He played bongo drums in strip clubs. He cracked safes holding nuclear secrets at Los Alamos — not with tools, but with logic. He debated quantum mechanics with Einstein. He died in 1988 having reshaped how science is calculated, taught, and trusted. The infrastructure he left behind is still running.

Feynman had a test. If you could not build it from its reasons, you did not understand it. Not approximately. Not yet.

This was not a teaching philosophy. It was a personal standard he applied to himself first. At MIT, then at Princeton, he derived results that textbooks handed to other students as received fact. He needed to feel why something was true — not just confirm that it was. His professors found it irritating. The habit produced first-principles reasoning as a method, not a metaphor.

The method runs like this. Strip the problem to its actual structure. Ignore what authority says the answer should be. Follow the logic wherever it goes. Then check it against reality — not against consensus.

This is not unusual advice. What was unusual was that Feynman actually did it, on problems that mattered, at the highest level of difficulty physics offered in the twentieth century. And then he showed his work.

His father Melville started it. Walking through Queens with a young Richard, Melville would name a bird — "That's a brown-throated thrush" — and immediately add: knowing its name tells you nothing about it. Watch how it moves. Ask why. The name is not the thing. Most education reverses this. Feynman never let it.

By the time he arrived at Princeton as a graduate student in 1939, the habit was structural. John Wheeler, his advisor, recognized it immediately. Wheeler pointed Feynman at the hardest open problems in theoretical physics and stepped back. What came out, over the next decade, was a series of results so original that existing mathematical language could not fully contain them.

Feynman built new language. That is what the diagrams were.

Because the mathematics was breaking.

In the 1940s, quantum electrodynamics — QED, the theory describing how light and matter interact — was producing answers that collapsed into infinity. Calculate the energy of an electron. The equation returned nonsense. Not a large number. Infinity. Literally. Physicists knew the theory was pointing at something real. They could not make it compute.

Feynman's solution was renormalization — a technique for absorbing the infinities into the definitions of measurable quantities like mass and charge. The infinities did not disappear. They were systematically relocated. What remained was calculable. And the calculations were extraordinary.

QED, in its renormalized form, predicts the magnetic moment of the electron to eleven decimal places. The experimental measurement agrees to eleven decimal places. This is the most precisely verified prediction in the history of science. The technique that made it possible remains, in the view of some mathematicians, formally unjustified. It works anyway.

Feynman was not comfortable with this. He said so publicly. "The shell game that we play," he called it. He did not pretend the foundations were cleaner than they were. That honesty is itself a data point about how he worked.

The Feynman diagrams came from the same problem. QED calculations required tracking every possible way a particle interaction could occur — every path a particle might take, every virtual particle that might briefly exist and vanish. The bookkeeping was monstrous. Pages of notation for a single calculation.

Feynman replaced the notation with pictures. Each diagram represented a physical process. Each line, each vertex, each loop had a precise mathematical rule attached. The pictures were not illustrations. They were the calculation. You drew the diagram. You applied the rules. You got the number.

He introduced them at the Pocono Conference in 1948. The reception was skeptical. Freeman Dyson — then a young British mathematician newly arrived in America — recognized what Feynman had done. Dyson translated the diagrams into rigorous mathematical language that the broader physics community could accept. The field came around.

Physicists at the Large Hadron Collider in Geneva still use Feynman diagrams. Seventy-six years after Pocono. They may be the most productive scientific notation produced in the twentieth century. A picture that is also a proof.

In 1942, Feynman was twenty-four. The Manhattan Project needed theoretical physicists. He went to Los Alamos and worked alongside Hans Bethe, Niels Bohr, Enrico Fermi — the core of twentieth-century physics, assembled in the New Mexico desert to build the most destructive weapon in human history.

Feynman's contribution was computational. He organized teams of human calculators — this was before electronic computers — to run the numerical models that the bomb's design required. Bethe trusted him completely. Oppenheimer trusted him enough to introduce him to Bohr by name.

And then there were the safes.

Los Alamos kept its classified documents in filing cabinets and combination safes. Feynman opened them. Not violently. Not with stolen combinations. With observation and reasoning. He noticed that most people chose combinations they could remember — birthdays, recurring numbers, mathematically convenient sequences. He noticed that many cabinets were left unlocked or set to their factory defaults. He noticed patterns in how people carried their keys.

The safe-cracker method was not mischief, though Feynman enjoyed presenting it that way. It was a demonstration. Impenetrable systems have seams. The seams are usually human. Security that relies on complexity rather than actual closure is not security — it is theater. He left notes in the opened safes. The security officers were not amused.

The lesson applied beyond Los Alamos. Feynman spent his life identifying the seams in apparent impermeability — in physical theories, in institutions, in arguments that relied on their own authority rather than their actual structure.

In 1961, Caltech asked Feynman to revamp its introductory physics course. The existing curriculum was organized around coverage — work through the topics, hit the required material, produce graduates who knew the canonical results.

Feynman refused this structure. What he delivered instead, over two academic years, was an argument. Physics organized around ideas, not facts. Around the reasons things are true, not the catalog of what is true.

The Feynman Lectures on Physics — three volumes, published 1963 to 1965 — are the record of that argument. They are still in print. Since 2013 they have been available free online, with the permission of Caltech and the Feynman estate. They are among the most-read physics texts in the world.

The lectures are not easy. Feynman was not making things simple. He was making things honest. The distinction matters. Simple removes difficulty. Honest shows you where the difficulty actually lives and refuses to pretend otherwise.

He said explicitly: teaching is not the transfer of information. It is the construction of understanding. If a student cannot derive the result — cannot feel why it is true — then whatever has been transferred is not understanding. It is vocabulary. Useful, perhaps, but not the thing itself.

This argument challenged how universities teach science. It still does. The standard curriculum hands students canonical results and trains them to reproduce calculations. Feynman's model demanded that students encounter the reasons — that they build the edifice rather than inherit the furniture.

Most institutions have not adopted his model. The Lectures persist as a standing rebuke.

On January 28, 1986, the Space Shuttle Challenger broke apart seventy-three seconds after launch. Seven crew members died. NASA convened the Rogers Commission to investigate.

Feynman was sixty-seven and seriously ill with kidney cancer. He joined anyway.

The official investigation moved carefully, institutionally, at the pace that official investigations move. Feynman moved differently. He talked directly to engineers — not administrators — and asked them to quantify their confidence in the shuttle's components. The engineers gave him numbers. The numbers did not match NASA's public statements about mission safety.

The O-rings were the physical failure point. They were rubber seals in the solid rocket boosters. They were not rated for cold temperatures. The launch temperature on January 28 was 28°F. Engineers had raised concerns the night before. The launch proceeded.

At a commission hearing, Feynman produced a glass of ice water, a piece of O-ring material, and a C-clamp. He compressed a sample of the material, submerged it in the ice water, removed it, released the clamp, and showed the commissioners what happened: the material did not immediately return to its original shape. It had lost its resilience. It could not seal.

He said nothing dramatic. The demonstration said everything.

His dissenting appendix to the Rogers Commission report went further. NASA had been presenting mission risk to management as roughly one-in-one-hundred-thousand. Its own engineers, privately, estimated closer to one-in-one-hundred. Feynman named this gap directly. He called it a kind of self-deception — institutional pressure distorting risk assessment until the distorted assessment became the official one.

The appendix almost was not published. Commission leadership wanted it omitted. Feynman insisted. It appears as Appendix F.

His final line: "For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled."

This was his last great public act. He died on February 15, 1988.

Feynman said it in 1981, in The Pleasure of Finding Things Out: "If you thought that science was certain — well, that is just an error on your part."

This was not false modesty. It was not performance. It was the structural position he had held since he was a student in Queens listening to his father explain that the name of a thing is not the thing.

Epistemic humility — genuine uncertainty in a person with genuine mastery — is rarer than it sounds. Most experts perform uncertainty and practice certainty. Feynman inverted this. He knew more than almost anyone in his field. He was more publicly uncertain than almost anyone in public life. The combination was not a paradox. It was what serious inquiry actually looks like.

This becomes more important, not less, as trust in expertise fractures. The fracture is partly deserved. Institutions have claimed certainty they did not possess. They have suppressed internal dissent. They have allowed public relations to override private assessment. Feynman named all of this while it was happening — at Los Alamos, at Caltech, on the Rogers Commission.

What he modeled was a specific practice. Not cynicism about knowledge. Not the conspiracy register that mistakes doubt for revelation. Something harder: the ongoing discipline of holding what you know loosely enough that reality can correct it.

He also modeled something more uncomfortable. Brilliance and ethical failure lived in the same person. Feynman's treatment of women — documented in multiple accounts, including his own writing — sits alongside the diagrams, the lectures, the ice-water demonstration. This is not a footnote. It is part of the full record.

Erasing the tension would be the kind of dishonesty Feynman himself refused. Featuring it without naming it would be the same. He demanded honesty above comfort. That demand applies to him.

The deeper question is structural. Who gets to be publicly uncertain without losing credibility? Who gets to be wrong on the way to being right? Who gets to crack the metaphorical safe — playful, disruptive, uncontained — and have it read as genius rather than threat?

Feynman got to. The conditions that allowed it are worth examining.

If the next person working from first principles is already alive somewhere — taking apart a device, refusing the received answer, reasoning from the ground up — what are we doing right now to make sure we recognize her?