Consensus for its implementation was not established within a reasonable period of time. If you want to revive discussion, please use the talk page or initiate a thread at the village pump.
It contains the advice or opinions of one or more Wikipedia contributors. This page is not an encyclopedia article or a Wikipedia policy, as it has not been reviewed by the community.
Wikipedia uses the words "vote" and "poll" as quick shorthand for what we are actually doing. It doesn't help that votes and polls actually look similar, so when many people first encounter a poll, they believe they are being asked to vote. This causes many misunderstandings, and is in fact not the case.
Vote, by "voting" we mean a process by which people reach decisions in a majority based democratic system. Wikipedia is not a democracy nor a system of government, however, so we do not use voting as a process on en.Wikipedia itself, although there are very rare experiments.
Decisions are normally made by consensus: an agreement to a certain course of action by editors. Consensus is not quite the same as unanimity. Especially when working with "rough consensus" (which is necessary to get past minor issues somewhat more expeditiously), the differences are rather marked.
Consensus is normally reached through negotiation. Rarely, but often enough to be noticeable, negotiators might use polls as part of their toolkit.
A poll is a survey (a measuring tool) which determines the current state of a situation, with respect to consensus. It doesn't form consensus. It merely measures it.
In the same way that a ruler does not change the length of your finger or make a new finger (it merely measures it), a poll does not change or make consensus.
What a poll can do is give you insight into where people stand on an issue.
Polls are typically used in somewhat more complex situations, where it might be hard to otherwise get an overview:
Early or in the middle of a discussion, to determine where people stand, and who needs to negotiate with whom.
When we think consensus has already been reached, but we want to make sure we haven't accidentally missed any important opinions.
At one point in time, certain processes were also set up to use poll structured discussions. We still use these processes:
When we would like to make fairly simple decisions at a centralised location in a short amount of time.
Poll structured discussions may not be an optimal way to handle particular topics, and people are making efforts to slowly deprecate them. Unfortunately they have become somewhat deeply ingrained, so this will take quite some time.
Especially based on current insights on simplification and scaling, we don't recommend designing any new poll structured discussions at this moment in time.
Polling discourages consensus
Having the option of settling a dispute by taking a poll, instead of the careful consideration, dissection and eventual synthesis of each side's arguments, actually undermines the progress in dispute resolution that Wikipedia has allowed. This is a strength, not a failing, and is one of the most important things that make Wikipedia special, and while taking a poll is very often a lot easier than helping each other find a mutually agreeable position, it's almost never better.
Polling encourages the community to remain divided by avoiding that discourse; participants don't interact with the other voters, but merely choose camps. Establishing consensus requires expressing that opinion in terms other than a choice between discrete options, and expanding the reasoning behind it, addressing the points that others have left, until all come to a mutually agreeable solution. No one can address objections that aren't stated, points that aren't made.
Yes, establishing consensus is a lot harder than taking a poll. So are most things worth doing.
Definition
While a straw poll is not a substitute for discussion it can be a tool for probing opinions especially outside of article content discussions, so that one knows who to talk with to obtain a negotiated consensus. Straw polls have been used on Wikipedia for such purposes almost since the beginning of the project. Straw polls typically don't have opening and closing times. Instead, they give editors a chance to chip in with an indication of their opinion, together with a short summary of the reasoning that corresponds to that opinion. It's a good idea to keep a more detailed reasoning in mind given the likelihood that your opinon will be challenged. A call for a straw poll may trigger discussions instead—that's not a failure, it just means that the issue is not clear-cut, which is what the poll set out to determine in the first place.
Sometimes it's useful to take a survey of opinions on some issue, as an aid to achieving consensus and an indication of which options have the most support. Surveys should never be thought of as binding.
Be aware that Wikipedia is not a democracy: A straw poll is not a binding vote, or a way to beat dissenters over the head with the will of the majority. If a large number of people support one option but some don't, this doesn't mean that that's the "outcome". It means some people are disagreeing, and those people's objections need to be addressed!
Straw poll survey guidelines
Editors considering an article-related straw poll must remember that polling should be used with care, and should not invoke straw polls prematurely. Note that straw polling cannot serve as a substitute for debate and consensus; that no straw poll is binding on editors who do not agree; and that polling may aggravate rather than resolve existing disputes.
Straw polls regarding article content are often inconclusive and sometimes highly contentious. In order to have a chance of being productive, editors must appreciate the following:
The ultimate goal of any article discussion is consensus, and a straw poll is helpful only if it helps editors actually reach true consensus.
For that reason, article straw polls are never binding, and editors who continue to disagree with a majority opinion may not be shut out from discussions simply because they are in the minority. Similarly, editors who appear to be in the majority have an obligation to continue discussions and attempts to reach true consensus.
For the same reason, article straw polls should not be used prematurely. If it is clear from ongoing discussion that consensus has not been reached, a straw poll is unlikely to assist in forming consensus and may polarize opinions, preventing or delaying any consensus from forming.
Similarly, if a straw poll is inconclusive, or if there is disagreement about whether the question itself was unfair, the poll and its results should simply be ignored.
Once responses to a straw poll have begun, even minor changes to the phrasing of the poll are likely to result in an all out battle over whether the poll itself was fair. Consider proposing straw poll language several days prior to opening the actual poll to responses, and beginning the poll only once you have consensus on the precise question to be asked.
Essentially, article content is not subject to straw polling. People have been known to "vote" on a fact, which is ultimately pointless.
The purpose of a straw poll is to stimulate discussion and consensus. Editors should evaluate the explanations that the participants in a straw poll offer, and should see if those explanations help to develop their own opinions or suggest compromise. In this context, a few well reasoned opinions may affect a debate much more than several unexplained mere votes for a different course. It's the opinions we're looking for, not a show of hands.
In the context of Wikipedia articles, straw polls are most helpful, if ever, in evaluating whether a consensus exists or in "testing the waters" of editor opinion among a few discrete choices such as two choices for an article's name. Even in these cases, straw polls may never be understood as creating a consensus, but merely as one tool to measure where consensus stands.
Straw polls should not be used excessively. If a straw poll was called on an issue recently, there is usually no reason to call a second poll, even if you think that consensus may have changed or that the first poll was conducted unfairly. If you disagree with the "majority" opinion, simply remember that the straw poll is not binding and continue discussions. Of course, this is subject to consensus. (A quick poll as per "changed mind", "haven't changed" mind, might occasionally be helpful when you're working on a negotiated consensus... to make sure you're not forgetting people!)
The words "vote" and "voting" have a variety of connotations, but on Wikipedia they are defined as ballot-casting or majority voting, especially by older Wikipedians. For that reason, the use of the words "vote" and "voting" are considered incorrect when describing Wikipedia processes. Surveying and commenting are words closer to what we do on Wikipedia in the English language. The actual shorthand for straw poll serveying used on Wikipedia is simply "Polling".
Certain polling is frequent, other polling not so frequent
A number of formal processes, such as WP:AFD rely upon polling and help streamline the functioning of the project. In general issues are almost never resolved with a poll, merely where individuals stand on the issues is clarified.
Content issues are almost never subject to polling. Out of our 1.7 million articles, fewer than 0.1% were ever polled upon.
Deletion issues can be subject to polls on {{Deletion debates}}, but in most cases they are resolved through other processes, such as WP:PROD and WP:CSD.
Guidelines are never created through polling on them, only consensus about aspects of guidelines are gauged. Gauging whether policy should be enacted happens but not frequently.
Novice users sometimes think they should make a "motion" and "call for votes", but Wikipedia doesn't work like that.
Feature requests are not subject to polling, simply because the developers have more important issues to consider than popular demand, such as server load. The developers, of course, do not function in a vacuum, so while a poll is not going to affect whether or not a feature request is implemented or not, the discussion and comments made during a poll likely will.
Editor conduct used to be subject to polling in the past, via a system called Quickpolls, but this has been deprecated years ago because it generated more heat than light.
Admins, Bureaucrats and Arbiters are, in most but not all cases, given corresponding tools after a poll to gauge community approval.
Creating a survey
These guidelines provide a framework that may be followed when creating a new survey. These are not binding in any way.
Any Wikipedian may start a survey on any topic, but attempts to reach consensus are much, much, MUCH preferred, and should perhaps be followed even when it pains us most.
Consensus must be reached about the nature of the survey before it starts. Allow about a week for this process.
In general, surveys are to help gauge the degree of consensus on an issue, such as whether a particular article version appears to be POV or NPOV. Surveys should not be used for the purposes of "fact finding".
A deadline for the survey should be considered so as to resolve the issue in a timely manner.
Once started, the questions and wording in the survey should not change. However, if someone feels that the existing survey is seriously flawed, this is typically an indication Step 2 was not completed properly.
If the majority of opinion is in one direction, but a significant minority of people oppose it, work to find a solution that can be accepted by as many people as possible.
Multiple Support or Oppose sections in a survey make navigation difficult, and edit summaries which read (→Support - this option is best) make it difficult to observe on watchlists. Consider using unique headers, such as parenthetical headers used in the massive Admin accountability poll.
Survey etiquette
If you are posting on talk pages, asking experienced editors to give their opinion on an issue, make sure not to use language that may suggest bias.
Good: "Hey, Bob, could you tell me what you think about this discussion? I think your input could help."
Bad: "Yet another attempt to push POV-- please help!"
Sample survey
Note that this is purely a sample of one way to organise such a survey—different circumstances may call for different approaches.
Please sign your name using four tildes (~~~~) under the position you support, and please add a (hopefully brief and well thought out) comment. If you are happy with more than one possibility, you may wish to sign your names to more than one place. Extended commentary should be placed below, in the section marked "Discussion", though brief commentary can be interspersed.
Discussion resulting from the survey would go here. If there were a significant amount, it might be moved to a talk page instead.
Points that need to be integrated
This is a sort of semi-sandbox, where points that are yet to be integrated in the main text can be kept
Polls on article content should generally be avoided. They can lead to (accusations of) wikiality.
Polls on policy proposals are likewise a bad idea
Polls on feature requests are totally ignored by the developers. They are essentially a different community, and they frankly have no time to try to figure out what precicely goes on day to day on the pages of en.Wikipedia or any of the other 700 wikis run by the Wikimedia foundation. That and polls can't overrule the laws of physics. ;-) Developers do what is possible in the real world. This doesn't mean that developers aren't willing to help though! If you have a problem or a decent idea, try approaching them directly!
Polls are non-binding. (see under Consensus - no binding decisions)
As anywhere else, your opinion on a poll may be challenged. Be prepared to explain your opinion further, and be prepared to explain your logic.
If you think someone is holding an illogical view, feel free to engage them in good faith, and let them explain their views.
It is fair game to try to change a persons' opinion on a poll (but there are certain standards of politeness and decorum... can we describe them?)
Retroactively making a poll binding (e.g. "it's not binding but we'll do what the majority wants") is a hazard to be avoided
AFD et al. are not infrequently decided in favor of a minority with better and policy/guideline based arguments. (because logically worded arguments will advance consensus further. )
RFA/ArbElect are subject to interpretation by the closer
Some pages (RFA, *FD) use variants on polling, and also introduce the concept of Rough consensus. This means that instead of unanimous support or "I don't care", people make do with only say 80% support. This does not quite make these processes a vote, however (see above). and you can be caught by surprise if suddenly the whole poll morph into a discussion, or if you find yourself "badgered" about a valuable oppose position that you are holding. (It's an artifact: if there's an 80% cutoff, and you are currently opposing, and your opinion is reasoned and clear, your opinion can suddenly be worth as much as that of 3 or 4 other people! It's almost WP:BEANS to suggest it, but this is in fact your best bet at negotiating an improvement in conditions (improved behavior by candidate on RFA, or changes to a page that you feel are essential on *FD) .:-) Of course, you are not invincible, and you should beware of overplaying your hand, lest you be labeled a troll!
