> The author's system includes the weights and their tracks, but not the arm connecting them. But the arm is a part of the system. Are they not accounting for

> The author's system includes the weights and their tracks, but not the arm connecting them. But the arm is a part of the system. Are they not accounting for the centripetal force through the arm? (EDIT: Answering my own question, the preprint does include the text: > Notably, the robot’s ability to move does not violate the usual rule against movement without forces in three-dimensional space because the full three- dimensional dynamics in fact include normal forces on the robot supplied by the boom arm that confines it to the sphere. So the authors are (of course!) aware of this in 3D space. Their experimental setup is functionally no different from the swivel chair example.) > In order for you to demonstrate their phenomenon, your chair would have to change position (translate) from where it started, not just rotate in-place. No, by the same logic above, why can I not simply keep my upper arms stiff and declare them not part of the system? My lower arms and the weights are then "translating" through "space".