Sebens and Carroll (Br. J. Philos. Sci. 69(1), 25–74 1) propose that self-locating uncertainty, constrained by their quantum-specific Epistemic Separability Principle (ESP-QM), derives Born rule probabilities in Everettian quantum mechanics through a global branching model. This paper argues that their approach fails due to the loss of local amplitude information in global branching, particularly evident in an EPR-Bohm setup, where distant observers like Bob are assigned pure local states lacking the amplitude coefficients essential for Born rule probabilities. This flaw undermines the quantitative link to the Born rule, rendering their derivation empirically inadequate. Additional inconsistencies of global branching, including conflicts with decoherence dynamics, relativistic constraints, and limitations on superposition measurements, further weaken the model. Defenses invoking global amplitudes or benign non-locality fail to resolve these issues. This analysis underscores the need to reconsider branching mechanisms to secure a robust foundation for Everettian probabilities.
