I am a bit confused as to why you say there would be unpredictable outcomes if we used information of previous frames to compute new values in the current frame. It seems to me if the initial conditions and computation are always the same, the outcome would always be the same as well. For instance, if we could set the initial position of an object to 0,0,0, then in a position controller script set the object's X position of the current frame to the position of the previous frame's X position+1. How would the outcome be different each time the scene is played? At frame 10, the objects position should always be 10,0,0. Maybe Steve or someone can elaborate on this example to show what I am missing. It seems it would make animation easier in some cases if information of at least the previous frame's object parameters were known to the current frame's script.