The goal of this article is to discuss a recent conjecture of the two authors, which aims to describe the long time behavior of solutions to one-dimensional dispersive equations with cubic and higher nonlinearities. These problems arguably represent the single most important example where, even for small initial data, the nonlinear effects are stronger than the dispersive effects. Consequently, the outcome predicted by the conjecture depends essentially on the structure of the nonlinearity, precisely its focusing or defocusing character.