Sometime back I proved that for any undirected graph, it's possible to direct the edges so that the difference between indegree and outdegree of any vertice is at most 1.
Which leads to questions about which indegree-outdegree differences can be constructed.