Community getting worse DO SOMETHING RIOT

shadowwreaper·Mar 9, 2017·7 votes·1,948 views

"Language is fine. We don't penalize people for swearing, or using bad language." Riot tantram commented this on another thread.

I already commented on it so all im gonna say is Riot you need to severely take another look at the community again and reassess that statement. Because ive witnessed tons of reports go through just for someone cussing in general not at someone. Just trolls taking an opportunity to scream HES TOXIC REPORT REPORT just to be dicks. And nothings being done about them Same thing happened to me. i just cuss alot thats who i am. it does not make me malicious or toxic right? Yeah right.

28 Comments

Deep Terror NamiMar 9, 20177 votes

Because ive witnessed tons of reports go through just for someone cussing in general not at someone. Just trolls taking an opportunity to scream HES TOXIC REPORT REPORT just to be dicks. And nothings being done about them

What Tantram is saying is that the person screaming "HES TOXIC REPORT REPORT" is the one that is going to end up punished, not you. They can report you all they want, if you aren't normally being a negative/toxic player they won't matter.

Naked Baby ThiefMar 9, 20171 votes

Sounds great, but in practice isn't the truth. Look up the history of player behavior here. 99.99% of the time it is chat logs that result in bans. Very rarely does someone actually get looked at by a human being, and banned due to their actions.

Deep Terror NamiMar 9, 20174 votes

Sounds great, but in practice isn't the truth. Look up the history of player behavior here. 99.99% of the time it is chat logs that result in bans.

This post and my reply are specifically about verbal abuse and false reports for it, not gameplay offenses.

Very rarely does someone actually get looked at by a human being, and banned due to their actions.

Manual review is not necessary in the vast majority of cases; the IFS had less than a 1 in 6000 rate of false positives at launch, and has only improved since then. Yes, gameplay offenses are much harder to detect, but that's a discussion for a different thread.

Naked Baby ThiefMar 9, 20171 votes

Fair enough, but I would disagree about the need for manual review. The metrics for which they claim '1 in 6000 rate of false positives' are purely subjective, and can't possibly be substantiated quantitatively or qualitatively. In other words, it is a provably false claim.

How do I know this? Because it is a morality argument.

Even if they use the most advanced machine learning, it will fall extremely short of being 99.9998% effective (1 in 6000). If for no other reason than the subjective nature of morality, hostility, and toxicity.

If that is their claim, I wouldn't just question the integrity of their claim, but the integrity of those who made the claim. Even in the context of the most advanced machine learning alogarithms, which do not have the ability to take negative connotation, morality, and extremely complex use of words associated with hostility, into consideration, tend to be vastly less accurate than that.

So.... Sorry. But if at some point they claimed the systems were that accurate, they lied blatantly.

In fact, there are a few TED talks about exactly this, and the increasing importance of the human ability to discern morality, since machine learning cannot do so accurately.

Complex and interesting topic, given that machine learning has become so integral in so many industries, recently.

Deep Terror NamiMar 10, 20174 votes

Fair enough, but I would disagree about the need for manual review.

They still do manual reviews for fringe cases, and randomly to ensure system accuracy. Anyone that thinks the system made an error can request a manual review, but the conclusion is nearly always the same.

The metrics for which they claim '1 in 6000 rate of false positives' are purely subjective, and can't possibly be substantiated quantitatively or qualitatively. In other words, it is a provably false claim.

Actually, they hand reviewed every case at launch. This isn't a wild claim, it's an objective fact. If you think they are just lying about whatever they say there is no point discussing this further.

Naked Baby ThiefMar 10, 20171 votes

Impossible claims are impossible, this is a self evident truth. It is fine if you wish to make your own ignorance the end of a conversation.

It isn't me 'thinking', it is knowing the limitations of both technology and the very concept of morality itself. Moral arguments aren't logical arguments, and machine learning struggles with the the prior.

You are more likely to find an AI that can prove that God exists (Immanuel Kant) than to find one that can measure the morality of language consistently enough to be 99.9998% accurate, as you, and supposedly they, claim.

Deep Terror NamiMar 10, 20174 votes

I don't think you understand what it means by "1:6000 false positives". That doesn't mean it catches 5999 out of 6000 cases of negative behavior, that means it only flags 1 out of 6000 as negative when it shouldn't be.

Is that more clear? I have not said what you are trying to say I did.

Naked Baby ThiefMar 10, 20171 votes

Yes you condescending asshat, 99.9998% is the actual percentage of that ratio.

Is that more clear? Do you not understand how to convert ratio's to decimals?

Should I elaborate?

It is sad you can read what I have said so verbosely and fail to understand it

Deep Terror NamiMar 10, 20175 votes

I apologize that you're feeling insulted, that isn't my intention. You are arguing against something completely different than what I said.

Naked Baby ThiefMar 10, 20171 votes

I am not, and it most definitely was your intention.

How hilarious you would argue a machines ability to perceive morality when humans like yourself struggle with it so much?

Do you HONESTLY think someone doesn't know what a ratio is?

Sit there and answer that question. Be honest with yourself.

This is something that people learn in elementary education. Everyone knows what a ratio is, even most children.

So whether you 'intended' consciously to insult me or not, you set about to do exactly so with the words you chose.

It is either passive aggressive or an unconscious microaggression, but what it isn't, is you seriously thinking an adult that is explaining the complexities of defining emotions using math (IE machine learning computer alogarithms) doesn't know what a ratio is.

But do, write me a math formula for defining toxicity?

No? You can't? Exactly.

Also, just because you don't understand what I'm talking about, doesn't mean it isn't completely applicable to what you said.

Chances are your frustration in not understanding is what lead you to be so insulting.

Deep Terror NamiMar 10, 20173 votes

How hilarious you would argue a machines ability to perceive morality

The community decides what kind of behavior is a problem by reporting it, that is what drives the system. It doesn't need to "detect morality", players just need to report it enough that a pattern is formed.

I also wasn't arguing about morality, that was you.

Do you HONESTLY think someone doesn't know what a ratio is?

Just to reiterate, I said nothing about the definition of a ratio or the percentage conversion. What I was discussing was what a false positive is.

If you believe they are lying, I recommend taking it up with Riot Tantram (Lyte doesn't work at Riot anymore). Here is the quote from Lyte about the false positive rate;

{quoted}

The system analyzes in-game data like chat logs, and then assesses reports and tries to "validate" the reports. In addition, every report and honor in the game is teaching the system about behaviors and what looks OK or not OK, so the system continuously learns over time. If a player shows excessive hate speech (homophobia, sexism, racism, death threats, so on) the system might hand out a permanent ban to the player for just 1 game. But, this is pretty rare!

In terms of false positives, we recently flew in Player Support and Player Behavior team members from all around the world to hand-review thousands of chat logs, and we saw false positive rates in the 1 in 6000 range. So, we know the system isn't perfect, but we think the accuracy is good enough to launch.

Naked Baby ThiefMar 10, 20171 votes

The community doesn't decide, RIOT does.
Yes, I already address machine learning which is what Lytes quote pertains to, and you completely failed to undestand it.
1 in 6000 is a ratio. You are either purposefully being dense in order to protect your ego, or you are honestly too dimwitted to be having this conversation.

Deep Terror NamiMar 10, 20172 votes

You don't need to keep bringing up the definition of a ratio, for the fifth time, I was not telling you what a ratio is.

Naked Baby ThiefMar 10, 20171 votes

[{quoted}](name=Deep Terror Nami,realm=NA,application-id=ZGEFLEUQ,discussion-id=lAVoTHlG,comment-id=0000000000000000000000000000,timestamp=2017-03-10T10:28:37.025+0000)

I don't think you understand what it means by "1:6000 false positives". That doesn't mean it catches 5999 out of 6000 cases of negative behavior, that means it only flags 1 out of 6000 as negative when it shouldn't be.

Is that more clear? I have not said what you are trying to say I did.

This is you explaining what the ratio is, verbosely.

So you are either a liar, forgetful to the point where you cannot remember past 10 minutes ago, or are purposefully being dense.

that means

This you is describing a meaning.

that means

Two words you don't use... when not trying to explain the meaning of something.

I would wager on two of the three. You are a liar and purposefully being dense.

I have yet to define ratio to you, but let me go ahead and do that for you.

ratio ˈrāSHēˌō noun the quantitative relation between two amounts showing the number of times one value contains or is contained within the other.

SwainLaneWinGameMar 10, 20172 votes

Except, this isn't an argument about morality; this is an argument about a system programmed to detect toxicity. Morality does not factor into the equation here.

If you say "report 'soandso' for ____", the system flags that. If you say "die you cancerous f***ot", the system flags that. If you start spamming chat, the system flags that.

Certain phrases, keywords and chat behavior is processed through their algorithms and quantified to determine if a player's chat history over time meets or exceeds their standards of what is "toxic". If so, then a restriction or suspension is given.

If you go with the hardcore harassment and hate speech, you go straight to the perma.

Naked Baby ThiefMar 10, 20171 votes

Toxicity is a loose word which does little more than encompass a set of morality driven circumstances.

So saying that this isn't an argument about morality is provably false.

Lets look at your examples. "report sosandso for", They committed an action which the community deems wrong, ie, immoral. "die you cancerous f**ot*", obviously calling people this is immoral, and wrong.

"if you go with hardcore harassment" morally wrong. "and hate speech" morally wrong"

Morality driven argumentation.

And yes, if you spend 10 minutes here on the player behavior forum looking at posts, you find dozens of instances where peoples language was taken out of context completely, by a machine that cannot differentiate,

"DIE TEEMO!"

and

"KILL YOUR SELF"

In fact, there is an example of this on the front page of the player behavior forum at this very moment.

Again, I challenge anyone to define any formula for detecting the morality of any given sentence (or TOXICITY, which is synonymous with bad morality).

Because if you do, you will be rich. Pretty much instantaneously. (Because it doesn't exist, and people who study machine learning would KILL for this source code)

Wiggle Dat ButtMar 10, 20171 votes

[{quoted}](name=Deep Terror Nami,realm=NA,application-id=ZGEFLEUQ,discussion-id=lAVoTHlG,comment-id=0000,timestamp=2017-03-09T19:12:57.292+0000)

What Tantram is saying is that the person screaming "HES TOXIC REPORT REPORT" is the one that is going to end up punished, not you. They can report you all they want, if you aren't normally being a negative/toxic player they won't matter.

Everyone knows this never happens. The simple fact is that numbers matter, and if you get into a premade of 4 people and you don't do exactly what they say you are fked. Typically they will flame you each trying to "one-up" the other child in their premade trying to be cute and then if you say one single word back to them 4 people report you and you get punished while they get off scott-free.

ModPrandineMar 10, 20172 votes

[{quoted}](name=Wiggle Dat Butt,realm=NA,application-id=ZGEFLEUQ,discussion-id=lAVoTHlG,comment-id=00000001,timestamp=2017-03-10T22:30:57.143+0000)

Everyone knows this never happens. The simple fact is that numbers matter, and if you get into a premade of 4 people and you don't do exactly what they say you are fked. Typically they will flame you each trying to "one-up" the other child in their premade trying to be cute and then if you say one single word back to them 4 people report you and you get punished while they get off scott-free.

From Riot Tantram:

Actually, Erdrik is correct here.

A report flags a game for review. They do not establish guilt. The game is then reviewed for harassment, feeding, etc. This is why 'troll reporting' or 'gang reporting' do not work. False or inaccurate reports are thrown out.

You cannot ever be punished for saying nothing .. unless of course it's for something gameplay related like feeding, afking or cheating. But, you can never be punished for offensive language, hate speech or verbal abuse if you said nothing.

In short it doesn't matter if a trolling pre-made group reports you so long as you've done nothing against the rules.

Wiggle Dat ButtMar 11, 20171 votes

{quoted}

From Riot Tantram:

In short it doesn't matter if a trolling pre-made group reports you so long as you've done nothing against the rules.

I could mindlessly copy pasta riot garbage that is spewed from the anus of the absolute lord's of lying bullshit myself, but I am not a sheep and actually can think for myself. All it takes is seeing countless instances where the lone player is punished when put into premades. If you read my statement, I stated that when you have the premade of 3-4 people trash talk you, and you say one word back to them YOU get punished simply because their multiple reports flag you instantly, and nothing happens to you as your report doesn't count because you were already flagged by Riot Lyte's shitter system.

ModPrandineMar 11, 20171 votes

[{quoted}](name=Wiggle Dat Butt,realm=NA,application-id=ZGEFLEUQ,discussion-id=lAVoTHlG,comment-id=0000000100000000,timestamp=2017-03-11T00:33:24.017+0000)

I could mindlessly copy pasta riot garbage that is spewed from the anus of the absolute lord's of lying bullshit myself, but I am not a sheep and actually can think for myself. All it takes is seeing countless instances where the lone player is punished when put into premades. If you read my statement, I stated that when you have the premade of 3-4 people trash talk you, and you say one word back to them YOU get punished simply because their multiple reports flag you instantly, and nothing happens to you as your report doesn't count because you were already flagged by Riot Lyte's shitter system.

I did read your statement, and it's simply not true. The system only checks for consistent or extreme instances of negative behavior, so saying one word back to a trolling premade isn't gonna get you punished under normal circumstances. Also, while Riot isn't perfect, I'd much rather believe them and their proven track record than someone with a questionable track record.

shadowwreaperApr 5, 20171 votes

Lemme tell you why thats bullshit. Simply put ive gotten reported several times for language only. Ya say one cussword and suddenly your bot lane goes into all chat and says report this toxic ass plz. AND THEY SAY SURE. not even asking for second side they just willingly report anyone who fuckin says the words report

Deep Terror NamiApr 5, 20171 votes

A little cussing here and there won't get you punished, they can report you all they want and it won't do anything if your behavior doesn't warrant it. If they are harassing you about it, they are the ones at risk of being chat restricted or banned for their actions. Just mute them and ignore it as best you can.

FrostmorneMar 9, 20172 votes

toxic toxic toxic. but no word on the trolls and feeders each and every game. get a life and stop being so soft!

scazzmanMar 9, 20172 votes

The sad truth, Riot DOES NOT CARE about toxicity, they have a robot designed to detect to detect chat because it gets results, and it's only there to make it look like theyre actually doing something. and of course they arent just going to go out and say it

ReckfulsBaldSpotMar 10, 20172 votes

Their banning system is almost as pathetic as the community that does the reporting. I've played since season 2 and had my fair share of banned accounts and I agree this community has gotten not only worse but the people who do the reporting have gotten 20x worse because they know how to avoid getting banned but still provoke players. The people riot bans the most aren't the toxic players they ban anyone who types excessively or anyone who gets flagged by a scumbag premade group queue. It's pitiful that they're still banning accounts and nobody but the "toxic" people care about the scam they obviously run on anyone who types or cares about how they spend their time. Disgusting community disgusting player service.