Entanglement

Bohr treated the two measurements as aspects of a single indivisible phenomenon rather than as two separate events with something passing between them. That is why no signal is needed and none is available. The standing objection is not that this is wrong but that it declines to answer: the correlation is predicted exactly and explained not at all.

Because measurement never picks an outcome, there is no distant selection to explain, and supporters argue this makes the account genuinely local. The far particle's state is unchanged by your measurement; what changes is which version of you is reading which record. The comparison of notes is an ordinary, slower-than-light meeting.

The guidance equation for one particle depends on the instantaneous configuration of the other, at any separation — explicit nonlocality, exactly what Bell showed no theory of this kind can avoid. You still cannot send anything, because the outcomes on either wing alone are randomly distributed. The unresolved cost is relativistic: the theory seems to want a preferred notion of simultaneity.

In GRW-type models a spontaneous localization affecting one particle collapses the joint state, so the far particle's properties settle too. This is nonlocal in the way Bell requires, and building a relativistic version is an open research problem. As with all collapse models the payoff is that the dynamics differ measurably from standard quantum mechanics, so experiments can keep testing them.

For a QBist the entangled state is a single agent's set of gambles about two future experiences. Learning A's result revises the odds for B, exactly as drawing one card revises the odds for the next, with quantum theory supplying the discipline. Critics accept that this removes the spooky mechanism and ask what the world must be like for those particular odds to be the right ones.

Rovelli's move keeps everything local by giving up absolute facts. Each interaction settles values relative to the systems taking part; comparing two accounts is itself an interaction, and it happens where the records meet, at no more than light speed. What has to be abandoned is the intuition that there is a single true story of the pair that both observers are sampling.

Within a single consistent family the pair's behaviour is unremarkable probability. Griffiths argues that the appearance of nonlocality comes from combining incompatible families, which the single-framework rule disallows, so no influence need be posited. Critics reply that the freedom to choose frameworks is itself unexplained, and whether this dissolves the puzzle or relocates it remains contested.