IRC (InternetRelayChat) clients have usually supported “ignore lists”, just as newsreaders have supported so called “kill files”. This is a variant of IgnorePeople as applied to IRC.

On a particular IRC channel, you’ll usually find people who are “operators” (have ops), or can get operator status upon request. Operators can ban people using pattern matching on their nick name, their username, or their hostname or IP number. (BanningUsers)

If the operators are unwilling to interfere, however, it’s free for all. (RadicalInclusiveness)

One interesting problem is when the operators have higher personal standards then the ones they are willing to impose on the channel. There’s a certain UnwantedPerson? that aggravates them, without justifying a ban. Every other user can put this obnoxious person on their personal ignore list.

But if an operator puts the obnoxious person on their personal ignore list, then they can no longer watch over the channel (no PeerReview). In effect, they’re disowning part of the channel, refusing to take responsibility.

Furthermore, some IRC clients ignore not only to all messages from the ignored users, but try to figure out which messages where meant for the ignored users and ignore those messages, too.

Thus, other (trusted) people getting involved in a fight with an obnoxious person will no longer tip off operators, because they cannot see the messages from their friends to the ignored persons.

Once you start populating the ignore list, it’s easy to add more names to it, since many people might take the FlameBait left an obnoxious person. The operator is ignoring all messages from the obnoxious person, and all messages obviously meant for the obnoxious person. That leaves enough useless messages by people involved in the flaming, not obviously addressed to anybody in particular. To the operator, it looks like spam. And the easiest way out is to add those people to the ignore list.

And so it grows, until the only way to actually reach operators is via a clandestine network of quiet friends who never get involved in discussions. Weird.

See Also



Trying to kickban Xah Lee from #emacs hasn’t worked. Xah Lee is a famous UseNet troll on the various language groups. He just keeps changing nicks, and the operators on #emacs have been unwilling to ban his entire Internet Service Provider (ISP). The result is that I and many operators I like have put him on their ignore list, and that I have started adding all the people that like arguing with him to the ignore list as well. Now the channel seems fine to me, but I bet it seems like a weird madhouse to newcomers.

Maybe when people sign on to the IRC channel in question, the could see a link to a wiki site in the intro/welcome message that takes them to a page that describes the issue, and why channel ops put people on ignore lists, and best practices to deal with trolls/griefers/FlaimBait??

The problem is that the kind of person who engages with trolls just isn’t aware enough to bother reading let alone following the advice left on a wiki pointer. As supporting evidence I call on this paper, which is a keystone element of my personal philosophy: “Unskilled and Unaware of It: How Difficulties in Recognizing One’s Own Incompetence Lead to Inflated Self-Assessments” (PDF). I’ve become convinced over the last few years that the solution to this & many related problems in social systems lies in the realm of ReputationSystems?, which strengthen the influence of the most competent members of the group & implemented properly, can overcome this effect. I’m not sure exactly how it would work in an IRC environment but I suspect there’s a way to do it that’s a softer form of filtering than the black-&-white banning solution, that relies on the collective judgement of the group.

TimKeller 2006-10-07 02:48 UTC

Interesting thought. I hadn’t really associated reputation systems with handling trolls, but now that you mention it, you don’t need to have a system that is good enough to measure each and everyone’s reputation. It just has to be good enough to indicate the trolls.

Unfortunately, grafting a reputation system onto an existing protocol and no way to introduce new clients to a large part of the population, the system will be as bad as personal ignore-lists: The same people “in the know” will be able to take advantage of it, where as all the innocent newcomers will be swamped by the trolls.

Tim, I agree with you totally that the trolls won’t bother lookign at the wiki. But, the non-troll newcomers might look at it and think “oh, I get it, the people who actually use this network in non-troll ways are ignoring the trolls”. And, they might follow suit. If everyone who wants to get value out of the IRC network looks at the example of what existing people are doing, then they can “tune in” to the same “frequency” (setting trolls to ignore).

However, don’t let that be taken as an argument against reputation systems like what you describe, because I agree with you that reputation and trust metrics are very much needed to help different networks retain their collective value (plus to help people find the valuable networks). That is related, IMHO, to your SocialNetworkSecurity concept. So, my wiki “welcome new users, here’s the best way to get the most out of this IRC channel” suggestion is more of a small and easily applicable social “patch” to the existing that I have seen work in other environments and virtual communities. it’s a sharing of knowledge by the people who adminstrate the channel, for the people who want to get intended “value” out of it.

Define external redirect: FlaimBait UnwantedPerson ReputationSystems

EditNearLinks: WhatIsaTroll PeerReview TimKeller UseNet