Wikipedia talk:Three strikes you're out policy

Comments:
 * When you say "The first two blocks will then serve as a warning to the user", what exactly happens if a user gets one or two sysops blocking him? He can still edit, right? How does he know that he is "on the way to destruction"? --Stormie 02:50, Apr 9, 2004 (UTC)
 * Presumably he should be warned. A good sysop should do this before beginning this process. - Fennec 02:54, 9 Apr 2004 (UTC)
 * In that case, the user can receive a warning, like the "You have new messages" comment. It will say: You have one strike against you. We welcome your contributions, but will not tolerate trolling or vandalism." Danny 02:59, 9 Apr 2004 (UTC)


 * "trolling" can be subjectively assessed. What exactly is to constitute "trolling"? If a user is obviously vandalizing, and only has been vandalizing, then the vandalizing component of this policy is already covered by the existing blocking policy (which exists somewhere...) Dysprosia 02:55, 9 Apr 2004 (UTC)
 * Under current blocking policy, a combination of trolling and good edits could potentially prevent a user from being blocked despite clear damage to the Wikipedia. -Fennec 03:05, 9 Apr 2004 (UTC)
 * But what exactly are we supposed to measure possible trolling against? Maybe we need some guidelines on this? Dysprosia 03:08, 9 Apr 2004 (UTC)


 * I understand the problem this is trying to solve, but this isn't the solution. It means that in a time when there are fewer admins on a site (these do exist), a lone admin reverting vandalism must chase a vandal around the Wikipedia until other admins show up at RC.  If you were to somehow make a distinction between vandalism and trolling (which is effectively impossible - one person's vandalism is another person's NPOV additions), I'd support (barely) for trolling only. R ADICAL B ENDER  ★  03:07, 9 Apr 2004 (UTC)
 * Right now, there are plenty of sysops around, and, if necessary, more can usually be found on IRC. It seems, though, that you are saying the suggested policy is too lenient, when it is coming in response to those who say that the existing policy is strict. Danny 03:38, 9 Apr 2004 (UTC)
 * I took the liberty of correcting you: mIRC != IRC. GAAH! :) Why not just refer to the IRC channel? ;) Fennec 14:40, 9 Apr 2004 (UTC)
 * Whether or not more sysops are on IRC should be moot. That's more work for me to have to go round up others to handle something that isn't broken now: a vandal goes amok, he gets a warning, if he continues vandalizing articles, he gets a 24-hour ban.  Easy as pie.  Under the proposal, a vandal goes amok, he gets a warning, he continues vandalizing articles, I have to track down other admins (if they're not around) or go onto IRC and put out an APB, wait for others to ban the vandal too (continuing to rollback articles helplessly as I wait) and then he gets a 24-hour ban.  This is adding an extra layer of superfluousness where one is not needed.  Why are we wanting to make it harder to stop vandalism?  This proposal shifts the ongoing battles against vandals from an offensive posture to a defensive posture.  It ties our hands while we await others.  I'll give you another example.  Some anonymous vandal has been defacing List of characters from The Simpsons and other assorted articles off-and-on for weeks now.  However, the edits, which are all false, look somewhat legitimate, and thus most admins don't even notice (I do because I've been watching the user like a hawk).  Thus, the only person to notice the incorrect edits is me.  This is a perfect example of where this proposal fails.  Under the proposal, I couldn't keep banning the user (he's been warned multiple times) and instead I have to keep rolling the user's pages back over and over until either someone notices or I cry out for help.  This means I waste more time chasing a vandal around until he actually gets banned rather than doing something useful.  This also means that a coordinated vandalism attack would, while not rendering us completely impotent, would mean that everything becomes three times as hard.  You're solution might solve the problem of trolling, but only at the expense of hampering vandalism efforts.  Thus, as a solution, I find it totally unacceptable.  R ADICAL B ENDER  ★  15:36, 9 Apr 2004 (UTC)


 * Given how hard it is to get any consensus here, this will either lead to trolls being left to work until finally three admins can be found to block them, or to users being blocked by three, then unblocked by three others, etc.. This also means that certain problem users who have sysops backing them can effectively never be blocked. &mdash; Jor (Talk) 10:59, 9 Apr 2004 (UTC)


 * As Danny says, there are usually plenty of sysops around, and the difference with this policy is that it doesn't take a near consensus -- it shouldn't be hard to get three sysops when there is a case of true trolling or vandalism. One of the issues here is that many sysops find themselves spending all of their time dealing with trolls and vandals, but many sysops are increasingly hesitant to block even clear vandalism because it inevitably causes a kerfuffle: people (sometimes including sockpuppets and sub-trolls and sub-vandals) post a request at Review of admin actions, they call quickpolls to censure the sysop, and talk about sysops "abusing their power." I think we need to do something like this to restore a balance and deal with the ever-increasing amount of trolling and vandalism, while making sure that sysops don't become "lone rangers" dispensing what they personally perceive as justice to make up for the fact that it is so hard to achieve consensus in marginal questions. And regarding those people who can always get three sysops to unblock them, something beyond this policy, such as mediation or arbitration, will have to be used in any case. -- -- BCorr | Брайен 12:41, Apr 9, 2004 (UTC)

---

Regardless of the vote, someone needs to code the feature for it to become available. If nobody codes the feature, the vote itself is meaningless.--Eloquence* 13:14, Apr 9, 2004 (UTC)
 * Do we need it to be coded? We could just have a Three strikes page (or maybe a Users for deletion) for those in question to be listed, a la Quickpolls. - Fennec 13:54, 9 Apr 2004 (UTC)

---

Quickpolls don't seem to be working. Let's give this a chance, even if we start with a Three strikes page and doing this manually. -- Seth Ilys 14:09, 9 Apr 2004 (UTC)

Nobody can define trolling in a way which is anything more than a point of view. There is a tendency to accuse people of "trolling" if they express an unpopular point of view. Vandalism is something else, but "troll" is just a word for someone that you don't agree with. The Trolls of Navarone 14:15, 9 Apr 2004 (UTC)
 * Here here. JRR Trollkien 14:24, 9 Apr 2004 (UTC)
 * It has been suggested that you two are each other's sock puppets. Fennec 19:55, 9 Apr 2004 (UTC)
 * Trolls of Navarone, you claim there is a tendency to accuse people of trolling if they express an unpopular POV. Please support these claims. I'd like to see 10 examples of this taking place in Wikipedia. Please supply the evidence. Kingturtle 20:07, 9 Apr 2004 (UTC)

I don't understand this... please clarify the following:
 * what duration of blocks are we talking about? Are you saying that three sysops need to agree to give a a logged-in user a 24 hour time out, or are three sysops enough to block a logged in user permenantly?
 * As I see it, 3 sysops can block until an unblock comes into effect. That will also require 3 sysops. Danny 14:46, 10 Apr 2004 (UTC)


 * how does this interact with quickpolls?
 * My feeling is that QuickPolls are not proving effective. They are often being misused, while in many instances, the process can be too lengthy. Furthermore, we are having a constant problem with trolls and sockpuppets voting on QuickPolls. I am proposing this as an alternative to QuickPolls with the caveat that QuickPolls should stay around in more controversial cases. Danny 14:46, 10 Apr 2004 (UTC)


 * if three sysops decide within a few minutes of each other that a user needs to be banned, how do the first two "bannings" work as a warning?
 * This kind of happened last night in a case of vandalism. Four of us banned a persistent vandal within a few seconds of each other. In that case, I think it is legitimate that we do not have extensive warnings and actually shows how this new proposal can be effective in instances of vandalism. In the case of trolling, I believe that people will take more time before making their decision. It will also be possible to warn a user before resorting to a strike, and I hope people will do that. Danny 14:46, 10 Apr 2004 (UTC)


 * would there be a page (similar to the quickpoll page) where this is all recorded (along with notes on the relavant user talk page) or would it rely on us just leaving notes on their talk page (which they are likely to blank)?
 * Haven't thought of that, but it is certainly a good idea. Danny 14:46, 10 Apr 2004 (UTC)


 * how long do these warnings last before they expire? - i.e. sysop A warns in january, B in february, C arrives in april... does the user then get banned?
 * I would say 24 hours, though I would not be adverse to it being longer. Danny 14:46, 10 Apr 2004 (UTC)


 * if it's going to be an informal system of posting warnings rather than a coded one (like Eloquence mentions), we're going to have to keep a good eye out to make sure that all the people involved are sysops, unless the page is protected so you have to prove you're a sysop to edit it (there are what 200 sysops now, I don't know them all)?
 * Yes, only sysops would be able to issue strikes, just like now, only sysops can issue bans. I see it as a function instead of the ban function that sysops have already. Danny 14:46, 10 Apr 2004 (UTC)


 * if two sysops have warned, what happens when a third sysop arrives who disagrees? Can they add their vote and strike out one of the warnings?
 * I don't think so. Only after a ban goes into effect will it be possible to unban. If a sysop. On the other hand, this proposal is also worth considering and could be a good idea. Danny 14:46, 10 Apr 2004 (UTC)

Questions, questions... :) fabiform | talk 14:18, 9 Apr 2004 (UTC)
 * Here are some answers. Certain details will still have to be worked out, but I think it is worth trying this system to deal quickly and effectively with trolling and vandals and to eliminate all the clutter that exists on the QuickPolls page, which I believe is detrimental to Wikipedia at large. Danny 14:46, 10 Apr 2004 (UTC)


 * Thanks Danny. I supported quickpolls, but they don't seem to be working in practice.  They're a wonderful forum for trolling (ugh).  fabiform | talk 14:56, 10 Apr 2004 (UTC)

I agree with this comment. JRR Trollkien 16:42, 9 Apr 2004 (UTC)

This policy would require ad hoc changes to the Mediawiki software and is not needed. If this ever becomes a real problem (meaning a policy violations problems, not a "trolling" problem), I suggest we expand Quickpoll User remedies to mirror Sysop remedies:


 * 1) First violation: don't use a quickpoll
 * 2) Second violation: Vote for "official" warning
 * 3) Third violation: Vote for 24hr ban
 * 4) Fourth violation: Vote for 1wk ban
 * 5) Fifth violation: Vote to ban for 1 month or until arbitration decree, whichever comes first.

JRR Trollkien 14:24, 9 Apr 2004 (UTC)

I don't think more bureaucracy or mock-ups of the modern legal machinery are going to solve anything, and in fact invite abuses. It seems to me like the mission of the encyclopedia is pretty simple, and any given user is either contributing useful information or not. I also believe that those who don't contribute (the so-called "vandals") are unlikely to ever contribute, so a sysop can just ban that person.

I think this sort of pseudo-formal procedure is unnecessary. Also, "trolling" is and always will be a subjectively-defined term, and while banning suspected "trolls" is the knee-jerk revenge reaction, they get sorted out over the long haul. Jeeves 16:49, 9 Apr 2004 (UTC)


 * The trouble is that they don't 'Troll-Hunts' are used by the cabal to suppress criticism and enforce their ideas of what Wikipedia should be. A vandal is a vandal, a trolls is just someone who you don't agree with. Not a reason to ban. The Trolls of Navarone 16:54, 9 Apr 2004 (UTC)


 * I would remind this user of What Wikipedia is and is not, just in case. - Fennec 19:47, 9 Apr 2004 (UTC)
 * Under this policy, also note that while it only takes 3 sysops to block a user, it only takes three to unblock a user. As I heard someone say on the IRC channel: If a user does not have even 3 sysops out of 250+ to support his unblocking, then there's probably something wrong. - Fennec


 * A while ago, people were saying similar things but "If a user does not have even one sysop out of 100+ to support his unblocking..." The times they are achanging. Martin 21:01, 9 Apr 2004 (UTC)

What was the original policy text? It has been fiddled with to what purpose I am not clear but I'm curious if it was fiddled back or changed from the original proposal? - Tεx τ urε 22:33, 9 Apr 2004 (UTC)


 * original version


 * I restored the proposal as it was when Danny started the vote. Point 2 was truncated and point 3 was removed by The Trolls of Navarone at 22:22, Apr 9, 2004. Someone did fix point 2, but left point 3 out. I'm particularly annoyed that it was changed after people started to vote on it. -- BCorr | Брайен 15:08, Apr 10, 2004 (UTC)

Alternative idea: troll reform
First, I do not agree with the notion that quickpolls are not working. Some important policy changes (such as the requirement that someone uninvolved must start a quickpoll) have only been made recently. There are also still some misunderstandings as to the situations in which a quickpoll can be used. I designed the system specifically with the goal in mind to deal with three long-standing problems:
 * edit wars
 * sysop abuse
 * emergency situations in which quick action is needed

Quickpolls were created primarily as a response to growing concerns about developers and sysops acting too unilaterally in enforcing Wikipedia policy. They were not created to deal with trolls.

Have quickpolls succeeded in ending edit wars? Of course not. But that is not so much a problem of the polling system, but of the revert policy itself, which is still too vague and easy to game. People who would revert 10 times a day before the 3-revert-policy revert 10 times over 4 days now. A prolonged edit war is still an edit war. That is something we need to deal with.

As for trolling, that is a much more difficult situation. Trolling is a kind of all-purpose term. If I add controversial information about a person to a page, someone who likes that person might call me a "troll", only trying to cause problems. If I am new to Wikipedia, and maybe a little bit arrogant, the other person, a sysop, might quickly convince two regulars that I am a troll and "strike me out". I doubt that someone else would go to the effort to help me if they also need two other people first to support me. "Somebody else's problem."

No, I don't think that's the way to deal with trolling. I have an alternative suggestion. It seems to me that trolls like to hang out primarily on the Wikipedia: pages, on trolling-related articles etc. How about this: a new option for quickpolls to put users into an article work probation. A requirement would be that the contributor has not done any serious article work recently. If the quickpoll is successful, for the duration of the probation period, all edits outside the article space can be reverted. If the user re-reverts, he can be instantly banned, permanently if necessary. Exception would be legitimate attempts at communication, but this would be at the judgment of other users.

Inside the article namespace, normal rules would apply. Users would make all attempts to deal with the suspected troll in a friendly manner. However, they do not have to do so infinitely. After a number of at least 50 contributions, another quickpoll can be started. If the conclusion is that the user does not make attempts at serious contributions in the article space, a ban is the final option. I think this way we could force trolls to spend their time productively or go away. What say you?

One concern will be that this way we will harm Wikipedia by inviting trolls to work on it. But the goal here is really to turn trolls into useful contributors. 50 contributions isn't that much and should allow us to make a judgment.

I must also warn you that bans aren't all that effective on a technical level. We are likely dealing with a handful of trolls who keep coming back under different usernames. Getting a new IP is easy, and unless we definitely know who we're dealing with, we can't set up range blocks either (which are also very problematic, because they hit lots of innocents). How about a different route? Tell the trolls: You are welcome if you make serious contributions. Otherwise, go away.

Also, I would suggest prohibiting usernames which seem to imply that the user is deliberately trolling. Trollslayer seems OK, but JRR Trollkien is definitely over the line.

On a more general note, when I designed the quickpoll system, I built and improved it until a consensus (20 in favor - 2 against) could be reached. I do hope that this proposed system is not put into place on the basis of a simple majority.--Eloquence* 17:13, Apr 10, 2004 (UTC)


 * I think there's something to be said for disallowing usernames that indicate the user is not being a serious contributor with Wikipedia's interests at heart - user:Just a vandal, user:I hate Wikipedia, etc. Something for wikipedia talk:username, I guess. Martin 17:25, 10 Apr 2004 (UTC)


 * Or people who have the word "Troll" in their user name. Danny 17:26, 10 Apr 2004 (UTC)