A particular steady-state strategy of evolution with a small sized population is studied in this paper. After comparison with GA and ES, we specially focus our attention on the choice of genetic operators, and the way to apply them. Knowing that it is not possible to reach a universal heuristic able to choose the genetic operators and to manage them, we present methods where the genetic operators themselves are evaluated according to their performance. Thanks to those methods, we observe improvements in order to optimize classical problems.