Kaiserklein wrote:The average or the initial rating doesn't matter. Having a starting ELO of 1600 or of 0 doesn't make any difference. What matters is that pr takes into account the homecity level, the amount of games, the civs played, and probably some other bs (like you get your pr back faster when you've been inactive, for example), and it decays. While ELO just takes into account your current ELO and your opponent's, and doesn't decay, which is better because if you come back after a long time you'll naturally lose some ELO anyway. ELO is just a more objective rating system.
Actually the way I understand it is that pr starts at 0, and then quickly tries to determine what the actual ELO of a player is by inflating the rank changes for a little while. The average rating for pr is probably around 15 or so, at least its definitely more than 0. For ELO the average rating should be 1600.
The big problem with this is that on an ELO system with the size of ep's, where you get a top tier ELO if you reach 1800, it's actually beneficial to bash new players for rank, because beating a 1600 player will actually give a reasonable amount of ELO. If then, the influx of new players is reasonably large, this is going to distort the system pretty badly.
The way pr handles new players is probably slightly better for a ranking system with such a small playerbase and such a relatively large amount of new players. If the system will actively try to determine your real ranking for the first 10 games, and thus increase the rank you get from a win (but also decrease the rank the other player loses due to losing to a new player), the system will not be as exploitable. Similair systems are actually used in many competitive games, where the matchmaker will just take a few games to try to find what rank you really are, without letting other players profit from that.