What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 127 Stack Exchange communities.

I will confess, We'd noticed a bunch of flags with no complaints from anybody so I thought I'd test and see if something automated was happening with a series of progressively more offensive words. Fart, boob and one big-boy word. Fart was the lowest offender on the list. I deleted the one 4-letter word just in case we really did have a silent flagger who was profoundly offended by such things.

But seriously, fart?

If Stack is switching gears to Disney Land mode could somebody make an announcement or something so I can start working on a competing site where the assumption is that we're all at least teenagers?

Or did we just have somebody obnoxious screwing with the system?

Random errata: I never saw anybody getting flags but everybody else was seeing it.

share|improve this question
3  
No old farts were mistreated in the production of this comment. –  Lee Louviere Aug 2 '12 at 20:20
24  
I can assure you from very extensive testing that Chat does not have an automatic swearword filter. –  Pëkka Aug 2 '12 at 20:22
    
Okay, so somebody flagged and a mod actually made the call to boot me for 'fart' then? –  Erik Reppen Aug 2 '12 at 20:26
    
I don't know. It's weird - what you describe sounds automated somehow. Maybe a mod or dev can shed some light on this. –  Pëkka Aug 2 '12 at 20:26
    
I had just heard that somebody was spamming stars in the C# room so they had to lock it down somehow. Should I see if I can track down the list of active users at the time? –  Erik Reppen Aug 2 '12 at 20:27
6  
If 5 users agree with the flag, your message will be deleted and you'll be suspended from chat for 30 minutes "automatically". It's true a mod can validate the flag which is immediately binding, but I don't find it likely in this case, unless they saw it out of context and assumed you were spamming fart or something –  Ben Brocka Aug 2 '12 at 20:30
    
No, I'm a regular. It happened long after the test had scrolled off the page. I'm thinking some idiot has 4 bogus accounts. That said, I do find it pretty funny. But we were also noticing lots of flags with no actual complaints before then. –  Erik Reppen Aug 2 '12 at 20:31
    
I can still see the chat (JavaScript Chat). Esailija (also a regular) just claimed his comment had been logged as offensive by 14 users. I think it's time to ring the shenanigans bell. I believe he's referring to a message from this morning. –  Erik Reppen Aug 2 '12 at 20:33
    
You get suspended automatically if you a flag gets validate five more times than invalidated. Those flags were around a long time. –  R. Martinho Fernandes Aug 2 '12 at 20:44
1  
I marked the flag as valid. If it were in some context, i wouldn't have...but i see flagged messages consisting solely of "fart" and "boob", i'm gonna call it an immature attempt to annoy people. –  cHao Aug 2 '12 at 20:45
7  
Was it a loud one? We usually suspend for 15m, but if it's a really loud one... –  Yannis Aug 2 '12 at 23:26
2  
If Stack is switching gears to Disney Land mode could somebody make an announcement or something Here's your announcement. –  Yannis Aug 3 '12 at 1:40
1  
What ever you do, don't abbreviate Find And Replace Text! –  Niklas Aug 7 '12 at 14:58
add comment

2 Answers

up vote 23 down vote accepted

The primary problem is that other users from other rooms have absolutely zero context for what's happening. So when you log in to chat and you find "fart" and "boob" with four flags each from another room, the most logical conclusion is that you're a childish troll and the other room is desperately trying to flag you out so that they can return to normal conversation.

Also, there is no requirement for moderator intervention- once you meet the threshold, the temporary suspension is automatic. The threshold is at least five valid, plus I think it's about 1 per "not sure" and 3 per "invalid", or something like that.

In the full context I would never have validated those flags. But I didn't have anything like that, so I did. Realistically, it would only have taken one user to make those flags, and everybody else would have simply agreed with them.

The existing flagging system is known to be quite broken. This is just one example of how broken it is. It really needs to be replaced with per-room moderation by the room owners.

share|improve this answer
    
Yeah, what he said. :) –  cHao Aug 2 '12 at 20:47
    
This flagging system doesn't seem to be visible to me. All flags can be seen on entry to a room? I got booted long after that test scrolled off the page. –  Erik Reppen Aug 2 '12 at 20:47
    
10k users can see flags and mark them valid or invalid. –  cHao Aug 2 '12 at 20:48
    
@ErikReppen you only see flags that other people made on chat items that you did not make as well –  amanaP lanaC A nalP A naM A Aug 2 '12 at 20:48
    
Oh, my reps not high enough. –  Erik Reppen Aug 2 '12 at 20:48
    
Still, strange that stuff like this is coming up in C# and in JavaScript and on the same day. –  Erik Reppen Aug 2 '12 at 20:49
1  
@ErikReppen: Hardly the first time we've had flag trolls. Some guy once went through the entire transcript of the C++ room and flagged every use of the word "fuck". Until he ran out of flags, I think. –  DeadMG Aug 2 '12 at 20:50
2  
@ErikReppen It's also highly dependent on the room itself. In the C++ room, "offensive" words have a tendency to get starred rather than flagged. –  Mysticial Aug 2 '12 at 20:50
    
Okay. Fair enough. Is there anything you can't flag? –  Erik Reppen Aug 2 '12 at 20:51
    
We can see flags even from rooms we're not in. Flags get pretty much broadcast to all the users that can act on them, so theye often come with little context. –  cHao Aug 2 '12 at 20:51
    
As far as I know, there's a limit on how many flags you can issue or agree with. Apart from that, you can flag anything. And users from other rooms are asked to vote on flags, too. –  DeadMG Aug 2 '12 at 20:53
    
@Erik: Not really. I mean, there's no filter saying "sorry, this message is unflaggable". But i imagine (read: hope) there's some kind of penalty for casting too many flags that get marked invalid. Never tested it, or even cared to -- i don't flag for much. –  cHao Aug 2 '12 at 20:53
    
So I got blasted by a rep-based system I didn't even know about by people flagging a message in a room they weren't even in. Could the designers really not anticipate some UX issues there? –  Erik Reppen Aug 2 '12 at 20:54
    
The problem isn't so much that flagging is broken. The problem is that flagging has been broken for a very long time, and it's never been replaced, despite the fact that virtually all the chat frequenters mention this problem repeatedly on Meta. Perhaps the site's employees simply do not feel that the chat is worth the effort it would take to change the system. –  DeadMG Aug 2 '12 at 20:55
5  
@Erik: To be fair, though, in this case you did take it upon yourself to test the limits -- knowing full well that a bunch of flags were flying around. It's not like this was really out of nowhere. –  cHao Aug 2 '12 at 21:04
show 4 more comments

Your messages were flagged as offensive by 11 users and counter-flagged by 5 users. With the 11th flag, the difference between flags and counter-flags reached the threshold of six, causing automatic deletion.

I've looked at the data, and here's something interesting: Out of the 11 flags, 4 came from people who were in the room at the time. None of the counter-flags came from people who were in the room.

So counting only people in the room, the votes were 4:0 for offensive. Thus assuming flagging was kept on a per-room basis, the same thing would have happened (since in that case, the threshold would obviously be lower).

There is no evidence at all that

The existing flagging system is known to be quite broken

as DeadMG says. This claim is regularly made by a particular group of people who, let's say, tend to be very liberal in their choice of words, and who think that in their room, they should be allowed to curse as much as they want. Of course the Q*Bert rule holds in chat as well, but not everyone accepts that.

Looking at the recent flags, those messages that were deleted due to reaching the threshold should have been deleted, and those messages that were flagged but shouldn't have been deleted weren't, because they were quickly counter-flagged. This includes the flagging troll in the C# room that you mentioned – those flags were gone pretty quick. Unlike with your messages, nobody agreed that they were offensive (because they obviously weren't), and so nothing happened.

Despite popular Lounge<C++> opinion, the data clearly says the system is working remarkably well. That's not to say it's perfect of course, nothing is. But the oft-repeated claim the chat flagging system is horribly broken and needs to be fixed ASAP is entirely unfounded.

And finally coming back to your particular case, cHao says it well:

To be fair, though, in this case you did take it upon yourself to test the limits -- knowing full well that a bunch of flags were flying around. It's not like this was really out of nowhere.

Yeah, that.

share|improve this answer
2  
I think the automatic suspension contributes a lot to the perception that chat flagging is broken, it hugely amplifies the effect of false positives. Not being able to chat for 30 minutes is no big deal, but people react very strongly to such punishments if they perceive them to be handed out for no reason. There is no middle ground with chat flags, no option to just delete a message and tell people to keep the language clean, without punishing them. –  Mad Scientist Aug 3 '12 at 7:03
1  
The system looks broken to me too. It throws a contextless chat message at me, and asks me if that flag is valid. How should I know? Don't ask me a question if you don't give me enough information to answer it. –  CodesInChaos Aug 3 '12 at 8:56
    
@CodesInChaos: For reference, there's a link at the bottom of the flag dialog to the room where the message was sent -- and it'll take you right to the time period around the message. So you can get context if you want it. –  cHao Aug 3 '12 at 14:11
1  
4 came from people who were in the room at the time. remember when you were a kid and someone told a story about a haunted house and in the end it turns out it's the house you're currently in? That's how I just felt. –  Pëkka Aug 3 '12 at 19:54
add comment

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .