所有案件
{THIS_GAME_BUGS}
#46582: "ELO adaptation to the number of players is screwed badly"
#46582: "ELO adaptation to the number of players is screwed badly"
發生什麼事? 請從下方選擇
本站的某些功能異常(非遊戲過程中發生)
發生什麼事? 請從下方選擇
本站的某些功能異常(非遊戲過程中發生)
請檢查是否已有同課題案件
若肯定,請「投票」給這樁案件。最高票的案件將「優先」處理!
# | Status | Votes | Game | Type | Title | Last update |
---|
細節描述
• 如果可以的話,請轉貼螢幕顯示的錯誤訊息。
n/a• 請說明你當時想做什麼,你做了什麼,然後發生了什麼事
For instance:
boardgamearena.com/table?table=192563240
Hover the mouse over the 1st and and 4th player scores, do the math.
4 players games for downforce, if you do the sum, the ELO gains for the 1st player are multiplied in a hidden manner by 0.75
But the delta for the 2nd and 3rd players are not multiplied and the losses for the 4th player are not multiplied either.
To be 100% clear, this has nothing to do with the K factor. The problem isn't that the sum is not 0, I'm fully aware that this is normal. The problem is how the extra player count adapter is handled AND that it is not even shown or explained.
It is just plain wrong in the example above, as there can be any reason why the first player would have a 0.75 factor but the last player wouldn't.
--------------
But it is also wrong more generally speaking.
For instance, in a 6th player Race for the galaxy game, the player number factor is balanced I think (same for 1st and last player, if I remember correctly) but that still isn't ok. It should really be the same for all players. As it is, it is a big ELO tax on higher ranked players playing with lower ranked people which is not a good thing in terms of website dynamics; you do not want to discourage high ranked players from playing with lower ranked players.
As a 500+ player in a 6 players RFTG against average players, you gain like 4 ELO when winning (a 6 players game) and lose like 12 for being second. That is NOT normal and this because the ELO score for the first player is much much more reduced than the score for the second player.
---------------------
My solution:
1) The hidden players# factor should not be hidden and be DISPLAYED in the ELO change summary.
2) The hidden players# factor should be the same for everyone in the game regardless of their final position in the game.
• 你的瀏覽器是什麼?
Google Chrome v91
• 請轉貼未翻譯的英文字句。 建議將此錯誤的螢幕截圖上傳到 Imgur.com 並轉貼連結。
n/a• 這些文字存在於 翻譯系統 中嗎?若為真,其是否已被翻譯超過 24 小時?
For instance:
boardgamearena.com/table?table=192563240
Hover the mouse over the 1st and and 4th player scores, do the math.
4 players games for downforce, if you do the sum, the ELO gains for the 1st player are multiplied in a hidden manner by 0.75
But the delta for the 2nd and 3rd players are not multiplied and the losses for the 4th player are not multiplied either.
To be 100% clear, this has nothing to do with the K factor. The problem isn't that the sum is not 0, I'm fully aware that this is normal. The problem is how the extra player count adapter is handled AND that it is not even shown or explained.
It is just plain wrong in the example above, as there can be any reason why the first player would have a 0.75 factor but the last player wouldn't.
--------------
But it is also wrong more generally speaking.
For instance, in a 6th player Race for the galaxy game, the player number factor is balanced I think (same for 1st and last player, if I remember correctly) but that still isn't ok. It should really be the same for all players. As it is, it is a big ELO tax on higher ranked players playing with lower ranked people which is not a good thing in terms of website dynamics; you do not want to discourage high ranked players from playing with lower ranked players.
As a 500+ player in a 6 players RFTG against average players, you gain like 4 ELO when winning (a 6 players game) and lose like 12 for being second. That is NOT normal and this because the ELO score for the first player is much much more reduced than the score for the second player.
---------------------
My solution:
1) The hidden players# factor should not be hidden and be DISPLAYED in the ELO change summary.
2) The hidden players# factor should be the same for everyone in the game regardless of their final position in the game.
• 你的瀏覽器是什麼?
Google Chrome v91
• 請簡明精確地解釋你的建議,以便讓人一目了然。
n/a• 你的瀏覽器是什麼?
Google Chrome v91
• 當你不能動作時,螢幕上顯示什麼?(螢幕全黑?部份遊戲介面?錯誤訊息?)
n/a• 你的瀏覽器是什麼?
Google Chrome v91
• 遊戲規則的哪部分在 BGA 版本有所錯漏?
n/a• 這項違反規則之處可否在遊戲重播中看到?若可以是在哪步?(重播時左上角資訊)
For instance:
boardgamearena.com/table?table=192563240
Hover the mouse over the 1st and and 4th player scores, do the math.
4 players games for downforce, if you do the sum, the ELO gains for the 1st player are multiplied in a hidden manner by 0.75
But the delta for the 2nd and 3rd players are not multiplied and the losses for the 4th player are not multiplied either.
To be 100% clear, this has nothing to do with the K factor. The problem isn't that the sum is not 0, I'm fully aware that this is normal. The problem is how the extra player count adapter is handled AND that it is not even shown or explained.
It is just plain wrong in the example above, as there can be any reason why the first player would have a 0.75 factor but the last player wouldn't.
--------------
But it is also wrong more generally speaking.
For instance, in a 6th player Race for the galaxy game, the player number factor is balanced I think (same for 1st and last player, if I remember correctly) but that still isn't ok. It should really be the same for all players. As it is, it is a big ELO tax on higher ranked players playing with lower ranked people which is not a good thing in terms of website dynamics; you do not want to discourage high ranked players from playing with lower ranked players.
As a 500+ player in a 6 players RFTG against average players, you gain like 4 ELO when winning (a 6 players game) and lose like 12 for being second. That is NOT normal and this because the ELO score for the first player is much much more reduced than the score for the second player.
---------------------
My solution:
1) The hidden players# factor should not be hidden and be DISPLAYED in the ELO change summary.
2) The hidden players# factor should be the same for everyone in the game regardless of their final position in the game.
• 你的瀏覽器是什麼?
Google Chrome v91
• 你當時想採取哪個遊戲行動?
n/a• 你想觸發這個遊戲行動時做了什麼?
For instance:
boardgamearena.com/table?table=192563240
Hover the mouse over the 1st and and 4th player scores, do the math.
4 players games for downforce, if you do the sum, the ELO gains for the 1st player are multiplied in a hidden manner by 0.75
But the delta for the 2nd and 3rd players are not multiplied and the losses for the 4th player are not multiplied either.
To be 100% clear, this has nothing to do with the K factor. The problem isn't that the sum is not 0, I'm fully aware that this is normal. The problem is how the extra player count adapter is handled AND that it is not even shown or explained.
It is just plain wrong in the example above, as there can be any reason why the first player would have a 0.75 factor but the last player wouldn't.
--------------
But it is also wrong more generally speaking.
For instance, in a 6th player Race for the galaxy game, the player number factor is balanced I think (same for 1st and last player, if I remember correctly) but that still isn't ok. It should really be the same for all players. As it is, it is a big ELO tax on higher ranked players playing with lower ranked people which is not a good thing in terms of website dynamics; you do not want to discourage high ranked players from playing with lower ranked players.
As a 500+ player in a 6 players RFTG against average players, you gain like 4 ELO when winning (a 6 players game) and lose like 12 for being second. That is NOT normal and this because the ELO score for the first player is much much more reduced than the score for the second player.
---------------------
My solution:
1) The hidden players# factor should not be hidden and be DISPLAYED in the ELO change summary.
2) The hidden players# factor should be the same for everyone in the game regardless of their final position in the game.
• 當你試著這麼做時發生了什麼(錯誤訊息、遊戲狀態條訊息...)?
• 你的瀏覽器是什麼?
Google Chrome v91
• 問題發生在遊戲的哪一步?(當前遊戲指示是什麼)
n/a• 當你試著進行遊戲動作時發生了什麼(錯誤訊息、遊戲狀態條訊息...)?
For instance:
boardgamearena.com/table?table=192563240
Hover the mouse over the 1st and and 4th player scores, do the math.
4 players games for downforce, if you do the sum, the ELO gains for the 1st player are multiplied in a hidden manner by 0.75
But the delta for the 2nd and 3rd players are not multiplied and the losses for the 4th player are not multiplied either.
To be 100% clear, this has nothing to do with the K factor. The problem isn't that the sum is not 0, I'm fully aware that this is normal. The problem is how the extra player count adapter is handled AND that it is not even shown or explained.
It is just plain wrong in the example above, as there can be any reason why the first player would have a 0.75 factor but the last player wouldn't.
--------------
But it is also wrong more generally speaking.
For instance, in a 6th player Race for the galaxy game, the player number factor is balanced I think (same for 1st and last player, if I remember correctly) but that still isn't ok. It should really be the same for all players. As it is, it is a big ELO tax on higher ranked players playing with lower ranked people which is not a good thing in terms of website dynamics; you do not want to discourage high ranked players from playing with lower ranked players.
As a 500+ player in a 6 players RFTG against average players, you gain like 4 ELO when winning (a 6 players game) and lose like 12 for being second. That is NOT normal and this because the ELO score for the first player is much much more reduced than the score for the second player.
---------------------
My solution:
1) The hidden players# factor should not be hidden and be DISPLAYED in the ELO change summary.
2) The hidden players# factor should be the same for everyone in the game regardless of their final position in the game.
• 你的瀏覽器是什麼?
Google Chrome v91
• 請描述顯示問題。 建議將此錯誤的螢幕截圖上傳到 Imgur.com 並轉貼連結。
n/a• 你的瀏覽器是什麼?
Google Chrome v91
• 請轉貼未翻譯的英文字句。 建議將此錯誤的螢幕截圖上傳到 Imgur.com 並轉貼連結。
n/a• 這些文字存在於 翻譯系統 中嗎?若為真,其是否已被翻譯超過 24 小時?
For instance:
boardgamearena.com/table?table=192563240
Hover the mouse over the 1st and and 4th player scores, do the math.
4 players games for downforce, if you do the sum, the ELO gains for the 1st player are multiplied in a hidden manner by 0.75
But the delta for the 2nd and 3rd players are not multiplied and the losses for the 4th player are not multiplied either.
To be 100% clear, this has nothing to do with the K factor. The problem isn't that the sum is not 0, I'm fully aware that this is normal. The problem is how the extra player count adapter is handled AND that it is not even shown or explained.
It is just plain wrong in the example above, as there can be any reason why the first player would have a 0.75 factor but the last player wouldn't.
--------------
But it is also wrong more generally speaking.
For instance, in a 6th player Race for the galaxy game, the player number factor is balanced I think (same for 1st and last player, if I remember correctly) but that still isn't ok. It should really be the same for all players. As it is, it is a big ELO tax on higher ranked players playing with lower ranked people which is not a good thing in terms of website dynamics; you do not want to discourage high ranked players from playing with lower ranked players.
As a 500+ player in a 6 players RFTG against average players, you gain like 4 ELO when winning (a 6 players game) and lose like 12 for being second. That is NOT normal and this because the ELO score for the first player is much much more reduced than the score for the second player.
---------------------
My solution:
1) The hidden players# factor should not be hidden and be DISPLAYED in the ELO change summary.
2) The hidden players# factor should be the same for everyone in the game regardless of their final position in the game.
• 你的瀏覽器是什麼?
Google Chrome v91
• 請簡明精確地解釋你的建議,以便讓人一目了然。
n/a• 你的瀏覽器是什麼?
Google Chrome v91
案件沿革
2021年12月30日 20:26 • jackg_tor • 開發者尚未重現這項錯誤:
@Phoxtrot, I don't think that this is nearly as messed up as you think. As sprokitz said at the end of boardgamearena.com/forum/viewtopic.php?f=9&t=13955&p=49837&hilit=details+of+the+elo+computation#p49837, the final values appear to be correct but the intermediate calculations are not. If the K factor is over 20, it is shown reduced by the #players factor (0.75 for 3-player games, 0.6667 for 4-player and ? for > 4) but if the K factor is 20, that same #players factor is applied to the final sum. Either way, though, the final number works out the same.
為本案件添加內容
任何可能重現這項錯誤或了解你的建議之相關資訊,都請在此填寫:
- 其他同樣狀況的桌號/步數
- 按 F5 是否解決了這個問題?
- 問題是否發生了好幾次?每次都發生?時好時壞?
- 建議將此錯誤的螢幕截圖上傳到 Imgur.com 並轉貼連結。
更改案件狀態為:
bug?id=46582