Ranking up fast in cooperative games

[Ceaseless]

So no ranking at all.
Just a feature that creates new sort of nonsense, absurdities, and hostility, like one described in The Gang, or one that started this topic. Not being able to know what skill represents your teammates. And yet that no system is better for you then something else.

Nonsense, absurdities, hostility and abuses in competitive games are somehow ok. But imagine, not specified one that could be in coop games somehow prevent from even making proposal to create that system.

If you stop thinking that coop games are so much different then competitive ones, in terms of ranking, you could even use ELO one, with obvious modifications. Won't be that good as dedicated one, but much less work and still much better then no system at all.

The notion that cooperative games are no different from competitive ones is one great reason why I find hiding the cooperative ratings to be better at the present moment. You have people believe those numbers actually mean the same thing, and they end up taking them for something they're not. Cooperative games are a whole different beast from competitive games.

Getting points and not being able to lost them even if you lose is not ranking. And coops games are not only same as competitive ones, you still play to win, you still need skill to play better and face harder challenges. So you can have ranking that give estimation of that skill for you and your potential teammates. Exactly as in competitive games.

No, it isn't ranking. It's an acceptance that BGA doesn't have a great system for trying to really rank cooperative games, rather than pretending they do.

Cooperative games are not the same as competitive ones. Your team is playing to win, individual members contributions may vary, in some cases the team may only need 1 player that wants to win and the rest just following their lead. In other games it's about having pre-coded language, and other player synergy headaches. It varies game by game what sort of commitment is involved.

Grading the opposition is a whole different game too. In competitive games the opposition is graded by similar standards, tested against other players in the pool. An artificial system isn't. The very nature of the way the objectives challenge players are completely different and so the way to grade them isn't going to be the same.

[tarongot]

It's an acceptance that BGA doesn't have a great system for trying to really rank cooperative games, rather than pretending they do.

From this, and anything that goes from this I don't agree with you.
You keep looking for factors that could be included to rank skill in cooperative games, and use it as excuse to not actually make one. But those factors aren't need for simple average system, especially if you want one system for all coop games.
Just like you have one simple, not including factors like luck in dices/cards, unbalanced factions, board layout favoring a given player and others, win by 1% point same as win by 100%, and other.

[tbhp]

I mentioned the issues with Chess sites trying to rank bots, but I'd find it even harder to sort out for cooperative games when you add in things like teamwork considerations. Having certain players basically carry the game while others reap the rewards can be a real thing for some games. I'd consider have the ratings affected by who the teammates were, the stronger your teammates are, the less you should gain.

There are several ceiling issues, the commonly referenced boosting, but also the lack of a proper challenge ceiling in some cases. If all of the challenges are easy enough everyone's compared rating is how they just forever grind the wins for lack of something to really filter them. Artificial challenges do lack the natural balancing factor competitive games have as players will throw each other's ideas back and them and continue to refine them.

The difference with Chess is that this is a game that exists in regular competitive mode against humans (that's the normal way it's supposed to be played), so when you start including games against AI and rating players based on that, there needs to be consistency with their rating against humans. The rating they get against AI needs to eventually reflect what their level against humans is, otherwise it's going to create contradictions and people won't take the numbers seriously.

Things seem very different when giving people ratings in a board game that can only be played cooperatively. There is no other measure of skill that one could compare those numbers against. So there is no pressure for the chosen system to output "correct" ratings. People aren't likely to be complaining about it like they would if it was Chess and the ratings looked all wrong.

So while the issues you point could exist, they would be hard to detect and they probably wouldn't be considered a big deal by the community.

The ceiling issue could become notable, but then again the people who play coop games probably know that already if it's acknowledged that there is a more or less optimal way to play against the artificial opponent. So, reaching top Elo in those games wouldn't be seen the same way as we look at it in competitive games, because people would know that past a certain point everyone is at the same proficiency. It wouldn't be an issue with the system, it would just be a characteristic of the particular game and of how good the artificial opponent is at it. People would probably be okay with the ceiling, because they would already be aware it's an inherent part of the game they are playing. Just like how people are okay with Elos being lower when the game is more luck based. It's just what is logical and expected.

[tbhp]

Cooperative games are not the same as competitive ones. Your team is playing to win, individual members contributions may vary, in some cases the team may only need 1 player that wants to win and the rest just following their lead. In other games it's about having pre-coded language, and other player synergy headaches. It varies game by game what sort of commitment is involved.

Why does it matter that different coop games involve different things? People aren't gonna compare the ratings of different games anyway, since they are different games.

We already don't compare the ratings of different competitive games. We know that an Elo of 300 doesn't hold the same meaning in a high luck based game as it does in a low luck based game.

People only ever compare players in the context of the same game, never across different games. And as for the varying contributions issue, that's not anything new. People already complain about that in online competitive games like League of Legends where some will say their ranking is unfair because it's determined mostly by their teammates. But we all know that's not valid, because over time your ranking will reflect your real level, because you are the common denominator in all the games you play.

[FrankJones]

If the players win a co-op game, that should be a win for all players. If the players lose, or quit, or abandon, that should be a loss.

Is that not how the system currently handles this, for example, in Pandemic?

[Ceaseless]

Cooperative games are not the same as competitive ones. Your team is playing to win, individual members contributions may vary, in some cases the team may only need 1 player that wants to win and the rest just following their lead. In other games it's about having pre-coded language, and other player synergy headaches. It varies game by game what sort of commitment is involved.

Why does it matter that different coop games involve different things? People aren't gonna compare the ratings of different games anyway, since they are different games.

We already don't compare the ratings of different competitive games. We know that an Elo of 300 doesn't hold the same meaning in a high luck based game as it does in a low luck based game.

People only ever compare players in the context of the same game, never across different games. And as for the varying contributions issue, that's not anything new. People already complain about that in online competitive games like League of Legends where some will say their ranking is unfair because it's determined mostly by their teammates. But we all know that's not valid, because over time your ranking will reflect your real level, because you are the common denominator in all the games you play.

Implementation. Trying to sort out what on earth you're trying to do when setting up the rating system is a lot harder. I'm not familiar enough with League of Legends, to have any thoughts on that example.

I mentioned the issues with Chess sites trying to rank bots, but I'd find it even harder to sort out for cooperative games when you add in things like teamwork considerations. Having certain players basically carry the game while others reap the rewards can be a real thing for some games. I'd consider have the ratings affected by who the teammates were, the stronger your teammates are, the less you should gain.

There are several ceiling issues, the commonly referenced boosting, but also the lack of a proper challenge ceiling in some cases. If all of the challenges are easy enough everyone's compared rating is how they just forever grind the wins for lack of something to really filter them. Artificial challenges do lack the natural balancing factor competitive games have as players will throw each other's ideas back and them and continue to refine them.

The difference with Chess is that this is a game that exists in regular competitive mode against humans (that's the normal way it's supposed to be played), so when you start including games against AI and rating players based on that, there needs to be consistency with their rating against humans. The rating they get against AI needs to eventually reflect what their level against humans is, otherwise it's going to create contradictions and people won't take the numbers seriously.

Things seem very different when giving people ratings in a board game that can only be played cooperatively. There is no other measure of skill that one could compare those numbers against. So there is no pressure for the chosen system to output "correct" ratings. People aren't likely to be complaining about it like they would if it was Chess and the ratings looked all wrong.

So while the issues you point could exist, they would be hard to detect and they probably wouldn't be considered a big deal by the community.

The ceiling issue could become notable, but then again the people who play coop games probably know that already if it's acknowledged that there is a more or less optimal way to play against the artificial opponent. So, reaching top Elo in those games wouldn't be seen the same way as we look at it in competitive games, because people would know that past a certain point everyone is at the same proficiency. It wouldn't be an issue with the system, it would just be a characteristic of the particular game and of how good the artificial opponent is at it. People would probably be okay with the ceiling, because they would already be aware it's an inherent part of the game they are playing. Just like how people are okay with Elos being lower when the game is more luck based. It's just what is logical and expected.

Reminds me of this fun little piece. https://xkcd.com/2501/ Don't expect much from random peoples' understanding of a given subject. I would not expect people to have a general understanding of what the cooperative elo is supposed to be. Even here, you have people looking for cooperative games to be held as similar to competitive ones, which is obviously not going to work.

[tarongot]

I know ufm has argued in other threads about the +1 model being good for coop games. Meaning that you just get 1 elo for the game no matter the outcome. I was just reading how this is problematic in the game the gang because their is no separation between players from 1-99. So you can’t even exclude poor players from matches w/ advanced settings because it would require 100 games to be labeled as “average”.

I'm pretty sure this is topic You mentioned: https://boardgamearena.com/forum/viewtopic.php?t=32289. So thanks again for that info. I'm glad I found it, from ufm words it seems that Pandemic has exactly simple, average generic system that could be implemented in all coop games. That is consistent with my observations.
System based on same principle a ELO for competitive: to compare points of both players and give/lost points based on that. No more factors are really included in competitive games BGA system.

Reminds me of this fun little piece. https://xkcd.com/2501/ Don't expect much from random peoples' understanding of a given subject. I would not expect people to have a general understanding of what the cooperative elo is supposed to be. Even here, you have people looking for cooperative games to be held as similar to competitive ones, which is obviously not going to work.

Since that is mostly directed at me, I will answer. Seems like "person not having general understanding" is one that used those words in quote above. Seems it is obviously working in Pandemic. And you have gradation of players skill. You have players with 1000 games that are on average - strong ELO level. And you have one with less that have master level. Nothing more you really need.

So only reason that is not implemented in all existing coops, and demanded in new one is Prejudice against cooperative games from players like one quoted, developers like ufm, and possibly BGA itself.

[Ceaseless]

I know ufm has argued in other threads about the +1 model being good for coop games. Meaning that you just get 1 elo for the game no matter the outcome. I was just reading how this is problematic in the game the gang because their is no separation between players from 1-99. So you can’t even exclude poor players from matches w/ advanced settings because it would require 100 games to be labeled as “average”.

I'm pretty sure this is topic You mentioned: https://boardgamearena.com/forum/viewtopic.php?t=32289. So thanks again for that info. I'm glad I found it, from ufm words it seems that Pandemic has exactly simple, average generic system that could be implemented in all coop games. That is consistent with my observations.
System based on same principle a ELO for competitive: to compare points of both players and give/lost points based on that. No more factors are really included in competitive games BGA system.

Reminds me of this fun little piece. https://xkcd.com/2501/ Don't expect much from random peoples' understanding of a given subject. I would not expect people to have a general understanding of what the cooperative elo is supposed to be. Even here, you have people looking for cooperative games to be held as similar to competitive ones, which is obviously not going to work.

Since that is mostly directed at me, I will answer. Seems like "person not having general understanding" is one that used those words in quote above. Seems it is obviously working in Pandemic. And you have gradation of players skill. You have players with 1000 games that are on average - strong ELO level. And you have one with less that have master level. Nothing more you really need.

So only reason that is not implemented in all existing coops, and demanded in new one is Prejudice against cooperative games from players like one quoted, developers like ufm, and possibly BGA itself.

You say Pandemic has a system applicable for all cooperative games, yet it is that very system which Ufm has to insert an edit that no fewer than 3 different games switched out of that method, unless that was an oddly phrased way of saying they were fixing their post rather than noting an update in methodology. Either way it seems this method was not a popular one among the cooperative games. And this was all before BGA realized the system was a disaster and made their update to address that.

You say Ufm was biased against these cooperative games, yet they were one of the developers putting in the work to handle them.

[FrankJones]

I am confused.

Can anyone explain to me in basic terms the optimal way to calculate rankings or Elo for a fully co-op game like Pandemic?

The rankings and Elo for competitive games rely on Elo differential, which does not exist in a fully co-op game. My apologies if this was already explained in this thread.

[Jellby]

There's no optimal way. Whatever you do, many people (I guess more than 50%) will complain.

Some games give a fixed "reward" per game, win or lose. Usually +1, could be +10 or whatever (e.g. The Crew)
Some games give a fixed "reward" per victory. Usually +1, could be +10 or whatever (e.g. Thermopyles when it was solo only, I don't know now)
Some games give a pre-defined "reward" per victory, depending on the settings: higher reward for mode difficult settings (e.g. Regicide)

The main problem with all those games is that your rating never goes down, as long as you don't also get a "penalty" whenever you lose (or abandon or concede) a game.

Some other games do a "fake" elo calculation. They assign some elo to a virtual oponent based on the settings (higher for more difficult settings), and depending on the win/lose result and the players' elo, there will be an increase or decrease of elo at the end (e.g. Hanabi)

Personally, I favor the latter method, it's the only one that provides some evaluation of skill. It has been argued that this leads to toxic environment, cheating and many complaints. Developers and admins will know better, but I'd say those are simply consequences of cooperative games with strangers. If you want to eliminate those issues, the best (only?) way is to remove the game from BGA.