Haute Lumière

Commerce · II.01 · MMXXVI · daylight

La Bourse  /  Volume II  /  Nº II.01

From Scarcity to Abundance

Volume II — Foundations: The Paradigm and the Science

---

THE PLATE

---

THE LETTER

Volume I asked what you do on Monday. This volume owes you the reason it is true, and owes it in a form that survives a reader who does not want it to be.

That is a different job and it wants a different register. Nothing in these eleven chapters will exhort you. There will be no appeal to what the world could be if only we chose it, because an appeal of that kind is a request for agreement, and agreement given without evidence is withdrawn the moment somebody credible frowns. What is offered instead is an argument with its arithmetic visible, its sources named, and its weakest joint pointed out by the author rather than discovered by the critic.

We begin where every economics begins, because every economics begins in the same place. Lionel Robbins fixed the discipline's self-definition in 1932: economics is the science which studies human behaviour as a relationship between ends and scarce means which have alternative uses. That sentence has done eighty years of honest work and it is not wrong. It is underspecified, and the underspecification is load-bearing, because it leaves out the one question that decides what an economy is for: scarce in what sense?

There are at least four senses, and they behave so differently that treating them as one thing is the source of most of the disagreement between people who would otherwise agree. A thing can be scarce because the universe contains only so much of it. It can be scarce because what exists is in the wrong place, the wrong form or the wrong concentration. It can be scarce because it is in somebody else's hands and they are entitled to keep it there. And it can be scarce because we have built an institution whose function is to make it behave as though it were, when physically it is not.

Only the first of those is a fact about the world. The other three are facts about arrangement — and arrangement is the thing an economy actually is.

This chapter separates them, costs them, and says plainly which ones will not move no matter how well we think. There is at least one, and it is not the one usually named.

— The Editors

---

DISCOVERY

What is already working

Start where the evidence is strongest, which is with cases where a scarcity that looked absolute turned out to be positional, was treated as positional, and stopped being a constraint. There are more of these than the literature usually assembles in one place, and they are not obscure.

Light. Consider what a lumen used to cost. A candle flame converts fuel to visible light at roughly 0.3 lumens per watt; a kerosene lamp does slightly worse. A modern white LED luminaire delivers about 130 lumens per watt. That is a factor of 433 in conversion efficiency alone, before a single additional joule of fuel is found, extracted or burned. William Nordhaus's study of the price of light, which reconstructed the true labour cost of illumination from Babylonian sesame oil forward, found that the official price indices had missed almost the entire gain, because they priced lamps and fuel rather than light. The abundance was produced by changing the conversion, not by enlarging the stock — and the measurement system in place at the time could not see it happening.

Water, where somebody moved it. Iceland in 1970 heated its houses by burning imported oil; the geothermal heat under the island had been there throughout, and was not scarce in any physical sense. It was scarce in the only sense that was operating: it was not plumbed. Within two decades the great majority of Icelandic homes were on geothermal district heat. The resource did not change. The pipework did.

Food, and the sharpest case in the empirical literature. Amartya Sen's Poverty and Famines examined the Bengal famine of 1943 and found something the then-standard model could not accommodate: food availability per head in Bengal in 1943 was not markedly lower than in years without famine. People died because their entitlement to food collapsed — wages, prices and the particular bundle each occupation could command — not because the rice was absent. Sen's constructive counterpart is less often quoted and matters more here. The Maharashtra drought of 1972–73 produced a far larger proportionate fall in food output than Bengal had seen, and a small fraction of the mortality, because a public employment programme kept purchasing power in the hands of people whose crops had failed. Same physical shortfall, different entitlement structure, incomparably different outcome. That is positional scarcity identified and then addressed positionally, with the counterfactual supplied by the historical record rather than by a model.

Sequence data. In February 1996, at a meeting in Bermuda, the leaders of the public human genome sequencing centres agreed that all sequence assemblies above one kilobase would be released to public databases within twenty-four hours of generation, and that the sequence would not be patented. The Bermuda Principles were an unusual act: a group of people holding a good with essentially zero marginal reproduction cost declined to make it artificially rival. The downstream biology built on the resulting public record is the argument for it, and it is not seriously contested even by those who competed against it.

Copper, quietly, for a century. Copper does not degrade when recycled; refined secondary copper is chemically indistinguishable from primary. The UNEP International Resource Panel's status report on metal recycling rates places copper's end-of-life recycling rate above 50 percent — one of the highest of any industrial metal. Nobody legislated this into being. It happened because copper scrap has always been worth collecting, which is to say: the circulation was built by a price, not by a policy, and it has been running since before anyone had a word for circularity.

Five cases; one pattern. In each, a quantity that was described as insufficient turned out to be sufficient in a different arrangement — a different conversion, a different route, a different entitlement, a different licence, a different number of times round. None of them involved finding more of anything.

Hold that pattern lightly for now. The next movement is where it gets tested against the cases that do not yield.

---

THE ARITHMETIC

What works, what does not, and where the line sits

First, the distinction the whole volume rests on.

Two properties, and they are routinely conflated. A good is rival if one person's use diminishes what is available to another: a tonne of copper, a fish, an hour of a surgeon's attention. A good is excludable if it is practically possible to prevent someone from using it: a field with a fence, a file with a key.

Rivalry is a property of the good. Excludability is a property of the institution. That single sentence separates physics from arrangement, and it produces four boxes rather than the one the word "scarce" implies.

Elinor Ostrom's entire life's work lives in the top right box, and her contribution was never the claim that commons work. It was eight design principles specifying the conditions under which they do, which is also a specification of when they collapse. We inherit that discipline along with the authority.

The bottom left box is where the arrangement question becomes sharp, because a good that is non-rival and has been made excludable has been made artificially rivalrous. Nothing physical changed. A design file, a musical performance, a sequence, a formulation, a mathematical result: the ten-thousandth copy costs what the first one did to reproduce, which is approximately nothing.

Honour the case for making it excludable anyway, because the turn does not earn its force otherwise. Paul Romer's 1990 model put non-rivalry at the centre of why economies grow at all — an idea, once had, can be used by everyone at once, which is why accumulated ideas compound in a way accumulated capital does not — and then requires some excludability, because otherwise nobody funds the having of the idea. The patent is not a mistake. It is a deliberate, time-limited, welfare-reducing device accepted in exchange for a welfare-increasing incentive.

So the question is never whether to make non-rival goods excludable. It is how much, for how long, and to whom the surplus falls — and on that, a figure from the same author. Nordhaus's 2004 estimate of Schumpeterian profits concluded that innovators capture on the order of 2.2 percent of the total surplus their innovations generate. The other ninety-eight escapes into the rest of the economy. That is not a leak in the system. It is the system.

Second: the material constraint, computed honestly.

Now the other direction, because a chapter that only argued the arrangement case would be an advertisement. Take the three materials most often named in this argument and put the actual numbers on the page. All figures are from the United States Geological Survey's Mineral Commodity Summaries, January 2024.

  commodity        reserves Mt   resources Mt   2023 production Mt   ratio
  ---------------------------------------------------------------------------
  lithium                   28            105                0.180    3.75x
  copper                 1,000          2,100               22.000    2.10x
  phosphate rock        74,000        300,000              220.000    4.05x

Divide and you get the static life — how long the stock lasts at today's draw with no growth at all: lithium 155.6 years, copper 45.5, phosphate rock 336.4. Those are the numbers usually quoted and they are close to meaningless, because draw does not hold still. Let draw grow at g per year, and a stock with static life L is gone at:

        T  =  ln(1 + g·L) / ln(1 + g)

Lithium's draw grew 23 percent in 2023 alone. Take a conservative long-run 10 percent and run it:

  lithium, reserves       (28 Mt),  +10%/yr   ->   29.4 years
  lithium, ALL identified resources (105 Mt)  ->   42.8 years

Read that twice. Finding 3.75 times more lithium — not hoping for it, but counting every tonne the USGS has identified anywhere on earth at any grade — buys 13.4 years. Now hold the stock at reserves only and halve the growth rate instead:

  lithium, reserves       (28 Mt),  +5%/yr    ->   44.5 years

Halving the growth rate buys 15.1 years, which is more than quadrupling the resource base. The logarithm does not care how much there is. It cares how fast the draw compounds.

This is the honest negative, and it is the one abundance thinking does not dissolve. No reframing, no new institution, no better distribution, no appreciative question touches the arithmetic of exponential draw on a non-regenerating stock. A stock with r = 0 under a draw growing at g > 0 is a finite race and the finish line moves logarithmically with everything you find and linearly with nothing at all. Anyone who tells you otherwise is selling something, and it is usually a share class.

There is exactly one lever inside the arithmetic that behaves differently, and it is the reason this volume is not pessimistic. Circulate the material and you are no longer drawing on the stock at all; you are drawing on a flux you created. If a fraction p of material in service returns to service each cycle, the service delivered per tonne of virgin material is:

        M  =  1 / (1 − p)

        p = 30%   ->   1.43x
        p = 50%   ->   2.00x
        p = 80%   ->   5.00x
        p = 90%   ->  10.00x

Set that beside the geology. Every identified copper resource on the planet is 2.10 times reserves; copper's existing end-of-life recovery, which nobody planned and no ministry funds, is already above 50 percent, which is worth 2.00 times. The scrap trade has already done, at no capital cost to anyone, almost exactly what every undiscovered deposit in the USGS's global assessment would do. Raising recovery from 50 to 80 percent is worth 5.00 times reserves — more than identified and undiscovered resources combined. The non-linearity runs the wrong way from intuition: the last ten points of recovery are worth more than the first fifty.

Third: the position that energy cannot buy out.

Chapter I.01 established that accessible surface fresh water is 0.0075 percent of all the water there is. That figure is correct and it is not a ceiling; it is a price. Seawater fails to be fresh water only in the sense that separating the salt costs energy, and the cost is known precisely: the thermodynamic minimum is about 1.06 kWh per cubic metre, and the best reverse-osmosis plants run at roughly 3.5. Global freshwater withdrawal is on the order of 4,000 km³ a year. Desalinate all of it:

  at 1.06 kWh/m3 (the thermodynamic floor)    4,240 TWh/yr  =  0.48 TW
  at 3.50 kWh/m3 (today's best plants)       14,000 TWh/yr  =  1.60 TW

Against 19 TW of human primary energy that is 8.4 percent at today's plant performance — large, and entirely finite. Against the 173,000 TW of solar flux arriving at the surface it is 0.0009 percent. The water constraint is an energy constraint wearing a hydrological costume, and at a sufficiently low energy price it stops being a constraint at all.

And now the honest half, which is the second negative and the more instructive one. Desalination delivers water at the coast, at sea level. Agriculture is inland and, almost everywhere that matters, uphill. Lifting a cubic metre a hundred metres costs about 0.27 kWh at perfect efficiency and half again as much in practice; a thousand kilometres of pipeline costs more in friction, and a great deal more in capital and in the politics of right-of-way. Energy abundance converts a chemical position problem into a geographic one. It does not abolish position, and nothing does.

Fourth: the cut.

Here is the thing that is hiding in plain sight in the table above, and it overturns the ordinary reading of every number in this section.

You read "lithium reserves: 28 million tonnes" as a statement about the earth. It is not one. In the USGS's own definitions, reserves are that part of the identified resource which could be economically extracted at the time of determination — current technology, current prices, current law, ground someone currently has permission to dig. Every one of those qualifiers is economic or legal. None is geological. The stock of copper in the crust did not change when the copper price doubled; the reserves did, and they are reported in the same column, in the same units, as though they were the same kind of number.

Which means the sentence "we are running out of X" is, in almost every case anyone has ever said it, not a claim about rock. It is a claim about price.

And here the reader expects relief, and should not get it, because the inference runs the other way. A price event is exactly what running out feels like to anyone who cannot pay. The phosphate is there — 74 billion tonnes of reserves, more than 300 billion tonnes of resource, over three centuries of static life — and in 2008, when the phosphate price rose roughly eightfold in fourteen months, smallholders across sub-Saharan Africa farmed without fertiliser that season. The rock was not absent. It was priced, and it was elsewhere: Morocco holds 67.6 percent of world reserves, and in 2023 China mined 40.9 percent of world production. The USGS's own line on the commodity is the flattest sentence in the entire document: there are no substitutes for phosphorus in agriculture.

So the reassuring finding and the alarming finding are the same finding. Scarcity is positional — and position includes whose it is. The observation that almost no shortage is geological does not soften the problem. It relocates it, from geology, where nothing can be done, to arrangement, where everything can, and where the people who arrange things have interests.

That is the whole argument of this chapter, and everything after it is engineering.

---

DREAM

What becomes ordinary

In the economy that has taken this seriously, the first thing that is different is a vocabulary, and it changed because the accounts changed.

Nobody says "scarce" without a qualifier. The word has split into four, and the four are in the reporting: thermodynamically bounded, positional, entitlement-limited, institutionally enclosed. A procurement officer reviewing a supply risk states which of the four she is looking at, because the remedy is different in each case and the committee would ask. A material that is thermodynamically bounded gets substitution research and a circulation target. A material that is positional gets logistics and inventory. A good that is entitlement-limited gets a transfer, because that is what the evidence says works. A good that is institutionally enclosed gets a licensing conversation, and the conversation is about duration and reach rather than about principle.

The second thing that is different is that circulation rate is a line in the accounts. Every firm that puts durable material into the world knows its own p — the fraction of what it shipped that came back into service — the way it knows its gross margin. It is on the standing pack. It is forecast. When it improves, somebody's number improves with it. This turns out to be the single highest-leverage reporting change of the period, not because anyone was persuaded to care about materials, but because once p was measured the arithmetic of 1/(1−p) did the persuading without assistance.

The third is that non-rival goods are priced as what they are. A design, a protocol, a sequence, a clinical formulation: the questions asked of each are how long exclusivity should run, how wide it should reach, and who pays for the having of the idea, and those questions have answers that differ by domain rather than one answer applied everywhere by default. Some domains have short exclusivity and large public funding. Some have long exclusivity and none. Standards bodies and patent pools are ordinary infrastructure, the way clearing houses are ordinary infrastructure, and nobody finds it ideological that a good with zero marginal cost is distributed at close to zero price once its creation has been paid for.

The fourth is the quietest and it has the largest effect. When a shortage appears, the first question asked in the room is where is it, and whose is it — not how much is there. That question is faster, it is answerable from data that already exists, and it is right more often. It routes attention to warehouses, permits, contracts, entitlements and pipes, which is where shortages actually live.

None of this requires a new physics, a new ethics or a new political settlement. It requires four words where there was one, one number that is not currently counted, and the habit of asking the location question first.

---

DESIGN

The structure that gets there

The mechanism is a classification followed by a matched remedy, and it is deliberately mechanical, because the failure this design prevents is a good-faith argument in which two people apply the correct remedy to different senses of the same word.

The four-question protocol. For any claimed shortage, in this order.

1. Is the good rival? Does one person's use reduce what is available to another? This is a physical question and it has a physical answer. If the answer is no, skip to question four; you are looking at an institutional arrangement and no amount of production will address it.

2. Is the constraint thermodynamic? Is there a conversion, separation or concentration step whose energy cost is bounded below by the second law, and are we near that bound? Compute the ratio of current practice to the theoretical minimum, as this chapter did for desalination: 3.5 against 1.06 is a factor of 3.3, which is engineering headroom. A factor of 1.1 would not be. A constraint within a small multiple of its thermodynamic floor is real and will not yield to cleverness.

3. Is the constraint positional? Five positions, and each has a distinct remedy:

The fifth is the one most often skipped and most often decisive. Sen's finding about Bengal is a title finding, and the Maharashtra counter-case is a title remedy.

4. Is it enclosed? If the good is non-rival, the scarcity is manufactured by design and the design is negotiable. The question is then never abolition — it is scope, duration and who funded the creation. Write those three down explicitly; most disputes dissolve on the first.

Sequencing, and why this order. Rivalry first because it is cheap to establish and eliminates the largest class of errors. Thermodynamics second because it is the only genuinely immovable answer and you want to know early whether you have one. Position third because it is where most real shortages live and where most remedies are available. Enclosure last because it is the most contested and you should arrive at it having ruled out the others, which makes the conversation about the actual disagreement.

Governance. One person cannot own this classification, because the classification decides the budget and anything that decides a budget will be argued toward whichever answer funds the arguer. Two conditions hold it honest. The classification is published with its reasoning — not the answer, the reasoning, so that a disagreement is about an identifiable step. And whoever classifies is not whoever receives the remedy. That is the same separation an audit function has, for the same reason, and it costs about a day a quarter.

Where this design fails, stated plainly. It fails on goods that are rival in one dimension and non-rival in another and are therefore argued in whichever dimension suits. A road is non-rival below capacity and sharply rival above it. Spectrum is non-rival across non-interfering uses and rival within a band. Attention is rival absolutely. The protocol handles these only if the analyst states the operating regime first, and an analyst with a preferred answer will choose the regime that produces it. There is no clever fix for that. There is only the separation of who classifies from who benefits, applied consistently and checked.

---

DESTINY

How it holds when nobody is pushing

A classification scheme is a fragile thing to leave behind, because a classification with no consequence attached becomes a form that people fill in. Three conditions make this one self-sustaining, and they are structural rather than cultural.

The circulation number has to be in the reporting pack. Not a sustainability appendix — the standing monthly pack, beside gross margin, with a forecast and a variance. Anything reviewed monthly survives a change of management; anything reviewed by exception does not. And p is unusually well-suited to this: it is a ratio of two quantities most firms already measure for other reasons, and the 1/(1−p) arithmetic makes an improvement of three points visibly worth more than an improvement of three points in almost anything else on the page.

The classification has to have a budget consequence. If a material is classified as positional and the remedy is logistics, the logistics money should move. If it is classified as thermodynamically bounded, the substitution research should be funded and the growth assumption in the plan should come down. A classification that changes no line is decoration and will be abandoned within two cycles by people who are not being dishonest, merely busy.

The vocabulary has to reach people who never read the argument. This is the one that actually determines whether it holds. The four senses of scarce have to appear on forms, in templates, in the standing questions of a risk committee — in the furniture, where they are absorbed rather than taught. Ideas transmitted by persuasion decay at the rate people leave. Ideas transmitted by form fields do not.

Now the failure modes, honestly. This fails when the classification becomes a rhetorical instrument — when "that is merely positional" is deployed to mean "that is not a real problem", which is the exact opposite of what this chapter argues, since positional problems are the ones that kill people. It fails when the circulation number is gamed, which is easy: recovery rates improve beautifully if you narrow the denominator, and the fix is to define the denominator once, in writing, and to have somebody who does not benefit from the answer recompute it annually. It fails when a firm classifies its way into complacency about a stock that genuinely has r = 0 and a draw that genuinely compounds. And it fails, most commonly, when the analysis is correct and nothing downstream is connected to it — which is why the budget consequence is not optional and why the next movement is about an instrument rather than a practice.

---

DELIGHT

What it feels like

There is a specific, slightly guilty pleasure in the moment a shortage resolves into a location. You have been carrying a problem shaped like there is not enough, which has no handle on it and no place to put your hands, and then somebody asks where it is and the problem changes shape in front of you. It becomes a warehouse in the wrong country, a permit that expired, a grade that nobody costed, a contract that gave the title away in 2011. Problems with addresses are enormously more pleasant than problems without them, and the relief is not intellectual — it is physical, the drop in the shoulders that comes when a thing stops being weather and starts being work.

The other pleasure is slower and it belongs to the circulation number. There is something genuinely lovely about watching a material come back — the same copper, the same steel, the same glass — and knowing that the thing you shipped is still in the world doing its job, several owners downstream, and that you will see it again. It is the opposite of the feeling of selling something. It is closer to the feeling of lending a good tool to someone careful.

And then the smallest one, which is the habit itself. Once you have the four words you cannot stop hearing the undifferentiated one, and every time somebody says scarce you will find yourself quietly asking in which sense — not to correct them, just because the question has become interesting. That is what a paradigm actually is, from the inside: not a belief, a reflex.

---

OPERATIONALIZE THIS

At the level of finance

The classification becomes real when title does. Here is the instrument.

The structure: materials-as-a-service with retained title, financed by an asset-backed working-capital facility against the retained inventory.

You do not sell the material. You sell the service the material performs — lighting, flooring, cooling, uptime, packaging cycles — and keep legal title to the physical substance throughout. Philips Lighting, now Signify, contracted with Schiphol Airport in 2015 on precisely this basis: the airport bought light, the manufacturer kept the fittings. Interface's carpet programme and Xerox's remanufacturing operation are the same structure in older clothes. This is not novel finance. It is novel only in that it is usually justified environmentally when its justification is on the balance sheet.

The mechanics.

• What is sold. A unit of service, metered — lumen-hours, square-metre-months, cycles, tonne-kilometres. Priced per period, not per unit of matter.
• Title. Retained by the provider, stated explicitly, registered where the jurisdiction has a security-interest register. This is the clause the whole instrument hangs on. Get it drafted before anything else; retrofitting title retention into an existing supply contract is far harder than writing it in.
• Return logistics. A named counterparty, a named trigger and a named price for the returned material, agreed at the same time as the sale. A take-back obligation with no priced counterparty is a future write-off.
• The facility. Working capital advanced against the book value of material in service, at an advance rate set by the recovery fraction. A conservative opening advance rate is p × 0.8 of assessed residual value — if you recover 60 percent, advance against 48 percent, and raise it as verified recovery improves. That ratchet is what makes the lender's interest and the operator's interest the same interest.
• Term. One service life of the asset plus one recovery cycle. Not longer: you want the facility to clear on evidence rather than on faith.
• Security. The retained material itself, plus the assigned service receivable. Two securities, one of which is physical and recoverable, which is why this prices better than an unsecured green facility and should be presented that way.

The balance-sheet treatment. The retained material stays on your balance sheet as inventory or as property, plant and equipment depending on its form and your jurisdiction — it does not pass through cost of goods sold at the point of shipment, because it has not been sold. Revenue is recognised over the service period rather than at delivery. Two consequences, both of which need saying to your finance function before anyone hears about them elsewhere: reported revenue in the transition year falls even though cash economics improve, and working capital rises. Model both, present both, and present them first. A structure that surprises the CFO in month seven is dead whatever its returns. Depreciate the retained asset over its recirculated life, not its first-use life; that is a conversation about useful economic life, which your auditors have every year.

The counterparty. Begin with one large customer who already has a facilities-management relationship with you and an interest in moving capital expenditure off their own books — their motivation is the mirror image of yours, which is why this sells more easily than it sounds. The reprocessor who takes the returned material is contracted before the first unit ships.

The number that decides it. One figure, front page:

      p · (residual value per unit)  −  (reverse logistics + reprocessing
                                          + title administration per unit)
      ----------------------------------------------------------------------
                          virgin input cost per unit

If that ratio exceeds zero, every cycle of circulation is cheaper than buying the material again, and the instrument pays for itself out of the material rather than out of the customer. If it is negative, do not proceed on environmental grounds and do not let anyone else do so either — a materials loop that loses money each turn is a subsidy with a logo on it, and it will be cancelled in the first hard quarter, taking the credibility of the whole idea with it.

The first ninety days.

Ninety days does not produce a circular business. It produces a clause, a denominator, a counterparty and one signed contract — which is the smallest collection of objects from which the rest can actually be built.

---

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

1. Where in this organisation have we already solved a shortage by moving something, converting something or renegotiating something, rather than by buying more of it? What made that possible?
2. Which material or component comes back to us, in any quantity at all, and who built that route — was it designed, or did it grow because somebody found it worth doing?
3. Think of something we give away freely that costs us almost nothing to reproduce. What has it returned to us that we never counted?

Dream — what becomes possible

1. If everyone here could say which of the four senses of scarce they meant, every time, what argument would we stop having?
2. Imagine our recovery fraction were on the front page of the monthly pack. What would people start doing in the first quarter that they do not do now?
3. If we kept title to everything we shipped, what would we design differently — and which of those design changes would we want anyway?

Design — what we build

1. Which of our shortages has an address, and have we ever gone and looked at it?
2. Where are we within a small multiple of a thermodynamic floor — and where are we, in fact, nowhere near it and have stopped trying?
3. What is the smallest material loop we could close completely this year, all the way to a priced counterparty for the returned material?

Destiny — how it holds

1. What would have to be true for the classification to still be honest when the person who benefits from the answer is the one filling it in?
2. Who outside this team would notice first if our recovery number were being improved by narrowing the denominator, and have we asked them to watch?
3. What would we want still running here in ten years that depends on none of us being in the room — and what would we have to write down this quarter for that to be true?

---

WORKS CITED

Ayres, R. U. and Warr, B. (2009). The Economic Growth Engine: How Energy and Work Drive Material Prosperity. Edward Elgar.

Elimelech, M. and Phillip, W. A. (2011). "The Future of Seawater Desalination: Energy, Technology, and the Environment." Science, 333(6043), 712–717.

Georgescu-Roegen, N. (1971). The Entropy Law and the Economic Process. Harvard University Press.

Hardin, G. (1968). "The Tragedy of the Commons." Science, 162(3859), 1243–1248.

Hotelling, H. (1931). "The Economics of Exhaustible Resources." Journal of Political Economy, 39(2), 137–175.

Jevons, W. S. (1865). The Coal Question. Macmillan.

Nordhaus, W. D. (1996). "Do Real-Output and Real-Wage Measures Capture Reality? The History of Lighting Suggests Not." In Bresnahan, T. F. and Gordon, R. J. (eds), The Economics of New Goods. University of Chicago Press, 27–70.

Nordhaus, W. D. (2004). "Schumpeterian Profits in the American Economy: Theory and Measurement." NBER Working Paper 10433.

Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.

Robbins, L. (1932). An Essay on the Nature and Significance of Economic Science. Macmillan.

Romer, P. M. (1990). "Endogenous Technological Change." Journal of Political Economy, 98(5), S71–S102.

Sen, A. (1981). Poverty and Famines: An Essay on Entitlement and Deprivation. Clarendon Press.

Skinner, B. J. (1976). "A Second Iron Age Ahead?" American Scientist, 64(3), 258–269.

Solow, R. M. (1974). "The Economics of Resources or the Resources of Economics." American Economic Review, 64(2), 1–14.

United Nations Environment Programme, International Resource Panel (2011). Recycling Rates of Metals: A Status Report. UNEP.

U.S. Geological Survey (2024). Mineral Commodity Summaries 2024. USGS — lithium, copper and phosphate rock; reserve and resource definitions in Appendix C.

Weber, S. (2004). The Success of Open Source. Harvard University Press.

Note on figures. All reserve, resource and production figures are from USGS Mineral Commodity Summaries, January 2024, and are computed in lib/verify/II_01.py; the exponential-depletion model, the circulation multiplier, the desalination energy budget and the luminous efficacy ratios are computed in the same module and are reproducible there. Desalination energy bounds follow Elimelech and Phillip (2011); global freshwater withdrawal is taken at the order of 4,000 km³ per year from FAO AQUASTAT. Chapter I.01 established the solar flux ratio and the accessible-fresh-water share; this chapter builds past both rather than restating them.