I am thinking to write my Master's thesis around theorizing, and then implementing a PlayMaker or Kismet-like (building game logic by visually arranging FSMs) tool in Unity. The only thing I am still concerned about is the actual research question that I should pose.
I was kinda hoping that the more experienced game designers out there might know.
Update: What about reducing the use of visual programming to graphically designing FSM-Action-Transition flows, which can then be attached to game entities (very much like http://playmaker.com does it)?
Update 2: Regarding the evaluation of the approach: I was thinking that it would not be so practical to try to quantitatively prove that my tool is better than the competition. This is in fact not the essence of the thesis. The problem that I see myself trying to solve is more of the type of "how do we prove that the approach is easy to pickup and learn from people without specific programming knowledge" I could build a "test sandbox" - an unfinished puzzle level that each of the test subjects is supposed to complete by following a number of certain tasks. The taks will vary from simply adding a missing transition between two states, through correcting intentional mistakes in the logic, to designing entire state machines on one'e own. The evaluation will then be done based on the completion of the tasks.