Two muons are born in the decay of an eta meson and fly apart, linked by quantum entanglement. One experimenter, Alice, measures the spin of the negative muon, while another, Bob, records the direction of the positron from the decay of the positive muon. It would seem that the order of events should affect the correlations. But calculations show the opposite.

This is like a dance where partners switch roles, but the music stays the same. If Alice's measurement happens before the decay (t1 < t2), it projects the pair into a definite spin state, and Bob sees a distribution characteristic of a pure state with opposite spin. The violation of the CHSH inequality reaches its maximum value of 2√2. If the decay occurs earlier (t1 > t2), the correlations persist: events selected by Alice's result look as if Bob's muon had a definite spin even before the decay.

The canonical explanation for the second case relies on post-selection: the muon decay influences the probabilities of Alice's measurement outcomes without violating causality. But for spacelike intervals, the order of events is not fixed. Each observer is free to describe the correlations in their own way: as a collapse from Alice's measurement, as the decay affecting Alice's spin, or even as a retroactive change in Bob's spin. All descriptions give the same predictions.

This underscores that wave function collapse is a convenient computational device, not a literal description of physical reality. Previously, it was unclear how to interpret correlations in entangled systems when one particle decays before a measurement on the other. Now it is shown that correlations do not depend on temporal order, and different observers can give different but equivalent descriptions of the same experiment. There are no practical applications of this knowledge yet, but it deepens our understanding of quantum mechanics and its interpretations.