Do you honestly fail to see how if you use both of those algorithms that one of them will take precedence over the other if you are going to produce one result set? I think you are assuming that latency is measured to a fixed location. No, latency is measured between peers. A host is chosen based on who has the lowest latency to all match participants. If those match participants are chosen based on a skill bracket and not location then you are gonna get latency all over the place. It is simple: choose a pool of possible match participants based on location and peer latency. Then build a game based on skill and wait time. Boom we are back to decent games.