Interaction frameworks are the norm for prototyping, implementing and sharing user interfaces and interaction techniques. However, they often lack the flexibility to easily implement new kinds of interfaces and interaction techniques, since they were basically designed for implementing standard and normalized WIMP user interfaces. This forces programmers to rely on " hacking " in order to experiment with functional prototypes, and could drastically limit the range of scenarios where these prototypes will work. In my PhD, I study the interplay between people designing interaction techniques, and their software frameworks. My goal is to identify a number of fundamental features and requirements that programming languages and systems should support, in order to improve the flexibility of interaction frameworks for programming advanced interaction techniques.