Talk:Bans

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search

The only problem we have is that banned users are masking their identity by creating new logins and editing from ISPs that assign new IPs for each dial-up. Unmasking new user's IPs will rectify this situation. Then we will be able to wear-out these banned users by not accepting anything they write. Reverting 142.177s edits has worked very well - he went from dozens of edits a day a handful of edits a week. The only problem has been that one new user has been reverting the reverts - thus rewarding 142.177 for subverting the ban. That will only encourage him to edit more - just as your proposal would do. --Maveric149

Why do you have a problem with 142.177 editing the encyclopedia? MyRedDice
Because he is banned for making an implied death threat and for generally bad behavior. Keeping him around is far more trouble than his edits are worth. Or do you think that this type of behavior should be tolerated? --mav
How do you distinguish between good edits by 142.177 and good edits by other people using the 142.177.*.* domain? (which has 60,000 IPs in it, I believe...) MyRedDice
I let any edits by a 142.177 IP stand whenever I'm not convinced those edits were made by the banned user. But when a 142.177 IP edits the same type of article that the banned user is known to edit (anything dealing with green economics, for example) and uses the same type of wording the banned user has used then I become suspicious. So I place a note on the IPs talk page explaining that a person using the same IP range has been banned from Wikipedia and that somebody may mistake you (the 142.177.xxx.xxx user) for this banned person and may delete their edits. But 142.177 just can't help himself not to insult me and always blows his cover. This person is also heavily editing Disinfopedia (making it very obvious that he has a beef with Wikipedia and certain users of Wikipedia) so whenever a 142.177 IP makes a reference to Disinfopedia I regard that as evidence beyond a reasonable doubt that that person is the banned user. To date only a very few 142.177 IPs have not given an obvious indication that they were being used by the banned user. --Maveric149

How does unmasking IP addresses solve anything, if they get a new one every time they dial up? I can't see how any technical solution can enable us to recognise a banned user if they connect with a different IP address and a different user name after their ban. So to identify them we'll still have to follow new users around watching what they're doing, and we'll still have to rely on circumstantial evidence to identify them. And then we ban them, and then they sign in again with another IP address and another user name, and we are immediately right back to square one. So what have we gained by banning them? Nothing at all. In fact, we've lost out, because not only can the banned user come back with a new name, with just as much ease as when they signed up the first time, but also they could get in lots of bad edits under the new name before anyone twigs that it's the same person! So why not just let them sign in with the old name? That way, we know who they are at all times, and save ourselves the rigmarole of having to identify them over and over again!
Any arguments about "subverting bans" would then become irrelevant. People can stop having to be constantly paranoid about whether someone may or may not be some previously banned user, and just concentrate on the quality of their edits. If they're good, leave them, and if not, improve them - or revert them if they are unsalvageable. It seems to me that reverting bad edits and leaving good edits "punishes" badness and "rewards" goodness. People would soon get bored of posting naughty things if they were always reverted quickly, and they would probably eventually either give up and go away, or realise that they're better off posting good things, because the good things aren't reverted. So it would only encourage people to be good.


So generally I think Martin's essay sounds quite sensible. Except of course in that I disagree about Lir being a troll. :) Haha, I bet that last sentence is going to result in everyone disregarding this message as the rantings of a madman... -- Oliver P. 21:07 2 May 2003 (UTC)
But when we already know that a particular user intentionally inserts bias or wrong information in an article or otherwise causes a great deal of lost effort for other contributors then it is sensible to ban the person. Helga, for example, just could not learn to de-bias her edits. Many users felt obligated to follow her around and fix her bias - myself included. The reason we did this was to protect the quality of Wikipedia and to reduce the likelihood that Helga's edits would encourage other users to abandon NPOV (creating many bad examples). This was a contributing factor in the loss of at least two longtime contributors and the near loss of several others. Some people are simply anti-social and having them around tends to drive away people who hardly ever make an incorrect or biased edit.
(cutting in) "Banning" Helga appeared curiously ineffective in achieving the stated aim of not "having her around". She was banned on Sep 9, her last logged in contribution was Jan 25, and I'm told she then started making contributions from anonymous IPs. How did "banning" Helga help to avoid driving people away?
? Just because she edited for a few days after the ban does not mean that the ban failed. The ban did work because she was once a very heavy contributor and caused a great deal of extra work for anybody who cared anything about history. But now I don't have to check all the edits made by Helga every day. I can concentrate on other things. That is improvement. --mav
By my count it's a few months, not a few days... have I got the dates wrong? MyRedDice
I was less than clear. A few months after she was banned she edited for a few days. --mav
I personally don't know the details when it comes to dates with Helga, but when I first arrived at the Wikipedia last July, Helga was causing a lot of users a lot of trouble and wasting a lot of time; since she has been banned there's hardly been a peep from her. This is one case where a ban has been effective (and I speak as somebody generally sceptical of the efficacy of bans). --Camembert
(to self) - actually, there seems to be some confusion with dates here, and I'm not sure if it's me confused or somebody else: Helga's last logged in edit seems to have been Jan 25 2002 - my recollection is that she was editing from anon IPs after that but was not banned at that time - she was only banned after I arrived (probably around September 2002). Am I mistaken? --Camembert
Rats! Yep, you're right, Camembert - I screwed my dates up. I'll accept that it looks like Helga's ban was indeed effective. MyRedDice
Our policies are useless if they are not enforced and those people who constantly ignore and subvert policy should be banned - it is simple as that. Oh and unmasking IPs of new logged-in users will make it easier to see if a suspect new user is editing in the same IP range as a previously banned user. This essay is way too idealistic. --mav


Point is, I have been there for more than a year, I have no memory of a very community shared discussion on how to deal with problematic users, except for ultimate banning by Jimbo (I remember in particular Helga's banning) . In particular, I have no memory of any wide discussion leading to the decision that "any edit by a banned person had to be reverted on sight, whatever the quality of the edit" and that "any new article by a banned user had to be deleted without being put on the vote for deletion page, whatever the quality of the edit". I understand quite well the interest some could find to this option, but I have no memory having ever been able to give my opinion about it, and I can see it just ain't working. So, if I agree with "our policies are useless if they are not enforced", I question how and when this policy was ever discussed and decided. User:anthere
Yes there has been a great deal of discussion about this every time we have banned somebody. It is silly in the extreme to say otherwise. Did you not notice the disscussions? Also the onus of reverting is not on every contributor; We do not require every user to do vandalism patrol (but this is one way people can contribute). And it is working very well; have you seen Helga around lately? When was the last time you saw Clutch edit? Even 142.177 has drastically reduced his edits because he knows full well that any edit he makes is subject to deletion. And this policy was decided by Jimbo. --mav
I disagree. There has indeed been a great deal of discussion *before* a user was banned (e.g., how to deal with the bad edits, what about banning this user...), but very little discussion *after* the ban, when it was seen the ban could not work because of multiple IPs issues. The discussions which take place where only about "how to enforce better the ban (ie, banning a whole range of IPs, banning user name available by sysop, actually seing the IP of a user...). All these were discussed. What was *never* (I think) discussed is what do we do in case all these don't work (in other words, the ban user go on editing). For example, I totally disagree with the policy of instant reversion without any check of the quality of the article, + the addition of the comment to the banned user, go away. This go away is very very bad imho, and I am *absolutely* sure it was never discussed, never agreed together. It was not either decided by Jimbo. It would also be "silly" to pretend otherwise. User:anthere
Oh, I'm very realistic. I consider that with six billion people in the world, we will never be able to identify and punish all trolls. Therefore, instead of trying to rigorously identify all trolls, I feel the pragmatic approach is to consider how to minimise the damage they cause, and maximise any benefits. Hence, harm reduction. MyRedDice
The way to minimize their damage is to not let them edit. We simply don't have the human resources to waste on reviewing edits made by a user who is known to either input crap or who is otherwise not a team player. With that said, I've already spoken out against the Adam Family witch-hunts (which are a distracting waste of time IMO). I will give any "new" user the benefit of the doubt until they prove to be a nuisance. And if they also prove to be a previously banned user then they obviously have not learned a thing and the ban is still in force. Any banned user can change ISPs, create a new account and then simply behave to quietly get around the ban anyway. --mav

I'm going to quote from what I wrote before, because I still think it's true:

I'm not saying that you have to check every edit before reverting. I'm saying that if you do notice that an edit seems reasonable, there's no need to revert it.

I'm not suggesting that you have to review every edit made by a banned user. But sometimes I find myself reviewing an edit before I realise that it's made by a banned user. If I've already spent the time reviewing it, and I already know that it's a good edit, then why revert? Why shouldn't I save myself the time and keep the edit? MyRedDice

For people who care about the quality of Wikipedia we do have to check each edit of users known to input bad edits.
BTW you are not obligated to help enforce a ban on anybody. But at the same time you shouldn't knowingly undo efforts by people who are trying to enforce a ban because doing so encourages the banned person to edit more and it implicitly condones whatever bad behavior they were banned for. We are a community that is trying to write an encyclopedia and some users are simply a drain on the community and should be expelled. If they are not expelled then their actions will harm the quality of the community and therefore harm the quality of the encyclopedia. It is as simple as that. --mav



An idea, since you have proposed to ignore troublesome users: since Wikipedia is an application running on a database, could one of the developers create a program for use of the Sysops (or a designated groups of members) that would:

1. Lock out the IP address of the troublemaker,
2. Roll back all articles that troublemaker has changed, then
3. Release the lock on the IP address.

This query would also document every instance of its use, to discourage abuse.

The advantage of this tool would be to enable reverts at a much faster rate than the troublemaker could make changes, & thus increase the disincentive for making trouble.

Obviously, if the group consensus is not to ignore & revert -- this discussion has only begun -- then there is no point to writing this application. -- Llywrch 20:58 2 May 2003 (UTC)


OK, I'm confused. Martin's "Rainclouding troublesome users" section seems to argue for SoftSecurity bans (reverting all edits made by a banned user). I thought Martin was arguing for us to abandon bans alltogether but it looks like he is arguing for us to move away from HardSecurity IP/user names bans. Given how easy it is for a user to switch IPs or create new user names I would have to say that I don't have a fundamental problem with this (I still disagree though - IMO this is a techical problem and a technical fix or two can make these types of bans more effective and more targeted). However, HardSecurity measures would still be needed in emergency situations no matter what (like with the MIT vandal or the still hypothetical vandal bot). So, Martin, is it correct to say that you really only have mainly technical issues with the banning system and do not wish to abolish the concept of bans? --Maveric149

Not quite - but I agree that we are probably quite close. I don't want to abolish the concept of bans, but I do want to change it, hence the name change to "rainclouding". Key differences:
  1. Others may reinstate good edits made by the rainclouded user. You shouldn't revert such reinstatements without justifying it in the normal way (unless the reinstater is also rainclouded)
  2. Rainclouding is compatible with both "revert all edits" approaches and "revert the bad, keep the good" approaches.
  3. There is no need to revert edits instantly - it's compatible with a lazy approach to reverting
I guess the first point is the one that you'd be most unhappy with. However, I think it's crucial. Currently, your policy of reverting all edits clashes with my policy of keeping the good and reverting the bad. That makes me unhappy - I've spent some time reviewing an edit and deciding that it's good, and yet I can't reinstate it. Further, it creates unnecessary edit wars between "revert all edits" and "revert the bad" users that we can manage without. So I hope you'll consider this as a good compromise.
Of course, nobody should reinstate bad edits by a rainclouded user - doing so generates bad karma. MyRedDice
That makes zero sense. We already RevertBad edits of every damn article edit we review. How in the world is this type of "banning" different than normal PeerReview that is done on nearly every article every day? --mav

Let me try and clarify the difference...

  • Normal contributors get the following treatment:
    • Good edits - kept or improved further
    • Bad edits - generally fixed - some may be reverted
    • Talk edits - kept or replied to
  • Rainclouded contributors would get the following treatment:
    • Good edits - generally reverted - some may be kept or improved at our discretion
    • Bad edits - reverted
    • Talk edits - generally reverted - some may be ignored, moved, or anonymised
  • By contrast the current treatment of "banned" users is:
    • Good edits - reverted (sometimes reinstated, then reverted, then reinstated, then reverted, ...)
    • Bad edits - reverted
    • Talk edits - reverted or flamed

In other words, you're free to revert every edit by a rainclouded user - I'm happy to make that concession. All I ask in return is that you allow me the freedom to reinstate edits that I particularly like. That's not unreasonable, surely? MyRedDice

Fair enough. Then your idea could be applied as a last resort before a hard ban (like we have now). If your method is truely superior then we would hardly ever need to go to the next level (I seriously doubt this but I'm willing to experiment with the next ban). My prediction is that your method will not be able to work for any particular ban for long and will be quickly upgraded to a hard ban (for example, nearly every damn edit made by Micheal was plain wrong and had to be checked - it took a hell of a lot of effort to follow him around and evaluate his edits). --mav
Ok, I'm happy with that solution. I'd probably apply "rainclouding" before a user got to the "ban" stage - one advantage of softer solutions is that you can apply them slightly more liberally. The genius of my idea is that it becomes equivalent to a "ban" if everyone agrees that the user's edits are worthless. Ironically, if you have a consensus that a "ban" is necessary, then there's no difference between a "ban" and "rainclouding"! :) MyRedDice
I also see that you are spending a great deal of time moving talk comments by and about banned users onto that user's pages. This seems to me to argue against your premise that the current ban system is resulting in lost encyclopedic content - what about the content you would have created had you not been occupied by moving talk comments? Reverting edits made by a banned user only takes seconds - what you are doing takes a great deal of time. --mav
It looks like more time than it is because it's all in one go. But I did it at a time of my choosing, when I was in the mood for that kind of thing, rather than one of Zog's choosing. So at the end of it I feel rather pleased - whereas edit wars and reversion wars all ways leave me feeling frustrated.
Besides, it's not like my content is any good most days, anyway... :) MyRedDice


I appreciate your proposition Martin. Leaving an uncooperative user in the silence is an interesting option. That is precisely what a ban is, rejection of a community. Disinterest.

Current option of instant deletion is clearly not working. Plus, it is resulting from time to time (granted, not often) to the destruction of good edits. Plus, the current policy was perhaps not discussed.User:anthere

This is just one method to enforce a ban. But many (most?) banned users don't give a damn about other people so disinterest on talk pages will not stop them from doing whatever activity got them banned (Michael is a perfect example). And stop trying to pretend this policy was not discussed - we discuss a great deal about what to do with a troublesome user every time before a ban. --mav
Please also stop trying to say any of my comment is just silly or just pretense. You give the answer yourself, *before*. Now, show any place where was discussed and community approved that any banned user edit should be reverted on sight with the comment "go away". When my kids are naughty, I give them a time out; tell them I am not talking to them anymore, and that they may in their room for punishment. But, when they use the time out to "clean" the room, or to do any otherwise good action, I don't tell them "you bad boy, go away", and I don't put back the mess in the room. I do not forget they were bad, but I recognise the fact they can be good in their badness
I deeply regret Wikipedia is such a male-ish place. With a couple of more mom, I think it could softer and more tolerant place. Or maybe more social workers, who know there is good in any bad person, and learn to reward the goodness of a socially bad person, rather than punishing him again again and again. User:anthere
However, choosing RevertAll over RevertBad was a Jimbo decision, rather than something we came to by consensus, or some kind of compromise. Or so it seems to me. MyRedDice
No. RevertAll is used for users that we cannot IP ban for technical reasons (like 142.177 for example). These people are just as banned as the IP banned users but the only way to enforce the ban is to revert every edit (which should not have been made in the first place). This is SoftSecurity. (aside: it would be real nice if CamelCased words on meta automatically linked to MeatballWiki) --mav
My argument would be that it is SoftSecurity overzealously applied
The reason: It is a stand-in for HardSecurity when that method (IP ban) is not practical. I in fact use this for almost all but the most persistent vandals; reverting every edit they make (via the rollback feature) gives them the message that their graffiti will not be tolerated while at the same time not giving the Big Brother feel that an IP ban gives (which always should be a last resort after SoftSecurity has failed). IP bans, IMO, have a much greater likihood of evoking the "I'll show you" feeling in a vandal which increases the chances of them coming back with bigger and better "guns" (the much feared and hypothetical vandal bot that creates log-in after log-in and is on a rotating IP over several ranges creating dozens of edits a minute - like the MIT vandal times 100). --mav

I've placed some text at Bans and blocks that, in the fullness of time, after everyone here has ripped it to shreds and stitched up in a better way, I'd like to place at en:Wikipedia:Bans and blocks as our official policy statement. Currently we're following an unwritten policy, which is kinda dubious... MyRedDice


May I suggest that instead of having to decide whether to revert or keep edits of an exiled user, you move all his edits to his namepage? This is kind of like grounding the user. If you later forgive him, he can restore anything worth restoring. If you later forget him, you can restore anything worth restoring. However, if you allow the incorporation of an exiled user's occasional good edit into the corpus, then she will likely find it much harder to leave. After all, not only is it a mixed message, but her work is now being used. After she is long gone, you can incorporate her work without her noticing, of course. -- User:SunirShah

Hmm, this does require extra effort - both keeping edits and reverting edits take relatively little effort - moving takes a bit more, especially when Wikipedia is slow. It also is impractical in some cases - eg, when a banned users comments are necessary to retain the "flow" of a talk page, or when the edits are primarily typos/grammar. But it'd be good to mention it as an option - as long as people don't feel compelled to do it...
I think mixed messages are inevitable in a community the size of Wikipedia though - we've probably passed the tipping point where it's possible to have a single community view. MyRedDice

Martin, what is the current situation of this experiment ? Did you practiced it extensively ? Reactions ?

  • very positive reactions to moving ban talk to a /ban page - I think I've got more kudos for that on my user page than anything else I've done on Wikipedia!
  • I've been moving trollish comments on my user page to the user talk page of the person who made them, with few objections even from the person who made the comments.
  • Removing signatures from comments made by banned users has got some negative feedback - I think it's probably inappropriate on en.wiki because signing comments is the rule rather than the exception - so I've stopped doing that.
  • Modifying comments by removing the flamelike bits has had mixed feedback, but I still think it's a good idea, provided I clearly mark "(paraphrased)".
  • I've yet to try to implement "soft bans" and "hard bans" (as described in bans and blocks). This is probably the most contentious point, so I'm a little wary at the moment... MyRedDice
Perhaps, we should make a clear difference betwen what a hard ban and what a soft ban is. Not that I like the word soft ban either...but I can't think of another one (soft banning is not really about banning (kicking out, whether gently or not), it is about leaving someone alone in a corner...that is not the same thing at all. Ah! remind me of school expulsion (hard ban) and little boys left aside 'cause they didnot please the band leader (soft ban). :-(((

I was directed to this page by Mintguy as a result of a discussion we have been having about the correct procedure to observe regarding alleged spurious edits by Harry Potter. Is there already an agreed upon mechanism for lesser than banning procedures (like rainclouding?) What is the public page that attests that? -- Cimon avaro 15:59 3 Jun 2003 (UTC)

bans and blocks has a section on "soft bans" aka "rainclouds". AFAIK, it's been used precisely once (on en:user:No-Fx, prior to a hard ban). It's all still very much up for discussion - indeed, I'm about to edit it to reflect recent discussions about /ban pages... MyRedDice 21:37 3 Jun 2003 (UTC)



These are good ideas in my eyes . Ilyanep from the english Wiki