Multiverse

← Encyclopedia

“Multiverse” entered the language as a complaint. William James coined it in an 1895 address on whether life is worth living, and meant by it not a cosmology but a verdict: “Visible nature is all plasticity and indifference, — a moral multiverse, as one might call it, and not a moral universe.” Nature taken alone, he argued, underwrites no single order of value. The cosmological sense is a twentieth-century repurposing — scientific currency came only in the 1990s, after decades of fiction had made the idea familiar — and the migration is part of the story. A term coined for the world’s failure to add up was adopted by physics to name the most contested explanation of why the world adds up so suspiciously well.

The modern subject’s first discipline is to see that “the multiverse” is not one hypothesis but a family of logically independent ones; most confusion comes from running them together. Max Tegmark’s 2003 taxonomy, the standard sorting, distinguishes four levels. Level I is simply more of our own universe: if space is infinite and matter runs through all its configurations, every arrangement recurs — including, by Tegmark’s estimate, an identical copy of any given person about 10^(10^29) meters away. Same laws, same constants, different initial conditions — the one level Tegmark calls uncontroversial. Level II is the bubble universes of eternal inflation, in which the constants, the particle content, even the number of dimensions may come out differently from region to region. Level III is the many-worlds of quantum mechanics — a different kind of object altogether, taken up below. Level IV, Tegmark’s own “ultimate ensemble,” is the collection of all mathematical structures, each with different fundamental laws. The levels stand or fall separately.

Tegmark’s paper carries a precise characterization of Level III: the branches of the universal quantum state, he writes, “add nothing qualitatively new, which is ironic given that this level has historically been the most controversial.” The irony has not faded. His taxonomy simultaneously brought Everett’s interpretation into the multiverse conversation and made clear why it does not quite belong there — a theoretical move as important as any of the four levels themselves.

The bubbles have a definite pedigree. Alan Guth introduced inflation — a conjectured burst of exponential expansion in the universe’s first instants — in 1981, to explain the cosmos’s flatness and uniformity. In 1983 Paul Steinhardt showed that in the improved models inflation need not end everywhere — it stops in local patches while continuing outside them — and Alexander Vilenkin showed the behavior to be generic. Andrei Linde’s 1986 model gave the now-standard picture: inflation, once begun, never ends everywhere, endlessly budding pocket universes of which ours would be one. Guth stated the unsettling arithmetic himself: in such a cosmos, “anything that can happen will happen; in fact, it will happen an infinite number of times.” Eternal inflation is not, however, an inevitable consequence of inflation — not every inflationary potential behaves this way, arguably including those observation favors. And Steinhardt, who built the first working example, became one of its sharpest critics: a theory in which every outcome occurs somewhere, he argues, has stopped predicting anything.

Steinhardt also championed an alternative to the inflationary picture: the cyclic or ekpyrotic scenario, developed with Neil Turok from brane-cosmology. In that framework, the Big Bang is a collision between two membrane-surfaces in a higher-dimensional space, and the cycle repeats. The picture generates no ensemble of bubbles under different constants; it competes with eternal inflation for the same explanandum — a flat, uniform cosmos — without invoking multiple causally disconnected regions. Whether it succeeds in its own terms, and on what evidence, is a separate question; it stands here as a reminder that the inflating multiverse is not the only post-standard-model cosmology in play.

String theory supplied the variety. In 2000 Raphael Bousso and Joseph Polchinski showed that the theory’s quantized fluxes generate a dense spectrum of possible vacuum states — dense enough to include the absurdly small cosmological constant actually observed — and the stabilization mechanism published by Kachru, Kallosh, Linde, and Trivedi in 2003 made such vacua concrete, with quoted estimates running to 10^500. Leonard Susskind named the profusion the string landscape — a metaphor borrowed from evolutionary biology — and drew the moral plainly: such diversity, he wrote, “gives credence to the Anthropic Principle,” whether physicists like it or not. The combination is what the contemporary argument is mostly about — eternal inflation the engine, the landscape the menu: bubbles without end, each realizing a different vacuum, under different effective laws. The landscape’s physics is covered in the string theory entry; the multiverse argument here is about what the combination does and does not explain.

Hugh Everett’s many-worlds is routinely annexed to this family, and it does not belong to it. Everett’s 1957 “relative state” formulation answered a problem internal to quantum mechanics — the measurement problem — by proposing that the universal quantum state simply evolves and never collapses; the “worlds” of the version Bryce DeWitt popularized in 1970 are branches of that single state, multiplying with every quantum event everywhere, not just in laboratories. They are not remote regions of space; they are not located anywhere at all. The many-worlds interpretation is exactly that — an interpretation of a theory already in hand — whereas the cosmological multiverses are claimed consequences of theories, inflation and string theory, whose own standing is still in dispute. Tegmark, filing Everett’s branches as Level III, notes the irony that they add “nothing qualitatively new,” though historically they drew the most fire. (A minority proposal — argued by Bousso and Susskind, and independently by Nomura — holds that the two are at bottom the same structure.)

What work would the other worlds do? The serious answer is anthropic: if the constants vary across an ensemble, observers necessarily find themselves in the rare regions that permit observers, and apparent fine-tuning needs no further explanation. The reasoning has one famous success. In 1987 Steven Weinberg argued that if the cosmological constant varies across a multiverse, a typical observer should measure a small positive value: observers exist only where it is small enough for galaxies to form, and typicality puts the expectation inside that narrow window, not at zero. The 1998 supernova surveys then found cosmic acceleration driven by just such a value; the refinement with Martel and Shapiro had put numbers on the claim before the data came in. It is the best card the multiverse holds. The machinery underneath is less reassuring: it needs a principle of indifference over observers — typical with respect to which reference class is never clear — and philosophers still dispute whether the inference from fine-tuning to many worlds commits an “inverse gambler’s fallacy.” The Stanford Encyclopedia of Philosophy’s entry on fine-tuning quotes White (2000) articulating the objection directly: it is a mistake to suppose “that the existence of many other universes makes it more likely that this one — the only one that we have observed — will be life-permitting.” Fine-tuning’s full analysis, including its relation to design arguments, belongs to the fine-tuned universe entry; what the multiverse requires of it is the measure problem. Above all stands the measure problem: extracting predictions from an eternally inflating multiverse means assigning probabilities over an infinite ensemble that lacks the structure to define them, and different ways of taming the infinities give different answers. Tegmark has called it “the greatest crisis in physics today.” Until it is solved, the objection runs, a theory in which everything happens somewhere is compatible with any observation whatever.

The dispute over whether any of this is science broke into the open in December 2014, when George Ellis and Joe Silk published a comment in Nature under the title “Defend the integrity of physics”: exempting speculative theories of the universe from experimental verification, they argued, undermines science itself; the targets were multiverse cosmology, string theory, and “non-empirical” confirmation. Sean Carroll answered for the defense: multiverse models are weighed the way all models are weighed — by abduction, Bayesian inference, and empirical success — and are “utterly conventionally scientific,” cosmology being impossible without considering what may lie beyond the horizon. The argument has one empirical handle. In eternal inflation our bubble must occasionally collide with neighbors, and a collision could leave a disk-shaped scar on the cosmic microwave background. The first systematic search, by Feeney, Johnson, Mortlock, and Peiris in 2011 using the WMAP seven-year data, returned a clear conclusion: their model-selection results “do not warrant augmenting LCDM with bubble collisions.” Analyses of the sharper Planck maps found no significant trace either. Even this handle is loose: the scenario does not generically predict observable collisions — an early imprint would be inflated away — so a null result does not falsify it. Which is, the critics observe, precisely the problem.

The idea is old, and the temptation is to make it one idea, continuously held. Each historical plurality of worlds was in fact its own claim, with its own logic. For Leucippus and Democritus in the fifth century BCE, infinite atoms moving in an infinite void congealed by chance into innumerable kosmoi — world-systems owing nothing to purpose or design, forming and perishing forever; Epicurus made the doctrine canonical, and Lucretius gave it Latin verse. Anaxarchus is said to have told Alexander the Great the worlds were infinite, and Alexander wept, “for he had not yet conquered even one.” Plato and Aristotle held the world to be unique, and their verdict governed for a millennium. When plural worlds resurfaced among the 219 theses condemned by Bishop Étienne Tempier at Paris in 1277, the question was theological and modal — whether God could create more worlds than this one, not whether he had; Pierre Duhem’s old claim that this decree birthed modern science is no longer endorsed as stated. Giordano Bruno’s innumerable worlds, set out in 1584, were other suns with other earths within one infinite universe — nearer, in modern terms, to exoplanets than to disconnected spacetimes. That the worlds figured among the charges for which he burned in Rome in 1600 rests on scholarly reconstruction — the Inquisition’s final list of eight propositions is lost, and the weight the cosmology carried among them cannot be settled. And Fontenelle’s Entretiens sur la pluralité des mondes (1686), evening conversations with a marquise that became arguably the first classic of popular science, was speculation about inhabited planets — not a multiverse at all. Four episodes, four claims. None of them is the modern proposal of causally disconnected regions under different laws.

The plurality intuition appears in other registers that the site’s coverage touches separately. Hindu cosmology — addressed in the Hinduism: Yugas entry — conceives of vast nested cycles of creation and dissolution, manvantaras within kalpas, operating across scales that dwarf any single cosmic age; the parallelism with an inflating multiverse’s ceaseless production is structural, not genealogical, and the two accounts differ sharply in purpose and ground. Kabbalistic traditions describe worlds and olamot that precede and surround the present one, a cosmogonic plurality tied to the dynamics of emanation and contraction treated in the Kabbalah entries. In both cases, the comparison is a pointer to those entries, not a derivation: the traditions were not attempting what modern cosmology attempts, and the resemblance, though real, travels no theoretical content between them.

The modern esoteric and popular reception is a different matter. By the early twenty-first century, the multiverse had become a cultural idiom — parallel selves making different choices in branching timelines, accessible through consciousness alteration or psychic sensitivity. The New Age literature drew mainly on Everett’s branches, filtered through decades of science fiction, and treated the proliferating worlds as personally navigable rather than epistemically inaccessible. The physics community’s own publicity had something to do with this: Everett’s “worlds” carry an unavoidable anthropomorphism that bubble universes with different vacuum energies do not, and the casual annexation of Levels III and II into a single “multiverse” concept in popular accounts blurred the precise logical distinctions Tegmark’s taxonomy was designed to enforce. The reception is worth noting not because it converges with the science but because it illustrates how a term migrates: James’s “moral multiverse,” Everett’s quantum branches, and the landscape’s 10^500 vacua are three distinct objects that popular usage assembles into one, repeating at the cultural level the very confusion the physics must resist.

These episodes do not descend from one another; they share a question rather than a doctrine: whether what can be seen is all there is, and what kind of reasoning could ever say. The atomists asked it of space; Tempier’s Paris asked it of God’s power; the present argument asks it of science. That is the form in which the plurality of worlds has returned: no longer whether the worlds exist, but whether claims about them can count as knowledge. The question was already old when it traveled, among reconstructed charges, to the Campo de’ Fiori. Four centuries later it is on trial again; only the court has changed, and the charge is no longer heresy but unfalsifiability.

Scholarship

The foundational taxonomy is Max Tegmark, “Parallel Universes” (arXiv:astro-ph/0302131, 2003; published in Science and Ultimate Reality, Cambridge University Press, 2003). Tegmark’s later book Our Mathematical Universe (Knopf, 2014) extends the Level IV argument. Bernard Carr’s edited volume Universe or Multiverse? (Cambridge University Press, 2007) collects contributions from Barrow, Carter, Davies, Rees, Susskind, Weinberg, and others, covering the full range of positions and remaining the standard reference collection. Brian Greene’s The Hidden Reality: Parallel Universes and the Deep Laws of the Cosmos (Knopf, 2011) surveys all nine (his count) multiverse varieties for a general audience with unusual precision. The Ellis–Silk exchange — George Ellis and Joe Silk, “Defend the integrity of physics,” Nature 516 (2014): 321–323 — remains the sharpest statement of the falsifiability objection; Carroll’s response came in “Beyond Falsifiability” (2018, arXiv:1801.05016) and in posts at his site. For fine-tuning specifically, the Stanford Encyclopedia of Philosophy’s entry on fine-tuning (Friederich, first published 2017, revised 2026) is the current open-access treatment, covering Weinberg’s prediction, the inverse gambler’s fallacy, and the anthropic principle with full citation. The first systematic CMB bubble-collision search is Feeney, Johnson, Mortlock, and Peiris, “First Observational Tests of Eternal Inflation” (arXiv:1012.3667, 2011), whose negative result set the standard for all subsequent analyses.

→ Related: String Theory · Giordano Bruno · Fine Tuned Universe · Many Worlds Interpretation · Block Universe · Eternal Recurrence · Hinduism Yugas