The Coal Question

Every resource is on a countdown. We keep misreading what it’s counting down to.

In 1865, a twenty-nine-year-old English economist named William Stanley Jevons published a book that terrified the British Parliament. The title was dry—The Coal Question—but the argument was existential. Coal was the foundation of everything: the steam engines, the locomotives, the iron mills, the family furnaces. Britain’s industrial supremacy—its military dominance, its empire, its standard of living—rested on a single combustible rock.

And the rock was running out.

Jevons wasn’t predicting mines would go dark next Tuesday. He was extrapolating. British coal consumption was growing at 3.5% per year, doubling every twenty years. Known reserves were large but finite. An exponential demand curve hitting a finite supply doesn’t end well. It just ends. Jevons concluded that Britain’s industrial preeminence had an expiration date—and Parliament took him seriously enough to convene a Royal Commission to investigate.

Oil had been discovered in Pennsylvania six years earlier, but in 1865 it was still a curiosity—good for kerosene lamps, not much else. The internal combustion engine was decades away. But oil turned out to be better than coal in nearly every dimension—more energy-dense, more portable, more versatile, lower emissions. Then came natural gas, cleaner still. Then nuclear fission, which packed more energy into a kilogram of uranium than a thousand tons of coal. Then solar, which harvests energy from a source that will burn for another five billion years.

Jevons saw a countdown. He assumed it was counting down to the end of industrial energy. It was counting down to the end of coal.

The Pattern

Jevons was the first to formalize the error, but hardly the last to make it.

Over a century later, in 1972, the Club of Rome commissioned a team of MIT researchers to do what Jevons had done—but bigger. Using a computer model called World3, Donella Meadows and her colleagues simulated the interaction of population growth, food production, industrial output, resource consumption, and pollution. The result was The Limits to Growth, one of the most influential publications of the twentieth century.

The conclusions were specific and stark. At prevailing rates of consumption, gold reserves would be exhausted in 9 years. Mercury in 13. Tin in 15. Petroleum in 20. Natural gas in 22. The model’s “standard run” projected global economic collapse beginning around 2020. The book sold thirty million copies. Its specific resource predictions were, almost without exception, wrong.

Not because the authors were careless — the methodology was serious enough to provoke fifty years of both qualified debate and unjustified policy. The problem was structural, and it was the same one Jevons had: the model could project consumption curves with great precision. It could not predict the inventions that would bend them.

In 1972, it was reasonable to project that population growth would outstrip agricultural capacity. Thomas Malthus had made essentially the same argument in 1798, and for most of human history, he’d been right—population pressed against the food supply, and the result was perennial famine. What neither Malthus nor the MIT model could anticipate was Fritz Haber, who figured out how to synthesize ammonia from atmospheric nitrogen—literally conjuring fertilizer from thin air. Or Norman Borlaug, whose semi-dwarf wheat varieties doubled and tripled yields across Mexico, India, and Pakistan. India’s wheat harvest jumped from 11 million tons to 20 million in six years. Borlaug won the Nobel Peace Prize for work that would eventually save over a billion lives. Today, the planet feeds eight billion people. Caloric production is now higher per-capita than in 1972, when there were fewer than four billion of us.

Proved petroleum reserves in 1972 were roughly 550 billion barrels. After fifty years of continuous extraction and presidential candidates campaigning to “drill baby drill,” proved reserves now stand at 1.7 trillion barrels—three times what the model treated as the total supply. Hydraulic fracturing, horizontal drilling, and deepwater exploration kept unlocking oil that no model had anticipated. The model assumed fixed resources consumed by growing demand. Capitalism disagreed.

In 1980, the economist Julian Simon made the pattern into a bet. He challenged Paul Ehrlich—author of The Population Bomb, which had predicted mass starvation in the 1970s and 80s—to pick any five commodity metals. If their inflation-adjusted prices rose over the next decade, Ehrlich wins. If they fell, Simon wins. Ehrlich chose chromium, copper, nickel, tin, and tungsten.

By 1990, every single one was cheaper. Ehrlich mailed Simon a check for $576.07.

Ehrlich looked at commodity prices and saw a countdown to zero supply. Simon looked at the same prices and saw a countdown to zero cost.

The countdown to zero cost won. It always does.

The pattern is so consistent that it implies something radical about how commodities actually work. Over the long run, the price of a commodity doesn’t rise, it trends toward zero.

The Price of Everything

A commodity is a product nobody really cares about. Not because it’s useless—because it’s undifferentiated. Nobody cares where their copper comes from, who mined it, or what the company’s mission statement says. Copper is copper. It doesn’t have a brand, it has a spot on the periodic table. Which means the only thing distinguishing one producer’s copper from another’s is the price. And when price is the only axis of competition, every producer on earth is doing the same thing: trying to deliver the same product for less money.

That’s a race with only one direction. Nobody is trying to make copper more expensive to produce. There’s no artisanal copper market. Every engineer, every process improvement, every new extraction technique pushes the cost down. These gains accumulate permanently across every commodity—you can’t un-learn electrolytic smelting, you can’t un-discover horizontal drilling. Each efficiency becomes the new floor, and the next improvement builds on top of it. Aluminum was once rarer than gold—Napoleon III served state dinners on aluminum plates while lesser guests ate off silver. Then Charles Martin Hall figured out electrolytic smelting in 1886 and collapsed the price by 98%. Today we wrap sandwiches in it and throw it away. The metal didn’t change. The engineering caught up to the geology.

But that’s only the first force. The second is more powerful: substitution. People don’t actually want copper. They want what copper does—conducts electricity, carries signals, moves heat. The commodity is just a delivery mechanism for a function. And delivery mechanisms get replaced. Fiber optics replaced copper in telecom. Aluminum replaced steel in manufacturing. The cost of a lumen-hour—the basic unit of being able to see after dark—has fallen by a factor of 500,000 since the Babylonians were burning sesame oil. A day’s labor in 1800 BC bought ten minutes of dim lamplight. A day’s labor today buys 20,000 hours of LED illumination. When the incumbent commodity can’t get cheap enough fast enough, something else delivers the same function for less.

Two forces. Both permanent. Both one-directional. Competition within a commodity pushes the cost down. Competition across commodities replaces the expensive one entirely. The long-run price of every commodity is zero—either extraction gets so efficient its only real cost is the shipping, or something better comes along and demand collapses. There is no third option.

Solar energy is the proof running in real time, fast enough to watch. In 1977, a watt of solar capacity cost $76. Today it costs roughly $0.20—a 99.7% decline, and the curve hasn’t flattened. Every time cumulative production doubles, cost falls by another quarter. The sun delivers 10,000 times current global energy demand to the planet’s surface continuously. It doesn’t send an invoice. The sand used to make the panels is so abundant, we sweep it out of our homes. The only real cost is the engineering, and engineering costs fall on a learning curve. They always have.

The scarcity of the old thing financed the new one. Solar was a novelty until fossil fuels got expensive enough to make it worth funding. By 2024, new solar installations generated electricity more cheaply than existing coal plants in most of the world—not new coal plants, but ones whose construction costs are already sunk. Jevons’ coal question answered itself. Not with a better mine, but with a technology he couldn’t have modeled because the photovoltaic effect wouldn’t be explained for another forty years.

Reading the Clock Backwards

Every generation faces a coal question—a resource it depends on, running out, with math to prove it. Every generation makes the same error: reading a countdown to replacement as a countdown to catastrophe. But there’s a subtler error, and it does more damage: responding to the countdown with restraint instead of invention.

Antibiotic resistance is a coal question—and a case study in getting the response wrong. Bacterial evolution is outrunning our pharmaceutical toolkit. The WHO projects 10 million annual deaths from resistant infections by 2050, up from 1.3 million today.

Paging Dr. Jevons.

The crisis isn’t biological. It’s economic. Our response to resistance has been to use fewer antibiotics—prescribe less, restrict access, steward carefully. Burn less coal. And the predictable result is that we’ve made antibiotic development one of the worst investments in pharmaceutical R&D. You can’t recoup billions in development costs on a drug doctors are told to use as sparingly as possible. So they stopped developing them. The entire global antibiotic pipeline—every drug being developed by every company on earth—contains roughly 45 candidates. For context, oncology has over 6,000. Major pharmaceutical companies have largely exited the don’t-die-from-infections space. We told the market to use less, and the market stopped making more.

Meanwhile, the alternatives exist. Phage therapy—using viruses that target specific bacteria—has been practiced for a century in Eastern Europe and is only now entering Western clinical trials. CRISPR-based antimicrobials can be programmed to kill specific resistant strains. AI-driven drug discovery has compressed candidate identification from years to weeks. None of these were in the WHO projection, because projections model the constraint, not the response. But they’ll only arrive at scale if we let the price signal do its job—reward the solution instead of rationing the problem.

The Condition

None of this is automatic. The countdown to zero cost only runs if the people solving the problem get to profit from solving it.

Britain didn’t answer the coal question by burning less coal—it developed the technologies that made coal obsolete. Borlaug didn’t feed a billion people by convincing them to eat less—he bred wheat that produced more grain per stalk. The mechanism that resolves constraints isn’t restraint. It’s invention. And invention requires an apparatus—universities, research labs, capital markets, legal frameworks—capable of converting problems into solutions.

The apparatus works when we fund it at the scale the problem demands. We landed on the moon eight years after deciding to try. The Manhattan Project took three years from inception to detonation. Fusion energy has been “thirty years away” for sixty years—but total global investment in fusion research over those six decades is roughly what the Pentagon spends in a single month. The timeline is a function of funding, not physics. When we actually commit, the timeline compresses in ways the skeptics never model—because they’re extrapolating from the underfunded version of the effort.

The real risk is never the constraint itself. It’s dismantling the apparatus that resolves constraints. Defund basic research. Politicize scientific institutions. Strangle nuclear energy in permitting for four decades. Restrict capital formation so only the already-wealthy can invest in transformative technology. Do these things and the other countdown wins—the one to zero supply. Not because the constraint was unsolvable, but because we took apart the machine that solves things.

Zero

The scarcity never goes away. It moves up the stack as each constraint is resolved by something better than what it replaced. The thing you’re worried about today will become essentially free. Then you’ll worry about the next thing. That’s what progress feels like from the inside.

But here’s the part we keep getting wrong. When the countdown appears—when the math says the resource is finite and the curve is exponential—every instinct says to slow down. Use less. Conserve. Ration. Buy time.

That instinct is exactly backward. The rising price isn’t the crisis. It’s the funding mechanism for whatever comes next. Solar didn’t get cheap because “Just Stop Oil” activists finally glued their hands to enough surfaces. It got cheap because fossil fuels got expensive enough to make the investment worthwhile. Every dollar of scarcity pricing that accrues to the old thing is a dollar of incentive to build the new one. Eli Lilly isn’t going to spend two billion dollars developing a novel antibiotic that doctors are told to prescribe as rarely as possible. But a company staring at a market where existing antibiotics don’t work anymore—where resistant infections are killing ten million people a year—will absolutely spend that money, if we let the price tell the truth.

Rationing the old thing and funding the new thing feel like the same policy. They are opposites. One stretches the countdown. The other ends it.

Running out isn’t a reason to cut back. It’s a reason to surge forward.