³Planning might actually seem to be very easy in a simple illustration like in Figure 3.1, since all you need is to start at the goal state, and go backwards in the search tree until you arrive at the root; thus you have found the path from the root to the goal. The reason why this does not work in practice is that in reality there are many overlapping trees, each starting from a different root state, and each goal state can be reached starting from a number of different roots. So, you cannot go backwards because you don’t know which tree to follow. You can see this in the example of planning a route between two points in Paris: It may help a bit to start calculating backwards from Eiffel Tower, but you cannot just backtrack in a tree because the possible routes going “back” from the Eiffel Tower are as numerous as the routes you can start going “forward” from your current location; routes computed “backwards” from the Eiffel Tower can take you anywhere in Paris, not just your current location.