Guidance on padding and "friendly" play

Originally posted by z00t
My suggestion would be :-

- for rated games close family/friends should not be played more than 10 times

I see several problems with this. Firstly, why should close family or friends be treated differently than any other players? I don't play differently if I play a friend than if I play a stranger. There would only be a problem if someone would deliberately let someone else win, which most family members or friends won't do any more than a stranger would, and which a stranger might do if xe gets some favours for it. Secondly, if you think it's important whether people are close family or friends, how would RHP know? Would we have to send in a list of close family and friends? Thirdly, to set a number doesn't make sense at all if you consider that some people play hundreds of games, while others only play very few. If you play 500 games, 10 games against the same person are not very much. It would make more sense to go by percentage of games. Then again, I don't have a problem at all with someone joining RHP to play a couple of friends and nobody else.

I'd like to see some sort of mathematical proof that if two players (A and B) played each other in 1/5th of all their games, and random members of the wider pool in the remaining 4/5ths, that their ratings would be skewed to any significant extent. Presuming A and B are both trying to win, I fail to see how A v B matches are significantly different from A v not-B, or not-A v B matches.

Ratings are only supposed to show your strength relative to the pool you are operating in; they are not an absolute measure, and should not be treated as such (or taken too seriously).

It's really no one's business if two people want to always play each other, or even if they choose to have these games rated; a natural limit will kick in at some point, anyway.

I suppose there MIGHT be a case for not listing the ratings of those who only ever play a handful of opponents alongside others who play in the wider pool. The justification would be that they were, in effect, playing in a different, smaller "sub-pool". But this could only apply in VERY extreme circumstances in which there was a significant mathematical issue. In the 1/5th case, I don't think there could be.

I don't even see why this is an issue anymore. Padding is when user LETS the other user win the game. It doesn't make a difference if I play 1100 players the rest of my life, and I don't see the difference if it's 20 different 1100 players or the same 1100 player, I'll win most, lose some... and for each lost game I'll lose several points. This will all average out to my normal rating wherever that may be.

Padding is when someone is losing on purpose, and that is a job for the game mods to decide. There is no need to limit games against any given user.

P-

Originally posted by dottewell
I'd like to see some sort of mathematical proof that if two players (A and B) played each other in 1/5th of all their games, and random members of the wider pool in the remaining 4/5ths, that their ratings would be skewed to any significant extent. Presuming A and B are both trying to win, I fail to see how A v B matches are significantly different from A ...[text shortened]... ; they are not an absolute measure, and should not be treated as such (or taken too seriously).

Read http://en.wikipedia.org/wiki/ELO_rating_system you do not understand what a rating is for or what rating pools are. You also don't consider the impact of the effect on the rating pool if a significant majority does this.

Originally posted by z00t
Read http://en.wikipedia.org/wiki/ELO_rating_system you do not understand what a rating is for or what rating pools are. You also don't consider the impact of the effect on the rating pool if a significant majority does this.

What are we supposed to read on this page that proves playing the same person over and over inflates a rating?

The only time inflation occurs is if one person is purposely losing the game.

P-

I agree with z00t on this issue....God forbid someone should want to play a lot of games with a friend/relative. 😵

Originally posted by Raven69
I agree with z00t on this issue....God forbid someone should want to play a lot of games with a friend/relative. 😵

Not quite as funny as when it was said in the other thread in General. Run along and play now.

*scruffs hair.

P-

Originally posted by Phlabibit
Run along and play now.

Make me. 😛

Originally posted by z00t
Read http://en.wikipedia.org/wiki/ELO_rating_system you do not understand what a rating is for or what rating pools are. You also don't consider the impact of the effect on the rating pool if a significant majority does this.

I understand the elo system perfectly well; that's why I'm asking you to demonstrate, mathematically, how someone who plays 1/5th of their games against one person - and the rest against random opponents from a pool of thousands - will have a significantly skewed rating compared to other people in that pool of thousands.

To be clear - I am asking for a worked example because I suspect your grasp of mathematics may be rather poor.

Assume both are trying to win their games against each other, and all their other games.

I can start you off if you like. Let's say player A is a strong player who would have an 1800 rating at RHP if his games against player B were not taken into account. Player B is a weak player and would have a 1100 rating if his games against player A were not taken into account. They are friends and actually play each other every fifth game at RHP. The rest of the time they play random opponents from the rest of RHP.

Show me the maths. Show me why Player A choosing to play Player B is mathemetically more advantageous to him/her than it would be to accept an open invite from a random 1100-rated player.

Originally posted by Raven69
Make me. 😛

Oooohhh, you!

😠

P-

Originally posted by dottewell
I understand the elo system perfectly well; that's why I'm asking you to [b]demonstrate, mathematically, how someone who plays 1/5th of their games against one person - and the rest against random opponents from a pool of thousands - will have a significantly skewed rating.

To be clear - I am asking for a worked example because I suspect your ...[text shortened]... to him/her than it would be to accept an open invite from another (random) 1300-rated player.[/b]

That is where the word intentional falls. Padding is someone letting the other guy win every game. Rather hard to prove, unless they are doing the fools mate variations every game.

P-

Originally posted by Phlabibit
Oooohhh, you!

😠

P-

Problem? 🙂

Originally posted by Phlabibit
That is where the word intentional falls. Padding is someone letting the other guy win every game. Rather hard to prove, unless they are doing the fools mate variations every game.

P-

I completely agree, but the point is even if you allowed his (wrong) definition of padding as "repeatedly playing a much weaker friend", the particular case he mentions would not stand up mathematically.

It is true that if I were a weak player and only ever played you here, and you were an absolutely abysmal player who lost every game, my rating would be artificially high compared to other ratings on RHP. But if I only played you every fifth game - and played random players from the whole of RHP the rest of the time - and if you did the same - then my rating, and yours, would be fairly closely in line with the rest of RHP.

Unless I am proved wrong. I'm not claiming to be great at maths.

Originally posted by dottewell
I completely agree, but the point is even if you allowed his (wrong) definition of padding as "repeatedly playing a much weaker friend", the particular case he mentions would not stand up mathematically.

It is true that if I were a weak player and only ever played you here, and you were an absolutely abysmal player who lost every game, my rating ...[text shortened]... ine with the rest of RHP.

Unless I am proved wrong. I'm not claiming to be great at maths.

I believe too many people are too concerned with there ratings. Is that why you are playing or is it for the enjoyment of the game. I could be accused of "paddiding" as I have only played some family members and have won all my games. I expect to win because I know I am a better player than they are and have been for many years. Should I not accept their challenge?

Originally posted by dottewell
I completely agree, but the point is even if you allowed his (wrong) definition of padding as "repeatedly playing a much weaker friend", the particular case he mentions would not stand up mathematically.

It is true that if I were a weak player and only ever played you here, and you were an absolutely abysmal player who lost every game, my rating ...[text shortened]... ine with the rest of RHP.

Unless I am proved wrong. I'm not claiming to be great at maths.

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. That would have a similar effect, wouldn't it? So should that be stopped as well?

To make it all fair and correct, we should probably let a computer program choose the opponents for us, making sure that the range of players you play is perfectly balanced. 🙄