In the theory of molecular light scattering and nonlinear optics, excited state damping is a significant consideration at frequencies near to resonance. Despite attempts to resolve a long-standing controversy over the propriety of such methods, there remains a dispute over the correct sign for the damping of antiresonant terms. Most established theory of Raman and associated light scattering employs a constant-sign rule at odds with a variable sign commonly used in nonlinear optics. However, by focusing on the polarizability it is demonstrated that arguments for the constant-sign convention vindicate standard Raman theory; flaws in the counterpropositions undermine the case for variable signing. It is also shown that a polarizability sum rule is valid only with constant-sign damping.