Born Classified

The trillion-dollar defense budget isn’t the expensive part

For all of the human lives spent winning World War II, it was two scientific achievements that principally ended it: the atomic bomb, and the Enigma codebreaking machine. By most accounts, the atomic bomb decided the war in the Pacific, while Turing’s codebreaker turned the tide in the European theater.

Both emerged from extraordinary science and extraordinary scientists. Both were classified at the highest levels. What happened next split the trajectory of the global economy in two.

The codebreaker machine was classified Ultra Secret—a designation above Britain’s highest standard classification. The machine itself, its specific architecture and implementation, remained classified and closely guarded for decades after the war. But Alan Turing’s foundational work—his 1936 paper On Computable Numbers, the theoretical bedrock of every computer ever built—and his related research, remained in the open literature. You can read it today, just as you could have during the war.

The bomb got classified too. But unlike the computer science, the Atomic Energy Act of 1946 classified not only the weapon—it classified the entire scientific foundation underneath it. The Act created a legal doctrine unprecedented in the history of knowledge: “born classified.” Certain categories of nuclear physics were declared state secrets from the moment of discovery—by anyone. A university researcher with no government funding and no security clearance who independently derived restricted results had, by law, produced a state secret before anyone in government had seen it. The knowledge was illegal before the ink dried.

This wasn’t intentional, or at least, it wasn’t strategic. Governments simply understood the power of bombs better than the power of bits.

A bomb is visceral—the mushroom cloud over Hiroshima was photographic proof that the physics underneath it was dangerous. Generals and senators didn’t need to understand neutron chain reactions to understand that the knowledge enabling them should never reach Moscow. Turing’s work was abstract. Math and logic. Theory based on a hypothetical thought experiment in which a machine reads an infinitely long strip of tape with 1s and 0s on it. So they classified the codebreaking equipment—that part they understood the importance of—but the idea itself was allowed to flourish in the public domain.

The Divergence

Modern computing is the result of eighty years of compounding breakthroughs on Turing’s open foundations. Transistors, integrated circuits, personal computers, the internet, artificial intelligence—each generation building on the last, each published, peer-reviewed, and available to every researcher and inventor on the planet. And available to you. The iteration loop that drives scientific progress—publish, challenge, replicate, extend—runs uninterrupted.

Physics, meanwhile, flatlined.

The most celebrated physics breakthrough of the last twenty years was the 2012 detection of the Higgs boson—a particle first theorized in 1964. Gravitational waves, detected in 2015, confirmed a prediction Einstein published in 1916. Significant accomplishments. But also confirmations of very old ideas, rather than the discovery of new ones.

The pattern holds across applied physics, too. The Boeing 707 entered commercial service in 1958, cruising at Mach 0.85. The Boeing 787 entered service in 2011, cruising at... Mach 0.85. More fuel-efficient, but the same speed, the same basic airframe, the same fundamental technology—fifty-three years apart. The biggest upgrade to the passenger experience was switching from CRT TVs to seat-back screens. And now airlines are removing those too, because your phone does a better job, thanks to computer science. Fifty years of Boeing, a prime defense contractor, shipping essentially the same commercial products despite decades of military advancements.

Materials science. Energy storage. Propulsion. Electromagnetics. Incremental refinement in secret, decade after decade, while the domains whose science stayed open compounded exponentially.

The Easy Answer

There is a serious, well-researched argument that this scientific stagnation is natural.

Physics had a golden age—roughly 1900 to 1960—that produced relativity, quantum mechanics, nuclear physics, and the solid-state physics behind the transistor. These were foundational breakthroughs: deep, accessible, and world-changing. The argument is that what followed was inevitably harder. The low-hanging fruit had been picked. Newton needed an apple. Now we need supercolliders. Tyler Cowen called the broader phenomenon the Great Stagnation. Bloom, Jones, Van Reenen, and Webb published “Are Ideas Getting Harder to Find?” and concluded that, yes, they are—across the board, research productivity is declining as each new discovery requires more people, more expertise, and more capital.

These are serious economists looking at real data. And the story is internally consistent: physics matured, the easy breakthroughs were made, and what remains is incrementally harder. The plateau isn’t a mystery. It’s a natural life cycle.

The trouble is what the explanation has to ignore to stay comfortable, not least of which is describing what Einstein produced as “low-hanging fruit.”

What the S-Curve Can’t Explain

If diminishing returns are natural—a feature of the science itself—they should hit every scientific field as the easy discoveries are exhausted. But they don’t. They hit physics, materials science, propulsion, and energy. They skipped biology, computer science, and every other domain whose foundational knowledge stayed in the open literature.

The National Institutes of Health spend roughly $47 billion a year on publicly funded biomedical research—published, peer-reviewed, available to researchers worldwide. Over the same fifty years that physics stagnated, biology produced genomic sequencing, CRISPR, immunotherapy, mRNA vaccines, and a revolution in molecular biology. No plateau. No diminishing returns. The pace of discovery accelerated. And it didn’t accelerate in isolation. Biology drew on breakthroughs in computing, chemical engineering, optics, statistics—every open field feeding into every other open field. Open science doesn’t just compound within a discipline. It compounds across the entire network, because a breakthrough published anywhere is available to researchers everywhere. Classification doesn’t just slow one field. It severs nodes from the network.

The “experiments got expensive” argument explains particle physics—you do need a multi-billion-dollar collider to detect the Higgs boson. But it doesn’t explain why materials science stagnated. Or energy storage. Or propulsion. Or applied electromagnetics. These are fields where the experiments aren’t prohibitively expensive and the commercial demand for breakthroughs is enormous. The S-curve landed selectively on the domains whose underlying science overlaps with defense classification, and left the open domains untouched.

Then there’s the timing. The physicists who built the bomb were trained in the pre-classification world—open publication, international collaboration, rapid iteration. The next generation, mentored by Oppenheimer’s contemporaries, also came up in that tradition. Both cohorts carried the knowledge base of open science in their heads. Then they retired in the 1970s. After that, every working physicist had been trained entirely within the classification regime. The open-science culture that produced the golden age wasn’t just suppressed by law. It retired out of the workforce one career at a time.

The S-curve theorists date the plateau to the same decade and call it nature. But the plateau doesn’t map to the exhaustion of discoverable physics. It maps to the disappearance of the last generation that practiced physics in the open.

What Leaks Through

Occasionally, classified technology escapes into the civilian world.

GPS was a military positioning system beginning in 1973. Civilian access was granted a decade later, with intentional accuracy degradation. Full precision wasn’t available until 2000. Twenty-seven years from invention to unrestricted civilian use. It has since generated an estimated $1.4 trillion in economic value in the United States alone. Fiber optics, developed for classified military communications, became the backbone of the modern internet. Radar, declassified after the war, produced microwave ovens, air traffic control, weather forecasting, and the foundation of modern aviation safety.

These aren’t anomalies. They’re samples. When declassified technology reaches the civilian economy, it produces outsized returns—because it enters a market that has been systematically starved of the underlying knowledge. GPS didn’t generate a trillion dollars in value because it was uniquely brilliant. It generated that value because it was one of the few things that made it into the economy. But the economic impact is only part of the story. Even once it was commercially available, it lost twenty-seven years of subsequent technological and economic compounding.

The talent tells the same story from the labor-market side. A gifted physicist motivated by hard problems might accept a lower salary—plenty of brilliant people choose academia over finance. But classification doesn’t just pay less. It strips away every non-monetary incentive that draws the intellectually driven. No publication. No peer recognition. No watching your work enter the world and reshape it. The choice isn’t between money and recognition. It’s between money and recognition, or neither. The disciplines where you can publish, build on others’ work, and see your ideas tested in the open attract the best minds. The disciplines where the best work disappears into a classified archive lose more of them every decade.

The Externality Nobody Measures

National defense is the textbook case for collective funding. No individual can opt out of the nuclear umbrella, so the cost is shared through taxation. The benefit—security, sovereignty, the freedom to build an economy without fear of invasion—is shared too. The principle is sound.

But the classification program has a non-tax cost that never makes it onto the ledger or into the conversation. The way defense R&D is implemented—classifying the science, not just the weapons—creates a compounding negative externality running alongside the positive one.

The physics presumably still advances inside the classified world, albeit more slowly. But the knowledge never reaches the civilian economy—never gets published, challenged, or extended by the engineers and entrepreneurs who would build products, launch industries, and drive the productivity gains that show up as increased wages, improving standards of living, and a broader tax base to support more R&D. This compounds the way interest compounds. Every suppressed breakthrough is a missing foundation for the next generation of breakthroughs. The drag isn’t the classified research itself. It’s the entire tree of civilian applications, products, and industries that never grew from seeds locked in a vault.

The defense budget is a trillion dollars a year. That’s the number we debate. The compounding productivity loss—eighty years of missing civilian applications building on missing civilian breakthroughs—dwarfs that number the way compound interest dwarfs principal. But nobody can put a figure on it, because we can’t measure the value of discoveries that were never made and industries that were never built. It will never appear on an appropriations request, because the cost doesn’t look like a cost. It looks like stagnation. It looks like S-curves. It looks like physics just getting hard.

The False Choice

The standard defense of born-classified is that it worked. Nine countries have nuclear weapons. Dozens more could have pursued them. The classification regime—alongside treaties, diplomacy, and deterrence—helped hold that number down.

But “worked” overstates the case. Pakistan acquired nuclear capability through A.Q. Khan’s espionage and smuggling network. North Korea built a weapon under the heaviest sanctions regime on earth. The classification of nuclear physics did not stop either program. We’re in a hot war with Iran because knowledge suppression doesn’t stop nuclear development. Any sufficiently motivated country can figure it out—the knowledge leaks, gets independently derived, or gets stolen.

What the classification regime actually achieved was delay. And delay is genuinely valuable when the subject is civilization-ending weapons. But delay is a different proposition than prevention, and it demands a different cost-benefit analysis. If born-classified prevented proliferation permanently, the sacrifice of eighty years of civilian physics might be a price worth paying. If it delayed proliferation—bought decades of time while the knowledge inevitably spread—then the question is whether eighty years of compounding civilian stagnation was a reasonable price for a slightly less fast nuclear proliferation. It might be, if we used the time to build a durable solution, but that is a much harder case to make.

The slowdown didn’t require classifying the science, anyway. We know this because we watched the other technology take a different path.

The United States developed cryptography, signals intelligence, and cyber weapons — all classified applications built on Turing’s open foundations. The NSA runs some of the most sensitive programs in the intelligence community, all of them downstream of computer science. Nobody classified computer science. The applications were classified. The weapons were classified. The underlying science stayed in the open literature, compounding for eighty years, producing a civilian economy worth trillions—while simultaneously producing the most sophisticated classified capabilities on earth.

Defense research could be handled similarly. Classify the weapon designs. Classify the engineering. Classify the enrichment techniques. But leave the physics—the foundational science that also underlies civilian energy, materials, propulsion, and a hundred other productive applications—in the open, where it could compound the way every other open science has. The Atomic Energy Act didn’t draw that line. It classified the knowledge itself. And the fields downstream of that knowledge, and our economy, have been paying the compounding price ever since.

Two technologies won the war. We opened one and got the modern world. We classified the other and got excuses.