The Accounting

A book that ends in hope has to earn it somewhere, and the only place it can be earned is at the beginning, in the ledger. So here is the ledger. It is not softened, and it is not inflated, and the second of those is harder than the first.

Inflation is the standard failure of writing about catastrophe. The largest available figure gets quoted because it is the most arresting, and the argument then rests on a number the author could not defend under questioning. This desk has spent a great deal of the year building machinery to catch exactly that move in its own work. The rule applied here is the one applied there: use the tightest figure that has a stated method, name the source, and mark the places where the honest answer is that nobody knows.

Who Is at Risk, and Where

The compilation below is from Working Paper 41, drawing on IPCC AR6, the World Bank’s Climate Risk and Adaptation Assessment, and NOAA projections. The population figures are observations. The timelines are not: they are projections, and they are the softest thing on the page.

Two thousand two hundred and seventy million. It is worth writing out, because the numeral slides past the eye at a speed the fact does not deserve.

What It Would Cost to Move Them

This is where the accounting stops being an inventory and starts being an argument, so the arithmetic is given in the open.

Moving everyone. At $100,000 to $300,000 per person for housing, infrastructure and livelihood, relocating all 2.3 billion costs $230 to $690 trillion, over a thirty-year window that would mean moving 77 million people every year. Global urban infrastructure has no capacity remotely near that. The total exceeds twice world GDP, sustained indefinitely. The verdict in WP41 is one word: impossible. Not difficult. Not expensive. Outside the space of things that can be done.

Moving some. Two to five hundred million people into high-altitude refugia in the Americas costs $100,000 to $200,000 per person — less, because the land exists and the infrastructure burden is lower — for a total of $20 to $100 trillion across thirty to fifty years. That is $0.7 to $3 trillion a year. The verdict there is different: resource-feasible, if subsidies are redirected. Politically implausible. Financially possible. Those are not the same sentence, and the gap between them is the subject of the rest of this book.

The Paradox on the Page

Everything necessary — decarbonisation, local adaptation, operator interventions, partial relocation — totals $2 to $6 trillion a year, between two and six per cent of world product. Against that, from SIPRI’s 2025 figures and the IMF’s 2023 accounting including externalities: military spending runs at $2.4 trillion a year and fossil fuel subsidies at $7 trillion. Redirected, that is $9.4 trillion a year available against a $2–6 trillion need.

The money exists. It is spent on other things. That is the whole of the paradox, and no part of it is a technical problem.

Which leaves three doors, stated in WP41 without euphemism:

A. Full decarbonisation, limited relocation of around 500 million, operator interventions.
B. Partial decarbonisation and a mass relocation attempt — which is to say, collapse.
C. Relocation of those who can pay, and operator-ordered defence for everyone who remains.

The paper says history suggests C, and proposes A. Both halves of that sentence should be read. It is not a document that mistakes its preference for a forecast.

What We Do Not Know

Now the other discipline, the one that is harder to keep.

The population figures above are counts. The cost ranges are estimates with a stated method and a factor-of-three spread that is printed rather than hidden. [OPEN] The timelines are the weakest column in the table, and this book will not at any later point treat them as though they were the strongest. A window of 2030–2070 is not a prediction that something happens in 2050. It is an admission that the year is not known within forty years, and the honest way to carry an admission is to keep carrying it.

So the doom dates are not stated here, and the reader should be suspicious of any book that states them. Not because the danger is smaller than claimed — the count is the count — but because a false precision about when is the exact move that lets a reader dismiss the whether once the date passes. Every failed deadline in the history of this argument has been paid for by the argument, not by the person who set it.

If You Do Not Accept the Table

Some readers will have stopped believing this chapter at the first row, and it is worth addressing them directly rather than writing the rest of the book around their absence.

Distrust of institutional science is not, in every case, ignorance. Some of it was earned, and a book that opens by calling it stupidity has lost the reader it most needed. Projections have been overstated in the press by people who never read the confidence intervals. Deadlines have passed without the announced consequence and nobody was held to account for having set them. This desk has spent the year cataloguing its own version of that failure — claims filed under evidence that did not support them — and is in a poor position to be indignant about anyone else’s scepticism.

So take the strongest form of the objection and grant it entirely. Strike the table. Every row, every cost range, every window. Assume the compilation is wrong in ways this author cannot see.

Here is what survives that deletion:

The gate and the fold do not commute. That is a theorem, it is machine-checked, and it is a statement about operators rather than about the atmosphere. Nothing in the proof mentions climate, and nothing in it would change if the climate literature were withdrawn tomorrow.

The asymmetry that follows from it — that a system in which one party holds the gate and another bears the fold is a system with the wrong operator order — survives too, and can be checked against cases the sceptical reader already accepts. A supply chain whose costs land four countries away from the purchase. A municipality that permits a development and inherits the drainage. A firm that books the revenue this quarter and the liability in someone else’s. The structure is the same structure, and it does not require you to believe a single projection about 2050.

And the fold geometry survives, because it is mathematics. Systems near a fold behave in a particular way whatever the system is made of.

If you accept the table, the theorem tells you what the response must look like. If you reject the table, the theorem is still true and still applies to whatever you do accept is happening. You are in the same book either way. That is not a rhetorical trick — it is what it means for a result to be domain-free.

The one position this book cannot supply an argument for is the one that rejects the table and declines to name what it accepts instead. That is not scepticism. Scepticism commits to something; it just refuses to commit to this.

Why This Is a Fold and Not a Slope

One structural claim, and it is tagged. [MODEL]

The intuition most readers carry is of a slope: things get worse in proportion to the cause, and a late correction buys back a late amount of damage. The catastrophe manifold developed in Working Paper 44 says the geometry is not that. There is a fold in the surface, and on the far side of it the system is not merely worse — it is in a regime where the ordinary relation between effort and result no longer holds. Determinism survives. Predictability does not.

This is the same structure named in Vilomah, in a register that has nothing to do with grief and everything to do with the same algebra. The gate K and the generative fold F do not commute. Applied in the correct order, a controlled transition; applied in the wrong one, the chaotic window. A fold is what a wrong order looks like when the system is a planet rather than a family.

The one piece of good news in the geometry is Prediction 1 of that paper: the fold is visible before it fires. A system approaching a fold announces it — in slowing recovery from small shocks, in variance that climbs before the mean moves. That is a testable claim and it is the reason Movement II is able to say anything at all. If the fold were invisible until it fired, this book would end here, and it would be a different book.

What This Chapter Is Not

It is not a forecast. Nothing here says what will happen; the numbers say what is at stake and what the response would cost, and those are different objects.

It is not despair, either, and the distinction matters more than it looks. Despair is a conclusion. This is an inventory. An inventory is the thing you take before deciding, and the deciding is the subject of everything that follows: that a catastrophe manifold has more than one basin, that order and timing change which one a system falls into, and that the single variable none of these models can close is the one the reader is holding.

The price of admission has now been paid. Whatever hope this book argues for from this point forward, it argues for having written the table above and left every uncomfortable number in it.

Sources and provenance: figures compiled in Working Paper 41, Planetary Triage: Operator-Ordered Survival, from IPCC AR6, the World Bank Climate Risk and Adaptation Assessment and NOAA projections; capital figures from SIPRI (2025) for military spending, the IMF (2023, including externalities) for fossil fuel subsidies, and the OECD for pledged climate finance. Impact-zone definitions and carrying capacities are in WP42, Refugia and Climate Impact Zones; the fold geometry and its three predictions are in WP44, Disaster Theory and the Climate Catastrophe Manifold; the doctrine of immediate action is WP43, The Gate Cannot Wait. The cost ranges are estimates and are given as ranges for that reason.

Next: the numbers describe a system with a fold in it, and a fold has more than one side. More Than One Basin takes up the second movement — why prediction is not destiny, why collapse forecasts have a long record of underpricing adaptation, and what it means that the models close every variable except us.