The many-worlds interpretation tackles one of physics’ hardest puzzles — why quantum objects seem to stop behaving like waves when they are measured — by arguing that they never stop at all: every possible outcome happens, but we only experience the branch we end up in.

The many-worlds interpretation begins with a refusal to add a special rule for measurement. That is its appeal, and also the source of its strangeness.

In ordinary quantum mechanics, an electron, atom or photon can be described by a wavefunction that contains several possible outcomes at once. When a measurement is made, we see one result. The textbook language says the wavefunction has collapsed. Many-worlds asks a sharper question: what if collapse is not a physical event at all?

On that view, the wavefunction never stops evolving as a wave. Every outcome allowed by the quantum state continues, but the observer becomes correlated with only one outcome in one branch. The wave has not vanished. We have become part of it.

The measurement problem is the starting point

The problem many-worlds is trying to solve is not that quantum mechanics fails in the laboratory. It is that the usual story uses two different rules. When no one is measuring, quantum systems evolve smoothly and deterministically according to the Schrödinger equation. When a measurement happens, the standard account says the system jumps into one definite result, with probabilities given by the Born rule.

That jump is the awkward part. What counts as a measurement? Does a detector count? Does a human observer have to read the detector? If the detector itself is made of quantum particles, why should it obey a different rule from the particle being measured?

The Stanford Encyclopedia of Philosophy entry on Everettian quantum mechanics describes this as the quantum measurement problem: if observers and measuring devices are physical systems like any other, they should also evolve according to the continuous quantum dynamics. A special collapse rule sits uneasily beside that.

Everett’s answer was to remove collapse

Hugh Everett III put the core move into print in 1957 in Reviews of Modern Physics. His paper, “Relative State” Formulation of Quantum Mechanics, proposed dropping the collapse postulate and treating observers as physical systems inside quantum mechanics, not as external agents standing outside it.

That sounds technical, but the consequence is direct. Suppose a quantum system is in a superposition of two possible outcomes. A measuring device interacts with it. An observer reads the device. If the quantum state never collapses, the combined state does not contain one outcome only. It contains a branch in which the device records one result and the observer sees that result, and another branch in which the device records the other result and the observer sees that one.

Each observer-copy has a normal-looking experience. None sees a blurred mixture of outcomes. From inside a branch, measurement looks like the selection of a single result. From the outside, if such an outside view were possible, the full wavefunction still contains all the branches.

That is why the interpretation is often called many-worlds, though many physicists prefer Everettian language because the “worlds” are not extra planets or parallel stage sets. They are effectively separated components of one quantum state.

Decoherence explains why branches stop talking

The modern version of many-worlds leans heavily on decoherence. Decoherence is not a speculative add-on invented for this interpretation. It is a well-developed account of how quantum systems lose observable interference when they become entangled with their environment.

Wojciech Zurek’s 2003 Reviews of Modern Physics article, “Decoherence, einselection, and the quantum origins of the classical”, is one of the standard references. The basic idea is that a system does not interact only with a measuring device. It also interacts with air molecules, photons, vibrations and surrounding matter. Those interactions spread phase information into the environment.

Once that happens, interference between the different outcomes becomes effectively inaccessible. The branches behave as though they have separated. Nothing dramatic needs to snap. The ordinary-looking world emerges because the alternatives can no longer recombine in any practical way.

This is where many-worlds differs from the cartoon version. The interpretation does not say a new universe pops into existence whenever someone looks at something. It says the universal quantum state keeps evolving, while decoherence makes different outcome histories cease to interfere for observers inside them.

The price is a larger reality

Many-worlds buys mathematical simplicity at the cost of ontology. It keeps the wavefunction and the Schrödinger equation, and it removes collapse. But it then asks readers to accept that all the other outcomes are real too, even though no observer in one branch can communicate with the observer in another.

That is why the interpretation remains divisive. Supporters see it as taking quantum mechanics seriously without adding a vague measurement boundary. Critics argue that it shifts the puzzle rather than solving it: if all outcomes happen, what exactly makes one branch feel probabilistic, and how should the usual Born-rule probabilities be understood?

There are proposed answers, including decision-theoretic arguments and branch-weight accounts, but they are still debated. Many-worlds is not a laboratory result in the same sense as interference, entanglement or Bell-test violations. It is an interpretation of a formalism that already predicts those results.

What it does and does not claim

The interpretation does not say that every imagined possibility occurs. It says that outcomes with non-zero amplitude in the quantum state continue in the full wavefunction. Nor does it say that people can move between branches, send messages to them, or use them as a science-fiction escape route.

It also does not remove the strangeness of quantum mechanics. It relocates it. Instead of an unexplained collapse at measurement, many-worlds gives a continuously evolving wavefunction in which observers find themselves in definite branches. The puzzle moves from “why did the wave collapse?” to “why do we experience only one branch, and how do probabilities arise when all outcomes occur?”

That is the cleanest way to understand its appeal. Many-worlds tackles the measurement problem by refusing to treat measurement as a special physical process. It says the wave never stopped. We only stopped being able to see the other parts of it.