Physics · Pedagogy · The Character of Knowledge

Richard P.
Feynman

1918 — 1988 · Far Rockaway → Los Alamos → Caltech

Nobel laureate, reinventor of quantum electrodynamics, author of the path integral, prophet of nanotechnology and quantum computing, drummer, safecracker, and the twentieth century's greatest explainer of physics. This guide surveys what he built, how he thought, and why both still matter.

"What I cannot create, I do not understand."
Chapter One

A Life in Motion

from radio repair in Far Rockaway to the Rogers Commission

Richard Phillips Feynman was born in Far Rockaway, Queens, in 1918. His father Melville, a uniform salesman, taught him the difference between knowing the name of a thing and knowing the thing — a distinction that became the engine of his entire intellectual life. As a boy he fixed radios "by thinking," ran a home laboratory, and rederived mathematics for himself in notebooks with private notation. That habit never left: throughout his career he insisted on reconstructing every result from scratch, in his own terms, before he would believe it.

The character behind the physics. The bongo drums, the strip-club office hours, the Tuva obsession, the safecracking — these are not decoration on the science. They are the same trait expressed differently: an absolute refusal to accept that anything, from a lock to a Lagrangian, was off-limits to direct personal investigation.
"I was born not knowing and have had only a little time to change that here and there."
Letter, 1988 · quoted in Perfectly Reasonable Deviations from the Beaten Track
Chapter Two

The Sum Over Histories

every path, all at once, each carrying a spinning arrow

Classical mechanics says a particle takes one path from A to B — the one that makes the action stationary. Feynman's 1948 reformulation of quantum mechanics says something stranger and more beautiful: the particle takes every path. Each path contributes a complex amplitude of equal magnitude whose phase is the classical action along that path, measured in units of ħ. The probability amplitude to go from A to B is the sum of these contributions over all conceivable histories.

K(b,a) = ∫ 𝒟x(t) · e iS[x(t)] / ħ,   S[x(t)] = ∫tatb L(x, ẋ, t) dt The Feynman kernel: integrate e^(iS/ħ) over all paths x(t) from a to b

Why does the classical world emerge from this democracy of paths? Interference. For paths far from the classical trajectory, the action varies rapidly from one neighboring path to the next, so their phase arrows point every which way and cancel. Near the stationary-action path, neighboring paths have nearly identical phases — their arrows align and reinforce. When the action is enormous compared with ħ (a baseball, a planet), the cancellation is so ruthless that only the classical path survives. Quantum mechanics doesn't contradict Newton; it explains him.

Blackboard I — Sum over histories

Random paths from A to B. Each path gets a phase arrow e^(iS/ħ); the chain on the right adds them tip-to-tail. Slide ħ down and watch the classical path take over.

resultant |amplitude|:

Where it came from, and what it bought us

The germ was a 1933 paper by Dirac remarking that eiS/ħ "corresponds to" the quantum propagator. At a Princeton beer party, Herbert Jehle pointed Feynman to it; by the next morning Feynman had discovered that Dirac's "corresponds to" meant, for infinitesimal times, "is proportional to" — and built an entire formulation of quantum mechanics out of it. His 1942 thesis and the 1948 Reviews of Modern Physics paper made it rigorous enough to use.

The payoff was enormous. The path integral made Lorentz invariance manifest where the Schrödinger picture obscured it; it generalized effortlessly to fields, where "sum over paths" becomes "sum over field configurations"; and it became the natural language of gauge theory, critical phenomena, and string theory. Modern lattice QCD is, quite literally, a Monte Carlo evaluation of a Feynman path integral on a supercomputer.

Three formulations, one theory. Schrödinger's waves, Heisenberg's matrices, and Feynman's paths are mathematically equivalent. But equivalence isn't interchangeability: each makes different problems easy, and Feynman argued that a physicist should carry all available formulations, because "every theoretical physicist who is any good knows six or seven different theoretical representations for exactly the same physics."
Chapter Three

QED and the Diagrams

the jewel of physics, drawn in cartoons

Quantum electrodynamics — the theory of how light and matter interact — was in crisis in the 1940s. Every calculation past the first approximation produced infinity. Between 1947 and 1949, Feynman, Schwinger, and Tomonaga independently tamed the infinities through renormalization; Feynman's route was the most idiosyncratic and the most transportable. He replaced pages of algebra with pictures: spacetime diagrams in which straight lines are electrons, wavy lines are photons, and every vertex where they meet contributes a factor to the amplitude.

A Feynman diagram is not a cartoon of what "really happens" — it is a term in a perturbation series, a bookkeeping device so effective it restructured how physicists think. Each diagram translates, by fixed rules, into an integral; summing diagrams of increasing complexity refines the prediction. The method spread from QED to nuclear physics, condensed matter, gravity, and beyond. It remains the lingua franca of particle theory.

Blackboard II — Electrons trading a photon

Time runs left to right. Toggle between electron–electron scattering (Møller) and electron–positron annihilation. The wiggly line is the virtual photon doing the work.

Renormalization: subtracting infinity, honestly

The infinities of QED arise because an electron interacts with its own field. Renormalization absorbs those divergences into the measured mass and charge of the electron: the "bare" parameters in the equations are unobservable, and once everything is expressed in terms of measured quantities, every prediction comes out finite. Feynman was never fully at peace with it — he called it "a dippy process" and suspected it hid a deeper flaw — but the numbers were, and remain, unarguable.

QuantityQED predictionStatus
Electron anomalous magnetic moment (g−2)/2≈ 0.001 159 652 18…Agrees with experiment to about one part in 10¹² — often called the most precisely verified prediction in all of science.
Lamb shift in hydrogen≈ 1057 MHz splittingThe 1947 measurement that triggered the renormalization race; matched by theory within experimental error.
Hyperfine structure, positronium, muon g−2All computed diagrammatically; the diagram expansion for g−2 now runs to five loops and over ten thousand diagrams.

For this work Feynman shared the 1965 Nobel Prize with Schwinger and Tomonaga. Characteristically, he claimed the real prize was "the pleasure of finding the thing out, the kick in the discovery, the observation that other people use it."

Chapter Four

The Little Arrows

all of QED, explained with a stopwatch — from the 1985 book

In QED: The Strange Theory of Light and Matter — four public lectures given in memory of Alix Mautner, who had asked him to explain his physics without mathematics — Feynman reduced the entire theory to one procedure: for every way an event can happen, draw a little arrow whose angle is set by an imaginary stopwatch that spins as the photon travels; add the arrows head-to-tail; square the length of the final arrow to get the probability. That's it. That is quantum electrodynamics.

His flagship example is partial reflection. Shine light on a sheet of glass and about 4% reflects from a single surface. But from a two-surface sheet, the reflected fraction cycles between 0% and 16% as the glass gets thicker — a fact that baffled Newton. In arrow language it is transparent: one arrow for reflection off the front surface, one for the back. The back-surface arrow's angle grows with the round trip through the glass; the front arrow gets an extra half-turn. When the arrows oppose, near-zero reflection; when they align, 16%.

Blackboard III — Partial reflection by a sheet of glass

Drag the thickness. Left: the two arrows and their sum. Right: probability = (length of final arrow)², tracing Newton's mystery curve from 0% to 16%.

reflection probability:

The deeper point of the book is epistemological. The arrows are not a metaphor for a mechanism; they are the theory, and the theory declines to say more. "The theory of quantum electrodynamics describes Nature as absurd from the point of view of common sense. And it agrees fully with experiment. So I hope you can accept Nature as She is — absurd."

"Physics is like sex: sure, it may give some practical results, but that's not why we do it."
Widely attributed · in the spirit of the 1965 Nobel lecture
Chapter Five

Beyond QED

cold helium, broken mirrors, and what's inside a proton

Feynman's reputation rests on QED, but three other contributions would each anchor a distinguished career on their own.

Superfluid Helium (1953–58)

Below 2.17 K, helium-4 flows with zero viscosity. Feynman built the microscopic quantum theory of this phase: he explained the roton excitation spectrum from first principles, showed that circulation is quantized in units of h/m, and predicted quantized vortex lines — later observed directly. He treated the many-body wavefunction with the same visual audacity he brought to QED, imagining atoms trading places in a quantum liquid.

V−A Weak Interaction (1957–58)

After parity violation stunned physics in 1957, Feynman and Murray Gell-Mann proposed that the weak force couples through a "vector minus axial-vector" current — meaning it acts only on left-handed particles. Published in 1958 (with Marshak and Sudarshan arriving at the same structure independently), V−A became a pillar of the electroweak theory. Feynman ranked it among the things he was proudest of: a law of nature he could claim, in part, as his own.

The Parton Model (1969)

When SLAC fired electrons into protons and saw them ricochet as if off hard grains inside, Feynman supplied the picture: at high energy a proton behaves as a bag of nearly free point-like constituents — "partons." The model explained the scaling seen in deep inelastic scattering and was later identified with quarks and gluons, becoming the computational foundation for every prediction made at particle colliders since, including the LHC.

Quantum Gravity & Ghosts (1962–63)

Feynman was among the first to treat gravity as a quantum field theory of spin-2 gravitons. In working out loop diagrams he discovered that unitarity requires fictitious "ghost" particles circulating in closed loops — a result that prefigured the Faddeev–Popov ghosts essential to the quantization of all modern gauge theories.

A style, not just a list of results. The common thread is visual, particle-first thinking pushed into domains ruled by formalism — plus a insistence on computing real numbers. Colleagues called it unfair: Bethe said there were two kinds of genius, the ordinary kind you could imagine being much smarter than yourself, and the magicians. Feynman, he said, was a magician.
Chapter Six

The Great Explainer

the red books, and the ethics of understanding

Between 1961 and 1963, Caltech asked Feynman to reinvent its introductory physics course. The result — transcribed, edited, and published as The Feynman Lectures on Physics, the famous three-volume "red books" — is probably the most influential physics text of the twentieth century. It is freely readable online today, and it remains what it always was: not a syllabus but a demonstration of what it feels like when a first-rate mind rebuilds all of physics in front of you.

What made the pedagogy different

Feynman taught from phenomena toward formalism, never the reverse. Volume I opens not with units and vectors but with the single most information-dense sentence he could construct: if all scientific knowledge were destroyed but one sentence survived, it should be that all things are made of atoms — little particles that move around in perpetual motion, attracting each other when they are a little distance apart, but repelling upon being squeezed into one another. Everything else, he claimed, follows from imagination and thinking applied to that.

His standard for understanding was constructive: at his death, his last blackboard carried the line quoted in this guide's header — "What I cannot create, I do not understand." The corollary he practiced on students and colleagues alike: if you cannot explain an idea in plain language to a bright beginner, you do not yet understand it yourself. That test circulates today as the "Feynman technique" — choose a concept, explain it simply on one page, find where you stumble, return to the source, repeat.

The books beyond the Lectures

WorkYearWhat it is
The Feynman Lectures on Physics (3 vols.)1964The red books: mechanics and heat; electromagnetism and matter; quantum mechanics.
The Character of Physical Law1965Seven Cornell Messenger Lectures on what physical laws are and how nature "uses" mathematics.
QED: The Strange Theory of Light and Matter1985Quantum electrodynamics for the general reader, via the little arrows of Chapter Four above.
Surely You're Joking, Mr. Feynman!1985Anecdotal autobiography, dictated to Ralph Leighton between drumming sessions.
What Do You Care What Other People Think?1988Arline, the Challenger investigation, and the value of doubt.
Feynman Lectures on Computation1996 (posth.)His 1980s Caltech course: reversible computing, thermodynamics of computation, quantum computers.
"The first principle is that you must not fool yourself — and you are the easiest person to fool."
Caltech commencement address, 1974 · "Cargo Cult Science"
Chapter Seven

Seeds of the Future

two talks that founded two fields

1959 — "There's Plenty of Room at the Bottom"

At an American Physical Society dinner talk at Caltech, Feynman asked why we couldn't write the entire Encyclopædia Britannica on the head of a pin — and then answered, in detail, that no law of physics forbids it. He sketched miniaturized machines building smaller machines, atom-by-atom manipulation of matter, dense information storage, and the biological cell as an existence proof of molecular machinery. He put up two $1,000 prizes: one for a working motor smaller than 1/64 inch (claimed within a year, to his mild disappointment, by conventional craftsmanship), one for text at 1/25,000 scale (claimed in 1985 by electron-beam lithography). The talk is now cited as the founding vision of nanotechnology — decades before the field had a name or the instruments, like the scanning tunneling microscope, to pursue it.

1981 — "Simulating Physics with Computers"

At an MIT conference on the physics of computation, Feynman posed a sharp question: can a classical computer simulate quantum mechanics efficiently? His answer was no — the memory required to track a general quantum state grows exponentially with the number of particles. Then the characteristic Feynman turn: if nature is quantum mechanical, build a computer out of quantum mechanics. "Nature isn't classical, dammit, and if you want to make a simulation of nature, you'd better make it quantum mechanical." The 1982 paper, with his 1985 follow-up on quantum mechanical computers, made him a founding figure of quantum computing — the field that today pursues exactly the quantum simulations he specified, and in which his name attaches to the standard path-integral-inspired techniques of the trade.

Also in the file: at Los Alamos he organized human and punched-card computation into what was effectively pipelined parallel processing; in the 1980s he consulted for Thinking Machines on the Connection Machine, working out the router equations for its hypercube network with differential equations no one expected to apply; and his lectures on computation treat Bennett's reversible computing and the thermodynamic cost of erasure — Landauer's principle — years before these became mainstream concerns.
Chapter Eight

Ice Water

the Challenger inquiry, 1986

On January 28, 1986, the Space Shuttle Challenger broke apart 73 seconds after launch, killing all seven crew members. Feynman — already ill with the cancer that would kill him two years later — was recruited to the presidential commission. He did what he always did: went around the official channels, talked directly to the engineers, and ran his own experiments.

At a televised hearing on February 11, he compressed the technical heart of the disaster into thirty seconds: he clamped a sample of the solid rocket booster's O-ring rubber, dunked it in his glass of ice water, and showed that at 32 °F the material lost its resilience — it would not spring back to seal the joint. Challenger had launched on the coldest morning in shuttle history, against engineers' warnings.

His personal appendix to the commission report went further than the demonstration. Management, he found, had estimated the odds of losing a shuttle at 1 in 100,000; the engineers who built it put the figure nearer 1 in 100. That gap — a thousandfold divergence between institutional optimism and engineering judgment — was for Feynman the real cause. His appendix closes with a sentence that has outlived the report: "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled."

Chapter Nine

The Character of Scientific Knowledge

doubt as a discipline, not a mood

Feynman's philosophy of science was unsystematic on purpose — he distrusted philosophy of science roughly as much as he loved science — but it was coherent, and it can be stated in his own recurring themes.

Guess → Compute → Compare

His one-slide account of scientific method, from the Messenger Lectures: first we guess a new law; then we compute its consequences; then we compare with experiment. "If it disagrees with experiment, it's wrong. In that simple statement is the key to science." It does not matter how beautiful the guess is or how smart the guesser.

Cargo Cult Science

From the 1974 Caltech commencement address: research that has the form of science — instruments, journals, jargon — but lacks "a kind of scientific integrity, a kind of utter honesty, a kind of leaning over backwards" to report everything that might make you wrong. The image: Pacific islanders building bamboo control towers, waiting for the planes that never land.

Satisfactory Ignorance

"I can live with doubt and uncertainty and not knowing. I think it's much more interesting to live not knowing than to have answers which might be wrong." Approximate answers, degrees of certainty, and open questions are not failures of science; they are its normal, honest condition.

Names Are Not Knowledge

The lesson from Melville Feynman's bird walks: you can know the name of a bird in every language on Earth and know nothing whatever about the bird. Vocabulary is a map index, not the territory. The test of understanding is always the ability to reconstruct, predict, and explain — never the ability to label.

These are working rules, not aphorisms. They governed how he refereed his own results (the appendix to his 1957 superfluidity paper lists the ways he might be wrong), how he handled the Challenger evidence, and how he taught. If there is a single Feynman doctrine, it is that integrity toward nature is a craft skill — learnable, practicable, and prior to genius.

Chapter Ten

The Octave Lab

recreate the blackboards numerically — GNU Octave, no packages required

Per the house rule of this series — what we cannot create, we do not understand — here are the two central demonstrations of this guide as short GNU Octave scripts. Paste them into the Octave REPL on the MacBook and compare with Blackboards I and III.

path_integral_demo.mGNU Octave
% Sum over histories: free particle from A to B in 1D.
% Perturb the straight-line path with random Fourier modes; each
% path contributes exp(i*S/hbar) with S = (m/2)*integral(v.^2)dt.
% Watch the resultant amplitude localize as hbar shrinks.

function path_integral_demo(hbar)
  if nargin < 1, hbar = 1.0; end
  m = 1;  T = 1;  Nt = 200;  Npaths = 4000;  Nmodes = 8;
  t  = linspace(0, T, Nt);
  x0 = t;                        % classical path: x = t (A=0 to B=1)
  amp = 0;  S_all = zeros(1, Npaths);

  for k = 1:Npaths
    x = x0;
    for n = 1:Nmodes               % sine perturbations vanish at endpoints
      x = x + (randn/ n) * 0.15 * sin(n*pi*t/T);
    end
    v = diff(x) ./ diff(t);
    S = (m/2) * sum(v.^2 .* diff(t));   % discretized action
    S_all(k) = S;
    amp = amp + exp(1i * S / hbar);
  end

  amp = amp / Npaths;
  printf("hbar = %.3f   |amplitude| = %.4f\n", hbar, abs(amp));
  hist(S_all, 60);  xlabel("action S");  ylabel("paths");
  title("Actions of sampled histories (min at classical path, S = 0.5)");
end

% Try: path_integral_demo(1.0), then 0.3, then 0.05 —
% coherence survives only near the stationary-action path.
partial_reflection.mGNU Octave
% Feynman's little arrows: partial reflection by a two-surface
% sheet of glass (QED, Lecture 1). Front arrow gets a half-turn;
% back arrow's angle grows with the round trip through the glass.
% Probability = |sum of arrows|^2, cycling between 0 and ~16%.

r = 0.2;                          % single-surface amplitude (4% = r^2)
phase = linspace(0, 4*pi, 800);   % round-trip phase ~ thickness

front = -r * ones(size(phase));           % reflection off front (+ half turn)
back  =  r * exp(1i * phase);             % reflection off back surface
P = abs(front + back).^2;                 % two-arrow approximation

plot(phase/(2*pi), 100*P, "linewidth", 2);  grid on
xlabel("round-trip phase / 2\pi  (∝ glass thickness)");
ylabel("reflection probability (%)");
title("Newton's mystery: reflection cycles 0% → 16% with thickness");
printf("max = %.1f%%   min = %.1f%%\n", max(100*P), min(100*P));

Both scripts are deliberately minimal — extend the first to two dimensions and you have the skeleton of a lattice Monte Carlo; add multiple internal reflections to the second (a geometric series in r·e) and you recover the exact Airy formula for thin-film interference.