Just because something is legal does not mean that it is moral.
@Burrower
Sorry, I thought that when delta rating is > 500, you earn almost nothing...
But it is not the case; here is the ELO Fide calculator:
ratings.fide.com/calculator_rtd.phtml
It appeares that, for example, a 2300 player winning against a 1800, even with k = 10, earns 0.8 points
Not negligible....
Regards
Sorry, I thought that when delta rating is > 500, you earn almost nothing...
But it is not the case; here is the ELO Fide calculator:
ratings.fide.com/calculator_rtd.phtml
It appeares that, for example, a 2300 player winning against a 1800, even with k = 10, earns 0.8 points
Not negligible....
Regards
It seems that this "cherry picking" issue would have a policy similar to that of sandbagging. Even know the process is kind of the opposite.
Virtuous poster #11 @Rrhyddhad makes a very appropriate point. Legality or rule compliance is definitely not the same thing as morality.
@Toadofsky stated "I don't see anything wrong with cherry-picking because it does follow the rules as the rules are currently written."
However, what's more important is the way the rules seek to maximise enjoyment for all users participating on this site. When persistent observations are made, such as suffering and system destabilisation inflicted by merciless and self-interest maximizing Cherry Pickers, then sometimes we may wish to review either the rules or the system. It's a scientific feedback loop where observation can update current theories/rules of fair play.
The Cherry Picker specifically chooses his strategy to exploit the lack of statistical validity in the way the glicko2 system handles extreme rating differences. In simple terms, this method of system exploitation works, which is why it is so popular. His intentions are not in the interest of the community.
Rules are best applied when they are formalized into quantitative criteria. On this note, I suggest quantifying a Cherry Picking Ratio (CPR) for each player. Win to loss ratio seems like an appropriate measure (with some adjustments for rating percentile). Since even if the player has improved significantly with time, his net long term choice of opponent rating difference will be integrated into this figure. For example:
(1) Win 90% of games or higher, clinical diagnosis of Cherry Picking and an appropriate punishment
(2) Win 10% to 90%, permissible.
(3) Win <10% of games, possibly also guilty of encouraging Cherry Picking by choosing to be a serial victim.
Tweak with the numbers as you please.
We must then adjust the significance of these figures depending on the rating percentile. The higher or lower the rating percentile, the more they should be adjusted, since the pool of opponents becomes smaller. For instance, a 2900 will need to settle for lower rated opponents, while a 700 will need to settle for higher rated opponents. Which is fair, because we don't also don't want the unfairness of very long wait times for high or low rated players. Therefore, we need to multiply the Cherry Picking Ratio by the percentile co-efficient to derive a finalized Cherry Picking Ratio (CPR) for each player, then apply the appropriate punishment if it is over the Cherry Picking Threshold.
Warm regards, Burrower 🙏
@Toadofsky stated "I don't see anything wrong with cherry-picking because it does follow the rules as the rules are currently written."
However, what's more important is the way the rules seek to maximise enjoyment for all users participating on this site. When persistent observations are made, such as suffering and system destabilisation inflicted by merciless and self-interest maximizing Cherry Pickers, then sometimes we may wish to review either the rules or the system. It's a scientific feedback loop where observation can update current theories/rules of fair play.
The Cherry Picker specifically chooses his strategy to exploit the lack of statistical validity in the way the glicko2 system handles extreme rating differences. In simple terms, this method of system exploitation works, which is why it is so popular. His intentions are not in the interest of the community.
Rules are best applied when they are formalized into quantitative criteria. On this note, I suggest quantifying a Cherry Picking Ratio (CPR) for each player. Win to loss ratio seems like an appropriate measure (with some adjustments for rating percentile). Since even if the player has improved significantly with time, his net long term choice of opponent rating difference will be integrated into this figure. For example:
(1) Win 90% of games or higher, clinical diagnosis of Cherry Picking and an appropriate punishment
(2) Win 10% to 90%, permissible.
(3) Win <10% of games, possibly also guilty of encouraging Cherry Picking by choosing to be a serial victim.
Tweak with the numbers as you please.
We must then adjust the significance of these figures depending on the rating percentile. The higher or lower the rating percentile, the more they should be adjusted, since the pool of opponents becomes smaller. For instance, a 2900 will need to settle for lower rated opponents, while a 700 will need to settle for higher rated opponents. Which is fair, because we don't also don't want the unfairness of very long wait times for high or low rated players. Therefore, we need to multiply the Cherry Picking Ratio by the percentile co-efficient to derive a finalized Cherry Picking Ratio (CPR) for each player, then apply the appropriate punishment if it is over the Cherry Picking Threshold.
Warm regards, Burrower 🙏
As I have suggested multiple times in this thread/topic, I think this solution could work and wouldn't require moderator activity:
github.com/ornicar/lila/pull/5708
github.com/ornicar/lila/pull/5708
Hey Toadofsky is there a forum page where I can read about RD? Dont quite understand it
Sure thing, a basic overview which I mostly agree with is here: lichess.org/faq#ratings
Further details can be found at http://glicko.net/glicko/glicko2.pdf
Further details can be found at http://glicko.net/glicko/glicko2.pdf
Yes certainly, there is such a regulation in FIDE that protects the highest rated people by treating the rating difference (in case of a loss) as a maximum of 400. However the inverse is not true and players with a huge rating difference do not gain the same amout of points from 1500 and 1900 rated people. You can test that out by yourself and play a couple of beginners from the lobby.
The thing he exploits from what I can understand is the steep learning curve and relatively small player base. Due to the fact that obscure variants are much less explored and understood the rating dynamic is a different one compared to chess.
The thing he exploits from what I can understand is the steep learning curve and relatively small player base. Due to the fact that obscure variants are much less explored and understood the rating dynamic is a different one compared to chess.
I really like the idea to reduce the RD threshold to 60. I guess it could be done for regular chess (except Rapid and Classical) as well.
Why is it a problem at all if someone prefers lower rated opponents? He will probably learn less, but I don't see any immorality in this. Or is there a different etiquette for antichess than for chess?
But don't change that Rd threshold again. It has been reduced from 60 to 50 only recently. If anything, it ought to be reduced further.
But don't change that Rd threshold again. It has been reduced from 60 to 50 only recently. If anything, it ought to be reduced further.
This topic has been archived and can no longer be replied to.