Generally, why and when should someone be blocked?
To prevent them causing further damage to the project, when they have been told, or should obviously know, that this is what they are presently doing. 'Damage' is usually phrased as 'disruption' and its meaning is writ large. In essentially all cases apart from those involving legal compulsions, however, the test boils down to: "are other editors/admins using unnecessary resources to deal with the actions in question". If the answer is "yes" then we are on course for a block in the face of persistence (past or future). In the case of "no", then education is the correct course if the actions are somewhat problematic, until we iteratively re-apply the test and conclude that requiring further education would give a "yes". This test applies right the way from simple vandalism, to edit wars, to persistent trolling, to arbitration remedies: other editors are having to waste their time and patience dealing with something that they should not have so to deal with. In cases of defamation and copyright problems, the blocks tend to based on a simpler test of "does this imperil the project?" and the answer is generally straightforward, one way or the other. Splash - tk 22:30, 7 May 2008 (UTC)
When would it be appropriate to protect a page?
The policy is quite detailed on this, and so refer WP:PROT. In general, semi-protection is a good tool where roving anon or non-autoconfirmed vandals are overwhelming the usual revert-block-ignore procedures. These procedures are expected to take a greater strain on Main Paged articles, and they are also expected to deal with routine, low-level vandalism to all articles. Full-protection is now used principally to put an end to edit-warring on a page where either blocks are inappropriate or are not working. In some cases, full protection without a block may be a means to initiate discussion, but the context is important and depends on what proportion of the active editors are misehaving. Some so-called high-profile templates also benefit from some degree of protection, though this practise is far too wide-spread. Protection is a last resort in most cases since it locks the entire world out of an article when the more targetted tools of rollback and block affect only those causing the problem and avoid violating the founding principles of Wikipedia more widely than absolutely necessary. It is never to be used to explicitly or implicitly discriminate among the good-faith editors to an article, whether they are anon vs. autoconfirmed or admin vs. non. Exceptions to this last are sometimes made via WP:OFFICE by specific named people.
When would it be appropriate to speedily delete a page? Splash - tk 22:30, 7 May 2008 (UTC)
At simplest, when the page in the judgement of admin meets one of the the criteria for speedy deletion are specified in considerable detail at WP:CSD, since these criteria have been decided by wide consensus and experience over a number of years and thereby explicitly and implicitly exclude a number of possibilities that may seem obvious but which do not enjoy wide support for on-sight, minimal-review deletion. More generally, the CSD deal with cases where the article is so poor or problematical that damage to the project (or its image) may result immediately if the article is not removed. More generally still, the criteria have expanded over the years to incorporate an increasing proportion of the most common single-shot deletion, entirely-predictable cases that would (and, indeed have) otherwise taken up a lot of admin/editor time. Thus a good criterion for speedy deletion is where the community has either by voluminous repeated practise or by acclamation declared something so, or (more recently) where purported legal concerns impel the swiftest actions. New speedy criteria have historically only arisen in cases where there are a large number of a case dominating a deletion process so as to demonstrate at high-frequency that the likelihood of anything other than deletion is very close to zero ([Insert evidence from Uncle G which lead to A7 initially]), and also to ensure that for the rarer easy-deletes there is still some scrutiny afforded by XfD in case of something unexpected arising by virtue of the community's lesser experience dealing with the topic.
In at-the-coal-face terms, it is useful to recall that joke articles and wilful junk are in fact vandalism, and can be deleted via G3 even though they may superficially not meet any of the 'A' criteria. Some people consider that WP:IAR comes into this answer. It has rarely done so in practise, since the surprising strength of the exclusions mentioned above — which arise from many community discussions over long periods — have tended to force the IARer to construct a basis in some policy or other that relates back to the nature of speedy deletion as described above. That is, they have not usually managed to successfully ignore all the rules, because they have had to go and make themselves some new ones to operate under.
How does one determine consensus? And how may it be determined differently on a talk page discussion, an WP:XFD discussion, a WP:DRV discussion, and an WP:RM discussion. Splash - tk 22:30, 7 May 2008 (UTC)
A consensus-based decision process is the rule of sociology and persuasion over statistics. It has a key characteristic that those in the group whom the decision goes against are likely nevertheless to largely be able to live with it (it is not necessary for every disappointee to live with it, though). This latter is helped by the eventual decision-maker clearly and persuasively setting out how they have interpreted the debate, but may also be achieved by acknowledgemet of sheer weight on the successful side. For this reason, 'consensus' can tend to occur in different statistical guises in different places, but nevertheless exist in all of them. People generally bow to the greater weight at a lesser ratio on AfD than on RfA, and so consensus may be manifest differently in the two cases. This 'weight' is not localised to the debate, however, and tends also on Wikipedia to import from policies and context which can have the implication of a 'silent body' of the community being considered even though not 'physically present' in the debate. Determining consensus then, is the process of evaluating whether people have come to general agreement on a point-of-view that is compatible with the core policies. Has the debate persuaded people to 'cross the floor' (this is one of the best determinants of consensus forming)? Has it educated the other side as to their flaws? Alternatively, have their arguments been exposed for what they are to general agreement? Can the debate be closed with such an explanation as will draw on the stronger side of the debate to let the other side see that they are not going to succeed at the present time? This process of argumentation can mean that the numerically-weaker side is handed success — this tends to be the case where their argument has been strengthened significantly by 'imported weight'.
Of course, a legitimate outcome of consensus processes is that no consensus is formed - noone is persuaded, the community remains riven and is not able to recourse decisive policies, neither side is evidently of sufficient momentum. On Wikipedia, in such cases, the default is usually that the status quo prevails, but may be challenged at a later time.
To deal with those specific examples in the question requires more details, but they can all be handled within the general framework I've described, with particular attention to the importing of weight from policy, custom and practise as they surround the respective processes. Sometimes, admins have sort of clubbed a debate to death by using their favourite policy as a weapon — relying massively on the imported weight notion, in effect. This has not worked well, since it generally means that the disappointees have two very good targets for their not-living-with (the admin's character invariably becomes one and the other is their interpretation and application of Pet Policy), and this has tended to result in greater total dislocation than a more subtle closure would have engineered. Splash - tk 22:30, 7 May 2008 (UTC)
User:JohnQ leaves you a message on your talk page that User:JohnDoe and User:JaneRoe have been reverting an article back and forth, each to their own preferred version. What steps would you take?
Assuming I decided I was usefully able to get involved in terms of judgement and time, usually though a mixture of counsel (including encouragement to use dispute resolution pathways), warnings, blocks and full-protection. The exact mix of the four depends really on how the edit warriors (have) respond(ed) to the earlier items in that list. Protection can precede blocks when the edit war is relatively low-grade, and the editorial behaviour not sufficient to justify blocks (see above). This can sometimes help discussion to be born. Protection may also come first if there are large numbers of people involved in the war and blocking them all is not practical or useful. Blocks can precede protection when the edit war is high-rate or real-time whether or not discussion is also occuring — edit warring is not tolerable at this level. Often 3RR will be triggered first, but sometimes people religiously stick to it ... every day for a week ... and need to be stopped after a warning to cease. In other cases, it is a judgement call as to which to do first if counsel and warnings do not work, as a function of the personalities and temperaments of the disputants, their number and the entrenchment of the sides (more entrenchment generally raises the spectre of more likely protection since blocks tend not to change minds). What one should not do is decide which version of the article is 'right', revert to it and protect it. This is against protection policy. Exceptions to this arise in the case of legal compulsions - copyright, defamation, etc. but blocks will also often be in order in these cases. Splash - tk 22:30, 7 May 2008 (UTC)
