Wikipedia talk:Consensus 

The project page associated with this talk page is an official policy on Wikipedia. Policies have wide acceptance among editors and are considered a standard for all users to follow. Before editing this page, please make sure that your revision reflects consensus. When in doubt, discuss on the talk page. Always remember to keep cool when editing. Changes to this page do not immediately change policy anyway, so don't panic.

This talk page is automatically archived by MiszaBot II.

Any sections older than 30 days are automatically archived. An archive index is available here.


Archive
Archives

"Consensus is a partnership between interested parties working positively for a common goal." -- Jimmy Wales

Contents


So what is consensus?

This key Wikipedia policy article as currently written crucially fails to specify what constitutes a consensus. Is it a unanimous or a majority agreement of some community ? And what is the relevant community ? Dictionary definitions of consensus typically say it is either unanimity or else majoritarian agreement. So it is clearly important to decide which it is. But such definitions also leave open the further question of whether it is at least a simple majority (i.e. at least 51%) or at least a great majority (i.e. two-thirds) of the relevant community. But the more basic problem here is the article’s failure to identify what the relevant community is.

Thus the Wikipedia fundamental policy of editing by consensus is surely in effect empirically empty, whatever all its rubrics about discussion and procedure ? —Preceding unsigned comment added by 158.143.134.44 (talk) 17:00, 23 October 2008 (UTC)

Adding another essay

I've created Wikipedia:What is consensus? to help the community understand what is a pretty tricky and nuanced concept. Consensus is definitely an art, and not a science. Feel free to chime in with comments. Feel free to WP:BOLDly change the essay I've created, although I might politely ask that you try to keep it in the same basic spirit I've tried to put forward. I'm even optimistic that some of what I've written in the essay is sensible and reasonable enough that it should be included right here at the policy page for WP:CONSENSUS. Randomran (talk) 01:38, 25 October 2008 (UTC)

Interesting, but founders on failure to define or identify what the relevant group is.--Logicus (talk) 17:12, 26 October 2008 (UTC)

Proposal

I would like this added to the policy. I think it is consistent with the policy as it stands, and some of you may feel it is implicit. I think it needs to be explicit. My intention is not to change the policy to but to clarify it, because Wikipedia is very different from other organizations that work by consensus.

At Wikipedia, consensus as an ongoing process, not a specific achievement. This is because Wikipedia is different from other organizations that address an issue, make a decision, and move on. Wikipedia is an ongoing project: we consider all articles to be works in progress, no article is ever finished, and people are editing articles all the time. Under these conditions "consensus" does not signal unanimity, but rather a process through which people with very different views and interests collaborate. Consensus-based edition requires above all else a commitment to work with others. This commitment in requires two other commitments: a commitment to our core policies, because no matter how different our views those policies provide a shared point of reference and a framework for working together; and a commitment to explaining the reasoning behind our edits, a willingness to listen to the reasoning behind someone else's edits, and a willingness to seek a compromise when both sides are complying with policy. Someone who does not participate in this process cannot use the concept of "consensus" to exercise a veto-power over others' edits. Someone who does not participate in this process cannot reject someone else's edit and use "there is no consensus" as an explanation. Only if one is part of a consensus-building endeavor, only if one demonstrates a commitment to collaborating and compromising with others, within the framework of our policies, does their objections to an edit count as a failure to achieve consensus. When editing is especially contentions a poll can be useful; if all editors on a page agree, save one or two, most editors would agree that there is a consensus. But if editors are evenly divided or close to it, there is no consensus. At that point it is a good idea to call for a WP:Request for comment in order to draw more people into the discussion.

If people think this is reasonable but take issue with the wording I invite others to edit it until we have, uh, a consensus. I know I could put this in an essay but I think many editors, especially newbies, need real clarity about how "consensus" does not give any single editor veto-power over edits to an article. This is a real problem for many newbies, and the policy should address it, this is my intention. Slrubenstein | Talk 13:52, 31 October 2008 (UTC)

I think that this proposal would require substantial work to make it useful and balanced, and I'm not entirely sure that it really adds anything.
Is it safe to assume that you're in a frustrating dispute at the moment? WhatamIdoing (talk) 01:18, 1 November 2008 (UTC)
well, I have to admit that I think something along these lines would be useful to implement as policy. I've seen claims of consensus without discussion used far too frequently as a form of page-ownership (cave-man logic, e.g.: "we got to this cave first: you can say it's not our cave all you want, but if you change anything we're going to poke you with a pointy stick"). it makes for well-written but often deeply biased articles. sad state of affairs when it happens. --Ludwigs2 02:25, 1 November 2008 (UTC)

Mostly disagree. This directly contradicts WP:SILENCE, WP:CCC. While it is common practice in some parts, it's also extremely pathological. Posting a request for comment on an already contentious poll is throwing oil on the fire. How hard are you trying to exceed Dunbar's number?

Typically when someone walks into a page via WP:BOLD they start out as the minority. By starting a poll, the on-page majority can freeze out the "newcomer". Over time, the number of these newcomers ends up larger than the people originally watching the page, and the disconnect with the community grows... until finally one of these (not so very) 'disruptive trolls' lists the page on *FD, and it gets deleted or shut down (most famous example is Wikipedia:Esperanza).

By calling too many people onto a page at once, the amount of organisation you need to do to keep them all working productively spirals out of control, and you will typically fail to reach consensus without a mediator. In this situation the people will end up roughly evenly divided due to simple statistics. It doesn't tell you anything about the situation.

These kinds of things make people believe that finding consensus must take months. In reality, if you're not making progress in a couple of hours, (or at worst days, if you happen to be in badly-aligned time zones) , something somewhere is b0rked.

summary:

Finally, all proposals fail. :-P We have sophisticated mechanisms for diffing texts and finding reliable middle ground. Be bold and use them.

--Kim Bruning (talk) 23:45, 17 November 2008 (UTC)

I don't have a problem with anything that Slrubenstein is proposing. However, I think it needs some work. Right now it doesn't seem to me to be written as policy. Rather, it is more like an essay. SLR: would you be willing to summarize in point form what you would like to see in the policy? I would certainly like an addition to the policy to make it more specific about what consensus is and how it is determined. I will propose something below. Sunray (talk) 23:54, 17 November 2008 (UTC)
Well, consensus is not determined by polls. Nor is it determined by proposals. :-) Don't propose please. Just Edit. Let's see what you come up with.
But watch out, because the whole "proposal process" concept (which is to say, make a proposal, hold a poll, request for comments, etc) has been shown to have failed more often than that it has succeeded, and there is simply no way around that. :-P I've been killing proposals for a proposal process for about as long as I'm on wikipedia. (I do it the honest way, mind you, by showing why it has never worked, isn't working now, and probably is unlikely to work in the foreseeable future). I'm not sure why people clamp on to it so strongly, but it's definitely a perennial proposal.
On the other hand, the "edit process" (currently described at WP:CONSENSUS if all is well) has several million success stories behind it. How about we look at the process that actually works, before we decide to promote the perennial proposal instead?
(of course, if you use the edit process to try to put forward that proposals and polls and exceeding dunbars number and possibly violating the laws of physics actually all work... there's some catch-22-like humor in that ;-) )
--Kim Bruning (talk) 00:45, 18 November 2008 (UTC) short version: If I 'win' you'll have learned something new about consensus and hopefully become more empowered (win-win); if you make me lose, we both lose, because you'll have disenfranchised yourself and others by promoting a severely suboptimal practice (lose-lose). Welcome to non-zero sum reality ;-) . And you're doing it simply by following the wrong process from start. See the talk page archives for details on consensus process, insofar they're not listed on the page. There's a reason certain processes are not listed here ;-)
But note that I do agree with much of the text slrubenstein puts forward. (To put it in arbcom language: The 'findings' are good, it's just the 'remedies' that kinda suck ;-) ). Let me ponder about how to phrase that. --Kim Bruning (talk) 00:56, 18 November 2008 (UTC)
That sounds great. I take your point about proposals. My particular concern would, perhaps, be better dealt with if I expressed it as a problem rather than a proposal. Since it is different than what SLR is talking about, above, I will start a new section. Sunray (talk) 01:18, 18 November 2008 (UTC)

Determining consensus

I've referred countless people to WP:CON, but I must say, it doesn't always help. It's great when you have reasonable editors editing relatively harmoniously. My problem with it is that has little to say about consensus decision-making in conflict situations. For example, by the time things get to mediation, we usually have folks with fairly entrenched positions. It would be nice to think that they could just keep editing until they get something they can all live with. But often they are at a complete impasse. Think two different ethno-religious groups at odds over some description. It may come down to a word or a phrase but it is deadlock. Here's the problem.

Problem statement: Achieving consensus in conflict situations can be challenging. It is often helpful to have a relatively straight-forward decision rule. Consensus is not necessarily unanimity and it is distinguished from majority rule. It is inevitably a super majority When there is a deadlock, it is helpful to have a decision rule, e.g., two-thirds, three quarters or four fifths, majority.

I believe it would be helpful to have a section in the policy that addresses the need for an agreed upon decision rule and introduces the notion of super majority. Comments? Sunray (talk) 01:43, 18 November 2008 (UTC)

Achieving consensus in conflict situations can be challenging, but there are a number of alternatives. Such as WP:RfC, WP:3 and if needs be WP:ANI. In addition we have our policies and guidlines WP:V, WP:RS, WP:NPOV and our policy here. In conflict situations were need Admins to step up to the plate and enforce policy and address the breeches. --Domer48'fenian' 08:59, 18 November 2008 (UTC)
If a policy needs enforcing on a regular basis, then perhaps it's not a policy (policies are our best practices documents. If you need to continuously force people to apply one such document against their will, maybe it really isn't such a good practice after all.) . Admins are not obliged to "enforce policy" (and are, in fact, encouraged to ignore all rules O:-) ) --Kim Bruning (talk) 05:13, 19 November 2008 (UTC)

Sunray is far from alone in his belief. However, a codified Wikipedia:Supermajority is the antithesis of Wikipedia:Consensus. Counting numbers and isolating your opposition is contrary to negotiation and compromise, which is necessary for consensus building. -- SmokeyJoe (talk)10:24, 18 November 2008 (UTC)

IIRC Supermajority voting was originally mentioned here, and deliberately removed. To repeat the ancient story, we do not allow voting in the main namespace to prevent people voting on content issues, where they might decide that 2+2=5 or π=3 . Such voting has been observed to undermine NPOV, which is a foundation issue. Not all issues are quite as clear cut as the examples, things can be quite subtle, and you might not notice you've made an NPOV violation until some expert comes along later.

We did allow voting (supermajority, simple majority, it doesn't matter since it's all voting) in the other namespaces. Of course, people ended up carrying over the voting systems from the other namespaces into the main namespace. Oops.

In the end, we ended up using consensus everywhere, just for the sake of consistency.

As it turns out, this was a good idea.

Now as for contentious issues, you can certainly reach consensus on an NPOV position, you just need to know how. (one place to start is Wikipedia:Writing for the enemy, for instance). In fact, NPOV and the wiki-consensus system were pretty much designed to fit together, IIRC. I'd have to look up the details on wiki:WardsWiki sometime :-)

--Kim Bruning (talk) 13:56, 18 November 2008 (UTC)

The statement that "supermajority is the opposite of consensus" is problematic. Consensus is not necessarily unanimity. When the decision is not unanimous, it is, by definition, a supermajority. I realize that we cannot go back to requiring a supermajoity of a certain amount across the board. I am wondering whether the policy should talk about the need for a decision rule, which would then be determined by participants in specific cases—for example, in a particular forum or case (on the talk page of a particular article). Sunray (talk) 17:29, 18 November 2008 (UTC)
I like the supermajority language because it emphasizes that unanimity will not always be achievable.
Fundamentally, the goal of unanimity assumes that all editors are rational adults with a primary interest in a properly referenced, neutral, encyclopedic article. This doesn't work in practice, as anyone that has worked on controversial topics can attest.
Imagine (a hypothetical example) an article like Schizophrenia and an editor that repeatedly deletes all positive mention of pharmaceutical treatments. You can "discuss" this to the point of putting tens of thousands of words on the talk page, but if a single editor's bottom line is that the article must never indicate even the smallest benefit from pharmaceutical treatments, then you can't actually reach a unanimous perspective. It is not possible for an article to simultaneously contain, and to not contain, a given piece of information.
If the editor then removes the "offensive" material each day with a polite talk page explanation, then you can't invoke SILENCE or declare that you've reached the sort of functional compromise in which the article isn't subject to an ongoing edit war. Ongoing edit war ≠ Consensus.
This page might benefit from some practical suggestions to the supermajority on constructing a "least offensive" version or acknowledging the controversy. To follow the above example, the appropriate information about pharmaceutical treatments should be included in the article, but it might also include a suitably referenced sentence about pharmaceutical treatments being opposed by various activists. While that would doubtless not be sufficient for our determined POV editor, it is something that the supermajority could do as a gesture of compromise while essentially imposing their version over the objections of the agenda-driven editor.
It also might benefit from an explanation that the point is to achieve consensus on how to best comply with Wikipedia's policies in the article, instead of how to most effectively promote various POVs. WhatamIdoing (talk) 20:47, 18 November 2008 (UTC)

Consensus is not unanimity, supermajority, majority, or anything that can be measured by numbers. Consensus is about agreement, about modifying the proposal, often watering-down, to take account of all valid positions.

Unanimity gives absolute power to every individual to block. Supermajority allows for the partitioning and ignoring of a minority view. These decision rules undermine consensus building. Consensus is about weighing arguments, recognising opposing viewpoints, and collective judgment on the debate. The conclusion of a consensus building debate will necessarily be NPOV, our most important policy, which is why consensus works so well for wikipedia (when it works).

I do not think it is a good idea to even mention supermajority. Instead, we could note that minority opinions can only be ignored after they have been defeated by logical debate. --SmokeyJoe (talk) 01:04, 19 November 2008 (UTC)

That's a useful way to put it. Logic wins out over majority or unanimity. ;-) --Kim Bruning (talk) 05:03, 19 November 2008 (UTC)

Would it be useful to look at how some consensus building organizations operate, and how they define consensus? I am thinking of ASTM International, for example. The essence of their process is (a) any voting member can vote and (b) negative votes must be accompanied by a written statement of reasons for the negative and remedies that would be acceptable. Addressing the reasons and adopting an acceptable remedy usually satisfies the negative voter. If not, there is an elaborate due process to ensure that minority voters are not being isolated, suppressed, bullied, alienated, overruled, or dictated to. I would say minority opinions can be defeated by logical debate; so can majority opinions, for that matter. Don't even mention ignoring. --Una Smith (talk) 05:35, 24 November 2008 (UTC)

Yes, it would be very useful. It would be useful both here and in the mainspace entries Consensus and Consensus decision-making. Agreed, don't even mention ignoring. --SmokeyJoe (talk) 05:46, 24 November 2008 (UTC)
In guidelines for its technical committees, ASTM International defines consensus as the judgment arrived at through the balloting and review procedures of these regulations (web link to "Green Book" PDF). Those regulations incorporate by reference several other documents, including Bylaws. Most of the Bylaws concern openness, transparency, and timeliness of the process, but among them is this principle: Careful attention to minority opinions throughout the process. That principle corresponds to the Wikipedia principle of NPOV. --Una Smith (talk) 15:03, 25 November 2008 (UTC)

When consensus fails its voting or authority rule

29-Nov-2008: The term "consensus" is, by considering the alternative, a unanimous agreement. It is not possible to get someone's consent when they don't. (end of proof []). When consensus fails, someone will be left out, probably by WP:3RR "double-teaming" against one person who can't out-revert an opposing group. However, ideally, a clear definition would note that, when consensus fails, either everyone stops, or the group consents to vote (as, at least, a consensus on voting), or an authority figure makes the decision. Any other squirrelly warping of the word "consensus" becomes pages and years (has it already been that many years?) of weasel wording. Please stop the weaseling of the term "consensus" to mean a "wannabe joint agreement, but not quite, in which case, we declare consensus because it's sooo wiki-sounding" (did I say that out loud?). A consensus is a total joint agreement, at the time, perhaps by compromise among everyone, but anything else is a failed consensus, resolved by some other method. After reaching consensus, when a person sneaks back (as many people have) and reverts the group decision (without user-talk), that is a "violation of the consensus" (or "false consensus" if the person agreed while planning the revert). Anyway, please stop writing "87 paragraphs" about reaching a consensus, when there isn't one. You'll never get consensus, with me, to accept warped definitions about consensus, no matter how "clarified" by endless complification. A "consensus" is a unanimous agreement (see proof above), else seek another decision-making method. -Wikid77 (talk) 17:14, 29 November 2008 (UTC)

Actually, in consensus forms of government, "consensus" usually means some kind of uneasy truce in which holders of a minority view are willing to accept, or at least not actively block, a compromise offered by the majority. Consensus does not mean that everybody agrees, it means that everybody disagrees to about the same extent. "Consensus" does not mean "unanimity." <eleland/talkedits> 21:46, 29 November 2008 (UTC)