We discuss a first version of a extensible study design for a generic image-sorting task. It allows for comparing qualitative and quantitative properties of different interaction styles (e.g., direct manipulation, CLI, tangible interaction), input modalities (e.g., mouse vs. touch screen), output modalities, and UI implementations. Therefore, the study design may be of use for designing reproducible and replicable studies. Study participants are asked to sort a set of 27 photos into five categories where each of the photos depict one distinct topic belonging to only one of the categories. We conducted a first pilot study using this design, comparing a desktop GUI, an interactive tabletop, and physical photos. Task completion time was significantly lower when sorting physical photos than in the other two conditions. The study results may serve as a baseline and show limitations of the preliminary design.