I hate to be overly simplistic but I find in these circumstances that IAR applies.

Just be courteous to all users involved, even those accused of incivility, and use the Socratic method. Question them about their actions in a way that suggests that you are not taking sides (which as an uninvolved administrator or editor should probably be the case anyway) and ask them about their assessment of the suitability of their behaviour. Usually when confronted with having to do a self-assessment most will agree to at least back off from the situation to get some head-space. Having a self-imposed break is much simpler and produces much better outcomes than having an administrator-enforced one.

I know that's a highly interpretive way of looking at things but if we over-think these things and try and put human nature into categories (not that Risker didn't do a damn fine job there) we'll just end up where we are now; constrained by policy and unable to tackle the reality of the situation.

Anyway, that's just my two cents. Feel free to shoot me for it.

From: gendergap-bounces@lists.wikimedia.org [mailto:gendergap-bounces@lists.wikimedia.org] On Behalf Of Risker
Sent: 28 October 2011 22:26
To: Increasing female participation in Wikimedia projects
Subject: Re: [Gendergap] the state of civility on en.wiki

There are a lot of challenges in being able to develop a consistent process of managing user behaviour.  Here are just a few that I've noticed over the years:

• User acting entirely within editing policy, although usually at the "bolder" end of the spectrum, being accused of behaving extremely inappropriately, often with the words "civility" and/or "courtesy" thrown in.
• Users relying on one editing policy to edit content in a way that could reasonably be predicted to arouse dissent, and then accusing other editors of "failing to follow policy" because they point to a different policy.
• Two or more users starting off with minor barbs (usually starting with allegations of policy/guideline violations and becoming increasingly personal), continued escalation over the course of several posts, then only one/a few of the involved users getting warned/blocked for "incivility". This one is particularly insidious, as it has the reasonably predictable effect of creating significant resentment on the part of those blocked (the now-sullied block log tends to be used as a club) whilst also appearing to support the behaviour of the non-blocked participants.  Both groups tend to feel the action justifies them continuing to follow the same behavioural pattern.
• Long observation of wiki-history indicates that systemic problems are rarely acknowledged, let alone acted upon, by the community unless one or a small group of editors exceeds usual behavioural norms to focus attention on the issue. To put it bluntly, it takes a lot of noise to get the community's attention on systemic issues long enough to address them, even partially.  This method has variable success, ranging from serious community discussions and policy/practice changes through blocking or otherwise sanctioning the users who raise the issues.  If not done well, the attempt at problem resolution devolves into discussions about the appropriateness of the initiator's behaviour rather than the underlying problem.  Initiators are regularly referred to as "uncivil". 
• The use of the term "collegial" to describe the editing milieu. Anyone who has spent much time in the academe will recognize a lot of the "problem" behaviours we see on our own project, particularly personalization of disputes, which is one of the major elements leading to the perception of incivility.  Indeed, some of our most significant problem areas involve editors with academic credentials behaving pretty much within the norms for their profession, i.e., pretty unpleasantly toward those who don't agree with their educated opinions.

In other words, as a community we create a climate where poor behaviour is the most effective means to motivate needed changes, where our policies and practices can be used as weapons both to support negative behaviour and also to "punish" positive behaviour, where the boundaries of unacceptable behaviour vary widely dependent on a large number of factors and enforcement is extraordinarily inconsistent, and where we openly claim to follow a behavioural model that *sounds* progressive but is in reality possibly even more nasty than our own. 

On reading far, far back into archives, it appears that "incivility" has been a problem almost since the inception of the project.  In the early days of the project, blocks and bans were almost non-existent, and huge amounts of time were invested in trying to "correct" behaviour (considerably more per capita than today, the community cuts its losses much earlier now than in 2002-04). In fact, blocks and  bans were very rare until the arrival of extensive trolling and vandalism in 2005-06, which led to the appointment of a massive number of administrators in 2006-07 in order to address these problems. 

None of this speaks to solutions, I know.  But it is important to put the discussion into a more historical context, and to recognize the flashpoints where incivility is often identified. 

Risker/Anne