[情報] 新系統-行為警報器,把玩家推回正軌

[colmash (col)]

http://na.leagueoflegends.com/board/showthread.php?t=3388973
原文網址(from riot lyte)
http://i.gbc.tw/gb_img/4/002/663/2663264.png
警告圖
http://lol.zones.gamebase.com.tw/forum/60895/topic/95853189?g=e775
素人自翻

Hey Summoners,


In a future patch, we'll be releasing a feature called Behavioral Alerts.
This system will detect spikes of negative behavior (i.e if a player is
having a particularly bad day or week) and fire an in-game client warning to
the player; we hope that these alerts will give players appropriate feedback
about their behaviors before they travel too far down the negative path. We
believe that these warnings will prevent a significant number of people from
ever being punished by player behavior systems; more importantly, we believe
that these warnings will make a difference in the large number of Neutral and
Positive players that have rare bad days of toxicity, but previously never
got feedback to nudge them back towards the positive.


在未來的更新中,我們將會釋出一個叫作"行為警報器"的新系統.這個系統將會偵測你的負
面行為的尖峰期(像是當玩家在某週或某天發生啥不好的事,並影響在遊戲內的行為),並在
遊戲畫面中跳出警告視窗來通知玩家.我們希望這些景告能給玩家針對之前負面行為適當
的反饋,在他們往完全負面行為前進前拉他們一把.我們相信這樣的警告能告阻止相當多的
玩家不被行為系統處罰,更重要的是,我們相信這個警告能讓因外在因素作出負面行為的玩
家有很大的改變,因為他們之前在被懲罰之前完全不會得到回應.
---


In the past few months, a few key insights have shifted our approach to
tackling player behavior in online communities. For example, we can use
Honor, Reports and other behavioral metrics at our disposal to classify the
types of players in League of Legends into four groups: Positive, Neutral,
Bad, Toxic. It turns out that only 1% of the active playerbase is considered
toxic---these are the players who are abusive to other players in a large
number of their games.

The data suggests that over 92% of players are Neutral or Positive.

At this point, many players are thinking,

“Lyte, you’re a crazy scientist. If more than 92% of players are Neutral or
Positive, why do a high number of my games seem toxic?”


在過去幾個月,我們追蹤觀察線上玩家社群行為的重點有些轉變.舉例來說,我們利用榮譽,
回報和其他行為度量工具來把玩家分成四種:正面,中立,負面,有害.從結果來看只有1%玩
家是真的算是有害類的,這些玩家是真的在遊戲中干擾其他人的玩家

資料顯示大概有92%的玩家是中立或是正面的.

看到這裡,會有很多人想問
"Ltye,你這個瘋狂科學家!如果有超過92%的玩家是中立或是正面的,那為什麼我的遊戲裡
面有這麼多的爛人?"


What we’re learning in the data is that context inside and outside a game
can twist behaviors-it can even create toxic behaviors in good people. A
player might have a bad day at work and get yelled at by their boss, then go
home to play some League of Legends. At this point, the player already has
high frustration levels or has a low tolerance for mistakes, and might end up
raging at his teammates for several games; this pattern of behavior can
create ripples in the community, and cause other players to have a bad
day-this behavioral effect is like a toxin that can percolate and spread
across hundreds and thousands of games. Given 10 players in a game, and the
randomness that comes from every player having a unique circumstance,
tolerance level and context, it’s now easy to see why a high number of games
seem toxic-in each game, 1-2 players might be having bad days in real life,
or in-game. So when a Neutral or Positive player starts down that negative
path, how do we stem it?



我們從資料中得知一件事,遊戲內外發生的事會扭曲玩家的行為,它甚至會讓一個正面玩家
作出負面行為.一個玩家可能在工作上不順利,被老闆兇了一頓,然後回家想要lol一場.在
這個時間點這個玩家心情已經在很糟的等級,而且很難容忍隊友的微小錯誤.這可能會導制
他會對他的隊友連續發怒好幾場.這種行為的模式會在玩家社群內造成擴散,讓其他玩家在
當時有壞的體驗,並讓這類的行為擴散到數百或是數千個遊戲之中.隨機抓10個玩家到同一
場遊戲內,每個玩家當下都有自己獨特的背景心態和容忍程度.你現在發現了為什麼好像每
一場都有爛人的存在.每個玩家都可能在真實生或或是人生中有不順利的時候,所以當有正
面或是中立的玩家開始往負面的道路前進時,我們要怎麼干涉他?


Research suggests that having strong feedback loops can have a dramatic
positive effect on behavior; in this case, the ability to quickly and
effectively deliver feedback to players as soon as they exhibit behaviors
that are out of line with the vast majority of players in League of Legends
could be enough to get these players back on track. An example of feedback
loops in action were Reform Cards, which provided more feedback to players
about why they were being punished by the Tribunal---this feedback improved
average reform rates by 7.8%.


研究顯示當有強力的回饋方式時,對行為會有相當大的正面影響.在這個範例中,快速並有
效的讓玩家知道他們之前的行為是有害並不讓大多數玩家所接受的可能有機會讓這些人回
到常軌.Reform Cards就是一個強力回饋的範例,它提供了玩家為什麼會被懲罰的明確資訊
,這讓玩家行為改善的程度提升了7.8%.


Like many other player behavior features such as the Tribunal and Honor
Initiative, Behavioral Alerts will take into account every player’s
behavioral reputation in the system; for example, there is a lot of math
behind the system to avoid firing alerts for false reports. In addition,
there are numerous mechanisms in place to completely ignore reports from
toxic players or those who abuse the report function to ensure that players
will never receive any alerts from these reports. We’re looking forward to
feedback from players and hope that these alerts will have a significant
impact on player behavior in League of Legends.

This feature is currently live on the PBE. Translations are still being
worked on for about half of the languages, and will be added in the future
before the feature goes live.


就像審判和榮譽系統一樣,行為警告會考慮每個玩家自己在系統中的評價.舉例來說,在這
系統背後藏有大量數學運算來避免因為不正當檢舉觸發警告,也有一些機制來避免因為有
害玩家的回報觸發你的警告,藉此來保證玩家不會收到錯誤簡舉產生的警告.我們期待玩家
的反應,並希望這些警告能告明顯的影響LOL中的玩家行為.

這個機制在PBE上線了,在這個機制完全上線前還有快一半的語言要翻譯,等全部翻完後才
會正式上線.

--
http://goo.gl/H4sdW 遊戲基地lol專區
http://goo.gl/iAyRJ 基地3D全造型預覽
http://goo.gl/W4DbX 基地英雄專頁
http://goo.gl/60OG0 基地lol攻略區,
http://goo.gl/BgfRZ 基地同人創作專區,歡迎投稿!
功能太多列不完啦!!

--
※ 發信站: 批踢踢實業坊(ptt.cc)
推 blackxo4u86 :統神:警告N支
推 LayerZ      :喵的，用遊戲做心理學研究，快去得諾貝爾獎啦
推 joe01231    :感覺是行為心理學家再做某種測驗 LOL
lyte是心理學背景沒錯啊XD

推 kindhearted :lol 心理學研究室....
推 AnEGG       :拳頭用心推
推 thedeathhero:92%是好玩家 另外8%就是爛的 算一下一場十人
→ thedeathhero:換句話說每場都有一個爛人的機會本來就很高呀
推 TarTo       :統神拜拜
推 edwdada     :riot真的很關注遊戲社群的氣氛 相關友情推
推 hihibaby999 :如果有用 那些正向勸架型的玩家怎麼會吃鱉呢???
推 eist        :這已經超越電玩領域了阿 太扯了XD
推 stuppi      :真的 心情不好的情況下玩LOL想要抒發 常常自己或隊友
→ stuppi      :打不好就很容易憤怒或自暴自棄
推 TZUKI       :感覺這樣發展下去就會出現sibyl system了
推 zero990314  :這是什麼神系統....
推 bellzebub   :有意義給推~
推 lpca        :瘋狂科學家XD
→ larcenciel2 :所以說92%其實也是蠻容易變成8%的,那8%的..沒道理阿
不對,是說92%的人,可能因為被老闆罵完後,行為跟8%的人一樣.
這種行為要快點阻止,不然可能會 一去不回+擴散出去

推 sparking    :打遊戲是要抒發情緒不是新聞 但能統計出來就神了..
→ sparking    :只是好奇依據的是什麼?
→ lpca        :有時本來好好的 結果有人嘴 大家就會開始吵起來
然後因為這場A吵到BCDE,B和E下一場會去吵其他人=>幾百幾千場都在吵,G_G

推 aresa       :這也符合長尾理論
→ lpca        :一人的行為蠻可能導致 整場遊戲的氣氛變差
推 fffffffox   :92%是好玩家，所以樓上至少有一名爛人?
我覺的爛人也不一定是每一場都爛,真的毒害玩家只有1%..XD

→ memy        :LOL版的Psycho-Pass… 有點威
→ lpca        :那1%我記得都超級糟的XD
※ 編輯: colmash         來自: 219.85.64.11         (04/30 15:46)
→ lpca        :就跟本板一樣 應該也有1%的釣客或是來亂的 哈哈
推 sonety      :這是心理醫師喔XD
推 potionx     :=====系統警告：不要隨便發怒了你去洗洗睡吧=======
推 edwdada     :簡單就是這段時間有玩家的回報會讓你知道 然後會過濾
→ edwdada     :釣客
推 qwertyuiop22:一場比賽遇到都是好玩家的機率根本不到五成阿
推 luckyman8078:超有用!!!
→ potionx     :如果92%算起來10個有9.2個，剩下0.8個可能糟糕，很高
→ potionx     :中立不知道多少比例是會回罵，對罵，很快素質就下降
→ qwertyuiop22:所以一場遊戲有負面玩家的機率根本高過五成= =
→ potionx     :如果算的很籠統，那平均起來大概是8成...
→ Vladivostok :反烏托邦系統XD
推 leo255112   :心理學家果然厲害@@
推 bird10259   :人本來就是這種習性阿 一個銅板敲不出聲響
推 leyi12      :系統不停告訴統神 你以前隊友要跟你說 洗洗睡了
推 qwertyuiop22:簡單算一下，遇到負面隊友的機率大概有三成
→ zeldazefac  :毒害的人竟然佔1% 超高的  假如有20萬玩家....
推 samhou6     :有點扯=口=
推 iamalam2005 :我以為要分成混亂善良、守序邪惡、中立...
推 ben91lin    :可能剛好也有92%的人是中立及負面的XD???
→ zeldazefac  :就有2000個毒害菁英
推 potionx     :其實要算上遊戲場數那些才準確，如果那1%玩整天的...
→ fifth       :戳率 = 1 - (0.92)^4
推 k020238     :為了不想被破壞體驗，只好跳出來當那8%
→ bloodknight :"景"告 > 警告
推 e031charles :1-0.92^9=0.5278..  遇到負面玩家的機率在一半以上
推 EvilPrada   :很多時候只是懶得檢舉那些嘴砲  並不是嘴砲只那麼少
→ vn509942    :這警報整場歐一歐一...肯定是它壞了(￣ㄧ￣ )
推 gekisen     :有心推
→ kerry0496x  :推  好人很容易變壞人  這是事實
→ kerry0496x  :而這系統就是要讓好玩家不要輕易變成壞玩家
推 sephiroth323:這8%是怎麼算的@@，逆風跟順風也一樣嗎?
推 catsondbs   :台灣神人Lyte
→ catsondbs   :      Ltye
推 michael0728n:這是真的還假的XDDDD
推 vi000246    :所以1%的人害到其他四個人 那四個人再害到更多人
→ kolubilly   :我覺得禁言才是根治之道。
推 widec       :真的假的 好像有點威
推 CaptainTeemo:不能在自家溫泉插眼寫on9了，哭哭
推 Jwar        :goood
推 suikanta    :專業戳戳可以掌控好次數避免被禁了...
推 z1233218    :錯字很多但是翻譯機給推
推 ChronoGate  :警告! 警告! 你的色相混濁了! (炸成肉醬
→ invander    :危險期？
推 evilture    :警告 偵測到12級幽能波型
推 ms0540017   :這篇感覺都可以寫篇paper找期刊發表了
→ luxifir     :有害玩家XD
推 magickevin  :怎麼不出遊戲馬上顯示被幾個人檢舉 這樣嘴砲一定很乖
推 icyfang     :Lyte應該是認知心理學家不是心理醫師，榮譽系統跟
推 tonyy801101 :如果只是文字的話 感覺功效有限
→ icyfang     :審判系統都是建構在心理學的機制下:D
推 rotusea     :太強了
推 hardyuse    :推  已收入情報區