real-time ability to change the orientation of the entire argument map so that premises appear above conclusions, left of conclusions, etc.
at least in philosophy, students are generally introduced to textual arguments with the conclusion at the bottom (under a line) and premises above and proceeding methodically downward toward the conclusion. It is good pedagogical practice to connect new material, e.g., how to graphically represent the inferential structure in arguments using an argument map, to material students have already learned. So, it would be pedagogically beneficial to be able to graphically represent arguments with MindMup in a way that students have already learned.
Ultimately, of course, there is nothing magical or obligatory about any particular convention for representing arguments or any particular orientation of an argument map. It is all about whatever offers the most clarity and works best pedagogically for a particular task. Sometimes that is with conclusions at the top, with the premises supporting from below; sometimes that is with the conclusions at the bottom, with the premises leading to them, and mirroring (many languages') top-to-bottom reading order; sometimes that is with the premises on the left and the conclusions on the right, mirroring (at least English) left-to-right reading order; etc. So, it would be pedagogically useful to be able to switch the orientation of the argument in real time to suit the pedagogical task at hand.
Indeed, one of those pedagogical tasks is to teach students that, though conventions are certainly useful, there is nothing magical or obligatory about any particular convention for representing argumentative structure, and that decisions about how to represent arguments should be driven by considerations of clarity and elucidatory power. The real-time ability to flip arguments around from left to right, and top to bottom can help focus students on the inferential structure of the argument itself, which is convention-neutral, as opposed to the conventions themselves.
Peter Murray commented
Implementing this idea would likely require adding something that I think is an independently good idea: a direction-of-support arrow built into the lines connecting premises and conclusions. With such arrows, the inferential relations would no longer be tied to the orientation conventions of the map.