The Soviet Union built an atomic bomb and starved its best biologist to death in the same decade. The difference between those two outcomes is the whole argument.

On August 29, 1949, the Soviet Union detonated its first atomic device at Semipalatinsk, four years after Trinity. The program ran under Lavrentiy Beria, which is to say under a man who could have its physicists shot and sometimes did. It worked anyway.

Six years before the test, in January 1943, Nikolai Vavilov died of starvation in a Saratov prison cell. He was the foremost plant geneticist alive and had assembled the largest seed collection in the world, and he had been arrested for believing in genes. Trofim Lysenko, who did not believe in them, ran Soviet biology instead. In August 1948 the academy formally banned Mendelian genetics from Soviet science. Soviet agriculture and Soviet medicine paid for that decision for the next forty years.

Same state. Same terror. Same decade. It built the bomb and it could not do biology.

Anyone who claims that freedom is a requirement of human progress has to account for that pair of facts before saying anything else. Tyrannies produce results. The Soviets reached orbit first. Nazi Germany built the V-2 and the first operational jet fighter. The proposition that coercion simply forecloses achievement was refuted by the twentieth century, and restating it more firmly does not make the century go away.

The explanation lies in the difference between those two Soviet programs, and it is the most useful thing anyone can understand about how human beings stay alive.

The Animal With No Instructions

In June 1861, Robert O’Hara Burke and William John Wills starved to death at Cooper Creek in central Australia, in country where the Yandruwandha had lived for thousands of years. They were not short of food. They were eating nardoo, a fern whose spore cases the Yandruwandha ground into cakes, and which carries an enzyme that destroys thiamine unless the plant is processed correctly. The Yandruwandha processed it correctly. Burke and Wills did not, and died of vitamin deficiency with full stomachs. John King, the third man, lived because the Yandruwandha took him in and fed him.

Nothing in Burke’s body knew what to do with nardoo. Nothing in a Yandruwandha body did either. What separated the men who lived from the men who died was knowledge, held by one group and absent in the other, and no quantity of hunger or effort or courage could generate it on the spot.

That is the biological fact, and it is worth stating precisely, because the loose version is false. Other animals are not mindless automata. Crows manufacture hooked tools and modify them for the job. Chimpanzee communities pass down local traditions of termite-fishing and nut-cracking that differ from valley to valley. The true claim is narrower and far more consequential: every other animal can survive on behavior it never had to learn, and no human population anywhere can. Strip a wolf of its culture and you still have a functioning wolf. Strip a human of his and you have a corpse.

We are the animal with no niche. No environment fits us, because our adaptation is to modifying environments, and modification runs on knowledge that has to be created, held, corrected, and handed on. Fire. Tanning. Water that will not kill you. Which plants are food, which are poison, and which are food only after six hours of leaching. None of it is in the genome. All of it was discovered by somebody, and every generation that survives is one that received the stock and added to it.

The process that produces it has a shape. You notice a problem. You guess at an explanation. You test the guess against reality. Reality kills most guesses, and the survivors you keep provisionally, until something better arrives. Conjecture, criticism, correction. It compounds, because each solved problem exposes problems that were invisible before it was solved. Fire makes cooking possible; cooking makes a smaller gut and a larger brain possible; and neither outcome was in view when someone first kept a flame alive overnight.

That last clause is where the whole difficulty lives. The process runs toward things that cannot be specified in advance.

What the Ministry Could Not Order

Which returns us to Beria.

An atomic bomb in 1945 was a known object. The Americans had built one, the physics was published or stolen, and Klaus Fuchs handed over the Los Alamos design. The Soviet state could point at a target and spend lives until it hit, because someone had already established that the target existed and roughly where it stood. Coercion is adequate to that job. Give a frightened physicist a specified problem and sufficient resources and he will frequently solve it.

Genetics offered no such target. Nobody could tell Lysenko what the answer was, because nobody knew it. That was the entire point of doing the work. So the question of which biology was correct got settled the only way a coercive system can settle an open question: by authority. The man with power declared the answer, the state enforced it, and because the answer was wrong there was no longer any mechanism for finding out.

That is what the twentieth century demonstrates. Directed effort toward a known destination survives tyranny, sometimes brilliantly. Undirected discovery does not, because central direction requires knowing what to direct toward, and the entire value of a discovery is that nobody knew.

The Soviet Union matched an American bomb. It could not have produced the personal computer, and no plan could have called for one, because a plan is a statement about what you already know you want. When Bardeen and Brattain were investigating semiconductor surface states at Bell Labs in December 1947, nobody had requisitioned a laptop, a mobile phone, or a global network. They were answering a much smaller question. The answer turned out to contain the century.

Freedom earns its place here epistemologically. It is the condition under which an open question gets settled by reality rather than by whoever holds the authority to end the argument.

The Second Person on the Beach

A man alone on an island needs no rights. He needs to think and to act, and nothing stands between the thinking and the acting. He can be wrong, and reality will inform him, and that exhausts the feedback his situation contains.

Rights become necessary the moment a second person walks onto the beach. Now there exists a way for the first man’s thinking to be severed from his acting that has nothing to do with whether he is right. The second man can stop him. Take what he built. Require that he ask before he tries.

This is what makes rights something other than a convention a society may adopt or decline according to taste. They are the answer to a specific problem, and the problem appears at a specific moment: when one person’s mind can be prevented from reaching reality by another person rather than by an error. Property recognizes that a man keeps what his thinking produced. Contract recognizes that two minds can bind themselves without a third forcing them. Free exchange is how one man’s solution reaches the people who have the problem it solves.

More people means more of this, not less. Society does not soften the requirement. It is the reason the requirement exists.

Markets are the part of this that gets misdescribed most often. A market is a criticism mechanism. Millions of guesses about what people need are made at once, reality answers each one in the form of whether anybody will trade for it, and the guesses that fail stop consuming resources. Prices carry those answers to people who never met the person who asked the question. It is the same conjecture-and-refutation loop that runs in a laboratory, running across an economy instead, at a scale no committee could hold in its head.

Which is why interventionism does its damage epistemically before it does it economically. Subsidize a failing enterprise and you have not merely misallocated money. You have suppressed a result. Reality was in the middle of answering a question, and the state stopped the answer from arriving.

Permission

The threat to this in 1950 wore a uniform. The threat now wears a lanyard and asks for a risk assessment.

The contemporary case against discovery never announces itself as a case against discovery. It says: proceed, by all means, once you have demonstrated that it is safe. Pause the models until we understand them. Establish the framework before deployment. Show us that the thing you have not built yet will not cause harm.

Put that way it sounds like prudence, and prudence is a virtue. But examine what is being required. Demonstrating in advance that a novel thing is safe means knowing its effects in advance, and knowing its effects in advance means having already done it. The demand is to produce the knowledge the experiment would have produced, before running the experiment. Where the requirement has teeth, it forbids whatever is genuinely new while permitting whatever is already understood—which is to say it permits precisely the directed work that flourished under Beria and forbids precisely the undirected work that Lysenko’s state could not do.

The cost is invisible by construction, which is what makes it so easy to impose. A refused grant generates no headline. The discovery that was not made leaves no remains.

Sometimes the cost is visible and gets imposed anyway.

In the mid-1990s researchers at the Swiss Federal Institute of Technology finished a decade of work on a modified strain of rice. They called it Golden Rice. By inserting two genes, one from maize and one from a soil bacterium, they engineered a grain that produces beta-carotene in the part people eat. Vitamin A deficiency affects roughly 250 million children. Between 250,000 and 500,000 of them go blind every year, and half of those die within twelve months of losing their sight.

Golden Rice worked. Independent testing confirmed it was safe, repeatedly, over two decades. It still is not widely deployed. Regulators across Asia, under pressure from organized opposition invoking the precautionary principle, delayed and restricted and in many cases blocked it. Activists destroyed the Philippine field trials in 2013. One analysis in Nature Biotechnology put the cost of the delay in India alone at 1.4 million life-years annually.

Nobody had to prove Golden Rice was dangerous. The children going blind were the status quo, and the status quo never has to justify itself. That is the asymmetry the whole permission regime runs on: existing harm carries a weight of zero and speculative harm carries infinite weight, and the arithmetic is performed in the dark.

The other face of it is the discovery that nearly did not happen. In 1995 the University of Pennsylvania demoted Katalin Karikó after her grant applications on messenger RNA were rejected again and again. Blame the reviewers and you miss the mechanism. They were performing the job the system asks of reviewers, which is to judge in advance whether unproven work will pay—the one judgment that cannot be made about work whose value is still unknown, which is the only kind of work that matters. Karikó continued without rank or funding. The nucleoside-modification results she published with Drew Weissman in 2005 are what the mRNA vaccines were built on. She took the Nobel Prize in 2023.

The system said no and the system was wrong, and the sole reason anyone knows it was wrong is that one person disregarded it. There is no way to count the times nobody did.

The Bill

None of this makes discovery harmless. The century that produced antibiotics produced Zyklon B, and the same physics that ended one war designed the device fired at Semipalatinsk. Every real capability can be turned. Anyone selling you a creative process without a body count is selling something.

The argument is narrower and harder to escape. We have no other way to live. The nardoo stays poison until somebody works out the leaching, and there is no version of this animal that skips the step and eats safely on instinct. Every generation inherits a stock of solved problems and a fresh supply of unsolved ones, and the second category is not optional. Conditions change. Old solutions expire. A society that has stopped producing new knowledge is not holding steady; it is drawing down a reserve, and the reserve was never large.

So the choice was never between risk and safety. It is between the risks of finding out and the risks of not knowing, and only the first kind can be corrected after you see it.

Vavilov starved in a cell because a man with authority closed a question that was still open. A permission regime performs the same act with better manners, on the same kind of question, and calls the result care.