The last few years have seen a wealth of new nonlinear propagation modeling results appear in the literature, especially regarding coherent systems operating in the absence of optical dispersion compensation. One of the most prolific lines of research, though not the only one, has been that of improvements and upgrades to the GN-model, which have also led to the socalled EGN-model. In addition, many specific aspects of nonlinear propagation, including format and symbol-rate dependence of non-linearity generation, long-correlated nonlinear phase and polarization noise, the effect of co-propagating amplified spontaneous emission noise and distributed amplification, and still others, have been focused on and several new related results have been published.This has been a very positive trend but, from the viewpoint of the end-users, such as system and network designers, this large body of new knowledge may have been found difficult to sort out. The question of when and whether more sophisticated models are truly needed in any given system/network scenario, for a given set of accuracy and computational complexity constraints, then naturally arises. This paper tries to address this practical issue and provide indications regarding possible effective solution to varied end-users' requirements.