Specifying and programming graphical interactions are difficult tasks, notably because designers have difficulties to express the dynamics of the interaction. This paper shows how the MDPC architecture improves the usability of the specification and the implementation of graphical interaction. The architecture is based on the use of picking views and inverse transforms from the graphics to the data. With three examples of graphical interaction, we show how to express them with the architecture, how to implement them, and how this improves programming usability. Moreover, we show that it enables implementing graphical interaction without a scene graph. This kind of code prevents from errors due to cache consistency management.