The complication arises when we are using the term (|r_n-r_e|)^{-1}. To solve this we need to use a multipole expansion.
This multipole expansion also allows one to break the problem down to more manageable pieces, giving rise to different orders of contributions in the process. Identifying these orders allows one to possibly simplify the work by using the fact that some of these orders are zero.