Material slope on spherical worlds

Sorry to bump and old issue, but I am looking for some clarification of this issue as well. When Deathrey said that the up vector varies by location on the sphere, how can that be expressed in a material graph?