With this new punishment system....

Arcade Andrew·May 21, 2015·2 votes·642 views

never going to talk again
keeping all chat enable and team unmuted
no matter what they say I wont chat unless they do something good in which ill say "gj x" or "gj team" those will be the only two phrases I say (aside from "GGWP/GLHF")
No matter how nice they are or how good they do If I see them do ONE thing negative such as using a curse word "fuck" I am reporting them

now that theres a new punishment system that relies on scanning for code words and ignores context riot is saying its ok to report people for saying "fuck" even if its "fuck yeah!"

24 Comments

OhBoyItsaMegamanMay 21, 20156 votes

[{quoted}](name=Arcade Andrew,realm=NA,application-id=cIfEodbz,discussion-id=jIg8mFGE,comment-id=,timestamp=2015-05-21T20:18:32.379+0000) 4) No matter how nice they are or how good they do If I see them do ONE thing negative such as using a curse word "fuck" I am reporting them

That will reduce your report weight to nothing, and the system will ignore your reports.

now that theres a new punishment system that relies on scanning for code words and ignores context

No, it doesn't.

riot is saying its ok to report people for saying "fuck" even if its "fuck yeah!"

No, they're not.

Arcade AndrewMay 22, 20151 votes

Actually thats exactly what the new system does,

(read the comments on the link on the homepage menu before you downvote lol)

they said it scans for key words (such as "GG easy") and when it finds those words it punishes you for them

disregarding context whatsover

meaning my reports will all be valid

OhBoyItsaMegamanMay 22, 20154 votes

Directly from Lyte on reddit: http://www.reddit.com/r/leagueoflegends/comments/36sgex/soo_anybody_gotten_banned_yet_willing_to_share/crh5d7s

The system doesn't just look for a list of words, and in fact tries to learn what is negative or positive over time and creates it's own list of words and phrases. Every time a player reports or honors someone (and in the future, every vote in the Tribunal system), the machine learning engine will try to learn from the data and see if it needs to update what a word or phrase means, or what the context is. If you want more information about the methodologies you'll want to read up on "machine learning," and "NLP or natural language processing." We've basically combined all of these methodologies with player feedback (which in League is billions and billions of data points) to create this system.

And http://www.reddit.com/r/leagueoflegends/comments/36ow3s/new_player_reform_system_heads_into_testing/crh580c

Context is one of the hardest things for machine learning system to solve, and we're doing our best to improve the system's ability to understand context. The system is capable of learning over time too, so if "easy gg" becomes a positive phrase in the future, players will no longer be punished for it. We're hoping to get the system to a point where it can learn phrases and how the meanings change over time, and learn the contexts for phrases and how that changes over time.

I'm not thrilled that you have to go to Reddit to find this information, but it is quite clear that a.) this isn't a dumb system that issues blind punishments based on words absent of context, and b.) if no one else but you is reporting people for saying "fuck yeah", the system will never consider that a punishable phrase.

Arcade AndrewMay 22, 20151 votes

I mean I saw several red comments on the main website article comment section saying it would scan for certain words to deem if its a false report or not for phrases such as "GG easy"

OhBoyItsaMegamanMay 22, 20152 votes

[{quoted}](name=Arcade Andrew,realm=NA,application-id=cIfEodbz,discussion-id=jIg8mFGE,comment-id=0001000000000000,timestamp=2015-05-22T20:32:26.552+0000)

I mean I saw several red comments on the main website article comment section saying it would scan for certain words to deem if its a false report or not for phrases such as "GG easy"

That's because players in general consider "GG easy" to be an offensive phrase. People get reported for saying "GG easy". When Tribunal was up, Tribunal judges often punished players for games that they ended with "GG easy". When Riot manually reviews accounts to issue restrictions or time bans, "GG easy" can earn you a punishment.

If you don't like the fact that this system identifies "GG easy" as a potentially punishable phrase, then you don't have a problem with the new punishment system. You have a problem with the League of Legends playerbase.

Arcade AndrewMay 22, 20151 votes

you misunderstood the post I said that the reds said it SCANNED for phrases (such as "gg easy")

and punishes them if they find said word or phrase

so you could be saying "ggeasy" to your team (the enemy team would never see it, being gg easy is toxic tword the enemy team) or you could say "gg easy game boys!" in team chat) the enemy reports you because you said "fuck yeah :p" when somebody asks you if you play udyr alot, somebody else got offended by the word fuck and reported you or maybe your team did, system scans. sees "gg easy" cant determine context (that is want in all chat and wasnt ever said to the enemies only your team) and punishes you

also lets note something here: why are we allowing a community of children create the rules a community that gets butthurt over anything and then we send a robot to ban/punish people for no abiding to 10 year olds sensitive nature

league is failing in this concept because they dont have a clear defined set of rules, they just said "lalala let the players decide"

which means that in theory whats "positive and toxic" can continually change, meaning one day "gg easy" could be seen as positive, by the community but the system hasnt learned that yet so you see people getting punished for it and vice versa, saying "GGwp" could be deemed toxic by the community but its honorable in the systems eyes so nobody is getting punished for it

OhBoyItsaMegamanMay 22, 20151 votes

so you could be saying "ggeasy" to your team (the enemy team would never see it, being gg easy is toxic tword the enemy team) or you could say "gg easy game boys!" in team chat) the enemy reports you because you said "fuck yeah :p" when somebody asks you if you play udyr alot, somebody else got offended by the word fuck and reported you or maybe your team did, system scans. sees "gg easy" cant determine context (that is want in all chat and wasnt ever said to the enemies only your team) and punishes you

How can you say it can't determine the context? It has access to the entire chat log. Along with who muted who, who reported who, etc. And it has millions upon millions of games to reference to figure out whether it's a positive or negative statement.

Arcade AndrewMay 23, 20151 votes

did you even read the second part of this of WHY it cant learn context (besides it being a machine and not a human programed to key in on certain phrases)

it cant learn context because if you let the community set the rules of whats toxic, context is always changing, one day gg easy might be an honorable thing to say, and ggwp will be toxic, you never know

for example in china/korea "silver trash" is a toxic phrase but in NA "silver trash" isnt, it falls under neutral behaviour

SmokedAlmondsMay 23, 20151 votes

[{quoted}](name=Arcade Andrew,realm=NA,application-id=cIfEodbz,discussion-id=jIg8mFGE,comment-id=00010000000000000000000000000000,timestamp=2015-05-23T05:27:37.956+0000)

did you even read the second part of this of WHY it cant learn context (besides it being a machine and not a human programed to key in on certain phrases)

it cant learn context because if you let the community set the rules of whats toxic, context is always changing, one day gg easy might be an honorable thing to say, and ggwp will be toxic, you never know

for example in china/korea "silver trash" is a toxic phrase but in NA "silver trash" isnt, it falls under neutral behaviour

if you consider there to only be 3 options: positive, neutral, and negative that could be true, But if it were more of a spectrum it doesn't have to be. Its not like overnight everyone is going to decide something that was fine the previous day isn't ok anymore. What you would see is slowly over time people would report a given behavior either slightly more or slightly less.

I really wish we could get more details about how it works even though i understand why we cant.

OhBoyItsaMegamanMay 23, 20151 votes

[{quoted}](name=Arcade Andrew,realm=NA,application-id=cIfEodbz,discussion-id=jIg8mFGE,comment-id=00010000000000000000000000000000,timestamp=2015-05-23T05:27:37.956+0000)

did you even read the second part of this of WHY it cant learn context (besides it being a machine and not a human programed to key in on certain phrases)

it cant learn context because if you let the community set the rules of whats toxic, context is always changing, one day gg easy might be an honorable thing to say, and ggwp will be toxic, you never know

for example in china/korea "silver trash" is a toxic phrase but in NA "silver trash" isnt, it falls under neutral behaviour

That's not what context means. I think that's what you're misunderstanding.

Context means either the words surrounding the statement, or the circumstances in which the statement was said. If taken out of context, "kill yourself" is a negative phrase. In the context of someone saying "if you don't get some lifesteal you will kill yourself on his Thornmail", it is not negative at all. If you ignore the context, you can't tell whether it's bad or not. If you pay attention to the context, you can tell. This program pays attention to the context. If it needs to figure out whether something is negative, it can look back through every game where anyone has ever said "kill yourself" in any context, and look for a pattern. For example, if "kill" and "___self" and "Thornmail" are said together, the player never gets immediately muted by other players and they rarely get reported. From that, it realizes that that is a context in which you can say that phrase.

No matter how much you insist, it is not "programmed to key in on certain phrases". That's not what they're doing. Okay, you're familiar with Google, right? Back in 2007, what would happen if you typed the phrase "hunger games" into Google? It would bring up some sites about hunger, and about games, and maybe dieting or games about eating. A year later, if you searched for that term, it would pull up nothing but pages related to the novel The Hunger Games. How did that happen? Did someone at Google reprogram their search so that it would understand that Hunger + Games together means the person is definitely searching for information about that book? No! The search engine paid attention to what people were clicking on after they searched. It noticed very quickly that a.) there are suddenly 10000x as many searches that have the phrase "hunger games" in them, and b.) all of those searches are clicking on pages that include the words "suzanne", "collins", "novel", "book", "katniss", and so on. Not a single Google employee took the time to fix Google so that it would understand this. Since it is paying attention to the behavior of everyone who uses it, it catches on very quickly when the relevance or meaning of a phrase changes.

No one programmed this bot to punish people for saying "GG easy". It looked at chat logs for more games than you or I can possibly imagine, and came to that realization on its own by just looking at patterns. If the playerbase drastically changes their mind on something, this algorithm will also change its mind.

Arcade AndrewMay 23, 20151 votes

and the community can end up saying any time the words "kill yourself" are used its toxic, no matter what the context. the system can not learn at a quick enough pace to keep up with the community, for example

it used to be funny to say "fill or feed/troll or afk" ect in champ select and people would laugh

nowadays you get raged at or reported for "threatening to troll"

OhBoyItsaMegamanMay 23, 20151 votes

Okay, let's take an example and run with it. Robin Williams committed suicide earlier this year. The day before that, telling someone "You should pull a Robin Williams" would be a somewhat meaningless phrase with no negative connotation. After his death, it would be very offensive and certainly deserves punishment. In the span of a day, there is a phrase that instantly, across the whole NA community, goes from neutral to very toxic. Rioters who check on post-game reports understand this immediately. Rioters who respond to support tickets understand this immediately. The automated chat scanner doesn't pick up on it for a little while until it has happened in a sufficient number of games.

So what? What's the problem?

Arcade AndrewMay 23, 20151 votes

Thats the problem right there, the scanner cant keep up with the change in what is deemed toxic and not toxic, it will always flowlike that and without human interjection the scanner wont be able to keep up

SmokedAlmondsMay 23, 20151 votes

[{quoted}](name=Arcade Andrew,realm=NA,application-id=cIfEodbz,discussion-id=jIg8mFGE,comment-id=000100000000000000000000000000000001000000000000,timestamp=2015-05-23T20:34:12.511+0000)

Thats the problem right there, the scanner cant keep up with the change in what is deemed toxic and not toxic, it will always flowlike that and without human interjection the scanner wont be able to keep up

Based on what exactly? You just think it cant and are stating it as fact.

The rate at which it adapts would most likely follow the rate which people change their view. If slowly people start thinking "your mom" jokes are bad then the reports on that kind of thing are only going to go up slowly. If like someone massive changes the meaning of something like the robin williams example then people are going to be reporting that significantly higher overnight so it would likely change faster.

WickedfatesMay 23, 20151 votes

{quoted}

I'm not thrilled that you have to go to Reddit to find this information, but it is quite clear

well said

Junsui AoiMay 21, 20152 votes

Technically, they do discourage foul language regardless since there are kids that play this. And chat filter doesn't catch everything when people misspell it.

Granted, I hope they have an appeal system, since I'm sure there will be several misjudgments.

PhifeMay 21, 20151 votes

lmao, just because someone younger than me plays the same game i cant act my age?

I'll use my vocabulary as i like, even if its "FUCK YEAH team, we just gave them an ass fucking that team fight!"

Arcade AndrewMay 22, 20151 votes

now you cant, lol thats everyones point nobody likes this

Thanotos OmegaMay 23, 20151 votes

Umm even without the curse words that breaks the rules, pretty sure your not allowed to talk down to the other team

PhifeMay 23, 20151 votes

I wouldn't even say that in all chat.

Arcade AndrewMay 22, 20151 votes

1 in 6000 was riots rate of misjudgements

THATS REALLY BAD NUMBERS 1 in 6000? thats really bad...

ZombiemasterMay 22, 20151 votes

So, where exactly did they say that this new system will ALWAYS punish you if you say some of the words/phrases? And where does it say that it ignores context?

Arcade AndrewMay 22, 20151 votes

check the comments on the main article a bunch of reds kept saying that it scans for key words or phrases that the community deems toxic (in this case racism is toxic) so the word(s) "nigga/nigger" would be flagged by the system,

but what if you are saying " thats my nigga!" after a good play?

BakedLotionMay 22, 20151 votes

The thing is, some people are so butthurt they'll report you for saying something like "What are you doing? You shouldn't be doing that" and then the system determines thats toxic and it'll punish you. It's so dumb. Just let actual people read the shit.