Unscreened forces in the quark-gluon plasma?

We study the correlator of temporal Wilson lines at nonzero temperature in $2+1$ flavor lattice QCD with the aim to define the heavy quark-antiquark potential at nonzero temperature. For temperatures $153\mathrm{MeV}\le T\le 352\mathrm{MeV}$ the spectral representation of this correlator is consistent with a broadened peak in the spectral function, position, or width of which then defines the real or imaginary parts of the heavy quark-antiquark potential at nonzero temperature, respectively. We find that the potential's real part is not screened contrary to the widely held expectations. We comment on how this fact may modify the picture of quarkonium melting in the quark-gluon plasma.