• Please go to our Discord or SteamRep for trading. Bans and appeals go HERE

Sir Trooper

Gore-Spattered Heavy
Mapper
We can easily make it have very few false positives with time but I'm personally unsure about it being automated as for now, better would be to have it ping admins on discord whenever it pastes pieces of players chat and they can check the full context

I just went through 2000 lines of chat samples and flagged 11 as toxic from my personal view examples are "I hope you die in a ditch", "ez dom", "I hope all your family members get slowly dismembered infront of you", "I hope all your pets will suffer", "I hope they all will die painfully", "Kill that fucking child", "Noob down", "Big nose like israeli", "fag", "ez kill", "fuck that trash".

I feel like processing enough chat we can make it way more reliable than the wikipedia based AI, which is decent on its own but not good enough. I muted 7 people today using it but lots of those were racism, homophobia etc not just toxicity
You might as well just disable the chat lol
 
  • Haha
Reactions: Jymbles

gala

TF2 Admin
Contributor
You might as well just disable the chat lol
I don't think you understood correctly, saying that in chat does not automatically mute you, nor do we mute you for saying single lines like that without context. Saying stuff like that 10 times in 30-60 minutes would possibly trigger the recognition and it would be pretty clear your purpose of using chat is to bother others and you'd be better off not speaking at all
 

Sir Trooper

Gore-Spattered Heavy
Mapper
Yeah but what if your saying it whenever your friend dies, and she/he does it in return? We don't see it as toxic because we know they are not a noob but we do it cause it fun to do, it's like banter
 

gala

TF2 Admin
Contributor
Yeah but what if your saying it whenever your friend dies, and she/he does it in return? We don't see it as toxic because we know they are not a noob but we do it cause it fun to do, it's like banter
I didnt really understand this but it's not difficult to click stats of the player and look at the chat message context, scroll down and see if the player is always like this, how other players react to him, are they taking it as a joke, how many votemutes the person has to see that others are bothered by him etc and see whats going on
I wouldnt start muting people without checking those
 

Bottiger

Administrator
The hope is that it can be automated, because admins can't be online 24/7. But if it isn't accurate enough then it won't be.

The AI can and will make mistakes just like any human can. But if we require multiple detections, the error rate will go down drastically.
 

JayDee

Gaben's Own Aimbot
Contributor
We can easily make it have very few false positives with time but I'm personally unsure about it being automated as for now, better would be to have it ping admins on discord whenever it pastes pieces of players chat and they can check the full context

I just went through 2000 lines of chat samples and flagged 11 as toxic from my personal view examples are "I hope you die in a ditch", "ez dom", "I hope all your family members get slowly dismembered infront of you", "I hope all your pets will suffer", "I hope they all will die painfully", "Kill that fucking child", "Noob down", "Big nose like israeli", "fag", "ez kill", "fuck that trash".

I feel like processing enough chat we can make it way more reliable than the wikipedia based AI, which is decent on its own but not good enough. I muted 7 people today using it but lots of those were racism, homophobia etc not just toxicity
"noob down" is toxic??
"ez kill"?????
 

Rand

Uncharitable Spy
Contributor
We can easily make it have very few false positives with time but I'm personally unsure about it being automated as for now, better would be to have it ping admins on discord whenever it pastes pieces of players chat and they can check the full context

I just went through 2000 lines of chat samples and flagged 11 as toxic from my personal view examples are "I hope you die in a ditch", "ez dom", "I hope all your family members get slowly dismembered infront of you", "I hope all your pets will suffer", "I hope they all will die painfully", "Kill that fucking child", "Noob down", "Big nose like israeli", "fag", "ez kill", "fuck that trash".

I feel like processing enough chat we can make it way more reliable than the wikipedia based AI, which is decent on its own but not good enough. I muted 7 people today using it but lots of those were racism, homophobia etc not just toxicity

"ez dom"

I guess Victor is being flagged a lot. 😁
I don't think you understood correctly, saying that in chat does not automatically mute you, nor do we mute you for saying single lines like that without context. Saying stuff like that 10 times in 30-60 minutes would possibly trigger the recognition and it would be pretty clear your purpose of using chat is to bother others and you'd be better off not speaking at all

I can't speak for Trooper, but I think part of what he was saying is that YOU flagged those quotes going through the chat samples. Wouldn't that mean that the AI looks at chat like "ez dom" and "ez kill" more stringently? I am not trying to be an ass with my input here. I understand what you said about frequency and looking up a player's chat logs. Toxic is one thing, battle banter is another.
 

Temnost

Scarcely Lethal Noob
Contributor
I agree with this but it shouldn't be left to a knee jerk reaction. There is a line that's crossed when Banter because toxicity and I feel like that line moves around a lot depending on which players are in the server at the time.
 

Bottiger

Administrator
I know it is a hard pill to swallow but the way you act right now or what you consider to be "battle banter" might be toxic.

Some of the things you all claim to not be toxic could easily get you chat banned on Overwatch or Valorant.

Being toxic is not black and white, it varies between people and completely opinion based.

And due to the fact that there are always people on reddit and youtube claiming our servers are toxic, while there's basically ZERO people complaining about other communities with a toxicity rule being anti free speech... tells us that our previous stance of just letting the players moderate themselves is wrong.

My own opinion is that people should just personally mute those that are being "toxic" but it is clear that nowadays people demand the server operators to mute these people for them.

You should be glad that we're looking into using an AI for this too, it won't have a bias towards any particular player, it won't be inexperienced, and it doesn't have bad days.
 

BadmanoftheBrine

Banned
Contributor
>could easily get you banned in Overwatch

Spamming medic can get you banned in OW, meanwhile in TF2 people cosplay nazis as the medic and dont get banned by valve (valve dont own community swrvers, technically, so Im not counting anti nazi rules). If youre following OW logic youre going to kill your servers. Theres a reason people barely simp for those girls anymore.
 
  • Dumb
Reactions: Metroidz

herp 105.4

Rage-Inducing Forum Troll
Contributor
FF2 Balancer
ZERO people complaining about other communities with a toxicity rule being anti free speech...
that is because we (at least me) dont play on those servers and keep in mind they get labelled with other things, like being full of pedos or full of rems

o yeah and also those people who say skial is toxic are right to an extent, but its all circumstance because if someone plays bad and you are then called "shit", thats toxic but true

please for the love of god stop listening to reddit as they are filled with snowflakes who think criticism is toxic
 

Hematite

Somewhat Threatening Sniper
Contributor
We have added a rule targeting toxicity now.

We are also experimenting with using an AI to automatically detect toxicity. However the data it was trained on is a lot less tolerant to swearing than most gamers (wikipedia).

In order to make it more accurate, we will need people to go through hundreds maybe thousands of lines of chat, and label them toxic on a scale of 0 to 1. Is anyone willing to do something like this?

You'd get a text file like this and you'd be expected to put a comma and the number at the end.
Hey Bottiger,

How will this help with voice communication?

That said, I'm willing to help.

Edited to add: I think we need more than just a person or two running this exercise. As we've all agreed, toxicity is a sliding scale, so it would make personal biases less severe.
 

Bottiger

Administrator
We don't have the resources to process all the voice chat on all servers so it won't help with voice chat.
 

indecisive

Unremarkable User
Anyone who gets genuinely offended over some mean words on the internet needs to get a grip. Let players do whatever they want, it's a FPS online multiplayer game not a preschool class. Of course if they try to do something actually bad like doxxing then that's where the line draws. Otherwise, let people have their fun like the COD players once did.