Many-Worlds interpretation

In plain terms

Quantum mechanics describes a system by a state that changes smoothly and predictably in time, and the same rules allow a state to be a superposition, a combination of several possibilities held at once, such as a particle that's partly here and partly there. Yet a measurement only ever shows one result. The textbook handles this by adding a second rule, collapse, under which the state jumps to a single possibility at the moment of measurement, and Many-Worlds asks what follows if you refuse to add that rule.

Everett's answer was to take the smooth evolution as the whole story. After a measurement the apparatus and the observer are then themselves in superposition, with one component, called a branch, for each outcome, and Everett's move was to say that nothing is wrong with this. Every outcome happens, each in its own branch. The branches decohere, meaning that their interaction with the surroundings stops them from ever influencing one another again, so each branch contains an observer certain that their result was the only one. Nothing is added to the formalism, no collapse and no hidden variables, and that austerity is the position's appeal.

The cost surfaces at probability. If everything happens, you can ask what it can mean that one outcome was more probable than another. Counting branches gives the wrong answer, as a measurement with a seventy percent outcome and a thirty percent outcome produces two branches and not seven and three. The decision-theoretic repairs, which argue that a rational agent should bet as though the branch weights were probabilities, are ingenious and contested, and that's the famous open wound of the interpretation.

In CSD

CSD is single-trajectory, and this is a settled commitment recorded in the programme's non-goals rather than an open question to be revisited. There's one microstate, one complete state of the world, one history and one outcome. The outcomes that didn't occur are regions of the space of descriptions the trajectory didn't enter, and nowhere is there a branch in which they happened.

The contrast over probability is the instructive one. Everett's weight problem, the question of why observed frequencies match the branch amplitudes, the numbers the formalism attaches to each branch, doesn't arise here in that form, because there's always a definite state and probability expresses ignorance about it. The question becomes which measure weights the regions, where a measure is a rule that assigns a size to each set of states, and that question has a proved answer, the unique measure left unchanged by the symmetry of the state space. Where the Everettian literature argues decision theory, this programme points at a uniqueness theorem. This is a positional entry, and no theorem asserts it.

Mathematically

There's no formal content, as the corpus contains no theorem about the rival itself.

The formal contrast is that outcome weights in CSD are volume ratios on one constraint surface, so the measure problem is that of identifying a measure on a single space, discharged by uniqueness under symmetry, rather than that of justifying a weighting across many worlds, which is where Everettian effort concentrates.

Related entries

• Copenhagen interpretationCopenhagen is the oldest position on quantum measurement and the one that declines the question, as measurement is where the theory stops and not something it explains.
• Bohmian mechanicsBohmian mechanics is CSD's nearest neighbour, deterministic, with definite states, and openly non-local, and the dispute between them is over where the probabilities come from.
• Spontaneous collapse (GRW)Spontaneous collapse theories change the equations of quantum mechanics and can be killed by experiment, and CSD, so far, can't be, and its own documentation says so.
• Single trajectoryThere's one world and it follows one path, so probability lives in how experiments are repeated and not in how the world evolves, and theorems rather than wording keep the two apart.
• Measurement problemQuantum mechanics contains two incompatible rules of change and no instruction for when to switch between them, and CSD replaces the switch with theorems about what no dynamics can do.