than a line. A green zone facing a five- to tenfold population increase over thirty years is facing a sovereignty problem before it is facing a water problem: indigenous communities, national governments, and arriving migrants, with incompatible claims and no framework for adjudicating between them. The paper’s instruction is blunt and correct — sort this now, not during crisis — because the alternative is that it gets sorted during the crisis, by whoever arrives with the most force.
A plan to move millions of people into land that other people live on, written by the people doing the moving, is a plan with a name in history, and the name is not refuge.
This book cannot solve that and should not pretend to. What it can do is refuse the framing that makes it invisible. The refugia are not capacity awaiting occupants. They are places with prior inhabitants who did not cause this, whose consent is not a formality, and who are entitled to be parties to the arrangement rather than a constraint listed in a table. Any use of the word refuge that skips past that has already become the thing it was trying to prevent.
[OPEN] How consent at that scale is actually obtained, and what a receiving community is owed by those it receives, is not answered anywhere in this corpus. It is the largest unaddressed question the refugia strategy raises, and it is a question for law, politics and the communities themselves, not for an operator algebra.
Two Hundred Years Out
The twenty-year column is bad and the two-century column is worse, and WP42 prints it rather than stopping at the horizon that flatters the argument. [OPEN] Everything in this section is a projection two centuries deep and should be read with the humility that deserves.
At two and a half to three and a half degrees stabilised, the Andes hold — buffered by altitude, glacier-fed water replaced by aquifer by around 2150. The Mountain West holds and then contracts, because aquifer depletion becomes the binding crisis at roughly the same date and forces a downsizing from five million to two. The Pacific Northwest crosses from orange to red: winter precipitation up fifteen per cent, summer down twenty-five, a metre of sea level, chronic flooding, and either massive restructuring of the cities or their abandonment.
And the Sahel goes red at three to four degrees with precipitation down forty to sixty per cent, which WP42 renders in one clause it does not soften: the Sahara expands to the Mediterranean coast, with aquifers exhausted half a century before that. South Asia goes red, the deltas uninhabitable, and ice nucleation ineffective at that scale by then — the gate having become local, exactly as the previous chapter said it would.
The Two Scenarios, and the Distance Between Them
WP42 sets the choice out as two pathways, and the gap between them is the closest thing this book has to a statement of what is at stake. [MODEL]
| A · Planned | B · Unplanned | |
|---|---|---|
| Relocation to 2050 | 300M, roughly 10M a year | 50–100M, chaotic and conflict-driven; receiving borders close |
| In-place defence | $5–10 bn a year of seeding across South Asia, the Sahel, Southeast Asia | $0.1–0.5 bn — pilots |
| Outcome | South Asia declines from 2 bn to 1.5 bn slowly and by plan; the Sahel still collapses; Southeast Asia partly sustained | Forcing outpaces adaptation; redistribution by famine, disease and conflict |
| Toll by 2050 | The difference between the pathways is put at 500 million to 1 billion people | |
[OPEN] Those figures are the widest-bounded numbers in this book and the ones most dependent on assumptions nobody can validate. They are printed because the shape is robust even where the magnitudes are not: the planned pathway and the unplanned one do not differ by a percentage. They differ by a number with nine digits in it, and the difference is decided by whether the arrangements exist before the movement starts or have to be improvised while it is happening.
Note also what Scenario A concedes. The Sahel collapses in both columns. Seeding is ineffective there at the scale required, and the planned pathway does not save it — it saves elsewhere, and is honest about where it fails. A strategy that claimed to rescue everyone would be a strategy nobody had costed.
If You Live in a Green Zone
WP42’s practical section is addressed to people who already live where everyone else will need to go, and it is worth carrying over intact because almost nobody in those places thinks of themselves as holding a position of consequence.
Secure water. Rights, storage, cisterns, aquifer access. Water is the binding constraint in every green zone in the table, and it binds around 2150 in both of them. What is arranged now is what exists then.
Learn to farm at altitude. Potatoes, quinoa, high-altitude grains, and the modern agronomy that makes them yield. The crop range is narrow and the knowledge is local and mostly held by the people already there — which is another reason the sovereignty question is not separable from the practical one.
Build for a five- to tenfold increase. Schools, clinics, housing, over thirty years. Infrastructure at that multiple cannot be improvised and is not expensive if started early, which is the same structure as the gate: cheap now, impossible later.
Settle the governance before the crisis. Who decides, who is owed what, on whose authority. Sorted in advance it is politics. Sorted during the arrival it is something else.
Which Makes the Work Specific
The value of an address is that it converts a disposition into a task. Not care about the future, but: the Altiplano’s binding constraint is the handover from glacier melt to aquifer, and somebody has to model it before the inflow rather than after. The Mountain West is aquifer-limited by the middle of the next century and its land title is fragmented, which is a legal problem before it is a hydrological one, and legal problems take decades. The Pacific Northwest is marginal, needs dikes, and imports its food.
None of that requires anyone’s permission to start on. It requires somebody to pick one and do the next real piece of work — which, by this point in the book, the reader will recognise as the only instruction that has been given at all.
[OPEN] Every capacity figure above is an estimate with wide bounds, and the two-century column is more uncertain than the twenty-year one by a large factor. They are given because a number that can be argued with is worth more than a gesture that cannot.
Hope, With Its Address Written Out
So: hope in this book means the Altiplano between 2,500 and 3,500 metres, and the aquifers under Colorado, and a water plan drawn before the people arrive. It means the correct operator order applied by whoever is in a position to apply it, early and incompletely. It means twelve to eighteen million places that exist, and the far larger number that has to be built, and the honesty to keep those two apart.
It is a smaller hope than the one most books offer, and it is the only one with a postcode.
Sources and provenance: zone definitions, carrying capacities and the twenty- and two-hundred-year projections are from Working Paper 42, Refugia and Climate Impact Zones; the partial-relocation costing is WP41, Planetary Triage. Capacity figures are estimates with wide bounds and are marked [OPEN]. The three-door framing is set out in The Accounting.
Next: The Wager — the whole of it in one breath.