In a recent paper, Ottinger and Beris assert that the celebrated Doi-Edwards (DE) model for entangled polymer dynamics is incompatible with the so-called "GENERIC" form of nonequilibrium thermodynamics. Simply put, they claim that the DE model for tube dynamics is inconsistent with the principle of virtual work. I show that with the correct choice of effective Hamiltonian, namely the full tube configurational entropy, the DE stress tensor does in fact result from a virtual work argument. (C) 2004 The Society of Rheology.