In this article, the author tries to show why we should keep improving simplified models, even if modern computers are able to perform huge amount of calculus, and that finite elements models are not as expensive as they were few years ago. A complete example of a simplified model aiming at estimating the seismic sliding of a dam during an earthquake is presented and compared to a finite elements model, with which transient nonlinear analyses are performed. Finally, the author tries to show how it is easy and useful to improve this kind of simplified methods, in order to take into account complex phenomena, such as the soil-structure interaction, or the effect of vertical vibrations of the fluid on the mean pressure applied on the upstream face of the dam.