Many-Worlds Interpretation

← Encyclopedia

Standard quantum mechanics was built with a seam down its middle. The Schrödinger equation evolves a system smoothly and deterministically through its superpositions, and then a second rule — invoked at measurement — has the superposition jump to a single outcome, chosen at random with probabilities fixed by the squared amplitude. Two dynamics, one applied between observations and the other during them, with nothing in the theory to say what counts as an observation or when the switch should be thrown. This is the measurement problem, and most of the work in quantum foundations is an attempt to dispose of it. The full technical anatomy of that problem — what decoherence actually explains and what it leaves untouched — belongs to the companion entry on the quantum measurement problem. Hugh Everett III’s proposal was the most economical: delete the second rule.

Everett completed the work at Princeton in 1957 under John Archibald Wheeler, and published it that July as the “‘Relative State’ Formulation of Quantum Mechanics,” in Reviews of Modern Physics (29: 454–462). His move was subtraction. Keep only the unitary Schrödinger evolution; throw out the collapse postulate entirely. What remains is a single universal wave function that never collapses and evolves forever deterministically — “pure wave mechanics,” in his phrase. The price is that an observer is now just another physical system inside the theory, subject to the same dynamics as anything else. When such an observer measures a system in superposition, the two become entangled, and the joint state evolves into a superposition of records: observer-saw-A with system-in-state-A, plus observer-saw-B with system-in-state-B, and so on, each term as real as the next. There is no single result. Each branch is a relative state — relative to the term in which the pointer reads A, the observer holds a definite memory of having seen A. The trick is then to recover the familiar appearance of one random outcome from inside such a branch. Everett’s stated aim was to “deduce the probabilistic assertions” of the collapse rule “as subjective appearances to such observers, thus placing the theory in correspondence with experience.” Collapse is not a physical event but the view from within a branch.

Everett’s doctoral path was harder than the tidy narrative suggests. The longer draft of the thesis, titled “Wave Mechanics Without Probability,” was submitted to Wheeler in January 1956. Wheeler asked for significant revisions before Everett could defend — partly out of concern that Niels Bohr, whose Copenhagen interpretation Everett was implicitly attacking, would reject the work. Everett left academia in the spring of 1956, before the formal defense, to take up defense analysis work in Washington; he never returned to academic physics. He spent much of his career at the Pentagon and later at Lambda Corporation, working on systems analysis and nuclear targeting. Peter Byrne’s biography, The Many Worlds of Hugh Everett III (Oxford University Press, 2010), draws on Everett’s papers and family interviews to reconstruct both the physics and the personal cost: the neglect of his 1957 paper during his lifetime, the alcoholism that accompanied his professional retreat from physics, and his death in 1982, at fifty-one. According to Byrne, Everett “never wavered in his belief” in the many-worlds reading, even as mainstream physics largely ignored it.

The famous name is not his. Everett never wrote “many worlds,” never spoke of splitting universes, and the popular image of branching parallel realities belongs to Bryce DeWitt, who introduced it to a wide audience in a 1970 Physics Today article, “Quantum Mechanics and Reality.” DeWitt credited the view to Everett, Wheeler, and DeWitt’s own student R. Neill Graham — the “EWG interpretation” — and made the metaphysics of literally splitting worlds its centerpiece: “This universe is constantly splitting into a stupendous number of branches… every quantum transition taking place on every star, in every galaxy, in every remote corner of the universe is splitting our local world on earth into myriads of copies of itself.” He did not hide his unease, recording the “shock I experienced on first encountering this multiworld concept.” There is a sharp irony underneath the popularization. In his personal copy of the description of these splitting, branching worlds, Everett wrote a single word in the margin: “bullshit.” The man whose formulation founds the many-worlds interpretation rejected the picture that made it famous. The careful name for the origin is relative-state, or Everettian, and “many worlds” properly denotes the DeWitt-and-after reading — one member of a later family that also includes many-minds, many-histories, and bare theories.

DeWitt’s tidy image of a discrete split at each measurement raises an immediate technical question, and it is the first of two that proponents themselves take to be genuinely open. A superposition can be rewritten in infinitely many bases; the same combined state of system and observer can be re-expanded as a superposition of entirely different “branches.” So what fixes the basis along which the world divides — why does it branch into definite-looking pointer readings and definite cats rather than into bizarre superposed alternatives? Without an answer, “all outcomes occur” says nothing, because it does not say which outcomes. This is the preferred-basis problem. Everett did not solve it, and for his limited purpose did not need to: he required only some decomposition exhibiting a record that matches experience. The modern reply is decoherence — the rapid, effectively irreversible loss of interference between the components of a superposition once a system couples to a large, uncontrolled environment. Decoherence is established, interpretation-neutral physics; every interpretation invokes it. In the Everettian setting it does double work. It dynamically selects a quasi-classical basis, because only certain stable “pointer states” survive constant environmental monitoring, and it renders the resulting branches effectively non-interfering, so each behaves like an autonomous near-classical world. On the mature version of the view, defended at length by David Wallace in The Emergent Multiverse (Oxford University Press, 2012), worlds are not fundamental furniture but emergent, approximate substructures of the one quantum state — “mutually dynamically isolated structures” that are robustly quasi-classical. A consequence follows that rarely survives popularization: there is no exact count of worlds. How many branches exist depends on the level of description and on how isolated one demands a world be. “Infinitely many universes, one per outcome” is a slogan, not the careful claim.

The second open problem is deeper, and it is the strongest standing objection to the whole program. If collapse is gone and every outcome genuinely happens, in what sense is any outcome probable? The Born rule assigns outcome A a probability equal to the squared amplitude. But if A and B both occur with certainty, each in its own branch, where does a number strictly between zero and one come from, and why that particular number? Lev Vaidman puts the puzzle at its sharpest: it is “senseless to ask: ‘What is the probability that I will get A instead of B?’ because I will correspond to both” successors. Everett’s own answer was a typicality measure: he imposed formal constraints until a unique measure over branches was forced — the squared-amplitude measure — and showed that branches typical by that measure exhibit the standard quantum statistics. But a typicality result is not yet a probability, and securing even that much requires extra assumptions tying the measure to what a rational agent should expect. The most-discussed attempt to close the gap is the decision-theoretic derivation of David Deutsch and David Wallace, which argues that the Born rule is not an added postulate but a theorem of rational choice. Deutsch’s 1999 paper, “Quantum Theory of Probability and Decisions” (Proceedings of the Royal Society A), stated the core claim directly: the probabilistic predictions follow from “the remaining, non-probabilistic, axioms of quantum theory, together with the non-probabilistic part of classical decision theory.” An agent in a branching world who satisfies a set of rationality and symmetry constraints is, on this argument, provably required to weight her bets by the squared amplitudes. Wallace’s subsequent papers (2003, 2007) and book formalized and defended the argument against early objections, and the resulting Deutsch–Wallace program remains the most technically developed approach to the probability problem within the Everettian framework. It is an elegant argument and a contested one. Critics including Adrian Kent, David Albert, and Huw Price charge it with circularity — that probability is quietly imported through the “rationality” axioms rather than wrung out of pure wave mechanics. Vaidman defends a different route, a “measure of existence” for each world paired with a postulate setting subjective probability proportional to it. A third approach, developed by Simon Saunders and collaborators in the volume Many Worlds? Everett, Quantum Theory, and Reality (Oxford University Press, 2010), pursues a branch-counting strategy refined by symmetry arguments. These are competing strategies, not a settled consensus, and whether any can deliver genuine probability from pure wave mechanics is the central unresolved question in Everettian foundations.

The popular imagination has fastened onto a darker corollary that deserves to be reported precisely, because it is so often distorted. The quantum-suicide thought experiment — a “quantum Russian roulette,” associated with Max Tegmark — pairs a quantum measurement to a lethal trigger. The reasoning runs: under many-worlds there is always a branch in which the experimenter survives, and conditional on remaining a conscious observer she experiences only surviving branches, suggesting she never witnesses her own death. This is a thought experiment, not a prediction and emphatically not advice. The physics literature treats it as a decision-theory puzzle, and its most careful treatment argues against ever acting on it. Vaidman’s Behavior Principle holds that one should care about all of one’s successor worlds in proportion to their measure of existence; because the death branches carry as much measure as the survival branch, a rational Everettian should treat quantum roulette no more favorably than the ordinary kind. His own statement is exact: “if I am terribly afraid of dying, I should choose classical roulette which gives me some chance not to die.” The argument depends, besides, on a particular reading of personal identity across branches and on the unresolved measure problem above, and it is untestable for anyone but the would-be victim, who could never communicate the alleged first-person evidence. It belongs to the speculative margin, sharply apart from any spiritual immortality of the soul.

None of this is what physics has been shown to be true. Many-worlds is one interpretation among several — alongside Copenhagen, the de Broglie–Bohm pilot wave, objective-collapse models, relational quantum mechanics, and others — and on present evidence they are empirically equivalent: no experiment to date distinguishes them. It is best described as the leading no-collapse rival to Copenhagen, the choice of a substantial minority. The polling is weak evidence, small and self-selected, but consistent in its order of magnitude. A 2011 conference survey of thirty-three participants put Everettian readings near eighteen percent against Copenhagen’s forty-two; a broader single-choice survey in 2016 found roughly fifteen percent for many-worlds against about forty-eight for Copenhagen; a 2025 centenary survey of more than eleven hundred physicists returned fifteen percent and thirty-six, with only a quarter of all respondents confident their favored interpretation was even correct. A respectable minority, stable to rising — not the majority view, and not a fringe one.

Deutsch has also argued that the Everettian structure of quantum mechanics explains why quantum computation is powerful: speed-ups arise because computation proceeds across exponentially many branches of the wave function simultaneously. This reading is contested — no interpretation is required by the mathematics — but it has shaped a strand of subsequent thinking about what quantum algorithms actually do. The technical account belongs to the companion entry on quantum computing.

The reception outside physics has followed a predictable trajectory. Jorge Luis Borges’s “The Garden of Forking Paths” (1941) explored branching time as a narrative device a decade and a half before Everett wrote his thesis, and that literary intuition lodged the idea of parallel selves in the culture before any scientific popularization began. The 1970s spread of DeWitt’s “splitting worlds” language gave the intuition a scientific-sounding warrant; by the 1990s, quantum branching appeared routinely in science fiction and parallel-timeline drama. In New Age and popular metaphysical writing, the move has typically been to read Everett’s branches as independently accessible parallel lives: the immortal self that persists across all timelines, the quantum basis for past-life memory, the physics of manifesting desired realities by “collapsing into” the preferred branch. This circulates widely in literature that treats consciousness as ontologically primary. The physics does not support any of it. Branches in the Everettian framework are not accessible from one another; no information crosses from branch to branch after decoherence has rendered them non-interfering. The many-worlds interpretation is defined by the universality of the wave function and the absence of collapse — nothing in that framework licenses the claim that one can choose, perceive, or “shift to” alternate branches.

One confusion is worth heading off, because the shared vocabulary invites it. The many-worlds of quantum mechanics is not the cosmological multiverse. Its worlds are branches of a single wave function of our own universe, coexisting at the same place and time, generated by unitary quantum dynamics; the inflationary multiverse concerns other regions of space, pocket universes with possibly different physical constants, predicted by cosmology rather than by quantum measurement. They are logically independent ideas that happen to share a word, and this site keeps them apart — though a minority proposal — Raphael Bousso and Leonard Susskind, and independently Yasunori Nomura — holds the two to be at bottom the same structure. The full taxonomy of multiverse levels — from Tegmark’s Level I through Level IV — belongs to the companion entry on the multiverse. What recurs across both, and across the anthropic debates over the cosmological constant, is the measure problem in another dress: the difficulty of defining probabilities over an ensemble in which everything that can happen does. Everett’s economy was to remove a postulate; the cost was to inherit, in the question of where the odds come from, a debt the interpretation has not yet paid. A single equation, left to run, and a margin note that called the rest of it nonsense.

Scholarship and primary sources

The original paper is Hugh Everett III, “‘Relative State’ Formulation of Quantum Mechanics,” Reviews of Modern Physics 29 (1957): 454–462. The suppressed draft, “Wave Mechanics Without Probability” (1956), was published in Jeffrey A. Barrett and Peter Byrne (eds.), The Everett Interpretation of Quantum Mechanics: Collected Works 1955–1980 with Commentary (Princeton University Press, 2012). The standard biography is Peter Byrne, The Many Worlds of Hugh Everett III (Oxford University Press, 2010).

Modern defenses: Bryce S. DeWitt, “Quantum Mechanics and Reality,” Physics Today 23.9 (1970): 30–35. David Wallace, The Emergent Multiverse (Oxford University Press, 2012). Simon Saunders, Jonathan Barrett, Adrian Kent, and David Wallace (eds.), Many Worlds? Everett, Quantum Theory, and Reality (Oxford University Press, 2010). David Deutsch, “Quantum Theory of Probability and Decisions,” Proceedings of the Royal Society A 455 (1999): 3129–3137.

Open-access reference entries: Lev Vaidman, “Many-Worlds Interpretation of Quantum Mechanics” and “Everett’s Relative-State Formulation of Quantum Mechanics” (Stanford Encyclopedia of Philosophy). Survey data: Schlosshauer, Kofler, and Zeilinger, “A Snapshot of Foundational Attitudes Toward Quantum Mechanics,” Studies in History and Philosophy of Modern Physics 44 (2013): 222–230; Sivasundaram and Nielsen, “Surveying the Attitudes of Physicists Concerning Foundational Issues of Quantum Mechanics” (2016).

→ Related: Quantum Measurement Problem · Multiverse · Quantum Entanglement · String Theory · Anthropic Principle · Quantum Computing