Robust Markov decision processes (RMDPs) are very nice but they are also… very robust. That’s why we propose a refined optimality criterion for policies in RMDPs, leveraging the notion of best-effort from game theory. I’m excited to present this joint work with Francesco Fabiano, Giuseppe De Giacomo, and Alessandro Abate this year at AAAI 2023! The preprint of our paper is already available on ArXiv.