Greg -
I realize that this will probably stir the pot a little, but......
I totally agree with the rating ceiling concept. However, I really think that new players should automatically begin at a set "initial rating" (the number itself is not important to my point here). Only those people who can "prove" that they have an established rating elsewhere, with a legitimate chess authority, would be allowed to start at a different rating.
Just my 2 cents worth....
Tim
Rating Ceiling
-
grandphish
- Posts: 4
- Joined: Wed Sep 24, 2003 11:56 am
I totally agree with the rating ceiling, however, if you are really serious about getting the ratings inflation problem under control, removing ALL ratings floors would a more effective action.
I offer the following analysis in support of removing all floors. I am using rhand as an example. I intend no criticism of rhand by picking him for the example. He is simply the player with the most number of games and therefore the effect of his games is greater than others with a small number of games.
rhand has played 5,856 games as of this writing. The average rating of his/her opponents is 1424. He/she has won 2,949 games, lost 2,089 and drawn 818. Using the following formula, I compute that his/her rating should be approximately 1483, based upon performance:
2,949 x 1824 (opponents +400) = 5,378,976
2,089 x 1024 (opponents -400) = 2,139,136
818 x 1424 (opponents rating)= 1,164,832
Total 8,682,944
Number of games 5,856
Average perforance 1,483
Since he/she has a rating floor of 2300, there is an inflationary addition of a minimum of 34.4 rating points for every game played against rhand.
Rating floor 2300
Computed strength 1483
Rating inflation 817
x 4%
Points added/game 34.4
My conclusion is that every game played against rhand adds 34.4 points to the opponents rating, on average, above what would be added/subtracted if there were no rating floor. For the games played by rhand alone, this has artificially added 199,104 rating points to the rating of others. (34.4 points per game x 5,856 games.) rhand accounts for less than 2% of the games on net-chess. I have no idea how many players have floors or how much their ratings are inflated over their actual performance. But I think we can agree that every player with an artificial floor is contributing to the inflated ratings with every game played. As long as these artificial floors are continued, the ratings will continue to rise artificially. Elimination of the rating floors will gradually slow or stop the problem as the previously overrated players fall to their natural rating based upon their performance. Since their ratings are inflated now, they will fall faster than normal toward their performance rating.
I have no rating floor, nor do I want a rating floor. I choose to have my rating represent my performance - nothing more, nothing less.
If we are really serious about establishing a credible rating system at net-chess, I urge everyone to support elimination of rating floors.
My apologies to rhand. I again emphasize that I used you only as an example of the problem and singled you out only because you are at the top of the pack in number of games played.
I offer the following analysis in support of removing all floors. I am using rhand as an example. I intend no criticism of rhand by picking him for the example. He is simply the player with the most number of games and therefore the effect of his games is greater than others with a small number of games.
rhand has played 5,856 games as of this writing. The average rating of his/her opponents is 1424. He/she has won 2,949 games, lost 2,089 and drawn 818. Using the following formula, I compute that his/her rating should be approximately 1483, based upon performance:
2,949 x 1824 (opponents +400) = 5,378,976
2,089 x 1024 (opponents -400) = 2,139,136
818 x 1424 (opponents rating)= 1,164,832
Total 8,682,944
Number of games 5,856
Average perforance 1,483
Since he/she has a rating floor of 2300, there is an inflationary addition of a minimum of 34.4 rating points for every game played against rhand.
Rating floor 2300
Computed strength 1483
Rating inflation 817
x 4%
Points added/game 34.4
My conclusion is that every game played against rhand adds 34.4 points to the opponents rating, on average, above what would be added/subtracted if there were no rating floor. For the games played by rhand alone, this has artificially added 199,104 rating points to the rating of others. (34.4 points per game x 5,856 games.) rhand accounts for less than 2% of the games on net-chess. I have no idea how many players have floors or how much their ratings are inflated over their actual performance. But I think we can agree that every player with an artificial floor is contributing to the inflated ratings with every game played. As long as these artificial floors are continued, the ratings will continue to rise artificially. Elimination of the rating floors will gradually slow or stop the problem as the previously overrated players fall to their natural rating based upon their performance. Since their ratings are inflated now, they will fall faster than normal toward their performance rating.
I have no rating floor, nor do I want a rating floor. I choose to have my rating represent my performance - nothing more, nothing less.
If we are really serious about establishing a credible rating system at net-chess, I urge everyone to support elimination of rating floors.
My apologies to rhand. I again emphasize that I used you only as an example of the problem and singled you out only because you are at the top of the pack in number of games played.
Totally agree with Tim and „grandphish“ (whom I know, as we’re both playing at playchess.de). The consequences of Tim’s suggestion: any new player joining netchess and giving his rating (ELO) with FIDE/USCF/ICCF has to give his full name and address, cause no way to play at those recognized chess authorities with an anonymous handle. Greg will have to check the new player’s rating according to the ratings lists published in the net by those authorities – those are official lists! As initial rating I’d suggest 1500.
I also agree with grandphish that the rating floor is to be dropped once and for all. You might at the utmost chose the “floor system” of the USCF, but I’d suggest to drop it completely.
The most dramatic change to guarantee a fair procedure for the ratings would be that the tournaments are offered by the admin and not by the players of netchess. Besides the private challenges all the organisations like USCF/FIDE/ICCF offer tournaments of different classes + the different championships and qualifications.
Best regards
Hans
I also agree with grandphish that the rating floor is to be dropped once and for all. You might at the utmost chose the “floor system” of the USCF, but I’d suggest to drop it completely.
The most dramatic change to guarantee a fair procedure for the ratings would be that the tournaments are offered by the admin and not by the players of netchess. Besides the private challenges all the organisations like USCF/FIDE/ICCF offer tournaments of different classes + the different championships and qualifications.
Best regards
Hans
-
gmiller
- Site Admin
- Posts: 1388
- Joined: Sun Mar 14, 1999 11:13 am
- Location: Jeffersonville, IN
- Contact:
I tried the "Everyone starts at 1500" and almost no one ever e-mailed me to have their rating set to their "real" rating. So we had 2100 and 800 players playing at 1500. Which equially caused problems. And back then I didn't require any proof, just an e-mail.
I seriously doubt the rating floor causes much of the inflated ratings. We currently only have 103 people playing a game who are currenly rated within 50 points of their rating floor. And I've decreased the floor so that it's 200 points minus the highest rating rather than minus 100, so that will nearly eliminate that as a factor. Additionally, I've lowered the maximum floor to 2100. Both of these are in line with what the USCF uses and I plan to adopt as much of the USCF rating system as possible.
I've also considered the following:
The idea of a "fudge" factor could work, but I think that would just PO a lot of people. And it's rather complicated to work into future ratings. I'd rather just tell people if they want to see how they compare to a USCF rating, muliply your rating by xxx (hopfully xxx isn't .0001).
I seriously doubt the rating floor causes much of the inflated ratings. We currently only have 103 people playing a game who are currenly rated within 50 points of their rating floor. And I've decreased the floor so that it's 200 points minus the highest rating rather than minus 100, so that will nearly eliminate that as a factor. Additionally, I've lowered the maximum floor to 2100. Both of these are in line with what the USCF uses and I plan to adopt as much of the USCF rating system as possible.
I've also considered the following:
- -Playing against a provisional player will have no affect on your rating.
-Only games against established players count towards getting an established rating (and probably lower the # of games required for an established rating).
-If a player remains within 100 points of their floor for xx days,months, etc, their floor drops 100 points.
-If a player ramains at their rating floor for xx games, their floor drops 100 points.
The idea of a "fudge" factor could work, but I think that would just PO a lot of people. And it's rather complicated to work into future ratings. I'd rather just tell people if they want to see how they compare to a USCF rating, muliply your rating by xxx (hopfully xxx isn't .0001).
Good first steps!
Nice job on the first steps to correct the problem.
I would like to be added to the list of people who have no rating floor at all, please.
The only reason for rating floors in the real world is to stop people from sandbagging and picking up money by playing in a 'C' class when they are an 'A' player, and that does NOT apply here. I have seen the argument that it protects your rating if you are gone for a long stretch and games time out, but I think that is totally bogus. If you are going to be gone, you should PLAN for it. If it's a true emergency (you're mother is dying in Tibet, and there won't be a net connection for 200 miles of water-buffalo ridden trail) you can always work something out one-on-one with Greg, but I think it should be an extreme exception. Otherwise, you should live and die by your games, good and bad.
Just my two cents...
-Thad
I would like to be added to the list of people who have no rating floor at all, please.
The only reason for rating floors in the real world is to stop people from sandbagging and picking up money by playing in a 'C' class when they are an 'A' player, and that does NOT apply here. I have seen the argument that it protects your rating if you are gone for a long stretch and games time out, but I think that is totally bogus. If you are going to be gone, you should PLAN for it. If it's a true emergency (you're mother is dying in Tibet, and there won't be a net connection for 200 miles of water-buffalo ridden trail) you can always work something out one-on-one with Greg, but I think it should be an extreme exception. Otherwise, you should live and die by your games, good and bad.
Just my two cents...
-Thad