As stated in the new trailer. Im guessing there will be a cluster of special EA servers that you can go to to match rank along with normal servers? 
-
-
Having had some limited experience (Thanks Zaedyn) with how some of this works, I'll explain how I think they'll do matchmaking.
Essentially, the matchmaking system is an AI learning program, if they've gotten something close to Microsoft's TrueSkill system used for Xbox Live matchmaking. You start out with an initial ranking which is in the middle of the pack, right square between the elite and the downright awful. You are thrown into a server with a bunch of average players. Depending on how highly your team is rated and whether you win or lose, you will either gain or lose points, the number of which is determined by how "good" the system ranks your team and the map bias which is computed based upon how likely a particular team is to win or lose a map.
So, your points fluxuate. Eventually, the system gets pretty good and can predict with a pretty good accuracy how "good" you are and whether your team will win or lose.
If I can find Zaedyn's doctorate thesis, I'll post it up here. Essentially, he wrote the above that I outlined for Wolfenstein: Enemy Territory and called it Active Team Balance. And it was a darned good system if I do say so myself (I may even show him this thread at some point, heh).
EDIT:
Found it!
Be forewarned, there's some math in that PDF... Lots of advanced math that I don't really understand, and my description of the above system (above the edit) proabably doesn't do the system justice.Statistics / Rankings Terms
There are several settings in ETPub that attempt to determine how "good" a player is in terms (hopefully) more meaningful than just XP or XP per unit of time. Here are the terms used and their definitions:
Kill Rating
How good of a killer the player is, based on how many other players the player kills, and their kill rating. In other words, killing players with a high kill rating increases the shooter's kill rating more than killing players with a low kill rating.
Player Rating
This is a measure of how much the player contributes to winning a map. This measure is calculated by seeing how many times this player is on the winning team after every map, and how good the opposing team was. Like kill rating, winning against teams with a high average player rating results in player rating increasing faster.
Win Probability
The probability that a team will win a map (based on the players' player rating, team size, and the map).
These statistics are gathered by etpub and the data is saved to local files on disk.
The etpub development team member responsible for implementing these player rankings and ratings is Josh Menke. He has been kind enough to begin working on an academic write-up to explain the mathematics and statistics methods behind these rankings. Due to time constraints the document is being gradually updated and expanded. The latest version can be seen at: http://axon.cs.byu.edu/~josh/etstats/update_bayes.pdf
Comment