The deep dive · one of eight answers to the same experiment

Many-Worlds

“Everything happens.”

Take the math at its word and the price is everything else. The interpretation for people who trust the equation over their own feelings — all the way down.

The train doors are closing. In one cut she slips through; in the other she's half a second late and watches it pull away.

Ten years later she's in a kitchen in Tokyo, laughing: "If I'd missed that train, none of this happens."

Ten years later she's on a rooftop in Barcelona, laughing: "If I'd caught that train, none of this happens."

Nobody's lying. Nobody's dreaming. Both mornings happened, both lives got lived, and each of her is certain she's the only one telling the story. The doors didn't choose. The doors did both.

Now shrink it.

The experiment that started all the trouble

Fire tiny particles at a wall with two slits, one at a time, with a screen behind it. One at a time matters: there's only ever one particle in flight. Common sense says each one goes through one slit or the other, and you get two piles. You don't get two piles. You get stripes — the pattern waves make when they ripple through both slits and overlap. So you put a detector at the slits to catch which one each particle uses. The stripes vanish. Two piles.

Everything on this page is one theory's answer to a single question: what, in the name of God, is going on between the gun and the screen?

Drag to orbit — two fingers on a touchscreen. Particles leave one at a time and land one at a time. The middle is deliberately not drawn — what happens there is the whole argument, and the lens is this theory's answer.

The idea

One equation runs the whole quantum world, and nothing in it ever says "collapse." Many-Worlds just takes that seriously. When an experiment could go two ways, the equation says both happen — so both do. The universe branches. You branch with it. The you reading this sentence is one of the versions, and every result this quiz could have given you was given, somewhere.

What it fixes

Everything, using nothing. No collapse, no special role for "observers," not one symbol added to the math. It's the cheapest interpretation ever proposed — measured in assumptions.

What it costs

Measured in universes, it's the most expensive idea in the history of ideas. Also: if everything happens, what did "30% chance" ever mean? Everettians have answers. They have *several* answers. That's the problem.

The story

In 1957 a Princeton PhD student named Hugh Everett III asked the forbidden question in its simplest form: what happens if we just delete the collapse rule? Keep the equation — the [[unitary]], smoothly-evolving Schrödinger equation — and add nothing. His answer: the equation already describes a measurement perfectly well. The measuring device ends up correlated with both outcomes. So does the physicist reading it. The mathematics doesn't describe one result happening; it describes both, each with an observer who saw it cleanly. Everett's move was to say: believe that.

Copenhagen's old guard was not amused. Bohr's circle dismissed the thesis; Everett got the message, finished a heavily trimmed version, and walked out of physics into Pentagon war-gaming. He died in 1982, at fifty-one, decades before his idea became respectable. It got its name — "many worlds" — from Bryce DeWitt, who championed it in the 1970s when almost nobody else would.

The modern version is more careful than the science-fiction picture. Branches aren't parallel universes that "split" with a pop; they're parts of one wavefunction that [[decoherence]] has made permanently unable to interact — practically separate histories, all equally real, all embedded in the same math. Nothing collapses, nothing is added, observers aren't special: you're just one more physical system that ends up on every branch, and each version of you finds a single definite past in their memory.

The honest open wound is probability. If every outcome happens, what were you saying when you said "30% chance"? The Oxford Everettians — David Deutsch, David Wallace — have a sophisticated answer involving how a rational agent should bet across their own branches. Others have different answers. The critics' point isn't that there's no answer; it's that there are several, and they don't agree, and for a theory whose whole sales pitch is "we added nothing," needing this much philosophy at checkout stings.

The argument

“An infinity of unobservable universes is the least parsimonious idea ever proposed.”
Count rules, not universes. I have one postulate and it's the equation you already believe. You're the one adding a collapse mechanism nobody has ever written down properly.
“It's untestable — the other branches can't be observed even in principle.”
I'm not a separate theory; I'm quantum mechanics, taken literally. Every interference experiment that keeps working on bigger objects is a test I keep passing. Find a real collapse and I'm dead. People are looking. That's testable enough.
“If everything happens, choices don't matter.”
The version of you that chooses well still lives a different life than the one that doesn't — and the weights aren't decoration. You're not off the hook. You're on every hook.

The experiment

Interference of 25,000-dalton molecules · 2019

The "both at once" trick was supposed to be for tiny things. Then it worked on atoms. Then molecules. In 2019 it worked on molecules of around two thousand atoms each — big enough to almost see — and the wave behavior showed up right on schedule. Every year the trick works on something bigger, and nothing in the math says it stops before it reaches you.

Fein et al., Nature Physics

The books

Something Deeply Hidden by Sean Carroll — the case for taking the equation at its word, written by the person at the party you'd actually enjoy arguing with. Then David Wallace's The Emergent Multiverse when you're ready to go professional.

Your neighbors

Pilot Wave — your fellow wave-realist — kept the wave, added a particle, skipped the branching.
Superdeterminism — the other fully deterministic picture — one history instead of all of them.

← all theories
Glossary 24 terms
wavefunction

The mathematical object that assigns a complex number to every possible configuration of a system.

Written ψ. It is not a wave in space like a ripple on a pond — for two particles it lives in a six-dimensional configuration space, not in the room. Whether ψ is a real physical thing or a bookkeeping device for what you know is precisely what the interpretations argue about.

See also: amplitude, ontic, epistemic

amplitude

A complex number whose squared magnitude gives a probability.

Amplitudes are what makes quantum mechanics quantum. Because they are complex, they can cancel — two ways of reaching the same outcome can add up to no chance of it happening at all. Classical probabilities can never do this.

See also: Born rule, interference

Born rule

The recipe that turns amplitudes into probabilities: probability equals the amplitude's magnitude squared.

Max Born added it in a footnote in 1926 and won a Nobel Prize for the footnote. Every interpretation has to reproduce it, and several struggle to explain why it holds rather than simply assuming it.

See also: amplitude

superposition

A state that is a combination of other states, with amplitudes attached.

Commonly mangled as "being in two places at once." More accurately: the system is in one perfectly definite state, which happens not to be a state of definite position (or spin, or whatever you are about to measure). The indefiniteness is relative to the question you ask.

See also: eigenstate, interference

eigenstate

A state with a definite value for some particular observable.

A state is only ever an eigenstate with respect to a specific question. Definite momentum means wildly indefinite position, and vice versa. There is no state that is definite about everything.

See also: superposition

interference

Amplitudes for different paths adding or cancelling, producing fringes.

The double slit is the canonical case: close one slit and the bright band at a given spot can appear, open both and it can vanish. Adding a second way for something to happen made it stop happening. No probability theory built on ordinary numbers does this.

See also: amplitude, decoherence

the measurement problem

Unitary evolution never produces a single definite outcome, yet we only ever see one.

The Schrödinger equation is linear and deterministic, so a measuring device interacting with a superposition should end up in a superposition of readings. It doesn't — you see one number. Every interpretation on this quiz is, at bottom, a different answer to this one problem.

See also: unitary, collapse, decoherence

collapse

The postulated jump from a superposition to a single definite outcome on measurement.

In textbook quantum mechanics it is simply an extra rule bolted alongside the Schrödinger equation, with no account of when it applies or what counts as a measurement. Interpretations either explain it, deny it happens, or make it a real physical process with its own dynamics.

See also: the measurement problem, spontaneous collapse

decoherence

Interaction with the environment rapidly destroying interference between branches.

Real and experimentally confirmed, and it explains why you never see a superposed cat. But it does not by itself solve the measurement problem, however often it is claimed to: it explains why the branches stop interfering, not why you end up in exactly one of them.

See also: the measurement problem, interference

unitary

Evolution that is reversible and preserves total probability — the Schrödinger equation.

Unitary dynamics never destroys information and never singles out an outcome. An interpretation that insists dynamics is always unitary (Many-Worlds) must therefore explain single outcomes some other way; one that admits non-unitary collapse must say when and why.

See also: the measurement problem, collapse

ontic

About what exists, independently of anyone's knowledge.

An ontic reading of the wavefunction says ψ is a real physical thing, as much a part of the furniture of the world as a field. Pilot Wave and spontaneous collapse are ontic about ψ.

See also: epistemic, the PBR theorem

epistemic

About what someone knows, rather than about the world itself.

An epistemic reading says ψ encodes an observer's information, so "collapse" is just updating your beliefs — no more mysterious than a probability changing when you look at a card. QBism takes this furthest. The PBR theorem is the main obstacle in its path.

See also: ontic, the PBR theorem, QBism

hidden variables

Extra facts beyond the wavefunction that would fix what actually happens.

The hope was that quantum randomness is like a shuffled deck — merely ignorance about details already there. Bell's theorem does not kill hidden variables; it kills *local* ones. Pilot Wave is a hidden-variable theory that survives by being frankly nonlocal.

See also: Bell's theorem, locality, pilot wave

locality

Nothing here is influenced by a choice made far away, faster than light could carry the news.

Give this up and you can keep definite pre-existing properties. Keep it and you must give up something else. This is the central trade in the whole subject.

See also: Bell's theorem, entanglement

entanglement

A joint state of two systems that cannot be written as one state for each.

The pair has a definite state while neither member does. Measuring one instantly fixes what the other will give — but you cannot use it to send a message, because each side alone sees nothing but noise until the results are compared.

See also: Bell's theorem, locality

Bell's theorem

No theory that is both local and assigns pre-existing values can reproduce quantum predictions.

John Bell showed in 1964 that such theories obey an inequality that quantum mechanics violates. Experiments — Aspect, then loophole-free tests in 2015, Nobel Prize in 2022 — come down on quantum mechanics' side. It is the closest thing here to a settled result, and it is what forces every interpretation to give something up.

See also: locality, hidden variables, superdeterminism

contextuality

An outcome can depend on what else you chose to measure alongside it.

The Kochen–Specker theorem (1967) shows you cannot consistently assign definite values to all observables at once, independently of context. Bell rules out locality plus definite values; this rules out context-independence as well, even setting distance aside.

See also: Bell's theorem, hidden variables

the PBR theorem

Under modest assumptions, the wavefunction cannot be merely information about a deeper real state.

Pusey, Barrett and Rudolph (2012) showed that if systems have real underlying states and independently prepared systems are independent, then ψ must be ontic. Escaping it means denying one of those assumptions — which is exactly what QBism does, by denying there is an underlying state to be ignorant of.

See also: ontic, epistemic, QBism

no-go theorem

A proof that a whole class of theories cannot reproduce quantum mechanics.

Bell, Kochen–Specker, PBR and Frauchiger–Renner are the big ones. Together they have closed every exit marked "common sense." This is why no interpretation is sane: sanity was ruled out, and all that remains is choosing which strangeness you prefer.

See also: Bell's theorem, contextuality, the PBR theorem, Wigner's friend

Wigner's friend

A thought experiment where an observer is themselves in superposition, as seen by someone outside.

The friend, inside a sealed lab, sees a definite outcome. Wigner, outside, describes the whole lab — friend included — as superposed. Frauchiger and Renner sharpened this in 2018 into an outright contradiction: you cannot keep universal unitarity, single outcomes and observer agreement all at once.

See also: the measurement problem, no-go theorem

pilot wave

Particles always have definite positions, guided by a real physical wave.

De Broglie proposed it in 1927, Bohm rediscovered it in 1952. It is deterministic, it has no measurement problem, and the double slit becomes an ordinary mechanism. The price is explicit nonlocality — the guiding equation depends instantly on the whole configuration.

See also: hidden variables, locality

QBism

Quantum states are an agent's personal degrees of belief, not descriptions of the world.

Short for Quantum Bayesianism. A wavefunction is your betting position; collapse is you updating on experience. It dissolves the measurement problem at the cost of denying that quantum mechanics describes anything observer-independent at all.

See also: epistemic, the PBR theorem

spontaneous collapse

Collapse is a real physical process that happens at random, all by itself.

Ghirardi, Rimini and Weber (1986) added a tiny random collapse term to the dynamics. One particle collapses about once every hundred million years; a cat, containing about 10²⁷ of them, collapses immediately. Uniquely among these, it is a different theory from quantum mechanics and experiments are closing in on it.

See also: collapse, unitary

superdeterminism

Denying that experimenters' choices are independent of the system being measured.

Bell's theorem quietly assumes the settings you choose are uncorrelated with the hidden variables. Drop that assumption and locality survives. The cost is that the correlation must have been arranged at the beginning of the universe, and that no experiment can ever be a fair test of anything.

See also: Bell's theorem, hidden variables