A semi-classical theory was re-derived in a consistent form for properly comparing direct and heterodyne detection as a function of wavelength. Plots are shown for example cases. We show that heterodyne should be better than direct detection for wavelengths longer than 3 microns, even with a bandwidth disadvantage, since direct detection is more sensitive to ambient temperature background than heterodyne detection. For interferometry the advantage of heterodyne is more pronounced in this case due to the smaller beam filling factors. When we include even the effect of surpassing the noise temperature quantum limit with a novel correlation receiver architecture (see paper 10701-94), the advantage of heterodyne detection becomes irrefutable.