Here's a proposal I made on [[Wikipedia talk:Administrators]], one week ago. I think that given recent events, it's probably a good idea for me to give it a bit more publicity.
To blame Anthere's current distress solely on the banning system would be grossly missing the point. However I think it was naive of me to think that allowing sysops to ban logged in users would have no political effects. Clearly policy alone is insufficient to maintain the situation we had before user bans came in.
This was part of a discussion regarding the danger of hijacked sysop accounts.
--START QUOTE--
Perhaps we could use another model. The model I'm thinking of would be equally applicable to banning and desysoping. We need a method of enforcing the consensus model without leading to destructive short-term unilateral bans. Perhaps a sysop should be able to pre-register their support for a given user, and if any sysop (other than themselves) is supporting the user, they cannot be banned or demoted. This "support" would expire shortly after the supporting user logs off, so if the situation changes suddenly, we won't be left with the situation of frantically trying to contact sleeping Wikipedians. -- Tim Starling 03:22, Oct 9, 2003 (UTC)
--END QUOTE--
I've been thinking since I wrote that, that perhaps the expiry time should be a bit longer, say 24 hours.
A bit of background: ordinary editing works by the consensus model, because any disgruntled individual is free to revert an article. For a reasonably stable article to result, there has to be a consensus. In the meantime, the article fluctuates to alternately reflect the opinions of the different parties involved.
Sysops can block logged in users, and sysops can also unban logged in users, hence on the surface it seems as if the situation is the same. For a stable user status to arise, there must be a consensus among sysops.
However, in the meantime, the user in question may be banned for short periods of time. This is emotionally very trying for the user involved. In fact, the two times it has happened so far, it has led to a contributor becoming very angry and leaving immediately.
Short term bans of users with an emotional investment in Wikipedia are extremely destructive. They cause contributors to leave. They should be prevented at all costs. We need a system for ensuring that there is a consensus in favour of such an action, before it takes place.
My model is certainly not the only one. It would be possible, for example, to require quorum or petition of sysops in order to create a ban. However, this slows down response times, which is bad for preventing Michael-style vandalism. In choosing the size of the quorum, there is a trade-off between speed, and the danger of allowing cliques or factions to act arbitrarily.
My scheme assumes that a ban of a known contributor will occur after some debate. Sysops involved in the debate who disagree with banning should register their support of the user in question, before those in favour of the banning tire of the discussion and decide to act unilaterally.
If a sysop acts unilaterally very early in the debate, or without any discussion at all, I imagine this would be seen by the community not as an attempt to circumvent a technical control, but as a reprehensible unilateral action with no community support.
Finally, I'll quote the only real response to my proposal on [[Wikipedia talk:Administrators]]:
--START QUOTE--
:: Nice idea, but I'd want to be sure that '''all''' sysops have a working email contact address. Talking about such issues publically might severely inflame the situation. If X knows sie's about to be banned... ouch. [[User:MyRedDice|Martin]] 19:22, 10 Oct 2003 (UTC)
--END QUOTE--
In response: I would hope that people would use the "support" function frivolously, on a whim. They shouldn't have to explain themselves. It could be used whenever bad things are said about a user, just to indicate emotional support. During a debate, some contributors feel as if everyone's against them. They begin to feel emotionally spent and may even consider leaving. I think wordless registration of support from a wide range of users could help morale.
-- Tim Starling.
We need -something- along those lines... your proposal sounds like a basic web-of-trust system... which isn't a bad idea...
-- Jake
Jake Nelson wrote:
We need -something- along those lines... your proposal sounds like a basic web-of-trust system... which isn't a bad idea...
-- Jake
Come to think of it, a simple, plainly visible log probably would have prevented the whole situation in this case. I assume Anthere unbanned "Mediator" before she started using it. If Ed had known Anthere's reasons for doing that, perhaps this wouldn't have happened.
Maybe the "web of trust" can be second priority.
-- Tim Starling.
Tim Starling wrote:
Come to think of it, a simple, plainly visible log probably would have prevented the whole situation in this case. I assume Anthere unbanned "Mediator" before she started using it. If Ed had known Anthere's reasons for doing that, perhaps this wouldn't have happened.
Maybe the "web of trust" can be second priority.
Very true. A public record, say [[Wikipedia:Ban log]] for consistency with the other logs... this would be a Good Thing.
-- Jake
Jake Nelson wrote:
Tim Starling wrote:
Come to think of it, a simple, plainly visible log probably would have prevented the whole situation in this case. I assume Anthere unbanned "Mediator" before she started using it. If Ed had known Anthere's reasons for doing that, perhaps this wouldn't have happened.
Maybe the "web of trust" can be second priority.
Very true. A public record, say [[Wikipedia:Ban log]] for consistency with the other logs... this would be a Good Thing.
Done, in the slow branch of CVS, so it might be a couple of weeks before it goes live. The following messages need translation: blocklogpage, blocklogentry, blocklogtext, unblocklogentry.
It's probably the wrong list to ask this, but is there a standard procedure for requesting translation of the user interface?
-- Tim Starling