Interpretations: Copenhagen, many-worlds, Bohm, QBism, collapse models
Under 1,400 metres of rock at Gran Sasso in central Italy, a block of ultra-pure germanium sits in a shielded vault, waiting for faint flashes of X-rays that ordinary physics says should not be there. If quantum superpositions collapse on their own, as some theories propose, the jostling this causes to electrons would make them glow, very slightly. In 2021 the detector reported the glow absent at the level the simplest such theory requires. Every other interpretation of quantum mechanics predicts the same numbers as every other. Only this family can be caught out by experiment.
On the plate: the same double-slit experiment, told several ways.
The problem, stated precisely
Tim Maudlin put the measurement problem as three claims that cannot all be true. (A) The wavefunction is a complete description of a system. (B) It always evolves by the linear Schrödinger equation. (C) Measurements have single, definite outcomes. Decoherence (Chapter 52) shows that (A) and (B) produce an entangled superposition of a detector reading “left” and one reading “right”, which contradicts (C).
So every interpretation must give up one. Choose a view: , , , . The plate marks which claim it rejects. Everything in this chapter is Debated unless marked otherwise.
Copenhagen and QBism: the wavefunction is not the world
The Copenhagen view, associated with Bohr and Heisenberg, treats the wavefunction as a tool for predicting what measurements will show, with the measuring apparatus described classically. Its strength is that it is what working physicists actually do, and it never gives a wrong prediction. Its weakness is the line between quantum system and classical apparatus, which it never defines.
QBism, developed by Christopher Fuchs, Rüdiger Schack and David Mermin, goes further: a wavefunction is an agent’s personal degrees of belief about what they will experience, and collapse is just the agent updating them. Its strength: no spooky collapse, no non-locality, since nothing physical changes. Its objection: it says little about what the world is like when no one is betting on it.
Many worlds: every outcome happens
Hugh Everett proposed in 1957 that (C) is false. The Schrödinger equation is never interrupted; when you measure, you become entangled with the particle, and there are afterwards branches in which you saw each result, which decoherence keeps from ever interfering. The plate shows the branches multiplying at each detection.
Its strength is economy: it is just quantum mechanics taken literally, with no extra rule. Its hardest problem is probability. If every outcome occurs, what does it mean that one has probability 0.9 and the other 0.1? Arguments by David Deutsch, David Wallace and others derive the Born rule from rational decision-making in a branching world; critics doubt the derivations are free of circularity.
Bohm: particles with real paths
David Bohm in 1952, rediscovering an idea of de Broglie’s, gave up (A). There is a wavefunction and also a particle with a definite position, moved by the wave with velocity . The plate integrates that rule for a double slit: the paths never cross, yet they bunch into bright fringes and avoid the dark ones. If starting positions are spread as , every prediction matches quantum mechanics exactly.
Strengths: definite outcomes and no collapse. Objections: the motion of each particle depends instantaneously on distant ones, as Bell’s theorem requires, which sits uneasily with relativity; and the trajectories are invisible in principle. In 2011 Sacha Kocsis and colleagues reconstructed average photon paths in a double slit with weak measurements and found curves that look Bohmian; critics noted that such averages need not be anyone’s actual path.
Collapse models: a testable change
Giancarlo Ghirardi, Alberto Rimini and Tullio Weber gave up (B) in 1986. Every particle’s wavefunction occasionally, at random, localizes to about 100 nm: about once every seconds, about 300 million years, for a single nucleon. In a pointer of nucleons, one of them collapses about every s, dragging the rest with it. Superpositions of atoms survive; superpositions of pointers do not.
Because the dynamics differ, the predictions differ. Random localizations heat matter and make charged particles radiate. The plate maps the model’s two parameters, rate and size. Experiments rule out regions where effects would already have been seen, and the requirement that collapse actually work for visible objects rules out regions below. A related proposal by Lajos Diósi and Roger Penrose, linking collapse to gravity, was tested in that germanium vault: Sandro Donadi and colleagues found too little radiation for its parameter-free version. Its variants, and the GRW proposal itself, are still open.
Theorems that constrain every view
Some results bind all interpretations. Established, as mathematics. The Kochen–Specker theorem (1967): measurement outcomes cannot all be pre-assigned independently of which other quantities are measured alongside. The PBR theorem of Matthew Pusey, Jonathan Barrett and Terry Rudolph (2012): if independent systems can be prepared independently, the wavefunction cannot be mere information about some deeper state; it must correspond to something real. QBism escapes by denying the deeper state exists.
Daniela Frauchiger and Renato Renner (2018) built a version of Wigner’s friend, in which observers are themselves described quantum mechanically, and showed that no theory can satisfy three natural assumptions at once: quantum mechanics applies to observers, different agents’ reasoning is consistent, and each measurement has one outcome. Massimiliano Proietti and colleagues tested a related Wigner’s friend inequality with photons standing in for friends in 2019 and found it violated. What each interpretation should conclude is argued over.
What physicists actually think
In 2025 the journal Nature surveyed more than a thousand physicists. About 36% chose Copenhagen, about 15% many-worlds, with smaller groups for Bohm, QBism, collapse and others, and many chose none. Only 24% were confident their choice was right. After a hundred years, the theory that predicts more accurately than any other in science still has no agreed account of what it describes.
That is not a failure of physics; it is a precise, open question, with one testable branch, and you now know the arguments well enough to hold your own in it.
What you now have
Maudlin’s trilemma, the strongest case and the strongest objection for Copenhagen, QBism, many-worlds, Bohm and collapse models, the experimental tests of collapse, the Kochen–Specker, PBR and Frauchiger–Renner theorems, and where the profession stands. This ends Part One. Part Two joins quantum mechanics to relativity, and the result is quantum field theory.
· · ·
Three questions to carry with you.
And one to work out. In GRW, each nucleon collapses at rate per second. For a pointer with nucleons, the total rate is per second. What is the mean time to the first collapse, in microseconds?
Sources: T. Maudlin, Topoi 14, 7 (1995). H. Everett, Rev. Mod. Phys. 29, 454 (1957). D. Bohm, Phys. Rev. 85, 166 (1952). G. C. Ghirardi, A. Rimini and T. Weber, Phys. Rev. D 34, 470 (1986). C. A. Fuchs, N. D. Mermin and R. Schack, Am. J. Phys. 82, 749 (2014). S. Kocsis et al., Science 332, 1170 (2011). S. Kochen and E. P. Specker, J. Math. Mech. 17, 59 (1967). M. F. Pusey, J. Barrett and T. Rudolph, Nat. Phys. 8, 475 (2012). D. Frauchiger and R. Renner, Nat. Commun. 9, 3711 (2018). M. Proietti et al., Sci. Adv. 5, eaaw9832 (2019). S. Donadi et al., Nat. Phys. 17, 74 (2021). Nature survey of physicists on quantum interpretations (2025). Checked by code: Bohmian trajectories integrated from the guidance equation for two spreading Gaussian slits (no crossings; fringe density matches |ψ|²); GRW numbers.