Mechanical Constraints as Common Ground

between People and Computers This thesis presents a new type of human-computer interface based on mechanical constraints that combines some of the tactile feedback and affordances of mechanical systems with the abstract computational power of modern computers. The interface is based on a tabletop interaction surface that can sense and move small objects on top of it. Computation is merged with dynamic physical processes on the tabletop that are exposed to and modified by the user in order to accomplish his or her task. The system places mechanical constraints and mathematical constraints on the same level, allowing users to guide simulations and optimization processes by constraining the motion of physical objects on the interaction surface. The interface provides ample opportunities for improvisation b