String breaking with Wilson loops?

A convincing, uncontroversial observation of string breaking, when the static potential is extracted from Wilson loops only, is still missing. This failure can be understood if the overlap of the Wilson loop with the broken string is exponentially small. In that case, the broken string ground state will only be seen if the Wilson loop is long enough. Our preliminary results show string breaking in the context of the 3d SU(2) adjoint static potential, using the L\"uscher-Weisz exponential variance reduction approach. As a by-product, we measure the fundamental SU(2) static potential with improved accuracy and see clear deviations from Casimir scaling.