Bell's inequality

Bohr's reply to EPR was that the two measurements are parts of one indivisible phenomenon, so asking what A's result was before B's was measured is asking a question the theory does not license. Nothing propagates; nothing needs to. Critics grant that this is consistent and object that it is silent exactly where an explanation was wanted.

Supporters — Vaidman among them — argue that Bell's argument never gets started here: the theorem assumes each run has a single outcome, and this interpretation denies exactly that. The correlations appear when the branches are compared, which happens at a meeting point reached no faster than light. Saunders and others contest the move; the nonseparability of the entangled pair remains on any reading, and critics add that the branching structure and the probabilities still need an account the bare formalism does not supply.

De Broglie–Bohm reproduces the quantum curve exactly, and it does so by being explicitly nonlocal — the guidance equation for one particle depends on the instantaneous configuration of the other, however far apart they are. No signal can be sent, because the statistics of either wing alone stay random. The tension is with relativity: the theory appears to need a preferred slicing of spacetime that relativity does not supply.

In GRW-type models a localization event affecting one particle localizes its entangled partner too, so the nonlocality Bell proved unavoidable sits in the dynamics themselves. Relativistic versions are an active research problem. Because the modified dynamics differ measurably from standard quantum mechanics, this family remains the one that experiments can genuinely constrain — and so far every test has tightened its limits rather than found it.

QBists take the Bell correlations as a constraint on any agent's gambles, not as evidence of an influence. Since the state is a personal betting book, no physical thing changed at the distant wing when you updated it. Critics accept that this dissolves the mechanism question and press the obvious follow-up: what is the world like, such that agents must bet this way and no other?

Rovelli's account keeps everything local by refusing absolute facts: each interaction establishes values relative to the systems involved, and the comparison of two records is a further interaction, one that travels no faster than light. What must be given up is the assumption that there is a single observer-independent list of what happened.

Griffiths argues that Bell's inequality requires reasoning across frameworks that the single-framework rule disallows, so the theorem does not force nonlocality on this reading — only a restriction on which questions may be asked together. The correlations remain, fully predicted. Critics find the framework-selection rule too permissive, and the debate over whether this dissolves nonlocality or renames it is unsettled.