Bell’s inequality, which provides a test of whether a system is quantum or classical, is usually derived on the assumption that there are no correlations between the experimenter’s choice of measurement settings or anything else that could causally affect the outcomes. This leaves open the possibility that an unknown cause affects both the choice of settings and measurements before each trial – typically as little as microseconds before in experimental tests. A new “cosmic Bell test” by Johannes Handsteiner of the Institute for Quantum Optics and Quantum information in Vienna and colleagues uses light from distant astronomical sources to produce the settings. The work pushes back the most recent time that any local-realistic influences could have produced the observed Bell inequality violation by about 16 orders of magnitude, or approximately 600 years.