Difficult
Possible ratings correction to stop FFA teaming
Sort:
I don't enjoy the idea.
Dec 13, 2019
0
#84
I don't enjoy it either. +3 doesn't give you enough rank, for the compensation of -1.
For -1 loss, it should be +4 for 1st minimum.
Why don't we just accept that game theory is part of the game, and as long as there is no direct collusion then that's cool? you're always gonna get the guys in the lead getting ganged up on by the guys below in the game as that is just game theory. Accept it it's what make 4p chess fun
Jan 8, 2020
0
#88
Teaming is not only in the FFA, but also in Solo. Usually, the opposites take first and second places. Can make the second place (usually the winner’s assistant) less profitable than the fourth place (the victim of teaming): 3 -1-x -1 -1 + x (for example 3 -1.5 -1 -0.5). This will possibly make the 4 player stage longer. It will also help to punish players who spoil the game by killing one of the players, while having a weak chance of winning against the remaining player(s) (and then proud of second place).
Arseny_Vasily wrote:
Teaming is not only in the FFA, but also in Solo. Usually, the opposites take first and second places. Can make the second place (usually the winner’s assistant) less profitable than the fourth place (the victim of teaming): 3 -1-x -1 -1 + x (for example 3 -1.5 -1 -0.5). This will possibly make the 4 player stage longer. It will also help to punish players who spoil the game by killing one of the players, while having a weak chance of winning against the remaining player(s) (and then proud of second place).
2nd < 4th sounds kinda weird. I wonder if we can give 1st more award in-game, such as enabling take-over (all pieces from the player you mated changes to your color) or just a "first blood" reward which gives 1st checkmate 15 more pts.
Feb 20, 2020
0
#91
1st checkmate premium is a very interesting idea.
And as for rating calculations, I recently suggested +3 / +1 / -1.5 / -2.5. 2nd still has something, but much less than 1st; the differences between 3rd and 4th and 2nd and 3rd are not huge and ppl thus are not encouraged to teaming nor to avoid being 4th at any cost, and would normally all play for win.
This is not weird. It's like a poker game. You may not risk: discard cards and lose less money, or risk supporting a bet and fight for victory, but in case of defeat lose more money.
This also corresponds to the logic of the game. Killing a player (at the stage of 3 players, this is more obvious here), you must be sure that you will defeat the remaining player 1 vs 1 (or have equal chances), if this is not so, then in essence you killed your last chance of victory and the possibility of victory for the killed player, t .e. you ruined the game for two players (yourself and the killed), this indicates your stupidity, and not that you play better than the player you killed.
This is less obvious with the first killed at the 4 players stage, but I think it works the same way (because the fewer players left, the more it is clear who has the advantage).
Here are my immediate thoughts about the 3 -1,5 -1 -0,5 idea:
It has some merits in that it makes it less bad to be the first player to die. However, it makes it more unclear what you're supposed to do in general, especially in the 4 player stage, because it's no longer purely good to get points. A player that doesn't believe he can win anymore can make sure he gets 4th and then resign, whereas in 3 -1 -1 -1 you have nothing to lose by playing on.
As for the 3 player stage, it might become more grindy and passive because nobody wants to try to win unless they they feel confident they will win.
Imo there's a strong correlation between playing well and being able to get a lot of points.
(This idea also makes me wonder what would happen if we made a completely inversed scale like -3 -1 +1 +3. All 4 players resign on move 1?)
About +3 +1 -1.5 -2.5, what makes this different from the old +3 +1 -1 -3? 3rd and 4th are closer to each other and very different from 2nd, so all the more reason to play like Teams and eliminate the flank players.
>> However, it makes it more unclear what you're supposed to do in general, especially in the 4 player stage, because it's no longer purely good to get points.
In any case, you need to earn points, all the same, everyone plays for first place, the difference between fourth and second place is a secondary thing. For the most part, this should influence the decision to checkmate or assist the checkmate (at the 3-player stage), this will be done by the player more reasonably.
>> A player that doesn't believe he can win anymore can make sure he gets 4th and then resign, whereas in 3 -1 -1 -1 you have nothing to lose by playing on.
Usually a player who no longer believes in his victory has small number of pieces and points and just stands and waits while others fight among themselves. While he is waiting, he is unlikely to earn more points, so he does not lose anything from expectation.
>> (This idea also makes me wonder what would happen if we made a completely inversed scale like -3 -1 +1 +3. All 4 players resign on move 1?)
Such a game already exists, called antichess
Feb 20, 2020
0
#95
Ok, so what about +4/0/-1.5/-2.5?
No more big difference between 2nd and 3rd and between 3rd and 4th. But 2nd still is better (so it's not -1 -1 -1).
A big incentive to play for win.
No incentive at all to be 2nd and therefore probably no teaming at all anymore (at least in anonymous mode). Especially once one of the players gets eliminated.
Indipendenza wrote:
Ok, so what about +4/0/-1.5/-2.5?
This is almost 3/0/-1/-2 which was formerly known as WTM.
Feb 21, 2020
0
#97
Ok, thx, and what were the results?
The general goal I suppose is:
- to limit/eliminate blatant teaming,
- to remove unusual incentives (like: "I do whatever I can to avoid being 4th because it hurts),
- to decrease dirty playing,
- to make the game fluid and logical.
At 3 0 -1 -2, there will also be teaming at the 4-players stage, since the second is better than the third and fourth. At the 3-player stage, playing for second place also makes sense, especially for a player who is left without opp.
Feb 21, 2020
0
#99
Well, that's inevitable. Otherwise we kill FFA once for all and we only play Solo, with no difference between 2nd, 3rd, 4th.
The problem being there that instead of trying to win, people simply cooperate to prevent ONE guy who seems momentarily stronger, to win. That's like in the famous Russian short story about the crabs, you necessarily know it. Вместо того, чтобы выбираться вместе, крабы будут пытаться помешать тому, кто первый.
We're simply obliged to still create incentives for being 2nd and to some extent 3rd. But not too strong, as it would encourage blatant teaming otherwise. That's why I wrote that the differences 2/3 and 3/4 shouldn't be too big; and 2nd should still maybe earn something.
What about , +3.5/+0.5/-1.5/-2.5?
It's still very important to be 1st; still not very interesting to assist the 1st in a hope of being 2nd as being 2nd gives almost nothing (but still it's not 0 like in the other model); being 3rd is not a tragedy; and being 4th is not that different from being 3rd that ppl could be pushed into a perverse and anti-game strategy "now I play whatever I can to avoid being 4th because the difference between 3rd and 4th is too large"...
Solo is +4 -1.33 -1.33 -1.33 wins/losses vs average. and yes it's hard to gain rating vs lower rated, but the same applies to FFA (or any game using elo/glicko for that matter)
We are currently using average of the 3 opponents individually for each player, meaning the 'average opponent' is different for each player. We are changing it to be 'average of all 4' and hence the same for all players. that should be balanced in terms of elo/glicko because it's now total 4 wins and 4 losses vs the same rating, not different ones.
This will make the system a bit milder, for example a solo win 1800 vs 3 1200s will be 4 wins vs 1350 (avg of all 4) rather than vs 1200 (avg of 3 opps). the losers will get 1.33 losses vs 1350 rather than vs 1400.