The overlap is not a waveguide

Golterman and Shamir falsely claim that a waveguide model modified by adding many charged bosonic spinors, in the limit of an infinite number of matter fields, becomes identical to the overlap if in the target theory every fermion appears in four copies. Their modified model would give wrong results even in the vectorial four flavor massless Schwinger model, while a dynamical simulation of this model with the overlap works correctly. In this note we pinpoint the error in the derivation of Golterman and Shamir.