A local ecology is not healthy if its health depends upon wasting geometry outside its boundary.
Local and Global Ecology ends on a corrective — size responsibility to the scale of the consequence — but it leaves a hard question underneath the whole diagnosis. To size responsibility to consequence, someone has to be able to see the consequence: to detect that cost is crossing a boundary at all. And the obvious way to see it is alarming. Giving one system access to the entire global ecology would mean removing the opacity between states, firms, and people that the surveillance protections exist to defend — universal legibility to a single observer, which is a worse catastrophe than the one it was meant to prevent. So the diagnosis seems to demand a cure the framework has already forbidden.
The way out is a distinction that turns out to be one of the more important in the whole model.
Observability, not transparency
The framework does not need global transparency. It needs enough global ecological observability to detect transferred cost — and those are very different things. Bounded ecological observability is the difference between seeing inside a system and seeing what crosses its edge. You do not need to see the machines, the contracts, or the trade secrets inside a factory to know how much electricity it draws, how much waste it emits, how much product leaves its gates, how many people it employs, and what downstream dependencies it creates. The interior stays opaque; the boundary effects are legible. The same holds for a company, a city, or a state: a system can estimate how much local ecological gain is being purchased by pushing cost into the larger field without knowing anything classified about what happens inside. The effects may be observable even when the interior remains opaque — and that single sentence is what lets ecological accounting coexist with the right to opacity instead of colliding with it. What gets measured is never the inside of a state, a firm, an agency, or a household; it is only what crosses their boundaries.
Boundaries, flows, and burden
Picture every local ecology — a company, a city, a nation, an institution — as a boundary with things moving across it. Not just money: energy, materials, labor, information, risk, dependency, ecological capacity, standing, jurisdiction, pollution, security, technological capability, and social instability all cross that line. Three questions then organize the whole picture. The boundary: what is counted as inside — a corporation, a city, a country, a regional alliance? The flows: what crosses it, and in which direction? The burden: who absorbs the downstream consequence? That last one is the one that matters most. A local ecology becomes suspect at the precise moment its benefit stays inside while its burden systematically crosses outward. At the crudest conceptual level — not a formula yet — this is an ecological balance: local gain, minus the cost exported to surrounding ecologies, equals something like net ecological contribution. It is a far richer instrument than GDP, which counts the gain and is blind to the export.
Ecological transfer
The measurable object beneath all of this deserves a name: ecological transfer — a change in which capacity, risk, burden, resource, or jurisdiction moves from one ecology to another. The first thing to say about it is that transfer is not inherently bad. Trade is ecological transfer. Migration is ecological transfer. Aid, teaching, and the spread of knowledge are all ecological transfer, and a field with no flows across its boundaries is isolated and dying, not healthy. The problem is a specific shape: asymmetric transfer, where one ecology gains while another absorbs a disproportionate burden, and its sharper form, extractive transfer, where a local ecology raises its own viability precisely by reducing viability elsewhere. So the framework does not answer a powerful actor by saying you are bad. It asks a colder and more useful question: what is the net movement of ecological capacity across your boundary — and is your local gain purchased by depleting the field that everyone else also depends on?
An example makes the shape concrete. A firm building advanced AI can create enormous local value — better models, jobs, profit, research, medical advances — while exporting unemployment into other regions, energy demand onto the grid, cyber risk onto everyone connected, dependency onto its users, political concentration onto the system, and strategic instability onto its rivals. At the scale of a state the same structure appears: a state that gains a decisive AI lead becomes locally stronger — better defense, higher productivity, greater wealth — while its rivals accelerate their own militarization, cyber instability rises, supply chains are weaponized, trust declines, and the shared time available for careful governance shrinks. Its local ecological capacity rose; it may have risen by reducing global ecological stability. That is the transfer, and it is why the accounting has to reach past local gain to local gain measured against global capacity destroyed — because those are not always the same number.
Ecological leakage
There is a name for the part of a transfer that hides: ecological leakage — cost that crosses a system’s jurisdictional boundary without being included in the system’s own accounting. Pollution leaks. Labor displacement leaks. Exported risk, surveillance externalities, and destabilization all leak, in the exact sense that they are real costs the system caused and did not count, because they landed on someone outside its circle. Leakage is the accounting boundary problem stated as a flow rather than a circle, and it converts into one of the framework’s most foundational rules:
An ecology cannot claim improvement by excluding significant leaked costs from its own performance measure.
That is the boundary-shaped restatement of no wasted geometry itself. A system that looks like it is getting healthier only because it stopped counting what it dumps over the fence is not getting healthier; it has relocated its geometry waste to somewhere it is no longer charged for.
The whole transaction
Traditional accounting is superb at one half of this and nearly blind to the other. It measures value captured by the participant with great precision, and it measures ecological burden transferred elsewhere hardly at all. Suppose an AI system cuts a company’s labor costs by ten billion. Traditional accounting records a success and stops. The framework’s accounting asks what the first ledger left out: how much unemployment was transferred to households, how much tax burden shifted to government, how much retraining cost moved to communities — and, in the other direction, how much of the gain became available to fund the transition, and how much new capacity was actually distributed back into the field. Only with all of it on the page are you looking at the whole transaction rather than one privileged corner of it. The pattern it exposes has a familiar name in the money case — privatized gain, socialized cost — but the real reach of ecological accounting is that the same shape recurs far beyond money: privatized security paid for with global instability, local informational efficiency paid for with a general loss of privacy, national strategic advantage paid for with arms-race pressure on everyone, corporate productivity paid for with the slow erosion of a society’s own competence. Making those transfers visible is the practical work; the moral judgment is a separate step, and it is left to people.
Who is allowed to measure
This is the dangerous part, and where the design has to be most careful. If a single global institution were to compute that this state is exporting too much ecological cost, the state would simply reject the measurement — and we would be back inside runaway strategic competition, now with a master scorekeeper nobody trusts. Worse, a single authoritative ecological metric — an “Ecological Transfer Index” that rates a country 72.4 and a company 43.8 and is treated as truth — is exactly the gatekeeper the framework exists to prevent: a number granted the authority to decide who counts. So the framework does not imagine one global accountant. It imagines distributed ecological accounting: governments produce their figures, universities produce others, independent institutions and NGOs produce others, AI systems model the same flows under differing assumptions, and international bodies aggregate without adjudicating. Disagreement stays visible on purpose. The measurement itself must remain contestable, because the framework’s deepest rule about numbers holds here with full force — no metric receives ontological authority. Ecological accounting is evidence for deliberation, never sovereignty over it; several models estimating the same transfer differently is not a bug to be resolved into one official figure, it is the thing that keeps the whole instrument correctable.
A realistic path
None of this requires — or wants — full global ecological access any time soon, which was the right instinct from the start: prying open the interiors of states and firms within a few years would cause more destruction than it prevents. Boundary-flow accounting, by contrast, is close to reach, because the world already measures trade, energy, carbon, finance, migration, disease, shipping, and military spending, and advanced models are unusually good at combining exactly those partial signals into a picture of what crosses a boundary. So the sequence is staged. First the concept has to be built — clear definitions of local ecology, transfer, leakage, burden, and capacity contribution — before anything is measured, so the instrument is not quietly shaped by whoever builds it first. Then it is prototyped where observation is already easy: AI labor displacement, data-center energy use, compute concentration, regional economic effects — domains transparent enough to test the method honestly. Only later, and carefully, does it extend toward the hardest cases — strategic AI competition, cyber-risk export, supply-chain dependency, military escalation — and even there the goal is never to abolish state secrecy but to measure enough of the boundary effects that the transferred ecology becomes visible anyway. This is years of patient work, not a switch that gets thrown.
The governing principle
This sharpens the hardest scenario considerably. Strategic competition was diagnosed as a race in which every competitor optimizes the ecology it controls while the shared ecology none of them controls grows fragile. With boundary accounting, that fragility stops being a mood and becomes a quantity someone can actually estimate. When a state announces that its AI program raised national ecological capacity by some amount, the framework can ask how much global ecological capacity the same program destroyed — through arms-race pressure, cyber instability, energy capture, supply-chain exclusion, or the erosion of shared coordination — and lay the two figures side by side. That is a far more sophisticated question than the race usually permits itself to ask, and asking it is most of the work. Underneath all of the machinery is a single rule, and it is strong:
No ecology may count as a gain what it achieves by unaccounted depletion of a larger ecology.
Which is only the boundary-shaped restatement of the sentence at the top of this page, and of the name of the whole project: a local ecology is not healthy if its health depends upon wasting geometry outside its own boundary.