Expressive power of a language is generally defined as the breadth of ideas that can be represented and communicated in a language. For formal languages, the expressive power has been evaluated by checking its Turing completeness. In a modeling process, apart from the modeling language, we have two other counterparts: the system being modeled and the modeler. I argue that faithfulness to the system being modeled and usability for the modeler are at least as important as the expressive power of the modeling language, specially because most of the modeling languages used today are highly expressive. I call faithfulness and usability together “friendliness”. I show how we used the actor-based language Rebeca in modeling different applications, where it is friendly, and where it is not. I discuss how the friendliness of Rebeca may help in the analysis of models and allows for system synthesis on the basis of models.