Wikipedia:Advice on closing discussions

This is an essay.

It contains the advice or opinions of one or more Wikipedia contributors. This page is not an encyclopedia article, nor is it one of Wikipedia's policies or guidelines, as it has not been thoroughly vetted by the community. Some essays represent widespread norms; others only represent minority viewpoints.

Shortcuts

This page contains recommendations for editors who want to learn how to effectively close and summarize discussions, particularly RfCs, on the English Wikipedia. It may also be useful for people who want to understand why a discussion was closed in a certain way, or why certain types of arguments are stronger than others. General information about RFCs can be found at WP:Requests for comment, and general information about closures can be found at WP:Closing discussions. Additionally, while this page focuses on RfCs, many of the principles are applicable when closing discussions in other contexts, such as for deletion discussions and on most noticeboards. It does not apply to specialized pages such as DYK and VA that have unique rules for closure.

Formal closures are a component of dispute resolution that help to resolve disputes by summarizing the results of discussions. They usually include stating the consensus (for, against, or lack thereof) for the issues under discussion, along with a few sentences explaining why this is the case. Most discussions that need closure are first listed at WP:ANRFC, which serves as a list of currently unclosed discussions.

Secondary goals in addition to resolving the immediate dispute include:

to allow editors to move on from specific issues to new ones
to provide an authoritative reference for future editors if the subject comes up again
to help editors understand why a particular result occurred
to preserve the authority, integrity, and transparency of the closing process, which allows effective dispute resolution to continue in future
in some cases, to formalize uncontroversial decisions as well (e.g. some RMs and AfDs)

Being a closer is a position of responsibility and trust, and should be approached both seriously and cautiously. Each closing statement should be neutral and well-written, and should only be performed after careful analysis of the discussion in question. A poor summary may be disruptive and can cause more problems than it solves. For especially contentious subjects, it can result in days or weeks of unnecessary debate before finally being overturned.

General principles

The first principle of closures: The closer’s role is descriptive, not prescriptive. In other words, the closer describes what other editors have decided on, and does not make any decisions for anyone else.

The goal of any closing statement is to determine and summarize the consensus reached by the editors in the discussion. The outcome is determined solely by the participants; the closer’s role is solely to find out what the participants have decided. If you personally make a decision about the issue at any point, as opposed to evaluating the contents of the discussion, then you are making a prescriptive closure. This is often referred to as a supervote because you’re placing your own reasoning above that of the participants.

The closer must be neutral and impartial, without allowing any of their personal opinions to affect the outcome. The view to be expressed in the closing statement should be what is best for the encyclopedia, as expressed by the will of the community. The influence of each argument on the final outcome should be weighted by the strength of the argument, with reference to existing consensus as represented by e.g. the Wikipedia policies and guidelines. Numbers play a role, but they do not determine the outcome on their own. For details on the meaning of consensus, which is a term of art unique to Wikipedia, see this explanatory supplement.

As a closer, you should have a full understanding of the discussion. This includes all policies, guidelines, and essays cited by editors, and you should reread any that you aren’t already intimately familiar with. Note that discussions about controversial topics or in Wikipedia space often have a long history which you may not be aware of. Also, editors will often imply policy-based arguments without specifically citing them. Keywords such as “neutral,” “undue,” and “reliable” may be present, but not always. You should also fully understand all the points of view being expressed, as well as enough of the background to understand the subject area, especially the points of disagreement (including but not limited to the RfC question). This may also include familiarizing yourself with the topic, especially for content disputes about academic or other technical subjects. In some cases, participants may have spent dozens or hundreds of hours of experience editing in the topic area, and since as an uninvolved closer you’ll usually be unfamiliar with the discussion, it is especially important to do due diligence. You will be expected to have the necessary background to effectively evaluate the evidence and arguments presented.

When exercises of judgement are required on the part of the closer, it is often referred to as closer discretion. Discretion only applies to judgements about the discussion, not judgements about the outcome, and must be fully justified by the discussion’s contents. The most visible judgements are about whether or not editors have reached consensus when discussions are borderline, since those are the cases most likely to attract disagreement from the participants. While the extent of discretion is limited, a closer’s judgement will normally be accepted by the community as long as it is reasonable and does not imply supervoting. More subtle uses of discretion include what to emphasize when writing the summary or the precise weight given to specific arguments when analyzing the discussion. There is also a small amount of latitude for using a closing statement to make suggestions, but you should do so cautiously, as befits the responsibility you’re taking on as a closer.

Not all discussions need a formal closure, and not all discussions that get closed need a statement that summarizes the result. For instance, if editors have already come to an agreement, formal closure may not be necessary. However, there are some advantages to doing it anyway, such as serving as a point of reference for future editors if the topic comes up again. In these cases, shorter closing statements with less or no explanation are more acceptable. On the other hand, in some circumstances, such as RM and AfD, uncontroversial discussions may still require closure. If a discussion is listed at ANRFC (or an analogous page for discussions other than RfCs), a formal closure with a summary is likely to be appropriate.

Determining your suitability as a closer

A closer should typically be an experienced editor. While there are no hard rules about what this means, the ultimate criterion is the trust of the community. This does not mean a closer has to be an administrator, but a minimum requirement is full understanding of Wikipedia’s core policies, and good familiarity with any pages cited in the discussion. You should be experienced enough to understand the arguments being made, and to be confident that reasonable (and even unreasonable) editors will agree that your actions are fair. Some closures may also require understanding of Wikipedia’s culture, dispute resolution procedures, behavioral standards, etc. You should have a complete understanding of the issues before proceeding, as described in the previous section.

Not everyone will be suitable to close any given discussion, even when they have the necessary experience. You will need to make sure you think you can be genuinely neutral. Personal opinions about the outcome, beyond wanting what’s best for Wikipedia, can make you an unsuitable closer even if you don’t meet the usual definition of being involved. If you have a personal opinion, be more cautious the stronger your opinion is. If you’re unsure, it is generally best to err on the side of caution, and the more contentious the discussion, the more cautious you should be. If the outcome is truly uncontroversial, closures by involved editors are permitted and even encouraged. However, this situation is uncommon, and even small degrees of personal investment can make it difficult to recognize when this is truly the case. If you want to influence the outcome, or you think the discussion has missed an important point, then you should join it as a participant instead.

The ideal closer for a discussion is one who honestly doesn’t care what the outcome is, only about the integrity of the encyclopedia and the RfC process.

You probably shouldn’t close a discussion if:

you find yourself wanting to close a specific RfC

you’re tempted to close an RfC because you think another closer might not come to the same conclusion that you would

you find yourself reviewing this section while trying to figure out whether you qualify to close a specific discussion

Note that the appearance of involvement is important separately from actual involvement, because it will reduce the authority of the closure and its statement. For example, if you’ve made major edits in the topic area, or expressed opinions with respect to it, then this may affect your suitability as a closer. The existence of evidence that could be used to argue involvement, even if the arguments would be spurious, may be relevant to whether you can effectively act as a closer. If there are openings for your actions to be easily challenged, it will distract from the resolution of the issue.

Like other discussions on Wikipedia, closures are subject to acceptance by the community. Many factors affect whether a closure "sticks," e.g., an uncontroversial closure will have looser standards. The likelihood of a close being challenged is based on a number of factors, including whether editors are likely to accept a) you as the closer, b) the end of the discussion at this time, and c) your closing summary. Ultimately, a close only has as much authority as the editors in the discussion (and more broadly, the editors of the community as a whole) will give it. While editors cannot simply ignore a poor close, such a close will eventually be corrected by subsequent discussion, and the only real question is how long it will take. Misjudging a discussion, especially in a controversial area, can lead to a great degree of wasted effort for everyone as your actions are appealed or otherwise debated by the participants.

This doesn’t mean you should be frightened of mistakes. If you’re new to formal closures, focus on discussions that are easier or more straightforward, and ask a more experienced closer to check your work. And if it seems like you may have made a mistake, be ready to revert or amend your closing statement, or to bring it to the community for review.

For some extremely contentious closes, especially one's with wide-reaching effects, it's best to have a paneled closure that consists of multiple (typically three) administrators closing the discussion.

Making decisions

Common results

The only true requirement for a closing statement is for it to fairly and accurately reflect consensus. Nevertheless, there are specific outcomes that are often used in particular circumstances. The most common results are consensus, consensus against, and no consensus. The close does not have to contain one of these specific terms, e.g. invoking WP:SNOW already includes it by implication. As a very rough guideline, no consensus results are usually in the range of 30–70 per cent, and the closer you are to the boundaries, the more the strength of the arguments becomes relevant.

There are also a few specialized outcomes that can occur when consensus cannot be evaluated or a closure is no longer necessary. These are not precise categories, and should be treated with common sense.

Not enough participation is a common result for especially controversial topics, or for topics with site-wide implications. For example, in topic areas such as abortion and the Israel/Palestine conflict, certain editorial approaches have already been codified by major discussions involving many dozens of editors, sometimes holding years of precedent, and modifying or reversing these decisions generally requires a discussion of equal or greater scope. In other cases, there may be no prior consensus but only a couple of editors commented: this can occur in obscure discussions, in complex RfCs where many questions are being discussed and not everyone addresses every issue, or if a new proposal is made near the end of an RfC and not many editors had the chance to evaluate it. If there is no support or no opposition, and it doesn’t conflict with any other part of the RfC, it may be appropriate to close it accordingly, with the caveat that participation was low. Otherwise, it may be best to say “not enough discussion to evaluate consensus,” or “no consensus due to lack of participation” (since saying only “no consensus” implies a balance of arguments, as opposed to an inability to determine whether there is a balance of arguments).

Local consensus occurs when taking the RfC result at face value would violate a broader community consensus. For example, this can occur when an RfC is in the wrong venue or the result would supersede a prior discussion with greater participation. This does not necessarily have to be alleged by the discussion’s participants, since part of your role is to evaluate the discussion with reference to the broader context of Wikipedia, including existing consensus. Claims of local consensus may also occur when a group of editors is uncharacteristically absent due to canvassing or lack of appropriate notification, especially if the missing group may have a specific type of expertise, or if there is reason to believe they may have different opinions from the participants. However, it is unusual for missing notifications alone to invalidate a discussion entirely. Since anyone can make (neutral, appropriate) notifications during the discussion, claims of local consensus from participants (who could have made the notifications themselves) are often rejected by the community if made after the fact.
Procedural closes: ideally, these are uncontroversial and are often early closures. They are often performed by involved editors, and only rarely appear at ANRFC, unless e.g. an involved closure might exacerbate an existing dispute. For example, duplicate discussions will normally be consolidated, one being closed with a reference to the other.
Closure is unnecessary: the discussion may be obsolete, the nominator may have withdrawn their proposal, a closure may be redundant because all the participants have already recognized the outcome, etc. That said, since these cases are generally unambiguous and easy to close, it is often more convenient to close them anyways and gain the benefits of having an easy reference to the outcome in future.
RfC is not well-formed: this can happen when the results are unclear because of the structure of the RfC. For example, the RfC might have no clear question, or editors might not have answered the question. However, it is unusual for this to be the only result, because it is normally still possible to evaluate editors’ opinions. For example, you might be able to answer some aspects of the question but not others. Caution: This can sometimes be confused with situations where you simply don’t fully understand the discussion, in which case you should leave it for someone else to close. Before using this outcome, make sure you both fully understand the discussion and why a full result is impossible regardless of how familiar you are with the subject.
Consensus cannot be evaluated: as with the previous case, it is very rare for this to be the only outcome. However, it may occur if issues such as canvassing, sockpuppetry, or POV pushing appear to have occurred to such a degree that it is impossible to determine the true result. Again, this is different from no consensus, which implies that there is a balance of opinion. Note that the same caution given for the previous case also applies here.

Examples of individual closures can be found via the archives of WP:ANRFC, although of course the quality of closures varies. Don’t expect that the closure should resolve everything, or everything that was discussed, or even anything at all. While you should try to find areas of agreement, you may not be able to, and you should not try to create a solution where there isn’t one. Also remember that a no consensus result can lead to a resolution, and does not imply that the discussion failed.

Strength of arguments

The strength of an argument determines the degree to which it influences the outcome, with stronger arguments being given more weight in determining consensus. On the other hand, in other cases it may be appropriate to reduce (“discount”) the weight of an argument or comment, so that it counts for less in the final analysis. In full discounting, the argument is removed from the analysis and not considered at all. These factors are not an exclusive list, but you should be able to justify your decisions. Every situation should be approached case-by-case, and everything must ultimately be based in consensus.

The most important factor affecting the strength of arguments is grounding in Wikipedia policies and guidelines. This includes arguments making sense with reference to policy, and the importance of the specific pages being cited. For example, NPOV arguments are stronger than style-based arguments. Similarly, an argument relating to the core concept of a policy is stronger than an argument relating to a minor aspect of it, or an argument with only a tenuous connection to the policy in the first place. In general, policies are more important than guidelines, and guidelines are more important than essays. However, a closer also needs to be aware of the uncodified aspects of the hierarchy: BRD has a strong consensus despite being an essay, RS and MEDRS have the strength of policy despite being guidelines, and so forth. Reading many closures and spending time in Wikipedia policy space can help with this. All policies and guidelines have strong consensus support, while essays are weighted according to the level of consensus supporting them (which can be almost none to as much as BRD). The difference between policies and guidelines is typically less important than the degree of consensus that supports each individual one, and the strength of any argument to IAR for one of them (see also: Wikipedia:You might be Wikilawyering if...).

Arguments based on misconceptions of policy/guidelines, or which do not contain policy/guideline-based reasoning at all, should not be considered in your analysis. For example, ATA reasoning should be disregarded. There are also a number of argument types or lines of reasoning that are generally accepted as spurious or as violations of particular policy/guidelines. In some cases, these have their own links (such as WP:CRYSTAL or WP:RGW), but not always. There are also specialized reasons for rejecting particular arguments that only apply in certain areas of Wikipedia, which is why ATA is divided into subpages, and closers may need to be aware of them.

Arguments based on a misreading or misunderstanding of the question should also usually be discounted. The degree of discounting depends on the severity of the issue, based on how much the editor appears to understand the question and what specific position they expressed. An argument that doesn’t address the question at all should be discounted entirely. This can be due to error, but e.g. it may also occur in complex RfCs where not everyone addresses every issue. Sometimes an off-topic comment may include one or two relevant components, but avoid inferring someone’s position on an issue based on their position on a related issue. On the other hand, if the only issue is that an editor said “support” when their comment clearly shows they meant to oppose (or vice versa), if they commented without explicitly supporting or opposing, or if they deliberately chose to address only one part of the discussion (for example, commenting only to include a rebuttal to another comment), then their comments should usually be given full weight according to the strength of their arguments.

Comments where the sole content is “per previous editor” (or similarly superficial reasoning) should contribute to the outcome only after more substantial comments have been considered. This is because it may indicate a lack of engagement or not having given the issue more than superficial consideration. Other evidence suggesting that editors failed to give the question much thought, such as an especially short comment or especially simplistic reasoning, may also be relevant here. Caution: in some cases, discounting superficial comments can backfire, especially if a single argument is overriding, or because editors feel that a proposal is sufficiently absurd that writing detailed responses would be a waste of time. For example, this sometimes happens when editors feel that an RfC is attempting to overturn core policy.

The degree to which arguments have been rebutted by other editors may be relevant, as long as the rebuttals themselves carry sufficient weight. If one group is responding directly to the other’s arguments but the other isn’t, that may be relevant to determining which group has better reasoning. On the other hand, if one or two participants are trying to reply to everyone who disagrees with them, others may simply not be taking them seriously or have grown tired of repeating themselves. The strength of a rebuttal is not directly affected by whether it is acknowledged by opposing editors, e.g. many editors never return to the discussion after commenting. POV pushing can also be accounted for through this and similar factors; it should not usually be considered as a separate factor, because of how difficult it is to classify it neutrally.

The number of editors who cited each argument is normally relevant, but it does not determine the outcome on its own. There are several different perspectives on what the precise role of numbers in determining consensus should be, and different closers may consider them differently. However, it is widely agreed that consensus is not a vote; no numbers can make up for an invalid argument, and an argument is not made stronger by another editor repeating it. Numbers primarily support strength of argument in the aspects of a dispute that require interpretation – for example, to determine the best application of a policy or guideline to a particular circumstance. They may show what types of reasoning are predominant in the community, and whether an argument that may be ambiguous in its grounding in policy is more broadly endorsed. They may also be used to determine which of two arguments with otherwise equal strength is seen as more important by the community.

Content-related factors

Since as the closer you aren’t judging the issue itself, the subject of the dispute does not directly affect the outcome. However, there can still be significant effects on the strength of arguments: for example, the number and quality of sources being cited has a major impact in content disputes. In some cases you may need to check the sources yourself to evaluate this, but be careful to avoid any analysis of reliability that could be controversial under policy. More generally, you may also require some familiarity with the subject. At minimum, this involves knowing which sources are reputable and which are not, in order to correctly evaluate due weight. Otherwise, you may end up giving credence to spurious arguments, which can lead to results that e.g. contradict established knowledge or expert consensus. For example, evaluating the reputation of specific experts or of academic journals may require specialized experience beyond that provided by policies and guidelines alone.

A background in the relevant subject usually makes closures easier in general. However, in some cases a certain amount of background may also be a requirement, such as to evaluate whether someone has misunderstood the subject matter. This is especially relevant for more technical subjects, such as the sciences, and for subjects where misunderstandings or misinformation are particularly common, such as fringe theories. For example, if participants are repeatedly using unfamiliar terminology, it may mean that you don’t have sufficient background to evaluate the arguments correctly. You may be able to remedy this by studying, or it may be better to leave the discussion for someone else to close. This issue is particularly obvious in extreme cases, such as for technical discussions in advanced mathematics, but in other cases it may be easy to miss. You are expected to have the necessary background to effectively evaluate the evidence and arguments presented.

Editor-specific factors

Consensus is not normally affected by the identity of the discussion participants, but there are several circumstances where it may become relevant:

SPAs, socks, and canvassed editors will normally get little to no weight. These can often be identified through their behavior in the discussion. However, since it’s still fairly easy to miss them, editors may include claims about this within the RfC, leaving the closer to make the final judgement. If the effect on consensus cannot be easily determined, it may be necessary to close the RfC without any result. Other evidence, such as certain patterns in !voting, can also indicate that these issues are present.
Editors who were blocked, banned, topic banned, or otherwise sanctioned during the RfC, especially if related to conduct in the topic area in question, may also be partially or fully discounted. Past sanctions may or may not be relevant, since they can indicate an editor’s past behavior. This is subject to judgement, e.g. a long-term editor with a 24hr block on an unrelated subject will probably get full weight, while someone blocked for socking should get none. Setting your preferences to identify blocked users (under Gadgets→Appearance), or using an equivalent user script (such as this one) can be useful for identifying editors to check on. For example, confirmed socks normally receive immediate indefinite blocks. However, these scripts do not detect all types of sanctions, topic bans being a major exception.
Some closers are willing to discount IP addresses, while others are not. This decision is partly philosophical, based in long-running Wikipedia debates about whether IPs should be treated as named editors, probable sockpuppets, inexperienced by default, or anything in between (e.g. see WP:IPs are human too and WP:IP addresses are not people). It may be best to avoid situations where a decision turns on this question, as it may draw you into philosophical arguments and distract from the merits of your close. The instincts for detecting SPAs and socks that many editors develop over time can be useful in avoiding the issue, since those issues supersede any debates on the merits of IPs.
If an editor indicates they only have a mild preference about the issue, it may be appropriate to discount according to the strength of their preference. On the other hand, if editors !vote “strong support” or “strong oppose,” it should only be taken as an indication of conviction and the importance they place on the outcome. It can’t be used to directly weight their comment, since otherwise everyone could use it to artificially give their comments more weight.

Additional considerations

As a closer, you should be familiar with the structure and context of the discussion. This includes checking any discussion or edit that gets referred to in the RfC, including edits to the RfC question itself. You may need to look at the timestamps to see which comments were made before and after each change. RfCs may also require you to consider previous discussions on the same or a related topic, especially if an earlier discussion was formally closed. At minimum, you should usually go through any previous discussions that have been linked within the text of the RfC. The primary guide for what you should understand is the expectations of the discussion participants. Additionally, if your closure is different from a previous close, you should have an idea that something has changed in the interim, even if it’s just the number of editors in the discussion. In contrast, if e.g. the previous discussion was relatively recent or the newer discussion has much lower participation, it may not be appropriate to overrule the first one (see also: forum shopping).

Arguments not made by the RfC participants should not typically be considered, because doing so implies that you’re considering your personal opinion, placing your judgement above that of the participants who had the opportunity to use those arguments themselves. Sometimes it can be appropriate (or necessary) to refer to a policy not mentioned in the discussion, but this is relatively uncommon. For example, outside of exceptional situations, you should not prescribe a course of action that nobody in the discussion has suggested, even as a compromise – if it wasn’t directly proposed by the RfC participants, the outcome would represent your personal opinion. Another part of not interjecting your own opinion is not making a close that is wider than the scope of the question (after all, if the editors didn’t discuss it, they couldn’t have come to a decision on it). Similarly, don’t draw a conclusion that is broader than what editors would actually have agreed on. However, note that the outcome may not necessarily be about the original RfC question. The discussion may have focused on a narrower, broader, or completely different question, or on several related questions at once.

You may also need to account for changes over the course of the RfC. Patterns in !voting often indicate improper behavior such as canvassing, but they can also be legitimate, indicating a change in consensus over time. If so, this should normally be clear: the pattern may represent new arguments being presented that the first group doesn’t have a good answer for, a wider selection of editors becoming aware of the discussion, the situation being changed by an external factor, etc. If editors change their opinion, this may be indicated by striking out previous comments. Editors may also concede certain points or come up with new ideas without changing their previous comments, in which case judgement may be required about the degree to which the previous comments are obsolete. For an RfC that is affected by external factors, such as when an article describes a changing situation, you may need to cross-reference comments with the situation at the time they were made. In this case, more recent comments should usually be weighted more strongly, since they account for more information (including the earlier comments). If an important change occurs near the end of the RfC, especially if discussion is still continuing, it may be best not to close and allow the editors to continue discussion.

Writing the close

The usefulness of RfCs is strongly affected by the way the close is written. Consider your wording carefully, and think of how different explanations could be interpreted by different editors. All editors should feel like their comments have at least been carefully considered, and even if they don’t agree on the outcome, reasonable editors should agree that the closure is a fair summary of the views discussed.

Some closers prefer to minimize explanation, while others try to explain in greater detail. That said, most editors will vary their choice depending on the situation, e.g. making a very short closure in cases where the outcome should be clear to everyone.

Advantages of minimizing detail: requires less effort on average closure, allows faster response time leading to a shorter RfC closure backlog, it is easier to tailor responses to editors who challenge you, it is harder for challengers to wikilawyer, you are less likely to be called on a mistake
Advantages of greater detail: it is more useful to the RfC participants, it will be seen as more authoritative, it is less likely to be challenged, it acts as a definitive reference for your reasoning if the close is appealed, incorrect decisions will be more easily remedied, it helps ensure that you understand the issues before closing

Every close is different, and calls for different approaches. Different closers have their own styles, and no close will have space for everything, so as the closer you will have to decide what components are most important in any given situation. Closure of borderline, controversial, or complicated discussions will usually require more careful consideration and a more detailed explanation. Most of this section is primarily important for such discussions, or in other types of specialized circumstances, although it applies more generally as well.

General

One of the most useful things you can include as a closer is to acknowledge the arguments of the editors you’re closing against, and make it clear why the discussion closed against them. Additionally, neutrally describing the arguments of each group can help demonstrate to each group of editors why there’s a legitimate question. Clearly specifying the issues under discussion can also help, especially if the RfC was unclear or confusing (e.g. the wording was ambiguous, editors interpreted the question in different ways, the intent of the question was different from what was literally written, etc).

The evaluation of argument strength is another common component in closing statements, especially if your close is different than what the numbers alone would indicate. In addition to being an important part of your explanation, the reasons for discounting particular comments may not be immediately obvious and superficial inspection may give the impression that you closed incorrectly. In these cases, extra care should be taken with the explanation: among other things, appearing to have made such closes without careful consideration may open you up to claims of supervoting.

Especially when new to closing, you should check your closure for any statements that could imply improper reasoning, such as comments that may seem prescriptive rather than descriptive or may seem to reflect a personal opinion. Similarly, you should ensure that any personal interpretations of comments in the discussion are fully minimized; if a minor amount of this seems unavoidable, and if leaving the discussion for another closer would be unlikely to help (e.g. if some comments are inherently unclear, or contain unspoken inferences that may require interpretation), it may help to explicitly point this out in your close. This also provides a relatively straightforward way to edit your decision if it turns out you misinterpreted something important.

Framing

A closure should normally be presented as an analysis of the arguments rather than editors. Doing this should be relatively straightforward, since it reflects the proper procedure (except when accounting for unavoidably editor-specific issues such as sockpuppetry), but incautious language may imply otherwise. Additionally, identifying specific editors is usually unnecessary and may give the impression of taking a side. Negative comments may lead to embarrassment or defensiveness if taken personally; positive comments are easier to include, but may give too much weight to an individual editor’s opinions or imply that you would support other comments by the same editor. Linking the comments without including a name is better, but can still have the same effect.

In some cases, it can help to use wording that avoids explicitly invoking consensus. For example, when there is clearly a consensus against a proposal, “softer” phrasing such as “not enough support to make the change” may be useful in several situations: a) to avoid discouraging the proposer, primarily for newer editors, b) when it’s arguable whether there is consensus or no consensus, if they would both give the same result due to no consensus defaulting to the status quo, and c) if any editors appear to be especially invested in the proposal, since it emphasizes the fact that they have to convince others to agree with them. Other alternative terms such as “agreement” or “support” may also be appropriate, e.g. if a subject is particularly contentious, or if editors are arguing about the meaning of the word “consensus.” The goal of this approach is to promote productive discussion and discourage unnecessary disputes.

Controversial discussions usually require particular care in framing. One approach is to start your statement by describing points of agreement before starting to discuss the points of disagreement. Benefits of this include demonstrating common ground, showing possibilities for future cooperation between editors, and helping them understand precisely where their opinions diverge. Another method is to emphasize something that potentially dissatisfied editors will be happy with, to preempt perceptions of having taken a side by partisan editors. Similarly, you can point out which related issues are outside the closure’s scope, which also helps discourage the editors who supported the outcome from expanding it too far beyond the exact RfC question. It can also help to emphasize any relevant caveats – for example, pointing out that new proposals or additional discussion on existing proposals can still continue, especially if editors were generally dissatisfied with most of the available options, or some approaches that didn’t get very much discussion.

In controversial cases, or cases where you think outside organizations (such as the media) may be watching, or just in cases where one or more editors are new to Wikipedia, it may be useful to expand on any of the principles discussed on this page. For example, it may help to specify that you’re responding to an ANRFC request, to make it clear that you’re there as part of a formal dispute resolution procedure, instead of just a random editor stopping by. It may also be useful to stress that you’re evaluating what the editors in the discussion have decided (rather than making a decision of your own), and to explicitly link to or describe WP:Consensus.

Suggestions

It can be a good idea to phrase something as a suggestion or recommendation rather than a requirement, if consensus is unclear or you’re uncertain of the basis for your statement. These points should be left open for editors to discuss on their own, and since no actions are being required you will also avoid challenges on those particular points. It can be especially helpful to give suggestions on ways for editors to proceed (given the result of the RfC), since there are often unresolved issues that will lead to future discussion. One approach is to specifically point these out, to show the participants possible ways they could continue to make progress.

Suggestions must be neutral, without implying that you support any particular option; your only goal in this context is to help move the discussion forward. For editors who may be discouraged by the result, including suggestions can make it clear that they should still feel welcome in the discussion and give them a path to continue forward. It also diverts attempts to re-argue the same issue by giving dissatisfied editors a direction they can focus on, emphasizing what has not been resolved by identifying closely related issues, thus helping the discussion progress. A direct comment that the result isn’t perfect (which is usually the case, Wikipedia always being a work in progress) can also help in this regard.

If the question could have been better written, especially if it clearly caused problems or if the wording was a subject of discussion in the RfC, it may help to include advice on writing questions to help the RfC initiator in the future. In some cases, it may help to include a link to a guidance page such as WP:WRFC.

Splitting and alternative questions

Even a discussion that may appear not to have any consensus (e.g. if there was no clear question) may still have consensus on some aspects of the issue. Similarly, there may be consensus on an issue that was not actually the question being asked, since the question the participants discussed may differ from what was literally asked in the RfC prompt. A good closer will also notice when the participants disagree over one point, but substantially agree on a different point that will allow them to move forward. This applies even when one point was not included in the original question, or was not directly under discussion in the RfC.

If there are several issues being discussed at once, it is usually helpful to distinguish between them and evaluate consensus on them individually. The same applies if there are multiple aspects to the same issue, in which case the question can be split into parts. For example, first finding a consensus that the current version should be changed, and then separately discussing what to replace it with, is a stronger chain of reasoning than simply evaluating a proposed new version. This makes it easier for others to follow and understand the issue, and can also codify some degree of progress in a discussion that would otherwise be a no consensus result. Finding consensus on one point can help by simplifying the subsequent discussion, even if the RfC itself didn’t directly lead to any changes.

More complicated situations can occur when there are more than two options under consideration. For example, editors may form a consensus against an existing version while being split between multiple proposals, meaning that no version has a consensus but the status quo is clearly unsuitable. In this case, the most cautious option is to only describe the relative level of support for each option without invoking the concept of consensus, and let the editors decide what to do with that information. However, it may be helpful to divide the options in a way that lets the participants move forward, as long as the split is reasonable and doesn’t seem arbitrary. For example, if two or more options are sufficiently similar to each other, it may be appropriate to group them together. More generally, consider what editors think of the options they didn’t support, and whether they’d accept one of them as a second choice (for instance, if two options are both changes in the same direction but to different degrees), or whether they would explicitly oppose it.

Other notes

If a change was added to the article shortly before an RfC was opened and the close is no consensus, it is often helpful to specify that this result means the page should return to the status quo ante. Otherwise, editors who are less familiar with Wikipedia might not understand the result, and especially in controversial areas it may lead to edit wars.
There may be a heated dispute about a side issue that does not affect the RfC question, even when the overall result is clear. In these cases, you can acknowledge the dispute but specify that it does not affect the result. This makes it clear that you didn’t ignore it, but that you also aren’t choosing an outcome on that dispute. While you do need to read derails and other back-and-forth discussions in the RfC, they do not usually affect the final outcome even if they’re on-topic, except to the extent that they might clarify previous comments or have procedural implications.
If IAR applies, you should make it clear which argument was used to justify it. Since IAR cases are often subjective by default, it may help to say IAR “could” be invoked rather than explicitly invoking it, since IAR is not usually the sole argument in favor of a particular outcome. A definitive use of IAR could make the outcome appear to be contingent on its validity, whereas different reasoning might avoid that.

After the close

After the RfC closure, you should aim to maximize both accountability and transparency. If the closure was requested at ANRFC, add the Done template in the relevant section and it will be archived by a bot. Since the request will remain on the page for a short time, this allows the request to briefly remain open in case the closure is challenged. You can also delay archiving to ask for advice or a second opinion from other closers in lieu of taking the challenge directly to a noticeboard.

You should not usually perform the results of closes yourself, unless specifically requested by the editors in the discussion. Since involved editors will usually know the subject better than the closer, they will usually know more about how best to carry out the result. Additionally, unless the RfC was entirely unambiguous, there may be remaining issues about precise wording or details of implementation. These need to be decided by the editors, even if they weren’t resolved in the RfC itself. Similarly, if there are behavioral problems such as editing against consensus, that is for the participants to resolve. Other benefits of this approach include:

You might otherwise be drawn into disputes over how the closure is implemented. For example, any subjective aspect to your edit that editors interpret as favorable to one group could give the appearance of involvement. Similarly, it helps avoid any temptation on your part to make any subjective changes or do more than what the RfC requested.
The authority you hold as a closer may give an authority to your chosen wording which shouldn’t be there. Likewise, when someone edit wars against the outcome, it may reduce the perceived authority of the close if the closer was personally reverted, and others may refrain from helping because they see you as having taken responsibility for content enforcement.

Outcomes that require use of admin tools are in a separate category, but since the use of tools is a binary option, there are fewer of the shades of grey in interpretation that are the reason for not performing closes. Another exception to not performing closes personally might be BLP, e.g. if the consensus is that some text is a BLP violation.

You can also choose to get involved in the article after the close. However, this should be done carefully, because if you seem too enthusiastic or too opinionated, it may raise doubts about the close itself. It may be a good idea to limit yourself to clarifications on the close, at least for a period of time after you perform it. Otherwise, you may find yourself trying to fill two roles at once and any statements you make about the closure may not be interpreted in the same way.

Addressing objections

Closers are generally allowed to modify or overturn their close. This is usually done in response to clarification requests in the RfC’s immediate aftermath. It should not be done arbitrarily, and since your subsequent clarifications may be interpreted as carrying the same authority as the original close, you should take the same amount of care with them. Avoid situations where you need to edit your close immediately after writing it, as this may imply that you didn’t consider it carefully. However, you should make sure you are willing to edit or reverse your closes if necessary, especially if new information comes to light. Some of the possible reasons include misinterpreting an editor’s comments, finding out that the discussion was manipulated by sockpuppets, etc. While no close will please everyone, encountering significant resistance may be an indication of a close that is poorly written or insufficiently explained. On occasion, closers who are challenged will return to ANRFC to request a second opinion, in lieu of the time and effort required for a formal closure review.

Note that non-admin status is not in itself a reason to overturn a close (see this request for comment). Nevertheless, non-admins often focus on closing less controversial discussions. There is a limited latitude for non-admins to close discussions that are potentially controversial – if the close is good then it will stand, but if not then you may find a greater level of disapproval than an admin would. The level of disapproval will be lower if there is a long backlog of unclosed discussions. Some non-admins prefer to declare their status in the closure, usually by adding the NAC template. This isn’t required for RfCs since the closer’s status isn’t supposed to make any difference, but it may be a good idea to include it in cases where the closer is normally expected to be an admin, such as deletion discussions. Additionally, non-admins should not usually close discussions for which the result requires administrative status to perform, if the discussion relates to the use of the admin tools (for example, many discussions at ANI), or if an admin was requested at ANRFC.

Treat every objection seriously, even if you don’t think it should be. An implied refusal to seriously consider or discuss an objection may give the impression that you didn’t do due diligence in closing, or that you’re allowing personal considerations to affect the integrity of the close. Make sure to carefully consider each one, especially if the points being made didn’t occur to you when considering possible objections to your close. Do not dismiss or discount an editor’s concerns solely because they supported a position that you closed against. However, note that introducing new arguments (or repeating ones that were already made) is not in itself an objection – those should have been introduced during the discussion, and used to convince other editors rather than yourself.

In return for your consideration, the objecting editor is usually expected to discuss the issue with you before filing a formal appeal (which occurs at WP:AN), and not to revert the close while the appeal is ongoing (although both of these points are sometimes ignored). This is not a bureaucratic step inevitably leading to appeal, and a talk page inquiry is just as likely to be a good faith clarification question. Even if the questioner already intends to appeal, their concerns can often be resolved more easily, and you should take this as an opportunity to avoid needing the bureaucracy of a full appeal, or at the very least to understand the concerns better so as to defend the appeal more effectively. In some cases, editing a word or two is enough to resolve the issue, and you should be willing to make the change as long as you feel the close still reflects consensus.

Appeals

A few rounds of discussion on your user talk should usually be enough for you to fully evaluate the objections and determine whether or not you’re going to change your closure. If not, you should invite them to appeal rather than continuing to discuss. Appeals can also be a good opportunity to get feedback on your work, even though that isn’t their primary purpose.

The review will generally evaluate whether the close is within discretion – that is, whether the close is reasonable, rather than whether it is the best possible close. As such, you shouldn’t feel that the close needs to be perfect. Appeals may also be based on procedural issues, such as the close being premature or the closer being involved. A record demonstrating what may be considered valid reasons for overturning a closure can be found at WP:AN/Closure_review_archive, which contains a history of past closure reviews.

For the most part, each closure should stand on its own, although you should write a response to the objections in the AN discussion to make it clear why you didn’t agree to make the requested changes. You can also expand on your closure as necessary to clarify your thought processes. Remember that the justifications, as for any aspect of a decision, must ultimately be based in the decisions and arguments of the discussion participants.

Act professionally, and be available to respond to questions. Remember that as an uninvolved editor you shouldn’t care what the outcome is, so it shouldn’t matter to you whether the close is changed or overturned. Note that in this discussion you’re just another editor, albeit one who is offering some special insight into the issue instead of !voting. If you find yourself taking anything personally or find it hard to assume good faith, that may indicate that you’re too involved to be an impartial closer and should have left it to someone else.