Are we ever going to change the system to stop hurting honest players vs quitters?

[RazorOz]

Maybe having one mega thread on this issue instead of countless small ones could bring more attention? This topic pops up all the time. I don't think anything will change but good luck. These are the accounts in which you will find opposition.

The "I don't care about elo and none of you should either." accounts

The "How dare you insinuate bga is anything less than perfect?" accounts

The "Just use the red thumb." accounts (they expect you to play hundreds of games and find every bad apple on the site) I'll put the "they got -karma what more do you want" accounts here also

The "It isn't a problem because your elo will get back to where it should be." accounts (I completely disagree w/ this. As you've stated you had this happen to you six times recently. Meaning your elo isn't a true representation of your skill because it is only accounting for your recent losses)

The "Bga compensates you properly in this situation." accounts - They may even direct you to the f.a.q page which says you get the % of elo depending on game progress. What they fail to understand is that most games inaccurately depict the correct game progress (mostly on the short side) and that with multiple players you lose elo to all players besides the quitter. So basically you get next to nothing

"I don't care about elo and none of you should either." isn't really opposition, because it's a strawman, not actually addressing any of the actual points. Quitters ruin the gaming experience and make games less fun, is there anybody here who's going to argue they're not here to have fun?

The red thumb system is fine, albeit we've heard in other contexts it's being abused (to avoid the best players in Arena), but it's nowhere near enough to deal with the problem. Even if you play 100s of games you won't come close to finding every bad apple, I wouldn't be surprised if I've red thumbed 100 players, and yet still they keep coming. And this is with me changing my setting as much as possible to try and avoid these players in the 1st place, I have my reputation levels set to >75%, or more than >85% if it's an invitational game, I untick beginners and apprentices from being able to join the game, not because I want to avoid them, but because I realised before I did this that 80% of the quitters were beginners or apprentices. I still cannot avoid a plentiful flow of people wanting to ruin the game though.

The thing with ELO purely relating to this issue (because obviously agree with you on the recent changes to the ELO system, and have joined the calls against it in that mega thread on the topic) is it disproportionately affects the best players at games, who are nearly all premium subscribers and some of the most active players. The reason being if you're player who wins the majority of your games, but will never quit when you lose, it means when your game is voided because of a quitter you're mostly going to be losing wins, while rarely is it voiding your losses. On top of that because of what I said about how ELO neutralisation currently works, it's literally impossible for you to gain ELO from a quitter in multiplayer games, even if the game was like 97% progression you will still gain 0.

[ChiefPointThief]

"I don't care about elo and none of you should either." isn't really opposition, because it's a strawman, not actually addressing any of the actual points. Quitters ruin the gaming experience and make games less fun, is there anybody here who's going to argue they're not here to have fun?

The red thumb system is fine, albeit we've heard in other contexts it's being abused (to avoid the best players in Arena), but it's nowhere near enough to deal with the problem. Even if you play 100s of games you won't come close to finding every bad apple, I wouldn't be surprised if I've red thumbed 100 players, and yet still they keep coming. And this is with me changing my setting as much as possible to try and avoid these players in the 1st place, I have my reputation levels set to >75%, or more than >85% if it's an invitational game, I untick beginners and apprentices from being able to join the game, not because I want to avoid them, but because I realised before I did this that 80% of the quitters were beginners or apprentices. I still cannot avoid a plentiful flow of people wanting to ruin the game though.

The thing with ELO purely relating to this issue (because obviously agree with you on the recent changes to the ELO system, and have joined the calls against it in that mega thread on the topic) is it disproportionately affects the best players at games, who are nearly all premium subscribers and some of the most active players. The reason being if you're player who wins the majority of your games, but will never quit when you lose, it means when your game is voided because of a quitter you're mostly going to be losing wins, while rarely is it voiding your losses. On top of that because of what I said about how ELO neutralisation currently works, it's literally impossible for you to gain ELO from a quitter in multiplayer games, even if the game was like 97% progression you will still gain 0.

I got a chuckle out of seeing a consecutive win deduction on the 1 person you should've gained positive elo from At least you received 0pts and didn't lose elo like what has been happening to others.

Another issue is that when a 100elo player or lower quits it is impossible to subtract their elo. There are players who notoriously enter lots of games that they can't finish and ruin elo for thousands of other players in the process while not taking a loss themselves.

There are some games that seem to generate these type of people more than others. While waiting on bga to fix this problem which could be never there are some things you can try in the meantime. I am unaware of the heat community but you may have to adjust your settings more. If it is a issue with the clock like how stone age age has longer turns towards the end of the game and the clock doesn't properly adjust you can put the speed to the slowest. If you can you need to get rid of average players as well. Try befriending players in the community. If you are in a game with at least 1 or two other certified players it limits the odds of someone entering the game who will engage in dirty tactics.

[h_illes]

(I'm copying my comment from another topic on the same issue.)

The ELO calculation for multiplayer games is currently carried out for each pair of players in the game, then added up for each player. So if player D quits, player A has a tie with player B, a tie with player C and a win over player D, which altogether might end up negative due to the ties.

As far as I can tell, the following fix should be pretty good: if a player quits and the game ends, only do the pairwise calculation between the quitter and the other players, but do not do the calculation among the remaining players. That way, everyone would gain some ELO from player D, but between players A, B and C, there would be no exchange of ELO.

[Tisaac]

(I'm copying my comment from another topic on the same issue.)

The ELO calculation for multiplayer games is currently carried out for each pair of players in the game, then added up for each player. So if player D quits, player A has a tie with player B, a tie with player C and a win over player D, which altogether might end up negative due to the ties.

As far as I can tell, the following fix should be pretty good: if a player quits and the game ends, only do the pairwise calculation between the quitter and the other players, but do not do the calculation among the remaining players. That way, everyone would gain some ELO from player D, but between players A, B and C, there would be no exchange of ELO.

That solution would still imply that if you are losing and you have the highest elo, you better try to boot someone to avoid big loss

[h_illes]

That solution would still imply that if you are losing and you have the highest elo, you better try to boot someone to avoid big loss

That's true, 1st player would not get ELO from 2nd. But they wouldn't tie either, so at least there will be no ELO loss for the winner. There also has to be a player to boot, which I imagine does not happen every game. But apart from those points, yes, it's not a perfect solution. The trouble is, 'losing' might be clear in some specific situations, but it seems very hard to define in general, especially when the game is not in the last 1-2 turns.

[alexwjb]

(I'm copying my comment from another topic on the same issue.)

The ELO calculation for multiplayer games is currently carried out for each pair of players in the game, then added up for each player. So if player D quits, player A has a tie with player B, a tie with player C and a win over player D, which altogether might end up negative due to the ties.

As far as I can tell, the following fix should be pretty good: if a player quits and the game ends, only do the pairwise calculation between the quitter and the other players, but do not do the calculation among the remaining players. That way, everyone would gain some ELO from player D, but between players A, B and C, there would be no exchange of ELO.

This is so obviously the correct solution I have no idea why it isn't implemented.

[alexwjb]

(I'm copying my comment from another topic on the same issue.)

The ELO calculation for multiplayer games is currently carried out for each pair of players in the game, then added up for each player. So if player D quits, player A has a tie with player B, a tie with player C and a win over player D, which altogether might end up negative due to the ties.

As far as I can tell, the following fix should be pretty good: if a player quits and the game ends, only do the pairwise calculation between the quitter and the other players, but do not do the calculation among the remaining players. That way, everyone would gain some ELO from player D, but between players A, B and C, there would be no exchange of ELO.

That solution would still imply that if you are losing and you have the highest elo, you better try to boot someone to avoid big loss

But this argument works the other way as well. If I'm a lower elo player in a losing position, I'm incentivised to boot in order to "tie" the higher elo player.

Ultimately the principle should be: you should never be hurt if someone else quits the game. The suggestion that the only elo calculations are made between the players remaining in the game and the quitter is the best imo.

[ChiefPointThief]

(I'm copying my comment from another topic on the same issue.)

The ELO calculation for multiplayer games is currently carried out for each pair of players in the game, then added up for each player. So if player D quits, player A has a tie with player B, a tie with player C and a win over player D, which altogether might end up negative due to the ties.

As far as I can tell, the following fix should be pretty good: if a player quits and the game ends, only do the pairwise calculation between the quitter and the other players, but do not do the calculation among the remaining players. That way, everyone would gain some ELO from player D, but between players A, B and C, there would be no exchange of ELO.

That solution would still imply that if you are losing and you have the highest elo, you better try to boot someone to avoid big loss

But this argument works the other way as well. If I'm a lower elo player in a losing position, I'm incentivised to boot in order to "tie" the higher elo player.

Ultimately the principle should be: you should never be hurt if someone else quits the game. The suggestion that the only elo calculations are made between the players remaining in the game and the quitter is the best imo.

The solution above is better than what we have now but it isn’t a fix. If I’m an expert and some average player quits this means I still only get 1.something elo. Are either of these possible solutions?

There are a few options. One is that all non quitters tie and split elo from the losing player equally. Also note in this situation that the quitter is losing 10 elo for the karma hit that goes nowhere. Instead those points can go to the players. So in a 4p game each player is already guaranteed 3.33 elo in addition to there equal share of whatever elo is split evenly. To be clear if a player has increased k factor they should still get their equivalent.

Another is depending on the game it can just carry on with the quitter coming in last and play resumes as normal.

[Ceaseless]

(I'm copying my comment from another topic on the same issue.)

The ELO calculation for multiplayer games is currently carried out for each pair of players in the game, then added up for each player. So if player D quits, player A has a tie with player B, a tie with player C and a win over player D, which altogether might end up negative due to the ties.

As far as I can tell, the following fix should be pretty good: if a player quits and the game ends, only do the pairwise calculation between the quitter and the other players, but do not do the calculation among the remaining players. That way, everyone would gain some ELO from player D, but between players A, B and C, there would be no exchange of ELO.

That solution would still imply that if you are losing and you have the highest elo, you better try to boot someone to avoid big loss

If the higher rated player was losing, they'd still be incentivized to kick the other player from a strictly ratings perspective so they'd lose less rating for a tie than for a loss. This update would not change their motivation as far as rating goes, it just makes the set-up make more sense.

[Ceaseless]

That solution would still imply that if you are losing and you have the highest elo, you better try to boot someone to avoid big loss

But this argument works the other way as well. If I'm a lower elo player in a losing position, I'm incentivised to boot in order to "tie" the higher elo player.

Ultimately the principle should be: you should never be hurt if someone else quits the game. The suggestion that the only elo calculations are made between the players remaining in the game and the quitter is the best imo.

The solution above is better than what we have now but it isn’t a fix. If I’m an expert and some average player quits this means I still only get 1.something elo. Are either of these possible solutions?

Yes, higher rated players should still be getting their standard minimal gains for a lower rated player's loss. The one adjustment I do think should go in their favor is removing the nonsensical percentage ratings BGA has. The quitter lost, full points should be awarded accordingly.

There are a few options. One is that all non quitters tie and split elo from the losing player equally. Also note in this situation that the quitter is losing 10 elo for the karma hit that goes nowhere. Instead those points can go to the players. So in a 4p game each player is already guaranteed 3.33 elo in addition to there equal share of whatever elo is split evenly. To be clear if a player has increased k factor they should still get their equivalent.

Definitely do not want that. You should get what you get for beating the player naturally, no more and no less. So those with more rating get fewer points and those with less rating get more points relative to each other. The rating the quitter lost for the karma deduction is also just a net loss, that's a penalty for them, not rating to be transferred to other players.