That paper is really neat. I mentally connect it to Bertsekas's recent ideas on how the k planning steps are actually one Newton step and k-1 value iterations.

www.sciencedirect.com/science/arti...