User:Andrewa/Consensus is consensus

Consensus is Wikipedia's fundamental model for editorial decision-making, according to Wikipedia policy.

In the past, the term community consensus has often been used to refer to historical consensus that has become incorporated into policy and guidelines. Similarly but not nearly so commonly, the term local consensus has been used to refer to the outcome of a particular discussion. Unfortunately, the use of these terms can lead and often has led to misunderstanding. In particular, community consensus is easily assumed to automatically overrule local consensus.

The position in terms of long-standing policy and practice is:
 * Community consensus does not overrule local consensus. This would be contrary to policy, and would remove one of the key mechanisms for changing policy and guidelines, see below.
 * Consensus can change.

It is never necessary to use either of the terms local consensus and community consensus in valid arguments, and these terms are best avoided.

Applying consensus
Often, the sticking point in a discussion is the interpretation of a guideline or policy in order to apply it to the current case, or the reconciliation of rules that appear contradictory in the current case. Consensus must be allowed to decide this, as it is all we have, and this is itself supported by policy.

On the other hand, to allow consensus to be disregarded on the grounds that it is local would allow each editor to apply their own interpretation to all policies, including even WP:consensus itself, leading to circular argument and potentially to complete disregard of policy.

Consensus works remarkably well. And in any case, it is the primary way decisions are made on Wikipedia. This is not negotiable.

When consensus is wrongly assessed
There are avenues of appeal and review for almost all consensus decisions, and these should be followed. No consensus decision can simply be ignored on the grounds that it reflects only local consensus.

Historical consensus
Policies and guidelines reflect the historical consensus of the community. These rules:
 * Reflect consensus in older discussions than those still undecided.
 * Reflect precedents set by these older consensus decisions.
 * Often (not always) reflect the views of a greater number of editors, and a greater depth of reflection, than the current discussion.

Or to put this another way, if we don't follow the rules, why have them at all?

Administrators are particularly expected to bear policies and guidelines in mind when assessing consensus. The arguments are evaluated, particularly in the light of the policies and guidelines that are cited, rather than just doing a head count of who is for and who is against a proposal.

Community consensus
The assessment of consensus through the lens of policy seeks to incorporate this historical consensus into every decision. But it is misleading to refer to these rules as community consensus, and therefore to regard them as binding in all cases regardless of other arguments. This as noted above would itself be contrary to policy. Generally it is better instead to cite the policies and guidelines that are relevant, as part of the process of reaching consensus.

On those occasions when it is desired to explicitly point out that these policies and guidelines are expressions of wider consensus, the term historical consensus allows this without the problems associated with the term community consensus. The terms have similar meanings in the context of Wikipedia, but the term historical consensus has less value in rhetoric, and so encourages dispassionate discussion.

Local consensus
The term local consensus should also be avoided. Consensus is always understood to refer to those editors who take part in a discussion, whether current or historical.

All consensus is local, in that not every member of the community participates in every discussion.

Consensus is consensus.

Finer distinctions
Another problem with the community/local distinction is that it greatly oversimplifies the nature of consensus in an organisation as vast as Wikipedia. There are not just two levels of consensus building, but many.

It's a pyramid structure. At the bottom of the pyramid, the most "local" levels, are discussions on individual articles. There are many of these, each of them affecting few articles, and decisions here affect only the relatively few editors involved with these particular articles.

At the very top are discussions concerning the basic policies. There are few of these, and decisions here may affect many articles and many editors.

Somewhere in between are talk pages of less fundamental policies, and of detailed guidelines, WikiProjects, Task Forces, categories, and so on.

But the general rule that the upper levels affect more editors than the lower ones is not reflected in the number of editors actively involved in the discussions there. Policy change discussions tend to attract relatively few contributors compared to say a controversial RM. Category talk pages tend to be particularly neglected, with very few contributors visiting or watching them.

There is within this structure a very loose ranking of authority. Core policies are the most authoritative, and user essays the least, this much is clear. Essays in the Project Namespace, and pages kept for historical reasons, are also low down on the authority scale, although even here, some essays are widely cited and this appears to reflect relatively wide support. But in between come other policies, guidelines, the Manual of Style, conventions established by WikiProjects, precedents (relevant as many guidelines recommend consistency), information pages, help pages, documentation on category pages, documentation pages transcluded into template pages but not into their expansions, other instructions commented into the source generated by substituted templates, other instructions generated by bots that update Project Namespace pages such as Requested Moves... many, many things! The ranking of these by their relative authority is not clear at all.

And there is simply no way to translate this complex structure into one that will rank the levels of consensus that decide the content of these pages (etc). Again, consensus is consensus.

Changing rules and changing consensus
When consensus does change, policies and guidelines need to be changed in order to reflect this. There are two ways in which this can happen:
 * Consensus can first be sought to change the rule, citing examples of decisions that this change would affect, on the talk page of that particular rule.
 * Consensus can first be sought for specific cases that appear to be valid exceptions to the existing rules. WP:IAR may (or may not) be invoked in these cases.

In practice, the second has been the more productive approach. It can arise in two ways:
 * Most commonly, a specific case arises in the normal course of editing, and is discussed first on the talk page of the article concerned.
 * In some cases, discussions (such as requested moves) are raised as examples, partly in order to demonstrate a problem with an existing rule.

Do not disrupt Wikipedia
Great care must be taken not to violate the disruption guideline when raising example discussions. In particular, if the proposer of the example argues against the change they propose there, voting to oppose a move they have themselves proposed for example, then this is a danger signal but not a show-stopper. While bad-faith proposals are clearly disruptive (see the guideline for examples), even good faith ones are often borderline. There must be a reasonable possibility that consensus in the example discussion will be to support the change to the article concerned. If there is no such expectation, and so no prospect that Wikipedia will be directly improved by the outcome of the example discussion, then raising the proposal is disruptive, and can result in blocks and bans.

However if there is reasonable doubt as to the outcome, and so a genuine prospect of improving the article, then the example proposal is not disruptive even if the proposer votes against it. Concrete examples can greatly help to develop consensus, and while it is far better to instead find such examples in existing discussions, raising specific cases is sometimes the best way forward. They are intrinsically controversial, but they focus discussion on the practical issue of improving articles.

Following up
Whenever consensus is established that it will improve Wikipedia if the existing rules are not applied to the case under discussion, the rules must then be updated to allow this exception. Otherwise, the rules will no longer reflect consensus, and then, again, why have them?

This also provides an automatic review process. Discussion moves from the talk pages of the effected article(s) to the talk pages of the policies and guidelines. This brings these particular cases to the attention of a different, and often wider and/or more experienced, group of editors.

If the changes are rejected, then the specific cases should be formally reviewed.

Policies

 * Consensus particularly #Level of consensus
 * Ignore all rules
 * Policies and guidelines particularly #Life cycle which describes the two ways of changing policy as listed above
 * What Wikipedia is not particularly #Wikipedia is not a bureaucracy

Guidelines

 * Requested moves/Closing instructions
 * Deletion guidelines for administrators

Examples

 * Move review/Log/2012 July 10 A single long discussion, expand it and search within the page for local consensus (6 hits, and rather than community consensus which gets 8 hits but all in the same contribution, which the first search also finds).
 * Talk:Temple Beth-El (Pensacola, Florida) A requested move to shorten the disambiguator was raised in view of claims by several editors at Talk:Temple B'Nai Israel (New Britain, Connecticut) that policy mandated the shorter form of disambiguation. When both the example and original RM were resolved in favour of the longer disambiguator, the rule change required was shown to be substantially wider in scope than just that suggested by the original RM which had started the discussion. Note also that several editors regarded this as borderline disruption.

Shortcuts

 * Community consensus
 * Local consensus

It's fascinating how few links there are to either of these redirects, three and six respectively (including this page in each case) at the time of writing. On the other hand, the phrase community consensus gets 2,471 current hits in the Wikipedia talk namespace, and a whopping 4,441 in the Wikipedia namespace. There are a relatively modest but still significant 501 and 508  respectively for local consensus, also at the time of writing of course.

The conclusion seems to be that many editors use these phrases both in the project namespace and on its talk pages, but very, very few use them as wikilinks to policy or guidelines (despite one of them having been there and stable for many years ). There may be for many reasons for this.