Contraction mapping is not really required for this phenomenon. Contraction would mean that any two points are closer after the crumpling/mapping than before. In fact contraction will guarantee that there is only one such point which lies over the other.
For the aforesaid property any form of crumpling that does not tear the paper would suffice. Provided no part of the crumpled paper extends beyond the boundary of the pristine sheet (or the city, in your example). The phenomenon relies of Browder's fixed point theorem.
If it helps to reduce one dimension: think of a continuous curve defined over a part of the x-axis [0, 10]. As long as the curve stays inside the box [0,0], [10,10] and every point in the [0,10] part of the x-axis is mapped, it would be impossible to avoid the diagonal. Just try it.
Incidentally, a generalization of the theorem above, called the Kakutani fixed point theorem underlies John Nash's proof (that won him the Nobel) of the existence of equilibria in games.
For the aforesaid property any form of crumpling that does not tear the paper would suffice. Provided no part of the crumpled paper extends beyond the boundary of the pristine sheet (or the city, in your example). The phenomenon relies of Browder's fixed point theorem.
If it helps to reduce one dimension: think of a continuous curve defined over a part of the x-axis [0, 10]. As long as the curve stays inside the box [0,0], [10,10] and every point in the [0,10] part of the x-axis is mapped, it would be impossible to avoid the diagonal. Just try it.
Incidentally, a generalization of the theorem above, called the Kakutani fixed point theorem underlies John Nash's proof (that won him the Nobel) of the existence of equilibria in games.