02 May 07
1 edit
Originally posted by NordlysI think that's what I'm asking Zoot to demonstrate - the difference between doing this, and doing what I've described above, and why latter should result in unfairly skewed ratings, while the former does not.
I believe you are right. Another point, someone could always play much weaker (or much stronger) opponents without playing the same person all the time.
If he can't do this, and is simply saying people shouldn't consistently play weaker opponents, well... that's absurd.
02 May 07
Also taking into account that constantly playing a weaker person, where that person is genuinely trying to win, the stronger player still risks losing a large number of rating points. When the rating difference is great enough, the rating change becomes very small when the stronger player wins. Not enough for someone to use this for padding. There are a lot more effective ways to do this.
There is a good reason why the other thread became a circus act. The idea is completely ridiculous.
02 May 07
1 edit
Padding is not only deliberate losing of games it can also be considered to arise from Selective pairing which can compromise the integrity of the rating pool.
Originally posted by Wikipedia article
The Elo system when applied to casual online servers has at least two other major practical issues that need tackling when Elo is applied to the context of online chess server ratings. These are engine abuse and selective pairing.
The Elo rating is not supposed to be a reward system it is a statistical system. That is why here for example a provisional player starts off with a 1200 rating but cannot select the ratings of his opponents. Otherwise he would play against 2000+ opposition and end up with a high rating after 20 games.
However after those 20 games if he then selectively paired, the rating system is in danger of being compromised.
Prof z00t
02 May 07
2 edits
Originally posted by z00t[/b]Did you read the section on selective pairing from that page? I don't see where it says "low rated player you always beat".
Padding is not only deliberate losing of games it can also be considered to arise from [b]Selective pairing which can compromise the integrity of the rating pool.
Originally posted by Wikipedia article
The Elo system when applied to casual online servers has at least two other major practical issues that need tackling when Elo is applied to th ...[text shortened]... if he then selectively paired, the rating system is in danger of being compromised.
Prof z00t
P-
02 May 07
Originally posted by z00tThe "problem" of selective pairing is simply that, in systems where one can choose one's opponents, one can choose in such a way to maximise one's chance of ratings gain.
Padding is not only deliberate losing of games it can also be considered to arise from [b]Selective pairing which can compromise the integrity of the rating pool.
Originally posted by Wikipedia article
The Elo system when applied to casual online servers has at least two other major practical issues that need tackling when Elo is applied to th ...[text shortened]... if he then selectively paired, the rating system is in danger of being compromised.
Prof z00t
One does this by playing over-rated players and avoiding under-rated players.
It's not a question of whether you know the other person or have played them many times before. Yes, if you know their game you might have a feel for whether they are over- or under-rated at a particular point. But it is just as easy to assess the strength of a "stranger" at RHP (e.g. by playing through their games, looking at their rating history, etc.).
02 May 07
Originally posted by dottewellI don't think zoot wants anyone playing anymore after reading that. If you are playing chess, you are padding!
The "problem" of selective pairing is simply that, in systems where one can choose one's opponents, one can choose in such a way to maximise one's chance of ratings gain.
One does this by playing over-rated players and avoiding under-rated players.
It's not a question of whether you know the other person or have played them many times before. Yes, if ...[text shortened]... ger" at RHP (e.g. by playing through their games, looking at their rating history, etc.).
P-
02 May 07
Originally posted by PhlabibitYou don't still don't get it do you? You are supposed to play with different people who are :-
I don't think zoot wants anyone playing anymore after reading that. If you are playing chess, you are padding!
P-
A) - higher than you
B)- at your level
C)- lower than your level
The majority of the games should be in class B. If you are instead playing with D)- someone much lower ranked than you continuously, the distribution is skewed.
That is just like a GM who to get a title norm organises a mickey mouse competition, hand-picks his opponents and claims he is a GM. Was it not an exec of the USCF who resigned when confronted over "alleged rating manipulation"? You are in the US you should know better.
02 May 07
Originally posted by z00tWhat happens to your rating when a lower rated player beats you? Your rating goes down. The only way it can be padding is if that player agrees to let you win every game.
You don't still don't get it do you? You are supposed to play with [b]different people who are :-
A) - higher than you
B)- at your level
C)- lower than your level
The majority of the games should be in class B. If you are instead playing with D)- someone much lower ranked than you continuously, the distribution is skewed.
That is just like a GM ...[text shortened]... hen confronted over "alleged rating manipulation"? You are in the US you should know better.[/b]
you gain 5
you gain 4
you gain 3
you gain 4
you gain 3
you gain 4
you gain 5
You lose 40
P-
02 May 07
Originally posted by z00tI think you are taking ratings a little too seriously.
You don't still don't get it do you? You are supposed to play with [b]different people who are :-
A) - higher than you
B)- at your level
C)- lower than your level
The majority of the games should be in class B. If you are instead playing with D)- someone much lower ranked than you continuously, the distribution is skewed.
That is just like a GM ...[text shortened]... hen confronted over "alleged rating manipulation"? You are in the US you should know better.[/b]
Eg. Am I right in thinking that if two people subscribed to the site and only exclusively played each other using the rating system only for bragging rights that the admins should remove that feature from their games. Not very good business sense me thinks???
Try to remember that your RHP rating is just an indication of your chess strength relative to the people you have played, nothing more, nothing less.
03 May 07
1 edit
Originally posted by z00tNo; what matters is whether the lower rated players are over-rated, otherwise the minimal gain is offset by the risk of a major rating hit if you lose.
You don't still don't get it do you? You are supposed to play with [b]different people who are :-
A) - higher than you
B)- at your level
C)- lower than your level
The majority of the games should be in class B. If you are instead playing with D)- someone much lower ranked than you continuously, the distribution is skewed.
That is just like a GM ...[text shortened]... hen confronted over "alleged rating manipulation"? You are in the US you should know better.[/b]
And it makes no difference whether you know that person concerned, and play them repeatedly, or not - provided they also play regularly in the wider pool and are not over-rated.
It's manipulation if you deliberately bring it about that they are over-rated, then take advantage of that fact.
For example, regarding the Robert Tanner case you mention:
Sloan’s complaint alleges that Tanner achieved the titles of master and “original life master” by playing repeated games in 1992 and 1993 within a closed group of friends who traveled together and gathered in remote places such as Teton, Wyo., Wendover, Nev., and Ceres, Calif.
Original life masters must hold a master’s rating for at least 300 tournament games.
Sloan said the players in the group shared Tanner’s home address, and none of the nine men except Tanner ever competed outside the group.
“They are quite clearly fake people,” Sloan said from his New York home. “They don’t exist.”
Tanner maintains he did nothing illegal. He said the matches actually occurred, and the ethics committee supported his claim that the opponents were real people.
However, Tanner acknowledges in his Sept. 15 response to the committee that he circumvented the “spirit of the regulations” in the early 1990s while working to regain the master title. He also acknowledges that none of the opponents mentioned in the complaint were legitimate masters or even expert-level players.
Tanner said the opponents were friends he met while serving a church mission in Europe, and the group took remote hikes together when the friends visited the United States. He said the matches he submitted for rating were casual games played “under the stars” against opponents he has lost contact with over the years.
“In retrospect, although nothing illegal was done, it was manipulative,” Tanner wrote.
Manipulating the ratings system can be done when a highly rated player intentionally loses or earns a draw against an inferior opponent, which artificially bolsters the opponent’s rating. Within a closed group, an advanced player can then gain points by sweeping a field that includes several overrated players.
http://www.eastvalleytribune.com/story/80855
03 May 07
Originally posted by XanthosNZOne draw? One measley draw? Is that all you gave him?
Guess the jig is up.
http://www.timeforchess.com/core/viewpublicgames.php?p1name=Churchley&p2name=XanthosNZ&showgamescode=A&Search=Search
Mind you, at that time I suppose you were probably only rated about the same as him, weren't you?
Notice that I lost both of the unrated games here...
http://www.timeforchess.com/core/viewpublicgames.php?showgamescode=A&p1name=mokko&p2name=c99ux
