While equilibrium free energy differences can be obtained from simulations of non-equilibrium systems, such estimates are often hampered by convergence difficulties similar to those that plague traditional perturbation methods. In the nonequilibrium context, dissipation is the culprit behind poor convergence. I will discuss a general strategy for addressing these difficulties, in which dissipation is reduced by adding non-physical terms to the equations of motion. When these terms are chosen appropriately, the efficiency of the free energy estimate can improve dramatically. After sketching the general features of this strategy, I will illustrate its application using specific examples, and will discuss its relation to other strategies that are similar in spirit.