The Treaty That Worked

Section 1The question before us

Katherine Richardson, who led the team that quantified the planetary boundaries, explains what crossing one means with an analogy worth borrowing. Blood pressure above 120 over 80 does not guarantee a heart attack. It raises the probability. So you bring it down.

That is the entire epistemic posture of this paper. We are not going to tell you the world ends on a date. We are going to tell you that six measurements are outside the range within which human civilization developed, that this raises risks by amounts that can be estimated, and that a reasonable person responds to a bad reading by treating it rather than by arguing about the thermometer.

The argument about whether it is real has consumed the argument about what to do, which is a luxury the people having it can afford and the people who come after them cannot.

This paper does not solve any of the six. It proposes the only thing an American organization can usefully propose about problems that no country can solve alone: a mechanism. And it observes that the mechanism exists, has universal ratification, and has one clean success to its name.

1.2   And the reason this cannot wait

The Montreal Protocol was signed in 1987. Developing countries completed their phase-out in 2010. That is twenty-three years from signature to global compliance, on the easiest environmental problem humanity has ever faced — one class of chemicals, a few dozen manufacturers, and a substitute already on the bench.

Estimates of when the Amazon crosses the threshold at which it stops generating its own rainfall cluster in the late 2030s and 2040s.4

A treaty begun today, moving at the speed of the fastest environmental agreement ever concluded, arrives at roughly the moment the forest stops being a forest.

We put that at the front rather than in the footnotes because it is the whole case. Every argument in this paper about mechanisms, amendments and funds is an argument about a process that takes decades, applied to problems measured in decades, with no margin left for the years we will spend deciding whether to begin.

Section 9.6 states the objection in its sharpest form and concedes we have no good answer to it. There is only one response available to anyone who takes the arithmetic seriously, and it is not an argument. It is a date. The work has to start before the evidence is comfortable, because by the time it is comfortable the instrument no longer reaches the problem.

This is the part of the paper that is a call rather than an analysis. The generation that wrote the ozone treaty acted on a thirteen-year-old paper, against an industry that disputed it, about harm nobody could see. They were not certain. They were early, which is the only way anyone is ever useful about a problem like this.

1.3   Our own share of the failure

Conservation is a word that belongs to the tradition this series draws on. The instinct to husband a resource, to leave a place better than you found it, to distrust anyone who consumes capital and calls it income — that is not borrowed from anybody. It is the same instinct this organization applied to the federal balance sheet in Issue Paper No. 1 and to corporate balance sheets in Issue Paper No. 16.

It was given away. Over roughly thirty years the environment became somebody else's subject, and that tradition's contribution narrowed to disputing whether each finding was true. In the process it stopped making the arguments that were its own — that pollution is an uncompensated taking from a neighbor, that a firm should pay the costs it imposes, that a nation drawing down an aquifer is eating its seed corn.

There is a second admission, and it is specific. Almost every consensus in this subject is now treated as contested because one question inside it is genuinely contested. Nobody disputes that nitrogen runoff creates dead zones. Nobody disputes that an aquifer that does not recharge is finite. Nobody disputes that particulate pollution damages lungs. Those were never in question, and they were allowed to be swallowed by an argument about climate attribution — which cost the ninety percent everyone agreed on in order to keep fighting about the ten percent they did not.

Section 2The measurements

In 2023 a team led by Richardson published the first quantification of all nine planetary boundaries in Science Advances. The boundaries are not predictions. They are the ranges within which the Earth system has operated through the entire span of human agriculture and civilization.1

Six are outside that range. We set out five of them below, and a sixth in Section 3, at the level of detail the evidence supports. For each, the question we are interested in is not what should be done — it is why no country can do it.

2.1   Biogeochemical flows — the worst one, and it is about food

Roughly 190 trillion grams of nitrogen are applied to the world's cropland each year, against an assessed safe level around a third of that. It is the most badly transgressed of the nine.2

And roughly half of it never reaches a plant. Estimates of global nitrogen use efficiency range from about 42 to 46 percent, and the figure has fallen since 1961 — by 22 percent on one accounting. A fifty-year budget for maize, rice and wheat found that 52 percent of total nitrogen output was lost to the environment.2 The remainder volatilizes, leaches or runs off, producing dead zones now reported from more than 400 systems, affecting a total area of more than 245,000 square kilometers.2 The European Nitrogen Assessment put the cost of nitrogen pollution in the EU alone at €75 to €485 billion a year — one to four percent of the region's GDP — against direct farm benefits from nitrogen inputs of €20 to €80 billion.2

Two things make this difficult beyond any nation's reach.

Half the world's food is grown with synthetic nitrogen. This is not a pollutant to be eliminated; it is the input feeding roughly four billion people. Any proposal treating it as waste is proposing famine, and the target is efficiency rather than abstinence.

And phosphorus is a different problem wearing the same coat. Nitrogen is 78 percent of the atmosphere — the constraint on supply is energy, not scarcity. Phosphorus is mined, has no substitute in agricultural production, and is concentrated in a handful of countries, which is why the European Union lists phosphate rock among its critical raw materials. Nitrogen is a pollution problem. Phosphorus is a depletion and geopolitical problem. A single country can regulate its own farms and affect neither.2

We note one working example, because the problem is not intractable. Denmark has more than 60 percent of its land farmed and one of the most intensive export-oriented livestock sectors in the world. Across a series of national action plans beginning in the mid-1980s, its nitrogen surplus fell from 490 kilotonnes in 1985 to 278 in 2008, nitrogen efficiency rose from 27 percent to 40, and leaching from the root zone fell from 334 kilotonnes to 159 — approximately halved, while crop yields were maintained and animal production increased by around 30 percent.2

The groundwater confirms it. In the youngest oxic groundwater — nought to fifteen years old — 44 percent of measurements show significant downward nitrate trends, against 9 percent in water 25 to 50 years old.2 Point-source discharges fell 74 percent between 1989 and 2003.2

Two honest qualifications from the same research. The surplus fell by more than 50 percent between 1990 and 2011, but loadings to the aquatic environment fell only about 42 percent and ammonia deposition only 25 — because nitrogen already in the soil keeps moving for decades.2 And it took thirty years. Denmark is one country of 5.9 million people. The Mississippi drains thirty-one states.

2.2   Land system change and biosphere integrity

Both are transgressed, and the driver is overwhelmingly agriculture — the conversion of forest and grassland to cropland and pasture, and the reduction in the share of the planet's biological production left to everything that is not us.1

Section 3 treats the Amazon as the case, because it is the one where the irreversibility is measurable and the deadline is inside a normal human career.

2.3   Climate change

Transgressed, and we are going to spend less space on it than the reader expects — not because it is small, but because it is the one boundary where the public argument has already consumed all available oxygen, and because treating it as one of nine is more useful than treating it as the whole subject.

We would note only this, and it is a fact rather than a position. The single largest measured reduction in warming achieved by any international agreement was produced by a treaty that was not about climate at all. Section 4 explains why.

2.4   Freshwater change

Transgressed, and it contains the clearest case of a genuinely finite resource being consumed as though it were income. Fossil aquifers do not recharge on any human timescale; water drawn from them is gone in the way an ore body is gone. The Ogallala is the American example and the depletion is documented, though we have not assembled the figures to the standard this series applies and will not quote them from memory.

2.5   Novel entities

Transgressed — the category covering synthetic chemicals, plastics and materials released faster than their effects can be assessed. It is the newest boundary and the least quantified, and its inclusion is itself the finding: the assessment concluded that the rate of introduction exceeds the rate at which safety can be established.1

2.6   And the one we fixed

Stratospheric ozone depletion is inside the safe boundary. It is the only one of the nine that was crossed and has been brought back.

It is also the only one that was addressed by a binding global treaty. That is the argument of this paper and it is not a coincidence.

Section 3Why one country cannot do it: the Amazon as the case

We use the Amazon because it demonstrates every feature that makes these problems unreachable by national law, including the awkward one — that the country concerned has already proven it can act, twice, and that this still is not enough.

3.1   What is at stake, in numbers

The Amazon holds 150 to 200 billion tonnes of carbon — roughly fifteen to twenty years of global emissions. Its trees contain about a fifth of all carbon captured by vegetation on Earth.4

The forest generates its own rainfall by recycling moisture inland from the Atlantic. Lovejoy and Nobre originally put the threshold for that cycle failing at 40 percent deforestation; in 2018, accounting for warming and fire, they revised it to 20–25 percent. The 2021 regional assessment found 26 percent already deforested or degraded, with 20 percent irreversibly lost.4

Part of it has already flipped. Research supported by NOAA found the southeastern Amazon — about 20 percent of the basin, 30 percent deforested over forty years — has become a net carbon source, with dry-season rainfall down 25 percent and temperatures up at least 2.7°F.4

If the forest savannized, local rainfall would fall by up to 30 percent, with effects reaching Colombia and Argentina. Estimated economic damages run $957 billion to $3.6 trillion over thirty years — against a Gross Brazilian Amazon Product of about $150 billion a year.4

And it may take centuries to reverse, if it can be reversed at all.

3.2   Why national action is necessary and insufficient

Brazil has already done it. Twice. Deforestation fell 83 percent between 2004 and 2012 through enforcement, satellite monitoring and protected areas. The program was revoked in 2019 and clearing rose 60 percent. It was relaunched in 2023, and alerts for the year ending July 2026 came in at 5,119 square kilometers, the lowest since 2013.5

So the capability exists and the technique is known. Brazil's DETER system issues daily alerts on clearings as small as three hectares, feeding directly to enforcement agencies. Nobody needs a foreign commission to find the chainsaws.5

What no country can fix from inside is the structure of the incentive.

The land. Around 50 percent of Brazilian Amazon deforestation occurs on public land, particularly in what are called undesignated public forests — untitled lands belonging to no tenure category specified by law. These cover 56 million hectares, the size of Spain, and hold seven billion tonnes of carbon, almost a year of global emissions.6

The sequence is documented in the peer-reviewed literature and it is not subtle: claimants "invade public lands, deforest the area to signal occupation, request land titles to governmental agencies, and often lobby to modify land laws in favour of title acquisition." One assessment of a 2017 land law found it reinforced "a mechanism for acquiring land rights historically linked to deforestation, since land grabbers clear the forest to signal land occupation and claim land rights."6

By 2018, 2.6 million hectares of undesignated public forest had already been cleared — all of it illegally — and 11.6 million hectares had been registered as private property in the national environmental registry.6

The cattle are not the business. They are the cheapest available proof of possession.

The demand. When the United States imposed a punitive tariff in 2025, Brazilian beef exports to America collapsed by nearly 80 percent — and Brazil's global market share rose to 32 percent, because China absorbed the volume. Western consumption leverage over this problem was tested last year and found to be approximately zero.7

And the politics. The 83 percent reduction survived until an election. Everything achieved is one administration from reversal, which is not a flaw in Brazilian governance but a feature of governance generally.

That is the pattern across all six boundaries. The technical answer is usually known. The actor who must implement it bears the cost and shares the benefit with everyone, which is the oldest problem in political economy and the reason treaties exist.

Section 4The one that worked

The Montreal Protocol is the only environmental treaty ratified by every country on earth — 198 parties. It has eliminated nearly a hundred ozone-depleting substances. All 142 developing countries reached 100 percent phase-out of CFCs and halons by 2010. The ozone layer is on track to recover by mid-century.3

And although it was written about ozone, it avoided an estimated 135 gigatons of CO2 between 1990 and 2010 and up to 0.5°C of warming, delaying an ice-free Arctic summer by fifteen to forty years.3

Four features explain it, and each maps onto a failure elsewhere.

A substitute existed. Chemical companies developed replacements quickly, and DuPont — once it recognized regulation was coming — shifted from opposing the treaty to investing in the alternatives. The consumer's refrigerator worked the same and nobody was asked to give anything up.

Money removed the equity objection. The Multilateral Fund, created in 1991, has pledged nearly $5 billion and disbursed $3.9 billion paying the incremental costs of compliance in developing countries. Its Executive Committee has equal representation from developed and developing nations. That answered the standard objection — that poor countries were being asked to forgo growth to fix a problem rich countries created.3

Trade provisions solved free-riding. Non-signatories could trade controlled products only with a shrinking share of the world economy. More members meant more pressure on the remainder, which brought in more members. The mechanism was self-reinforcing.3

And it started narrow and ratcheted. It was not global at the outset — industrialized countries first, modest initial commitments, binding follow-up meetings that raised ambition as evidence accumulated. Countries that had already invested in compliance had no reason to leave.

4.1   And it has already been extended beyond its subject

This is the fact that makes the proposal practical rather than fanciful.

The Kigali Amendment of 2016 extended the Montreal Protocol to hydrofluorocarbons. HFCs do not deplete ozone at all. They are greenhouse gases, brought under an ozone treaty because the machinery — the fund, the assessment panels, the trade provisions, the ratcheting meetings — already existed and worked.3

Institutional credibility takes decades to build and cannot be conjured by naming a new body. This one has universal ratification, forty years of compliance data, a funded mechanism with balanced governance, and a demonstrated capacity to expand its own subject matter.

Section 5Reform I: Propose the amendment

The United States should lead a further amendment extending the Montreal architecture to additional planetary boundaries, taken one at a time, beginning with the one where a substitute is nearest.

The essential design constraint, which is also the answer to the strongest objection against this paper: each new subject requires its own amendment, separately negotiated and separately ratified. Kigali took years and is a treaty in its own right. The institution cannot extend its own jurisdiction, and no country is bound to anything it has not individually ratified.

That is precisely why it is a better vehicle than a new organization with a broad environmental charter. A body with general authority acquires scope over time. A body that must return to every capital for every subject does not.

We do not specify which boundary goes first, because the sequencing should follow the Montreal test rather than the severity ranking: begin where a substitute exists, because that is what made the first one work.

Section 6Reform II: Fund the mechanism

The United States should commit to a stated share of an expanded Multilateral Fund, on the existing model — paying the incremental compliance costs of developing countries, governed by a committee with balanced representation.

This is the part American readers will like least and it is the part that made the difference. Three point nine billion dollars turned the countries most likely to block the treaty into its most successful implementers.

The instruments this fund should be able to use are ones with demonstrated returns. Debt-for-nature exchanges, which restructure sovereign debt against binding conservation commitments and therefore bind across administrations. Enforcement capacity, since detection is solved and consequences are not. Tenure and registry work — designating the fifty-six million hectares in Section 3.2 is a records problem costing hundreds of millions against damages in the trillions. And productivity investment, because Brazilian pasture runs about 65 kilograms of beef per hectare, more than half of it degraded, and intensification could raise national production roughly 30 percent with no new land.8

The Amazon is not expensive to save. It is cheap, and nobody is paying.

Section 7Reform III: Use the trade provision

The amendment should carry the same trade architecture that made the original work: controlled goods may be traded freely among parties and restricted with non-parties.

This is the part with teeth, and it is the part a reader of this series will recognize. It is not a tariff, a subsidy, or an appeal to conscience. It makes membership worth more than exemption, and it does so arithmetically: each additional member shrinks the market available to the holdouts.

We note the honest limit. The European Union's deforestation regulation attempted a version of this unilaterally and has been delayed three times — to December 2026 — because supply chains could not trace their inputs and trading partners objected.9 The lesson is that the compliance machinery must exist before the deadline does, which is an argument for building it multilaterally and funding it, rather than for abandoning the instrument.

Section 8Reform IV: Fund credible communication, and do not run it

Where a change is visible to consumers, the mechanism should fund credible domestic scientific communication — and should never conduct it under its own name.

Montreal did not need this, because CFCs were invisible. The refrigerator worked the same. Nobody was asked to accept anything.

Anything touching food is a different matter, and there is a live demonstration. Seven American states — Alabama, Florida, Indiana, Mississippi, Montana, Nebraska and Texas — have banned cultivated meat, a product that was not for sale in any grocery store and had cleared federal safety review. Florida's governor described it as protecting consumers from "the global elite's plan to force the world to eat meat grown in a petri dish." Mississippi's ban passed unanimously, carrying fines and jail time. The Eleventh Circuit upheld Florida's law in March 2026.10

That is the reaction to a voluntary novel food nobody could buy. It should be read as a forecast.

Two conclusions follow. Prefer changes the consumer does not have to accept — a cheaper product on the shelf rather than a rule about what must be in their food. And where communication is necessary, the messenger cannot be an international body. A distant commission instructing people about their diet is the precise trigger for the reaction above. The evidence on persuasion consistently favors trusted local messengers over institutional ones, which means the fund's role is to resource domestic scientists and clinicians rather than to run campaigns.

Section 10What we are not claiming

We are not claiming the world ends on a date. Section 1 states the posture: these are risk indicators, and a bad reading is treated rather than argued with.

We are not claiming markets have failed. They have solved an enormous share of what was once thought unsolvable, and most environmental improvement in wealthy countries has come through growth and technology rather than restriction. We claim only that a market cannot price a forest nobody owns, an aquifer nobody meters, or an atmosphere nobody holds title to — and that this is a well-understood class of problem rather than an indictment.

We are not proposing a global government, and Section 9.1 explains the specific structural reason this design cannot become one.

And we are not claiming any of this is likely. A treaty amendment requires the consent of countries with no reason to give it, funded by a Congress with no appetite to fund it. We propose it because the alternative on offer is nothing, and because it has worked exactly once, which is once more than everything else has.

Section 11The argument you can carry

The paper compressed to what a person can remember and repeat.

And the one that indicts the argument's own side. Conservation belongs to that tradition — the instinct to husband a resource and distrust anyone consuming capital while calling it income. Thirty years were spent disputing whether each finding was true, and the arguments that were its own went unmade. Nobody disputes that nitrogen runoff creates dead zones or that an aquifer which does not recharge is finite. The ninety percent everyone agreed on was lost in order to keep fighting about the ten percent they did not.

Section 12Conclusion

In 1974 two chemists published a paper suggesting that a class of industrial refrigerants was destroying stratospheric ozone. The industry disputed it. The science was uncertain. The substances were useful, cheap and everywhere, and the harm was invisible, distant and slow.

Thirteen years later every industrialized country signed a treaty. Within twenty-three more, every developing country had eliminated the substances entirely, paid for by a fund the wealthy countries financed. Today the ozone layer is recovering, and the agreement has been ratified by more countries than any other environmental instrument in history — all of them.

Nobody gave up a refrigerator. Nobody rioted. The chemical companies made money on the replacement. And the treaty prevented more warming, as a side effect, than the agreements written about warming have prevented on purpose.

That is the entire proposal of this paper. Not a new organization, not a global authority, not a program of sacrifice. The thing that worked, applied again, to the next problem, one amendment at a time.

Six of the nine measurements are outside the range in which everything we have built was built. The only one we brought back is the only one we wrote a treaty for. A country that can read a chart should be able to read that one.