This isn't about a specific instance that has happened today but more a general idea relating to the way that handling chat flags works.

There are instances in chat where somebody will say something which, in context, is acceptable. However, due to the way the flag system works, only the single message that was flagged is shown to other users on the network, and this one message might be seen as unacceptable out of context.

One of the first things that people are supposed to do, in my mind, is gather the context of the comment that has been flagged before choosing to validate or invalidate it. Many people will join the channel that the flag originated from to see what is going on and make their decision, but many more will just hit either validate or invalidate.

This idea is to essentially enforce that you check out the context before flagging - to do this, would it be possible to make it so that chat flags can only be validated if you're currently present in the room that the chat flag originated from?

  • 6
    You can already see something is flagged, it has a blue box next to it
    – Flyk
    Jan 19, 2014 at 14:18
  • 4
    Chat flags need a huge overhaul either way.
    – user98085
    Jan 19, 2014 at 14:19
  • 2
    I think this is essentially a duplicate of this question, and also related to this.
    – FAE
    Jan 19, 2014 at 14:21
  • 5
    @rene The flags would remain anonymous - this change wouldn't expose anything that isn't already visible. Flagged messages have boxes against them that show the message which got flagged already. In addition, there are instances where a bunch of users join to check the context but the people who didn't join already validated the flag, so there are already instances of flags causing a bunch of users to join.
    – Flyk
    Jan 19, 2014 at 14:21
  • @FAE related for sure, this seems to be a unique proposal of how to fix it though.
    – Flyk
    Jan 19, 2014 at 14:22
  • Just to be clear: If I want to validate a chat flag I first have to join that chatroom? Or does only the current people chatting getting the flag?
    – rene
    Jan 19, 2014 at 14:26
  • 1
    @rene Everybody will still get notifications, but you'd have to join the channel to validate it - which would provide additional time for the people who would normally check context to check the context and would stop the people who just click "valid" on everything from doing so
    – Flyk
    Jan 19, 2014 at 14:28
  • @Flyk I just want to point out that I think it's only 10k (network) users that see the box of a flagged post.
    – MBraedley
    Jan 19, 2014 at 15:22
  • 1
    @badp as I said in the comments further down... "How do you programmatically determine that the X messages before and after are related to the message that got flagged? There are plenty of instances (especially in the larger channels) were there are two or more conversations going on at once." - That is, unless the chat flag system was changed to allow the selection of multiple messages at the point where you were flagging
    – Flyk
    Jan 19, 2014 at 15:47

3 Answers 3


Forcing everyone to actually enter the room is not ideal, I usually use the transcript to check up on the context to avoid the "invasion of blues" that otherwise happens.

But I don't think that this solves the actual issue. The main reason that context is often ignored when handing chat flags is that it happens so damn fast. The flags are shown to every moderator and 10k user in chat, and only the ones that act the fastest decide the outcome of the flag. If you take your time to investigate, someone else will have acted faster.

I'd still think that adding some more immediate context would be useful, the flag dialog should probably show some context around the message by default. But the main issue in my opinion is still that everyone gets shown the flag at the same time, and that only the fastest users determine the outcome of a flag.

  • 4
    This change would resolve the "fastest gun in the west" syndrome due to requiring the people who're going to validate the flag to join the channel first. Since there are already people who correctly join a channel (or view the transcript) to establish the context, these people will get the opportunity vote, rather than being beaten to the vote by the people who didn't.
    – Flyk
    Jan 19, 2014 at 14:52
  • @Flyk It's only one click more, that won't stop anyone that doesn't intend to read the context. Jan 19, 2014 at 14:54
  • 2
    The point is, of the people who're generally going to just click "validate", most of them are already not paying any real attention, they're not going to want to join the channel, validate, and then leave the channel, because effort. By forcing them into the channel to validate, you're going to be increasing the chances that they actually check the context out before flagging.
    – Flyk
    Jan 19, 2014 at 14:56
  • 3
    Is it feasible to set a timer on chat flags, perhaps requiring a 10-second or so duration of "think time" before a flag can be either validated or invalidated?
    – Yawus
    Jan 19, 2014 at 16:13
  • @Yawus Well, flags are made with suspension and removal of offending material in mind - that's supposed to be done quickly. A timer seems counter-intuitive.
    – user98085
    Feb 7, 2014 at 15:27
  • @FEichinger 10 seconds seems like a short enough time to get rid of offending material quickly and long enough for people to actually get some context about the flag.
    – Yawus
    Feb 7, 2014 at 15:44
  • @Yawus Or they'll hammer the button/wait like with every other "wait X seconds" message on the site. I'm not convinced. :/
    – user98085
    Feb 7, 2014 at 16:48

We've had this pop-up before yet the powers that be don't see it as a problem worthy of a fix. I hope that will change soon. When I've been in chat, I've had flags from rooms I never go in, and the flag (obviously) refers to one line of chat which has been flagged. I find it difficult to check the context of that flag with the current tools.

Now while some things can clearly be held as offensive without any further thought (e.g. obviously bigotry and slurs, direct personal attacks on people), it does raise some issues:

  • It can be difficult to know how to deal with some flagged items without knowing the context of that line within the current chat in that room (e.g. is the line "out of place" with the conversation around it or does it fit the topic and tone of conversation in that room at that time). I defy anyone to visit the Server Fault chat as a non regular and work out whether or not the chat is "normal" for the room...

  • Without context its difficult to know if there are a group of people in a chat being "uncivil" and flagging others who reply to them in kind. If x calls y a Unicorn's Butt then flags y for saying "did you really just call me a Unicorn Butt? Yeah well so's your face" and y gets banned for their reply while the original from x isn't also dealt with then I think y might reasonably feel aggrieved.

  • Context is king in knowing that something is abusive as well as knowing when it is not. I would suggest that someone talking about how good bacon sandwiches taste and how everyone should try one isn't likely to be offensive in ServerFault or Arcade chats, but could very well be so in the Islam and Jewish site chats.

My proposed change:

I accept that this suggestion isn't perfect and carries a risk of its own (abuse not being stopped as quickly as possible) but I suggest that you do not get flags for a chat unless you are a regular in that room or at least that "set" of rooms (e.g. where the room is part of a set of rooms associated with a site).

I hope that this will mean that the only people who see a flag will always have some context for the general culture of the room / associated site. It should certainly help stop the 'fastest gun in the west' problem to a large degree.

The next question is how would you decide that someone is a "regular" in a chatroom. Perhaps you would need to have received the "Outspoken" badge on your account for the associated site?

  • 1
    You can be a regular in a room without even being signed up to the site it is associated with, though
    – Flyk
    Feb 9, 2014 at 19:33
  • @flyk yep, there's no perfect compromise here. That's why I suggested the Outspoken badge (meta.stackoverflow.com/help/badges/138/outspoken) as the decider. That has its problems too of course but it might be a start.
    – Rob Moir
    Feb 9, 2014 at 19:38

There's a problem I see: the user going to validate the flag could be present by having the chatroom open in another tab, but not actively reading it; when (s)he is presented with the acceptalbe-in-context flagged message, the flag would still be acccepted in many cases.

A better solution would be to provide the n previous and the n subesequent lines when asking for a flag validation; this way it would provide some, well, context.

  • 1
    Who is to say anybody will actually read those "previous and subsequent" messages, then? I mean, you already assume people will, at any cost, not read the context.
    – user98085
    Jan 19, 2014 at 14:36
  • By reducing the effort needed to reach those messages ("enter the channel and scan for the exact messages location" vs. "you're already looking at it"), I think it's a reasonable assumption that it will increase the number of users reading that context. Of course there will be some users ignoring it altogether; you can't make some users read some text on screen no matter how important it could be, as anyone in IT probably already knows. Jan 19, 2014 at 14:40
  • There's not much effort involved. Enter the room, look at the highlighted message, read what happens around it. "n previous and n subsequent messages" often isn't enough for the context and would not really make any of it easier.
    – user98085
    Jan 19, 2014 at 14:43
  • 2
    How do you programmatically determine that the X messages before and after are related to the message that got flagged? There are plenty of instances (especially in the larger channels) were there are two or more conversations going on at once.
    – Flyk
    Jan 19, 2014 at 14:54

You must log in to answer this question.

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