Boards›Player Behavior

Overwatch system: The way to end griefing

Mudbloods·12/21/2019, 8:31:55 PM·3 votes·2,239 views

Dear Riot, After verbaly abusing the griefers and trollers that i seemed to encounter every single game (which one year ago got me perma banned) i started thinking about ways to catch these griefers who, in my opinion, are the ones who make the community toxic. Then i remembered the "Overwatch" function in CSGO. For those unfamiliar with the system, it basically lets experienced players review gamefootage to see if reports of hacking are actually true. Have you ever thought about implementing this for League? This is how i see it working:

You will have a computer function reviewing every report of griefing/inting very roughly (meaning that every case of griefing/inting will be caught, but there might also be some instances where it is not actually griefing slipping through).
Players above gold 1 with an honor level of at least 4 will be able to review games that have been flagged by the system.
After reviewing the game, the player will either flag the play as "not grief", "definetly grief" or "unsure"
After a game has been reviewed 10 times, and have been rated "definetly grief" 9 out of 10 reviews (with "unsure" not counting as a review) it will be sent to a Rioter, who will review the game and be able to punish the player accordingly.
To get players to review games, there could be made a weekly mission of reviewing 5 games which will grant an EXP or BE reward. Please let me know what you think about this proposition - constructive criticism is welcomed with open arms. And if you too would like a less toxic environment, please make sure to upvote this post. Sorry for my bad English Also: Please stop calling bad plays for "int" plays. Inting is short for intentional feeding and just because you play bad does not mean that you int. // Mudbloods - a concerned League player.

16 Comments

Imperial Pandaa12/21/2019, 8:54:14 PM10 votes

This idea comes up a lot so will just quickly address some points.

It is easier to judge CS:GO for griefing than it is for League. The community is largely divided on even defining exactly what griefing is. Is my Reksai support griefing or simply off meta? Is my AD Donger griefing or offmeta? Just some examples.
Honor resets at the beginning of the season. Meaning a few months won't have anybody reviewing.
Rank and Honor level does not mean good judge of griefing.
The Tribunal existed for just chat logs and I don't think it had a restriction and it was still months behind. This would be more so.

NF Remilia12/21/2019, 9:33:23 PM2 votes

And who is gonna judge and how? Imagine if iron player can judge challenger player. I notice almost all of the adc below plat thinks the support is trolling if they roam.... From rank to rank has huge difference. It will have easy cases but most of them will be hard to decide. Unexperience players can think its grief but someone that actually know what is he doing will not. Again my adc wanted to report me because i warded river (when we pushed enemy to their turret) imagine him judging challenger support and be like " oh yeah he leave his adc for 10sec thats afk , inting , griefing .. Perma ban this guy!" ... The system we have right now is not perfect , but it gets the job done and its way better than people's suggestions.

Kei14312/21/2019, 10:50:37 PM2 votes

{quoted}

9 out of 10 reviews (with "unsure" not counting as a review) it will be sent to a Rioter, who will review the game and be able to punish the player accordingly.

Fun fact, even without player's input, there already is a Rioter (infact, teams of them) reviewing disruptive gameplay reports. The hard thing is to be able to distinguish between intentional disruptive behavior and bad decisions/plays cos someone is either tilted, distracted, or just bad in general.

The process for disruptive gameplay reviews/punishments are already slow without the player's input. It will be even slower with the player's inputs, and speed was one of the main reasons why the tribunal got shut down.

Also, from what we see on these boards, people whom complain about intentional disruptive gameplay behavior happening close to every game typically are the ones that have it wrong.

FioraWillCarry12/22/2019, 3:19:11 AM2 votes

This is a bad idea simply because I will try to get every feeder in my game punished using this system. I will be biased as fuck. Every single time I get some noobshit going 0-10 in my games I will write that fucker up for griefing. I sincerely doubt you want somebody like me getting you banned for doing bad one game.

KFCeytron12/21/2019, 10:58:40 PM1 votes

Many years ago, LoL's behavior system used something called the Tribunal, comprising player volunteers who logged into a system that showed them chat logs from reported players. Those volunteers would then vote on whether to punish or pardon the reported player's case. This system's main flaw was that it simply took far too long due to LoL's huge playerbase. Participants would often be reviewing chat logs that were several months old, with a growing backlog. In addition to that, not every participant took the task seriously: some would spam the same verdict for every case without even reading the chat log, or even invert their verdicts on purpose.

The current system, called the Instant Feedback System, or IFS, is automated software that uses machine learning to determine what behavior should be punishable and when a player's behavior should be punished. It started with data from the Tribunal, and has been learning and adapting for years. It operates on the same basic principles as a spam filter: get a corpus of data (emails/chat logs), have humans categorize each item as acceptable or unacceptable, find patterns within each category, and then finally look for those patterns to automatically categorize new items without direct human evaluation. Each report is like clicking the "spam" button. When a new pattern starts to get lots of reports, the IFS recognizes it as a new form of toxicity.

The IFS is efficient and unbiased. The Tribunal was not.