The operational reality of quantum nonlocality

Does the remote measurement-disturbance of the quantum state of a system $B$ by measurement on system $A$ entangled with $B$, constitute a real disturbance -- i.e., an objective alteration -- of $B$ in an operational sense? Employing information theoretic criteria motivated by operational considerations alone, we argue that the disturbance in question is real for a subset of steerable correlations. This result highlights the distinction between quantum no-signaling and relativistic signal-locality. It furthermore suggests a natural reason why a convex operational theory should be non-signaling: namely, to ensure the consistency between the properties of reduced systems and those of single systems.