you're making it sound a lot less complicated than it is. Referencing/reworking/gutting old code, adding new menus, questions of balance and implementation ; the on/off suggestion has been made before, I don't think it a reasonable solution. It also takes the mind out of the automaton. Part of mastering the job is learning how to work with your puppet (despite having a rather frustrating AI that needs some fixing) and working with your puppet includes accepting that it has a psuedo-mind of its own.