Invention and Innovation: A Brief History of Hype and Failure

By Vaclav Smil

9 min read

Why does it matter? Because every technology promise you've heard this decade has an 1825 edition — and the pitch was identical.

Skepticism about technology is how Luddites get remembered. Enthusiasm is intelligence — the forward-leaning conviction that the next announcement represents genuine progress toward something the physics actually permits. To doubt publicly is to reveal yourself as someone who doesn't understand trajectories, who mistakes current limitations for permanent ones. Most of us have absorbed this so completely we've forgotten it's a choice.

What Vaclav Smil noticed is that the failures aren't random. They sort into three categories: harms that were knowable and institutionally suppressed, disasters that were genuinely unforeseeable, and impossibilities dressed as milestones. The first is an indictment. The second, a tragedy. The third is arithmetic in a press release. The inability to distinguish between them isn't a cognitive accident — it is a structural feature of who controls the conversation before the evidence arrives, and who absorbs the cost after.

Finish this and you have a test: which category does this belong to? The next announcement won't look the same.

The Hype Template Was Written Before Your Great-Grandparents Were Born

If you've watched the last decade of startup culture, you probably assume technology hype is a recent invention — a byproduct of venture capital's need for a story compelling enough to justify a valuation, of founder cults and press cycles hungry for the next moon shot.

It isn't. The script is at least two hundred years old, and Vaclav Smil has found the receipts.

In 1825, a group of London entrepreneurs floated shares in the London and Edinburgh Vacuum Tunnel Company. Their prospectus proposed a metal tube running 600 kilometers between the two cities, propelled by steam-powered vacuums, carrying passengers in just five minutes. Capital required: twenty million pounds sterling, divided into two hundred thousand shares at £100 each. The technology didn't exist. The materials science was decades away. None of that deterred the offering. The illustrator William Heath responded that same year with a satirical etching depicting the "Grand Vacuum Tube Company Direct to Bengal" (passengers being fired intercontinentally through a seamless metal tube), and the trade journal London Mechanics' Register reprinted the tunnel company's prospectus specifically "to throw ridicule upon some of the preposterous plans now before the public."

Nearly two centuries later, on August 12, 2013, Elon Musk released his Hyperloop Alpha paper: pods accelerating through low-pressure tubes, covering Los Angeles to San Francisco in thirty minutes. The media treated it as an entirely original vision. Smil traces the concept back to patents filed before the US Civil War, a working scale model demonstrated in London in 1912, and a Rand Corporation proposal for transcontinental links in 1972. The promoters changed. The idea didn't.

Smil's central argument, across a book cataloguing technologies that failed to become what their promoters promised: hype follows a template. Revolutionary speed. Universal access. Imminent availability — five years, ten at most. Capital raised before the fundamental engineering problems have been solved. Then the schedule slips, the physics wins, and the idea retreats into obscurity until the next promoter rediscovers it fresh.

The 1825 shareholders lost their money. The 2013 press releases aged badly too: Virgin Hyperloop One's flagship 2020 demonstration moved two passengers at 175 km/h on a 500-meter test track. Scheduled trains had exceeded that speed routinely since the 1960s. The tube idea is still alive. It has always been alive. That's not a coincidence: it's what happens when capital needs a story before engineering has one.

The Same Man Invented the Knowable Catastrophe and the Unforeseeable One

In April 1930, a mechanical engineer named Thomas Midgley stood before the American Chemical Society, held a flask of dichlorodifluoromethane — the compound marketed as Freon — to his face, and inhaled. He survived, as he knew he would. Then he turned toward a candle flame and exhaled slowly, extinguishing it. Nontoxic, nonflammable, near-perfect. The crowd saw a solved problem. They were right, for the moment.

What makes this scene strange is not the showmanship. It's who was doing it. Midgley was the same man who, eight years earlier, had driven the commercial adoption of tetraethyl lead as a gasoline additive. The same man who had insisted the product be marketed as "ethyl gas" (a name chosen to avoid mentioning lead, an element linked to neurological damage since antiquity). The same man whose employer's workers were dying of acute lead poisoning while the corporate position held that danger to the general public was essentially zero.

Smil's book is full of failed technologies, but it draws a sharp line between failure types, and Thomas Midgley is the sharpest illustration of the distinction. What he did with leaded gasoline belongs to one moral category. What he did with CFCs belongs to another. Most accounts of technology failure treat both as the same story: hubris, unintended consequences, the usual. They aren't the same story at all.

Tetraethyl lead was not an unforeseeable catastrophe. It was a knowable one, and the knowledge was present from the start. By December 1921, when tests at General Motors confirmed that a 1% TEL solution eliminated engine knock, the compound's toxicity was already established fact. Lead had been linked to neurological damage for two millennia. Alice Hamilton, a physician at Harvard Medical School, appeared at a 1925 Surgeon General's conference and stated the case plainly: no lead industry, she said, had ever operated without danger even under the strictest controls. She was not speculating. She was citing evidence already on record.

She was ignored. GM chose TEL over ethanol and benzene blends (both known to work) because Midgley's patent gave the company proprietary control over a low-volume additive that cost about a penny per gallon to deploy. The naming decision followed the commercial logic. Within two years of the December 1921 tests, TEL was at public pumps. Within three years, five workers at a New Jersey processing plant developed acute neurological symptoms and died. The industry commissioned a seven-month study of 252 workers, too short to detect long-term neurotoxicity, and used the inconclusive results to clear the path for continued production.

Between 1945 and 1975, the United States added approximately 4.7 million tons of lead to the environment through vehicle exhaust, peaking above 200,000 tons annually in the early 1970s. What finally ended US leaded gasoline was not a belated reckoning with Hamilton's 1925 testimony. It was catalytic converters, platinum-based devices that lead destroyed on contact. Removing lead was the price of cleaning photochemical smog. Children's blood lead levels fell 80% between 1976 and 1994 as a consequence. Algeria became the last country to ban leaded gasoline in July 2021, ninety-nine years after the first decisive tests.

The CFC story is different in a way that matters morally. When Midgley exhaled Freon-12 onto that candle in 1930, no available evidence pointed toward stratospheric ozone depletion: no published theory, no atmospheric measurement, no analogy in existing chemistry. The compounds were stable and chemically unreactive at the surface. In 1971, James Lovelock measured atmospheric CFC-11 and concluded its presence posed "no conceivable hazard." He wasn't being reckless; he was reasoning from everything then known. It took Sherwood Rowland and Mario Molina three more years, in a 1974 paper, to work out what actually happens: CFCs drift into the upper stratosphere where ultraviolet radiation breaks them apart, releasing chlorine atoms that catalytically destroy ozone — a single atom capable of eliminating 100,000 ozone molecules before leaving the stratosphere. Nothing in 1930 chemistry pointed there. The Montreal Protocol that followed in 1987 almost certainly prevented global catastrophe; without it, modeling suggests 67% ozone destruction by 2065.

The taxonomy Smil is building has two categories: harm that was known and suppressed, and harm that was genuinely impossible to anticipate. Midgley illustrates both, in a single career. One man, one lifetime, two entirely different relationships with knowledge and culpability.

Moore's Law Is an Anomaly That Infected Every Forecast That Didn't Deserve It

Moore's Law is not a law of nature. It is a description of transistor miniaturization on silicon, a phenomenon that compounded at roughly 35 percent annually for about fifty years, producing a ten-million-fold increase in semiconductor logic density. That curve invaded how educated people think about every technology, and it is almost entirely wrong as a model for anything involving molecules, heat, or chemical bonds.

The actual improvement rate for physical systems is 1 to 2 percent per year. Battery energy density in 1900 was around 25 watt-hours per kilogram. In 2022, after 120 years of engineering and materials science, it stands at roughly 300 watt-hours per kilogram. That is a 12-fold gain that sounds impressive until you do the annual math: approximately 2 percent, compounding steadily, for a century and a quarter. Steam turbine efficiency has followed the same gentle slope over the same span. Crop yields in sub-Saharan Africa gain about 1 percent annually. These systems improve. Just at the pace of chemistry and physics, not chip design.

Aviation kerosene contains about 12,000 watt-hours per kilogram. The best available batteries hold 300. That is a 40-fold gap. It is not one that clever engineers have been too lazy to close; it reflects fundamental electrochemical limits on what a battery can store. If energy density continued improving at 2 percent annually, reaching even a third of kerosene's energy content would take until roughly 2090. The aviation industry is supposed to start flying meaningful numbers of battery-electric aircraft by 2030.

The gap between that ambition and the physics is where Moore's Law intuition does its damage. The COP26 climate agreement required cutting global CO2 emissions by 50 percent by 2030. The actual decarbonization rate managed between 2000 and 2020 was 0.2 percent per year. Hitting the 2030 target requires 2.75 percent annually — a 14-fold acceleration, sustained for nine years, across the entire global economy simultaneously.

Physical systems can change. They cannot change on software timelines. A better algorithm ships in months. A better aviation fuel infrastructure requires new aircraft, new airport systems, new manufacturing supply chains, new training programs, and a fuel energy density that does not yet exist in any non-fossil form. Applying Moore's Law intuition to these problems doesn't produce ambitious targets. It produces arithmetic that ignores the physics, then deposits those targets into conference communiqués where they function as political commitments rather than engineering plans.

That thirty-five-percent-per-year transistor curve is the exception, not the template. Nearly everything else improves at rates that look boring by comparison — until you realize that "boring" means the gap between where we are and where the declaration says we'll be in nine years is not a question of effort or political will.

It is a constraint of matter.

The Most Important Inventive Decade in History Didn't Issue a Press Release

Imagine you're an investor in 1889, scanning trade journals for the next technology worth funding. You would have missed it. Not because you weren't paying attention — because the most consequential inventive decade in modern history didn't announce itself as one. No manifestos, no funding rounds, no conference declarations about disruption.

Between roughly 1880 and 1890, engineers working in separate countries produced the internal combustion engine in practical form, alternating current electricity generation and distribution, industrial-scale aluminum smelting, the steel-skeleton skyscraper, and Heinrich Hertz's laboratory demonstration of electromagnetic waves. Every one of those arrived in the same ten years. None arrived with a press release about what it meant.

Smil's accounting of that decade relocates the center of gravity. The smartphone in your pocket, the internet it connects to, the wireless signal it rides — all trace back to James Clerk Maxwell's 1860s equations describing electromagnetic phenomena. The derivation runs from Maxwell's equations through Hertz's laboratory proofs, through radio, vacuum tubes, and semiconductor physics, to the transistor Bell Labs assembled in 1947. Physics, chemistry, and materials science covered that ground across a century of incremental work.

Today the constraint isn't invention. Fewer than 10 percent of people in the world's warmest climates have air conditioning. The technology works, it can be manufactured at scale, and the deployment gap kills roughly 300,000 people annually. The US spends two and a half times more per grade-school student than Poland and ranks lower on international assessments. Students who can't read at grade level by fourth grade are four times more likely to drop out of high school; the lifetime earnings gap between a dropout and a graduate runs to roughly $300,000. Neither figure awaits a new technology. Both reflect deployment decisions, made or deferred, by people who already had the data.

Smil's argument across this book arrives here: the posture of waiting for the next breakthrough to solve civilization's largest problems has its priorities inverted. A breakthrough may arrive on its own timeline, which no amount of political will compresses. The deployment gap is already here, already measurable — and closing it requires neither a laboratory nor a funding round. It requires decisions. That is the story with the most arithmetic behind it.

Which Category Does It Belong To?

The Montreal Protocol paradox is the sharpest thing in this book. Smil argues for patience, for evidence, for arithmetic before ambition — then his single genuine success story is a 1987 treaty built on a 1974 Nature paper, acting before damage was fully measured, because potential harm was large and the cost of early action recoverable. That is not believing every press release. It is a skill: sorting any promise into one of three categories.

Knowable risk being suppressed: demand the evidence now. Genuinely unforeseeable mechanism: build early monitoring, move when the signal arrives. Physics-constrained impossibility in a funding deck: wait, because no political will compresses battery chemistry. The taxonomy rarely appears in the press release, so the work is yours. Once you've done it, you'll find that the technologies most capable of helping the most people already exist — simply, stubbornly, undeployed. The miracle you're waiting for is a choice. The battery chemistry is genuinely constrained. That's not what's stopping the air conditioning.