Public forum discussion.
3 Sep 2003
I am a bit surprised by the way the reatings system works.
For instance, Judy recently beat gwg_robot in a rated game of Golem's Word game (and she thoroughly deserved to win). As a result, her rating went up by 20 points, and gwg_robot's went down by 20 points. But Judy has over 400 played rated games, and gwg_robot only 23. So I would expect her rating to be pretty well established, and its to be still somewhat uncertain. Doesn't a result like Judy's win provide more information about its strength, than about hers? I would expect the robot's rating to fall by more than Judy's rises.
If you look at Judy's rating graph, it zigzags rapidly up and down. I don't think that this is becasue her play is volatile. I think it is because the algorithm is failing to recognise that some people's ratings are better known thatn others.
4 Sep 2003
My DEAR Nick,
My graph "correctly" shows my playing ability, and it looks like a craggy mountain range. I do have a lot of games where, frankly, I can't see the words for the letters...........but, I still try. I do not play for score (note unrated games) I play for fun. I was very pleased to beat the Robot, since the last time around he trounced me pretty good. You are obviouslly following the rating system more than I have been aware of it, and you may have some valid points. But, your comments about my ability, are simply just kind---------and I thank you for looking out for me.
10 Sep 2003
From what I have observed , I think the ratings system is designed to produce a snapshot of a players current strengh, it doesn't seem to matter if a player has played 1000 games or 10, your only as good as your last game.
10 Sep 2003
My Street Soccer rating just went from 1500 to 1482 and I have not yet completed a rated game. I am in several tournament games at the moment but why would my rating go down before having completed 1 tournament game? Any ideas?
Mike
10 Sep 2003
Berry, if you're only as good as your last game, then what would be the point of ratings in the first place? Mike, what happened in game #79686 ?
10 Sep 2003
Oops- I did not know how to find my own page of stats- I just did and I must have been so into that game that I did not realize it was already to move 52. I'm learning more about the site every day and I am getting pounded as a rookie should.
10 Sep 2003
Jan, I mean that the ratings arn't weighted, a player who has played 1000 games and is on a rating of 1600, would lose the same amount of points as someone who has the same rating after only 5 games, if they were to lose against the same opponent.
10 Sep 2003
OK. But the system isn't too bad since the rating of an established player doesn't change a lot after a match against a new player. On most places I have played on the net, new players start off with a fixed "established" rating.
10 Sep 2003
Any good rating system should try keep the inflation to an absolute minimum. If we would allow a system where one player could get more points than the opponent would loose, than at some point people will break through either a 10k barrier or through a -0 barrier. This is just a simple matter of mathematics.
Even good rating systems are subject to inflation. For example, it has been proven that the official chess rating, invented by Arpad Elo, is a victim of about 0.2% annually. This also explains that more people then ever before in the chess world have broken through a 2500 barrier. And that number will grow in the future.
The reason for this are low-rated people that drop out of play. Every player starts with a rating of 1000 or some other arbitrated number. A player quits playing chess at 900, which he lost at tournaments. So the chess community as a whole gains 100 points. This is where inflation comes in.
The are some more reasons inflation occurs, but maybe a mathmatician can explain those.