Sean Carroll on Quantum Mechanics, Many Worlds, and the Problem of Structure

A solo episode in which Sean Carroll makes the case that the Everettian (many-worlds) formulation of quantum mechanics is the most parsimonious reading of the equations, then confronts the open questions that remain: how does a bare, austere wave function in Hilbert space give rise to classical space, definite measurement outcomes, and the probability rule that physicists use every day?

Key ideas

1. Copenhagen is not a scientific theory — it is an IOU. The survey-dominant interpretation defines measurement as the moment the wave function collapses, but never specifies what counts as a measurement. Bohr and Heisenberg’s philosophical discipline (denying the reality of the wave function entirely) is at least consistent; the casual Copenhagenism of most physicists — accepting the wave function as real while invoking collapse without defining it — is not.

2. Everett removes the collapse postulate, which is a virtue. The Schrödinger equation, applied without exception, predicts that when an observer measures a quantum superposition they themselves enter a superposition. Many-worlds accepts that conclusion and identifies each branch as a separate world. Fewer axioms is a scientific virtue; the apparent profligacy of ‘many worlds’ is not extra ontology but a consequence of the single wave function everyone already accepts.

3. The problem of structure: classical reality must be derived, not assumed. The deepest open question in Everettian mechanics is not the mere existence of branches but how branches — and the space, particles, and locality we observe — emerge from the structure of Hilbert space at all. Carroll calls this the ‘problem of structure’: space, position, and separability are not built into the formalism; they must be shown to be emergent. Even committed many-worlders often resist this, because it means the world we navigate is not fundamental.

4. The Born rule is a self-locating probability problem, not a frequency problem. In Everett, all outcomes happen, so the standard ‘how often does this outcome occur’ question dissolves. Carroll’s preferred approach reframes it: given that you are about to become multiple copies of yourself, one in each branch, how should you apportion credence across branches? He argues that branch-counting alone gives the wrong answer, and that the correct credences follow from a decision-theoretic argument — rational agents who care about their future selves in all branches should bet in proportion to the squared amplitudes of the wave function. This is an active area of research, not a solved problem.

5. Locality in space is a derived, not fundamental, property — and perhaps not exactly true. Carroll and collaborators argue that for most choices of Hamiltonian, there is no local way to carve Hilbert space into subsystems at all. The cases where local subdivision works — where quantum fields at one point only affect their immediate neighbours — are rare and special. This suggests that the locality of our observed universe is an emergent feature, and raises the question of whether locality could be only approximately true in ways that are experimentally testable.

Related

• Sean Carroll — host and sole speaker
• Many-Worlds Interpretation — the interpretation defended and examined
• Quantum Measurement Problem — the problem Carroll argues Copenhagen fails to solve
• Sean Carroll on General Relativity, Quantum Mechanics, Black Holes and Aliens — companion Lex Fridman episode with broader coverage of Carroll’s physics positions
• Free Will — Carroll draws a brief analogy: free will and classical space are both indispensable in practice while absent from the fundamental equations