please edit this page if you want to
It seems to me that our current "zero tolerance" policy towards "banned" users does not have a success rate that we can particularly brag about. We've got DW (and co), 142.177.etc, Michael (and co), and so it goes on. We block them, they come back under new names and laugh at us. But what are we trying to achieve?
- Are we trying to ensure the integrity of the banning system?
- Are we trying to punish people for being evil? (or misguided, insane, anti-social, violent, or whatever...)
- Or are we trying to build an encyclopedia?
Our aim shouldn't be to "crack down" on trolls, for the sake of cracking down. Instead, we should practice en:harm reduction. When usenet lore talks of "don't feed the trolls" it means ignore them - not deliberately seek them out in order to ban them or revert them. Let's look at some harm reduction ideas for how to reduce the burden of troublesome users.
Harm reduction techniques 
I'm just thinking though some of this stuff, so my past actions are liable to be inconsistent with this! :)
Some edits were reverted under the concept that one should revert all edits made by a banned/troublesome user, including good edits. For example, edits to en:Chaco War and en:Amarya were reverted. We can improve efficiency by not reverting reasonable edits. Plus, the encyclopedia will then improve slightly in quality. 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.
A substantial number of edits were made to the user pages, such as en:user:Zog, etc. Less harm is caused by bad stuff here. So, we could be more efficient here by simply waiting a week, and then reverting the whole lot in one go. If Zog edits his user page five times a day, and we wait a week before reverting, then we've magically become 35 times more efficient, just by being lazy. In any case, getting into an edit war with Zog over a non-essential page like user:Zog encourages hir to come back.
By quickly making a [[/ban]] page we can save more time. No need to have the same discussion on a dozen pages, plus the mailing list. As a bonus, if the user in question wants to reform then such a page clues them in to what they're doing at no extra cost. Which means we don't have to waste time telling them what they're doing wrong, which brings me nicely on to the next header:
Troll talk 
Here's another way of reducing the time we spend dealing with trolls - don't talk to them. "Why don't you go find another sandbox to stomp in?" says one Wikipedian - every second spent writing that sentence was wasted time. Heck, it asked a banned user a question - it practically *invited* hir to respond. And, unsurprisingly, respond sie did - several times - and we wasted more time reverting hir. And then reverting hir deletion of the "sandbox" question a few times.
Flames of trolls are pointless too. We all know that trolls troll in order to receive flames. Therefore, telling them to "go away, troll" is likely to be counter-productive. Is there any evidence that sie will go away if asked? Perhaps somehow sie has got the mistaken impression that sie's welcome here, and all we ever had to do was suggest that he should depart and he'd pack up his trouble in hir old kit bag and leave, leave, leave? Trolls may be evil, but saying "Get thee behind me Satan" won't have a high success rate.
Ignore trolls. Don't flame them. Don't ask them questions. If you must communicate with them, do so calmly and briefly. If they flame you, take the fire out of their comment by rewording it - and then ignore it or give a minimal "thanks for your feedback" response. Alternatively, take their flame, cut and paste it onto the /ban page and say "this is not acceptable" - and then proceed to ignoring it. If it's a case of mistaken identity, then this'll become clear in due time, and then you can answer the question. If not, you've lost nothing.
Calling contributors "Trolls" feeds the syndrome. It is no less than calling a person "nigger", "kike" or "honky". It is a way to dehumanize the contributor and reduce them to an object to be kicked about thoughtlessly. If a member identifies their self as a troll, that might be okay, but even then, to accept their self-description tends to feed the disfunctional relationship. It might be fair to assume that people who identify others as trolls are here not to write an educational document, but to engage in personal power struggles to satisfy inner psychological urges.
In five words or less 
"Ignore trolls, don't ban them" Or "Keep good edits, revert bad"
Rainclouding troublesome users 
Banning people keeps Wikipedians happy. They say "X is a real smeghead". And chances are sie is. But just telling people "revert all hir edits"; isn't very satisfying. So instead, we block their IP, they get round the block by redialing their modem, and then we say "revert all hir edits".
I propose that we skip the troublesome "block" stage, and go straight to a mild form of "revert all hir edits". So at the top of the user's page, we'd write:
- This user is under a raincloud - see user talk:X/ban for details. All edits by this user may be reverted without loss of karma, though they do not have to be. (obviously this would be boilerplate text)
Basically, what we're saying is that you can revert this person without having to follow standard wikipetiquette. In other words:
- You can ignore them on talk pages (or move their questions to their user talk page, etc - see "Talking to trolls" above)
- You can revert them without having to give justification (though you can if you wish)
- Nobody will blame you for reverting good edits (though others may reinstate them if they think they're worth saving)
The above rules are a compromise between "revert all edits" approaches and "revert the bad" approaches. RevertAll people get to revert without wasting time examining edits, and the RevertBad people get to keep edits they particularly like without having to engage in edit wars with RevertAll people.
Note that adding such a block to the top of a user's page doesn't require sysop powers. So it's another way of making sure that we're immune to "clique" accusations, and make the process more open. If you don't agree that a user should be raincluded, then you can wipe the notice. In extreme cases, we can protect the user page so the rainclouded user can't wipe the notice. But even that might not be necessary - so try it without protecting first. Not protecting even though we can sends a clear message, I think - that we're so confident in the community and the power of the wiki way that we don't even need to protect the page.
Last paragraph comment :
You mention that adding the block to the user page does not require sysop power. That is quite true. But currently, banning a user is a "community" decision (so is it said). Your sentence seems to imply the decision to cloud an annoying user would not a community decision, but rather a decision that anyone could take as soon as he is *too* annoyed with someone (or at least, I think it will be understood as this). I think if the act of putting the notice is done by a sysop (unilateral action), then removed by a non-sysop (unilateral action), this could create a sort of edit war between the sysop and the non-sysop. Edit war in which the sysop will very likely win, if only by protecting the page, or by pressure.
Notice also, that the "way to be sure there is immunity to clique accusation" won't work if the page is protected, as the non-sysop will be officially put away from the process of deciding whether the annoying user will be banned or not. I fear this will somehow push the non-sysop in the "camp" of the annoying user
Well, sure anyone can make the decision to raincloud someone, but that decision gets peer reviewed just like every other edit. -- User:SunirShah
- I'd anticipate that disagreements would be thrashed out on the /ban page. If there was an edit war, it'd attract more editors, and it'd be fairly obvious whether or not there was a consensus in favour of rainclouding/soft banning. Unlike a hard ban, soft bans are not implemented using special sysop powers of protecting pages or blocking IPs, so if there's no consensus then they can't be enforced anyway. --myreddice