Chapter 5
§1 The Challenges of Consensus

¶1 Consensus: Any group in agreement about something whose opinion is the same as yours; antonym of cabal [i.e., those who disagree with you]. — Wikispeak

¶2 H. G. Wells thought the “World Encyclopedia” should be more than an information repository, it should also be an institution of “adjustment and adjudication; a clearinghouse of misunderstandings.” 1 Wells, “The Idea of a World Encyclopedia,” 921. Wikipedia certainly has its share of misunderstandings, some imported from the conflicted world it documents and some unique to its own undertaking. An example of a contagious real-world conflict is the “Creation-Evolution Controversy,” 2 Wikipedia, “Creation-Evolution Controversy,” Wikipedia, July 27, 2008, (visited on August 7, 2008). discussed in chapter 3. Also, political and ethnic differences are often mirrored at Wikipedia, prompting the formation of a “Working Group on Ethnic and Cultural Edit Wars.” 3 Wikipedia, “Wikipedia:Working Group on Ethnic and Cultural Edit Wars”, Wikipedia, January 23, 2008, (visited on January 23, 2008). There are also plenty of local “misunderstandings,” such as whether every episode of Buffy the Vampire Slayer deserves its own article. I raised this dispute earlier to illustrate two opposing philosophies at Wikipedia: inclusionism and deletionism. This issue, and the proliferation of articles, gave rise to an even more trivial — though no less bellicose — debate: If every television show episode has its own article, how should these articles be named so as not to conflict with other articles? This discussion reveals possible misunderstandings about consensus, and the difficulties of this decision-making practice in an open community.

¶3 In this chapter, I identify the difficulties of consensus decision making, and its meaning and practice for collaboration at the English Wikipedia. I consider this relative to insights from literature about consensus in other communities, including Quakers and the collaborators who built the Internet and Web using “rough consensus and running code.”

§2 The Case of Disambiguation

¶4 In the history of the encyclopedia much has been made of the attempts to organize knowledge and how that dream was eventually superseded by simple alphabetical order. 4 On the organization of knowledge, and alphabetization in particular, see Peter Burke, A Social History of Knowledge: from Gutenberg to Diderot (Cambridge, MA: Polity Press, 2000), 109-115; Daniel Headrick, When Information Came of Age (Oxford: Oxfo ... Wikipedia continued this trend by avoiding any formal organizational scheme and letting people simply name articles as seemed fitting; articles are then accessed via other pages, including user-created categories and search engines. 5 The new mode of dealing with information is described by Weinberger, Everything Is Miscellaneous. But a problem soon emerged: what happens when article titles conflict? A page’s title must be unique because it is also part of the Web address of the page. (Computer scientists call this a “collision” and it became increasingly common as the number of Wikipedia articles increased.) For example, what should the “Buffy” article contain? Should it be about the fictional character, Buffy the Vampire Slayer, or the related film, TV show, comics, or novels — and which season or episode, issue, or book? Plus, there have been a few notable real-life Buffys. How should Wikipedia distinguish between them all?

¶5 In the case of a collision Wikipedia will likely offer the reader a “disambiguation” link at the top of an article or a whole page with a list of links to more specific articles, or both. “Disambiguation in Wikipedia is the process of resolving conflicts in article titles that occur when a single title could be associated with more than one article. In other words, disambiguations are paths leading to different articles which could, in principle, have the same title.” 6 Wikipedia, “Wikipedia:Disambiguation,” Wikipedia, July 17, 2008, (visited on July 18, 2008). The “Buffy” article is in fact a disambiguation page that includes links to articles about the Buffy the Vampire Slayer film and television series, unrelated musical albums, biographies, and an astronomical object outside the orbit of Neptune. 7 Wikipedia, “Buffy,” Wikipedia, July 12, 2008, (visited on July 18, 2008). Furthermore, dozens of naming conventions have emerged that specify how to disambiguate collisions by qualifying the name with a parenthetical suffix, such as “Buffy the Vampire Slayer (film).”

¶6 However, must these disambiguators be applied in every case for consistency’s sake, or just those for which there is already a preexisting article? Answering this question, and thousands of others like it, is an integral part of Wikipedia collaboration. As the “Consensus” policy states, consensus is “how editors work with others,” it is “Wikipedia’s fundamental model for editorial decision-making.” Wikipedians are supposed to discuss and reason together, making use of verifiable sources and assuming good faith. “Policies and guidelines document communal consensus rather than creating it.” 8 Wikipedia, “Wikipedia:Consensus,” Wikipedia, June 1, 2009, (visited on June 10, 2009). However, in this case of disambiguation an interesting problem arose: there was disagreement as to whether there was consensus.

¶7 So then, what is consensus and how do you know you’ve reached it? This question eventually reached the Arbitration Committee: “a panel of experienced users that exists to impose binding solutions to Wikipedia disputes that neither communal discussion, administrators, nor mediation have been able to resolve.” 9 Wikipedia, “Wikipedia:Arbitration Committee,” Wikipedia, July 16, 2008, (visited on July 21, 2008). Jimmy Wales created the committee in 2004 because of the growth of the community and the corresponding number of disputes he could no longer personally attend to. Wales appoints members annually “based on the results of advisory elections” to arbitrate specific conflicts; however, the “ArbCom” seems to be evolving toward what might be thought of as Wikipedia’s high court in definitively interpreting — and some say making — Wikipedia policy. 10 For discussion of the extent to which the ArbCom interprets or makes policy see John Lee, “Re: BADSITES ArbCom Case about to Close,” wikien-l, October 17, 2007, ...

¶8 Presently, an ArbCom decision is documented on a wiki page in which disputants make statements and marshal evidence for their case. 11 “Evidence” is often densely sourced hyperlinks to various discussions and specific Wikipedia edits; I remove such references (e.g., “[32]”) from excerpts as they serve no purpose here. These are followed by the ArbCom members’ “preliminary decision” (e.g., to take the case and issue temporary injunctions) and conclude with a final decision that enumerates important principles of Wikipedia policy, findings of facts, remedies, and enforcement actions. Wikipedian Yaksha introduced the disambiguation case as follows:

¶9 This dispute is regarding whether articles for TV episodes which do not need to be disambiguated should have disambiguation. For example, Never Kill a Boy on the First Date (Buffy episode) has the disambiguation “(Buffy episode),” even though this disambiguation is not required. I believe we did reach consensus to follow the existing guideline of “disambiguate only when necessary.” The straw poll resulted in a supermajority (80%) support for “disambiguate only when necessary.” The discussion that followed supported this consensus. A detailed summary of the discussion, as well as four Request Move proposals all support the existence of this consensus. Given this, I (and others) begun to move articles which were inappropriately named. 12 Wikipedia, “Wikipedia:Requests for Arbitration/Naming Conventions”, Wikipedia, January 20, 2007, (visited on July 16, 2008).

¶10 However, not everyone agreed with the application of the use-as-needed policy and attempted to reverse the moving (or renaming) of articles. Wikipedian Elonka was of a minority that felt specific “WikiProjects” (i.e., pages and editors focused on advancing specific topics, such as a television series) should be able to use disambiguating suffixes consistently across their topic, whether needed or not. Putting aside “unethical tactics” that had been employed in the dispute, this minority that favored consistent-suffixes felt that “WikiProjects can set reasonable guidelines of their own.” “As television episode articles have been added to Wikipedia, most series followed the [use as needed] system, but many others chose to use a ‘consistent suffix’ system.”

¶11 Furthermore, the very process of deciding whether specific WikiProjects (e.g., television) could consistently use suffixes was troubled. Elonka noted that there was a poll, but its wording was confusing and contested “with multiple editors rapidly changing the wording and structure of the poll while it was in process.” Furthermore, “calls for a cleanly-run poll were belittled as ‘stalling,’ ‘immature delay tactics,’ and the ‘whining’ of ‘sore losers’ engaging in ‘borderline trolling’ who should just, ‘Give the fuck up, you lost.’ ” 13 Ibidem.

¶12 Had there been consensus on the naming of television episodes? Before returning to the details of this case, it’s best to first review the meaning of consensus and its seminal role in the development of the Internet.

§3 “Rough” Consensus

¶13 The Wiktionary definitions for consensus speak of “general agreement,” “without active opposition to the proposed course of action.” A more scholarly source gives a similar definition: consensus is overwhelming agreement “which does not mean unanimity.” 14 Wiktionary, “Consensus,” July 11, 2008, (visited on July 11, 2008); Lawrence E. Susskind, “A Short Guide to Consensus Building,” chapter 1 in Consensus Building Handbook, ed. Lawrence E. Susskind, Sarah McKea ... The encyclopedic article “Consensus Decision-Making” lists requirements of consensus that, if achieved, can also be considered benefits: inclusive (“as many stakeholders as possible”), participatory (“actively solicit the input and participation of all”), cooperative (“reach the best possible decision for the group and all of its members”), egalitarian (“equal input” with the “opportunity to table, amend and veto or ‘block’ proposals”), and solution-oriented (“emphasize common agreement over differences”). 15 Wikipedia, “Consensus Decision-Making,” Wikipedia, July 10, 2008, (visited on July 11, 2008). This is not unlike the meetings of one of the better-known practitioners of consensus: the Quakers. Michael Sheeran, a Jesuit scholar, writes of the history and practice of Quaker consensus in Beyond Majority Rules: Voteless Decisions in the Religious Society of Friends . In his study Sheeran notes nine features of Quaker meetings and decision making. 16 Sheeran is affirming nine characteristics first articulated by Stuart Chase, Roads to Agreement: How to Get along Better with Other People. (New York: Harper & Brothers, 1951), 51-52; Sheeran, Beyond Majority Rule, 51. One of those characteristics, central to the Quaker spiritual experience, has no analog in Wikipedia: silent periods at the start of meetings and when conflict arises. The characteristic of “small meetings” sometimes holds in the Wikipedia context for issues local to an article or project, but not at the larger scale. 17 Andrea Forte and Amy Bruckman, “Scaling Consensus: Increasing Decentralization in Wikipedia Governance,” in Proceedings of the 41st Annual Hawaii International Conference on System Sciences (Washington, DC: IEEE Computer Society, January 7–10, 2008 ... The remaining seven characteristics roughly parallel Wikipedia norms: unanimity and a lack of voting (e.g., “voting is evil”); pausing when agreement cannot be reached; participation by all those with ideas on the subject; listening with an open mind; facilitators, but no “leaders”; egalitarianism; and a factual-focus.

¶14 Therefore, consensus certainly seems like an appropriate means for decision making in a community with egalitarian values and a culture of good faith. Furthermore, this form of decision making has been central to online collaboration since the Internet’s start. Yet, while consensus might seem simple enough in theory, it is rarely so in practice, as is evidenced by the 1,176 pages of The Consensus Building Handbook . 18 Lawrence E. Susskind, Sarah McKearnen, and Jennifer Thomas-Lamar, Consensus Building Handbook (Thousand Oaks, CA: Sage, 1999). The history and challenges of online consensus, particularly this question of who decides when one has it, can be seen in the development of technical standards at the Internet Engineering Task Force (IETF) and World Wide Web Consortium (W3C).

¶15 Both of these institutions host technical working groups that develop standard specifications for Internet protocols (e.g., TCP/IP) and Web formats (e.g., HTML). They work primarily over mailing lists, though teleconferences and occasional face-to-face meetings are common. The IETF is one of the oldest existing collaborative institutions on the Internet — it can be said to have built the Internet. And the W3C, responsible for many Web technologies, might be thought of as an institutional fork resulting from, in part, frustration over the slow pace of work at the IETF. Much of this frustration was a result of trying to come to consensus over technical philosophical differences. One of the most contentious issues had to do with naming/identifying things on the Web — further evidence that naming things is not as easy as one might initially think. In this case the disagreement was about whether the string of characters one types into the address bar of a browser should be thought of as a stable identifier for that Web resource (i.e., URI) or just a locator (i.e., URL). 19 Berners-Lee, Weaving the Web, 62-63. (This distinction is confusing, might seem trivial to most, and has become less of an issue, but it was of great concern to those involved at the time.) And while the W3C, unlike the IETF, has a paying membership that helps support a full-time staff (to hopefully speed the work along) and has Tim Berners-Lee as director to lend coherence and direction to Web architecture, consensus decision making at the working group level was retained. 20 A history of the IETF, including its own sometimes troubled relationship with the traditional International Organization for Standardization (ISO) can be found in Andrew L. Russell, “’Rough Consensus and Running Code’ and the Internet-OSI Standards ... The W3C process document states: “Consensus is a core value of W3C. To promote consensus, the W3C process requires Chairs to ensure that groups consider all legitimate views and objections, and endeavor to resolve them, whether these views and objections are expressed by the active participants of the group or by others (e.g., another W3C group, a group in another organization, or the general public).” 21 Ian Jacobs, “World Wide Web Consortium Process Document,” W3C, October 14, 2005, (visited on October 14, 2005). Ironically, as the W3C matured, it too would be characterized as overly slow because of growing bureaucracy and the difficulty of achieving consensus in large and interdependent groups. Furthermore, Berners-Lee’s leadership role, which was intended to mitigate these problems and lend architectural coherence to the emerging standards, was occasionally challenged as not in keeping with the consensus practice of the working groups. 22 The W3C’s — and Berners-Lee’s — influence is questioned in Simson Garfinkel, “The Web’s Unelected Government,” Technology Review (November 1998), (visited on September 2, 2007). ... In turn, Jon Bosak, a “father” of XML, a data-markup and exchange format — and one of W3C’s most prominent successes — created an institutional fork for subsequent XML work. The Organization for the Advancement of Structured Information Standards (OASIS) jettisoned the ideas of appointing a director and consensus in favor of the parliamentary, and more clockwork-like, Robert’s Rules of Order.

¶16 In any case, it was at the IETF in 1992 that computer scientist David Clark characterized IETF collaboration: “We reject: kings, presidents and voting. We believe in: rough consensus and running code.” 23 David D. Clark, “A Cloudy Crystal Ball: Visions of the Future,” IETF, July 1992, (visited on March 26, 1999), 19. This “IETF Credo” would become one of the foundational aphorisms of collaborative culture on the Internet. Furthermore, this simple statement reflects the egalitarianism — and meritocracy — seen in previous chapters of this book. It also hints at a source of skepticism of some Free and Open Source Software (FOSS) developers toward Wikipedia since encyclopedic articles cannot be compiled and “run.” Returning to the question of whether 80 percent support for a Wikipedia policy constituted consensus, in the IETF Credo we also see the intriguing notion of “rough consensus” as documented in the IETF’s “Working Group Guidelines and Procedures”:

¶17 IETF consensus does not require that all participants agree although this is, of course, preferred. In general, the dominant view of the working group shall prevail. (However, it must be noted that “dominance” is not to be determined on the basis of volume or persistence, but rather a more general sense of agreement.) Consensus can be determined by a show of hands, humming, or any other means on which the WG agrees (by rough consensus, of course). Note that 51% of the working group does not qualify as “rough consensus” and 99% is better than rough. It is up to the Chair to determine if rough consensus has been reached. 24 Scott Bradner, “RFC 2418: IETF Working Group Guidelines and Procedures”, IETF, September 1998, (visited on March 26, 1999).

¶18 Yet, even 70 percent of a group, for example, as determined by a chair seems like a far cry from “general agreement without opposition” as described earlier. Indeed, to understand consensus one must consider a handful of issues including the character of the group, the constraints of time, the role of the facilitator, and group dynamics; all of which are made more difficult in the Wikipedia context.

§4 Deliberation and Openness

¶19 There are numerous methods for making group decisions; one might flip a coin, vote, or seek consensus — among others. Each has its merits and difficulties, and is more appropriate to some situations than others. Unlike the first two methods, consensus is not so much about quickly yielding a “yes” or “no,” but in arriving at the best possible solution. While the progress and the outcome of consensus are rarely assured, the focus is on the potential benefits of deliberation rather than the speed of the decision. (However, if consensus is achieved, the legitimacy of the decision will likely exceed that of a coin toss or vote.) As Wikipedia’s “Consensus” policy notes: “Achieving consensus requires serious treatment of every group member’s considered opinion…. In the ideal case, those who wish to take up some action want to hear those who oppose it, because they count on the fact that the ensuing debate will improve the consensus.” And even though polling may be a part of the consensus process, it is “often more likely to be the start of a discussion than it is to be the end of one.” 25 Wikipedia, “Wikipedia:Consensus,” Wikipedia, July 5, 2008, (visited on July 11, 2008).

¶20 But if consensus is a discussion, who is invited to the conversation? The IETF Guidelines notes: “It can be particularly challenging to gauge the level of consensus on a mailing list.” A mailing list probably has many more subscribers than actual active participants, and the number of messages is not a good indicator of consensus “since one or two individuals may be generating much of the traffic.” 26 Bradner, “RFC 2418,” section 3.3. Furthermore, the W3C makes allowances for notions such as quorum, supermajority, and members in good standing; these can be specified at the beginning of a group’s work in its charter. 27 Jacobs, “World Wide Web Consortium Process Document,” section 3.3. Wikipedia, and its topical projects, have no charter or formal list of members in good standing. It lacks many of the mechanisms other communities have to make the process of coming to consensus a little easier. Its openness is particularly problematic because it is susceptible to trolling and “forum shopping.” In the case of trolling, someone who simply wants to annoy others can ensure unanimity is never achieved and increase the chances that the group will collapse in frustration. With respect to forum shopping, the consensus policy notes: “It is very easy to create the appearance of a changing consensus simply by asking again and hoping that a different and more sympathetic group of people will discuss the issue.” 28 Wikipedia, “Wikipedia:Consensus (oldid=223708449)”. For example, during a dispute about removing articles on “marginally notable characters” it was suggested that a policy about such biographies be documented. Another Wikipedian responded: “The problem is that we have nowhere near consensus for such a policy. A large number of editors support it, and a large number of editors oppose it. Different specific cases have gone different ways, mostly depending on who showed up to the debate that day.” 29 Delirium, “Re: Writing Style (Was: a Valid Criticism),” wikien-l, October 7, 2005, (visited on October 7, 2005). Wikipedia consensus policy counsels that this “is a poor example of changing consensus, and is antithetical to the way that Wikipedia works” and turns, again, toward reasoned deliberation: “Wikipedia’s decisions are not based on the number of people who showed up and voted a particular way on a particular day; they are based on a system of good reasons.” 30 Wikipedia, “Wikipedia:Consensus (oldid=223708449)”.

§5 Time and Precedence

¶21 If consensus is a process whereby participants discuss and reason together, openness has another challenging implication beyond the question of who is contributing to the conversation: in an open and forever-changing group, how long might any decision be considered the group’s consensus? On first blush — and beside those Wikipedia “foundation issues” considered to be beyond debate such as neutral point of view — “consensus is not immutable.” 31 Wikimedia, “Foundation Issues,” Wikimedia, March 26, 2006, (visited on April 6, 2006); Wikipedia, “Wikipedia:Consensus (oldid=223708449)”; for mutability of consensus, see Travis Kriplean et al., “Community, ...

¶22 Consider a Wikipedia discussion related to how annoying it is when a bookmark or link to a Web page no longer works. Tim Berners-Lee, in the essay “Cool URIs Don’t Change,” writes, “Pretty much the only good reason for a document to disappear from the Web is that the company which owned the domain name went out of business or can no longer afford to keep the server running. Then why are there so many dangling links in the world? Part of it is just lack of forethought.” 32 Tim Berners-Lee, “Hypertext Style: Cool URIs Don’t Change.,” W3C, 1998, (visited on July 31, 2008). The (English) Wikipedia, understandably, suffered from such a lack of forethought when it succeeded beyond expectation and wanted a home for other language versions. What should the URI “,” which used to identify the English article on Chernobyl, point to now? If “” became a portal for all large language editions, and the English article was moved to an English namespace (i.e., “”), should the old URI “break” (i.e., return an “uncool” 404 error message) or redirect to the new location? Wikipedian Rowan Collins wrote, “I think talking of this as a ‘contract’ [to not break URLs] is somewhat overdoing it — it’s an important point that this was the compromise reached during a previous discussion, but unless there’s a *very* strong statement promising to uphold it ‘forever,’ we generally treat all consensus policies as re-negotiable.” 33 Rowan Collins, “Re: Multilingual Redirect,” Wikipedia-l, October 21, 2005, (visited on October 21, 2005). This is in keeping with the consensus policy, which states, “It is reasonable, and sometimes necessary, for the community to change its mind.” 34 Wikipedia, “Wikipedia:Consensus,” Wikipedia, August 23, 2007, (visited on July 16, 2008). Change is sometimes reasonable, but eternally arguing about the same thing is not, as Wikipedian Philip Sandifer notes: “I’ve been fighting with the same people over issues with reliable sourcing for well over a year, for instance, and yet those fights still continue despite, seemingly, a substantial shift in opinion away from the former hardline positions (things that included overbroad statements about blogs ‘never’ being reliable sources).” 35 Philip Sandifer, “Reflections on the End of the Spoiler Wars,” wikien-l, November 14, 2007, (visited on November 14, 2007).

¶23 As in many of the issues facing Wikipedia, Wikipedians must achieve a delicate balance, this time between rehashing tired issues and reconsidering vital ones: between “the need for open and fair consideration of the issues against the need to make forward progress.” On this point, the IETF and W3C have some means for judging the merit of an issue. First, working group charters are carefully constructed so as to focus on issues that are amenable to resolution within a specified time frame. Second, the working group chair has a critical job in summarizing and recording discussion; while “it is occasionally appropriate to revisit a topic, to reevaluate alternatives or to improve the group’s understanding of a relevant decision,” “unnecessary repeated discussions” can be avoided with careful records of previous arguments and conclusions. 36 Bradner, “RFC 2418,” section 3.3. Additionally, reasonable criteria for the consideration of an issue are articulated by the IETF Guide:

¶24 To facilitate making forward progress, a Working Group Chair may wish to decide to reject or defer the input from a member, based upon the following criteria: Old: The input pertains to a topic that already has been resolved and is redundant with information previously available; Minor: The input is new and pertains to a topic that has already been resolved, but it is felt to be of minor import to the existing decision; Timing: The input pertains to a topic that the working group has not yet opened for discussion; or Scope: The input is outside of the scope of the working group charter. 37 Ibidem.

¶25 Such criteria can be applied to not only newcomers, but also those who were present, but silent, or changed their minds. Despite all the focus on conversation and the cacophony that sometimes accompanies the consensus process, silence is one of the greatest challenges to successful decision making. In a working group, silence in response to a request for comments or objections is rarely a good thing; hopefully participants will at least say, “Sounds good to me.” While one would like to think that “silence implies consent, if there is adequate exposure to the community,” as Wikipedia’s consensus policy states, 38 Wikipedia, “Wikipedia:Consensus (oldid=223708449)”. this can be a risky inference. Instead, silence often indicates confusion or a lack of interest. (On working-group teleconferences one might hear members speaking to colleagues at their office, their children at home, and even snoring!) At the IETF and W3C this prompted a step in their processes called “Last Call”: before engineers begin seriously implementing and testing the specification, those who have the right to deliberate in consensus also have an obligation to make their views known. This sentiment is also famously captured in the Anglican wedding ceremony: before two newlyweds are married those who would object must “speak now; or else for ever hold your peace.” Or, in the less politic words of a Wikipedian, people should “either put up or shut up”: people don’t have to participate, “but when they don’t they should not moan when their voice is not considered.” 39 Gerard Meijssen, “Re: New Request for Cantonese Wikipedia: Vote at 29-6”, Wikipedia-l, September 25, 2005, (visited on September 25, 2005).

§6 The Facilitator

¶26 In many consensus-based communities a facilitator performs a number of tasks, the most important of which is positing a consensus statement. The articulation of such a statement and asking for objections is central to Jane Mansbridge’s definition of consensus in her comparative study of decision making titled Beyond Adversary Democracy ; she uses the term consensus to “describe a form of decision making in which, after discussion, one or more members of the assembly sum up prevailing sentiment, and if no objections are voiced, this becomes agreed-on policy.” 40 Jane J. Mansbridge, Beyond Adversary Democracy (New York: Basic Books, 1980), 32, (visited on October 25, 2005). This offering of an understanding and asking for objection is also part of the answer to the question of how a group knows consensus has been reached: when there are no objections. The IETF recommends that when the chair of the working group believes she discerns a consensus she should articulate her understanding of the consensus position and ask for comments; then, “It is up to the Chair to determine if rough consensus has been reached.” 41 Bradner, “RFC 2418,” section 3.3. A similar understanding exists at the W3C. However, determinations by the chairperson must be tenable to the working group and are reviewable by higher-ups, particularly after Last Call, in each institution (e.g., IETF area directors and W3C director and advisory committee).

¶27 Among Quakers, Michael Sheeran notes that facilitators are known as humble “clerks” — reminiscent of a Wikipedia discussion equating administrators with “janitors”; nonetheless, they can wield significant influence beyond their seemingly simple responsibilities, much as is often alleged in the case of Wikipedia administrators. (On this point, Mailer Diablo’s Second Law of Wikipedia is adapted from Stalin to read: “The Wikipedians who cast the votes decide nothing. The sysop/[bureau]’crat who count the votes decide everything.” 42 Wikipedia, “User:Raul654/Raul’s Laws (oldid=301373968)”. ) In his chapter on leadership, Sheeran notes, “The clerk’s responsibilities” might also serve as “devices for hidden control” with respect to setting the agenda (i.e., scheduling which issues are discussed and when), stating questions (i.e., in an even-handed manner), facilitating the discussion (i.e., encouraging participation and discouraging obstructionists), judging what is important (i.e., whether something is substantive or trivial), and judging the sense of the meeting (i.e., is there consensus?). 43 Sheeran, Beyond Majority Rule, 92-98. These devices might be used for good (e.g., structuring the agenda so that a working group gains momentum from easier issues) or ill (e.g., scheduling an unwelcome item at the end of a full agenda).

¶28 These communities benefit from the (hopefully) wise guidance of a trusted community member, be it a working group “chair” or meeting “clerk.” In much of Wikipedia decision making there is rarely any such formally identified resource at the start. (Though through requests for comments, mediation, or arbitration such a person might become involved.) The effects of this can be seen in the naming controversy with which I opened the chapter. Wikipedian Wknight94 summarized the issue in his statement this way:

¶29 A clear-cut case of supermajority consensus has become a nasty all-out war with a very vocal minority. A poll which is now visible here included a question of whether television episode articles should only be disambiguated when necessary…. The result was 26 people choosing to support disambiguating only when necessary and seven choosing to oppose. The poll was well-publicized. Nonetheless, a few members of the minority, mostly Elonka and occasionally MatthewFenton, have declared that there was no consensus and that the dispute is still open. The reason most often given is that the poll was modified several times while in progress. While that is true, it was mostly modified from a one-question poll with three choices to a two-question poll, each with two choices, and the meaning of the most contentious issue remained unchanged (not to mention Elonka herself modified the poll…) 44 Wikipedia, “Wikipedia:Requests for Arbitration/Naming Conventions (oldid=101952465)”.

¶30 Wikipedian Josiah Rowe spoke of a possible confusion about the meaning of consensus and the seeming triviality of the issue:

¶31 Of course, consensus does not mean unanimity, but as long as we were short of unanimity, Elonka (and one or two others) insisted that the poll needed to be re-run…. The core issue of this debate, how to name articles about television episodes, is really quite unimportant in the greater scheme of Wikipedia. I really don’t understand why the debate got to this point, and it saddens me that it has. Any resolution would be welcome. 45 Ibidem.

¶32 Beyond affirming the principle that one should abstain from personal attacks — particularly of a sexual nature — the ArbCom responded to this dilemma by focusing on the failure to “close” the discussion. 46 My selective portrayal of this case should not be construed as representative of the case or the participants. This focus on closure is puzzling in that most users believed the case was about naming policy and “about consensus — whether it was reached.” Wikipedian Yaksha continued: “The result i’m hoping for is just a declaration that we got consensus, and that people should respect consensus.” 47 Wikipedia, “Wikipedia Talk:Requests for Arbitration/Naming Conventions”, Wikipedia, February 12, 2007, (visited on July 16, 2008). Yet, the ArbCom characterized the issue as a procedural one, as noted in the affirmed principles of the decision:

¶33 1.3) After extended discussion, to be effective, the consensus decision making process must close…. In other, less structured, situations, as in the case of how to structure the titles of television episodes, there is no formal closer. Nevertheless, considering the alternatives proposed, the extended discussion engaged in, expressions of preference, there is a result which should be respected. Absent formal closing, it is the responsibility of users to evaluate the process and draw appropriate conclusions. 48 Wikipedia, “Wikipedia:Requests for Arbitration/Naming Conventions (oldid=101952465)”.

¶34 This is somewhat surprising and confusing. One must look closely in the decision’s “finding of fact” to see that consensus is presumed by the ArbCom (“a consensus decision was reached”) without stating how this conclusion is arrived at, and most importantly, how Wikipedians can convince a recalcitrant minority when this is the case? As Yaksha noted, he believed there was consensus and began to rename articles to disambiguate only when necessary, but “Elonka, however, claimed that there was no consensus, to move the articles, and that the moves were disruptive.” And so the “edit war” began. Wikipedians found the decision ultimately unsatisfying, and so in the following thread Wikipedian badlydrawnjeff asked who was responsible for knowing when to close, and ArbCom member Fred Bauder attempted to respond:

¶35 The final decision notes that “It is the responsibility of the administrators and other responsible parties to close extended policy discussions they are involved in.” What is a “responsible party?” What sort of expectation is it to close an “extended policy discussion?” At what point is it “extended,” and at what stage is it okay to throw in the towel? At an arbitrary moment or simply when the discussion becomes “disruptive?” Thanks. —badlydrawnjeff 22:15, 21 January 2007 (UTC)

¶36 An established and respected user who is not an administrator could close a discussion. An extended policy discussion is one in which most aspects of the question ha[ve] been discussed, alternatives considered, in short, a full discussion. Good judgement is needed to determine when consensus has been reached or when it is obvious there is no consensus. When the discussion becomes disruptive, more heat than light, it is probably past time to close the discussion and declare a result. Fred Bauder 22:38, 22 January 2007 (UTC)

¶37 So nothing really specific, per se? —badlydrawnjeff 01:21, 23 January 2007 (UTC) The subject does not lend itself to bright line rules. The question is whether the question has been fully discussed and a decision reached. Fred Bauder 01:52, 23 January 2007 (UTC)

¶38 However, it is still difficult to see how this situation could have been avoided. Wikipedian Ned Scott joined the conversation by noting that Bauder’s response wasn’t really useful:

¶39 It’s basically saying “If you think you’re right then say so and tell everyone to shut up.” Won’t everyone think they’re right in a discussion/dispute/etc? If the situation is reasonably clear one way or the other then we usually don’t have to resort to something like this to end it. The situations this is supposed to be helpful in are usually too unclear to actually use this. — Ned Scott 05:24, 24 January 2007 (UTC)

¶40 Wikipedia:Requests_for_arbitration/Naming_Conventions involved a matter where there was a consensus, but no closing. Based on lack of closing, an opposition party engaged in move warring. That was the problem we were trying to address. Fred Bauder 03:10, 25 January 2007 (UTC)

¶41 I guess that’s one way to look at it, but the solution offered still isn’t helpful. Nothing personal. — Ned Scott 04:27, 25 January 2007 (UTC)

¶42 Fortunately, Wikipedian Ace Class Shadow offered some useful advice while recognizing there is no perfect solution: one could use templates to mark that a discussion page is now archived, or one could “ask an impartial closer to do the deed, stating that you’ll respect their common sense judgment. To this day, I’ve only encounter[ed] one closing that, using this method, seemed at all inaccurate.”

¶43 For this dispute, it is not clear if this guidance could have convinced the minority supporting consistent-suffix use of the legitimacy of the “use as needed” policy. Therefore, beyond the censure on personal attacks and “given the existence of some uncertainty regarding how to determine if there is consensus in a particular case,” no punishment was proposed on any of the parties to the case “for past violations of policy.” 49 Wikipedia, “Wikipedia Talk:Requests for Arbitration/Naming Conventions (oldid=107562939)”. Such is the ambiguity and challenge of consensus practice, and a possible source of temptation to use some system of voting.

§7 Polling and Voting

¶44 Consensus is the preferred method of making decisions at Wikipedia. This is as much because of this method’s merits (e.g., discovering mutually beneficial solutions) as its alternatives’ demerits. While consensus can be difficult, Wikipedians frequently cite the aphorism that “Voting Is Evil.” 50 Wikipedia, “Wikipedia:Voting Is Evil,” Wikipedia, August 6, 2007, (visited on November 1, 2007). Voting is ...

¶45 Yet, as seen earlier, polling is an available technique within the consensus process. When a poll is taken on Wikipedia, individuals are invited to list their position under one of the specified options (e.g., A or B; accept, reject, abstain) with an explanation, which then might prompt further commentary and discussion. How is polling different from voting? While people may confuse polling with voting — or even speak of voting as “a quick shorthand for what we are actually doing” 51 Wikipedia, “Wikipedia:Straw Polls,” Wikipedia, May 23, 2008, (visited on October 3, 2008). — polling should prompt and shape discussion, rather than terminate it:

¶46 Wikipedia operates on discussion-driven consensus, and can therefore be regarded as “not a democracy” since a vote might run counter to these ends. Some therefore advocate avoiding votes wherever possible. In general, only long-running disputes should be the subject of a poll. Even then, participants in the dispute should understand that the poll does not create a consensus. At best, it might reflect how close those involved are to one. 52 Wikimedia, “Don’t Vote on Everything,” Wikimedia, 2007, (visited on November 1, 2007).

¶47 In fact, even polling is considered suspect — and “evil” — by some as it is thought to discourage consensus, encourage groupthink, be unfair, be misleading, and encourage confusion; 53 Wikimedia, “Polls Are Evil,” Wikimedia, 2007, (visited on November 1, 2007); Wikipedia, “Wikipedia:Polling Is Not a Substitute for Discussion,” Wikipedia, September 10, 2008, ... the botched poll in the disambiguation case is evidence enough of possible pitfalls.

¶48 To be fair, consensus doesn’t work well in all circumstances. It is best suited to small groups of people with some common interests and acting in and assuming good faith. It requires a community, not just an electorate. Jane Mansbridge, in her study of decision making, finds that groups with the largest number of interdependent friendships were those most likely to achieve a consensus that “did not paper over an underlying divided vote.” 54 Mansbridge, Beyond Adversary Democracy, 175. But as a group grows the “Community May Not Scale,” as is noted at Meatball (the wiki about wiki collaboration). Meatball’s consensus page states that as the size of the group (n) increases, so does the chance of conflict between individuals (n^2) and between subgroups (2^n):

¶49 Voting is one of the best ways to quantify opinions in a large group. Online communities that have a common goal will be continually in need of making decisions. In a small group of similar-minded individuals, a consensus decision can often be found by discussion. In such a group, VotingIsEvil. However, as the group grows larger, CommunityMayNotScale. For each member of the group, there’s a certain likelihood that he disagrees with one of the existing members. The likelihood of conflict between two individuals grows geometrically, according to MetcalfesLaw. The likelihood of conflict between two sub-groups grows exponentially, according to ReedsLaw. If individuals with contrary opinions are CommunityExiled, the group may succumb to GroupThink. If not excluded, they may delay consensus decisions indefinitely. Therefore, VotingIsGood. 55 Meatball, “VotingIsGood,” January 27, 2005, (visited on July 11, 2008). Meatball still uses the historic “CamelCase” wiki method of linking to other pages by concatenating the capitalized words of th ...

¶50 Clay Shirky writes that social software must be designed so as to protect a group from becoming “its own worst enemy” by finding a way “to spare the group from scale.” 56 Clay Shirky, “A Group Is Its Own Worst Enemy,” Keynote at O’Reilly Emerging Technology Conference,, April 24, 2003, (visited on June 27, 2007), 14. In fact, theorists and members of online community alike cite “Dunbar’s number” or “the Rule of 150” to indicate the challenges of community growth. (This idea of a maximum limit on the number of stable interpersonal relationships that we can maintain was popularized by Malcolm Gladwell in The Tipping Point ; he gives many examples including the Hutterites, a communal branch of Anabaptists, who split a colony in two once it reaches 150 members. 57 Malcolm Gladwell, The Tipping Point: How Little Things Can Make a Big Difference (Boston, MA: Little Brown, 2000), 181. )

¶51 Additionally, consensus is not a panacea for the difficulties inherent to group decision making. The Wikipedia “Consensus” article notes that consensus can take a long time, be frustrating in circumstances where there is little hope of agreement, and, when understood as unanimity, can give a self-interested minority veto power over group decisions. 58 Wikipedia, “Wikipedia:Consensus (oldid=223708449)”. Furthermore, the “Consensus Decision-Making” article notes that consensus is inherently conservative (i.e., preserves the status quo), susceptible to disruption, and can give rise to groupthink by which members suppress their own opinions for the sake of conformity or group harmony. This can even yield a paradox in which the group’s final position is held by few members, such as when members falsely, for harmony’s sake, support a “cascading” but minority position because it benefited from being expressed first. 59 Sunstein, Why Societies Need Dissent, 59.

¶52 And while voting may be appropriate in some circumstances, or at least a last resort if consensus fails, the openness of Wikipedia, again, contributes to the sentiment that voting “is evil.” Meatball notes “online voting suffers badly” because people can “stuff the ballot box” or bias the framing of the poll. 60 Meatball, “VotingIsEvil,” November 16, 2007, (visited on July 11, 2008). The “sock puppet,” a cousin of the “troll,” is an account used to “create the illusion of greater support for an issue, to mislead others, to artificially stir up controversy, to aid in disruption, or to circumvent a block.” 61 Wikipedia, “Wikipedia:Sock Puppetry,” Wikipedia, March 5, 2009, (visited on March 6, 2009). (Think of a literal sock puppet on your hand, agreeing with everything you say.) These types of problems are particularly prominent in the voting associated with “Articles for Deletion” (AfD, where Deletionists and Inclusionists duel) and “Requests for Adminship” (RfA, where bitter rivalries flourish). 62 Ben McIlwain, “We Need a Policy against Vote-Stacking,” wikien-l, May 4, 2006, (visited on May 4, 2006); Wikipedia, “Wikipedia:Articles for Deletion (oldid=128043238)”; Wikipedia, “Wikipedia:Requests f ...

¶53 While the distinction between polling and voting might blur as an argument grinds on and the possibility of consensus declines, the spirit and mechanics of the two are different. Consider that in consensus, if a group has not succumbed to groupthink, one should feel free to speak of and identify with a minority position. People will need to know whom to engage with and potentially learn from. Yet in democratic voting, the secret ballot has significant advantages. In a thread about a resolution of the Wikimedia Foundation Board, Danny Wool, a foundation employee, wrote that he found questions about who voted for or against a resolution troubling: “In a true democratic system, the secret ballot allows people to vote their conscience, rather than voting for popularity, material reward, fear of censure, and whatnot. A commitment to openness should not be misused so cynically.” 63 Danny Wool, “[Foundation-l] Breaking Promises (Was Re: Where We Are Headed),” Foundation-l, June 5, 2006, (visited on August 8, 2006). Additionally, consensus presumes good faith and sometimes sustains it; voting can operate without good faith and sometimes depletes it altogether. Mansbridge argues the differences between voting and consensus can be understood thus: “Voting symbolizes, reinforces, and institutionalizes division…. while a decision by consensus includes everyone, reinforcing the unity of the group.” 64 Mansbridge, Beyond Adversary Democracy, 10. This can also be seen in different Internet standards organizations. Granted, “politics” are present in every venue, but in consensus organizations participants might be more likely to show up with an open mind and willingness to engage others. In voting-based institutions, “stuffing the ballot box” can get out of hand. This was demonstrated most effectively in a case that merits a brief digression.

¶54 In response to the growing popularity of the OpenDocument Format (ODF) standard, an alternative to the proprietary format used by MS Word, Microsoft offered a new version of its formats through a “fast-track” international standard process. 65 Ryan Paul, “XML Spec Editor: OOXML ISO Process Is ‘Unadulterated BS’,” March 2, 2008, (visited on August 6, 2008). For supporters of open standards this could be a significant gain except that the format is said to be complex, vendor biased, that it does not make full use of other standards, and is difficult to fully implement. 66 Edward Macnaghten, “ODF/OOXML Technical White Paper,” May 2, 2007, (visited on August 6, 2008). When it came time to vote Microsoft is alleged to have pressed allies to acquire voting (“P”) memberships in an International Organization for Standardization (ISO) working group for the purpose of approving the format (i.e., identified as ECMA 376 / ISO 29500). It appears the standardization effort has been successful, although some claim the voting process was abused. 67 Scott M. Fulton III, “Evidence of Microsoft Influencing OOXML Votes in Nordic States”, BetaNews, August 28, 2007, (visited on August 6, 2008). ... And there has been an interesting — and unfortunate — side effect. In one of his final reports as chair of the responsible working group, Martin Bryan noted, “The second half of 2007 has been an extremely trying time for WG1. I am more than a little glad my 3 year term is up, and must commiserate with my successor on taking over an almost impossible task.” In addition to the short time frame, and the interdependencies and complexity of the task, the arrival of otherwise uninterested members made subsequent work impossible to complete because those who joined to vote on this single issue subsequently disappeared, preventing a quorum:

¶55 As ISO require[s] at least 50% of P members to vote before they start to count the votes we have had to reballot standards that should have been passed and completed their publication stages at Kyoto…. The days of open standards development are fast disappearing. Instead we are getting “standardization by corporation,” Something I have been fighting against for the 20 years I have served on ISO committees. I am glad to be retiring before the situation becomes impossible. I wish my colleagues every success for their future efforts, which I sincerely hope will not prove to be as wasted as I fear they could be. 68 Martin Bryan, “Report on WG1 Activity for December 2007 Meeting of ISO/IEC JTC1/SC34/WG1 in Kyoto,” November 29, 2007, (visited on August 5, 2008).

¶56 At Wikipedia, additional confusion arises about the gray area between consensus and voting. In voting systems, it is not uncommon for like-minded individuals to advocate, campaign, and horse trade (i.e., exchanging votes on different issues). 69 I discuss similar arguments over trading votes and advocacy that took place in the KDE development community when their bug tracking system adopted a feature allowing ... In a thread on the English Wikipedia list, Wikipedian Johntex advocated that “We Need to Recognize That Advocating Is a Basic Right”; he argued users should be able to “influence policy in ways that they believe are beneficial to the project,” including “building up groups of people who agree with you and who will help you bring about the beneficial change.” Advocating for a cause or recruiting those who already agree with you is “a Good Thing”:

¶57 Let’s stop insulting people by calling them “meat puppets” or “vote stackers.” Let’s stop confusing the issue by calling it “spamming.” It is not spamming. Spamming is indiscriminately notifying people that are probably not interested in the hopes that a few people will be. This is practically the opposite. Attempting to stifle advocacy is harmful to the consensus building process and it is harmful to the project. If we try to prohibit it, it will just be taken off-wiki, which would be a huge shame. 70 John Tex, “We Need to Recognize That Advocating Is a Basic Right,” wikien-l, May 4, 2006, (visited on May 4, 2006).

¶58 But others disagreed, finding these practices from the democratic sphere to be counter to consensus practice. Wikipedian Tony Sidaway responded, “We don’t do advocacy or campaigning on Wikipedia. Our decision-making processes are deliberative rather than democratic.” 71 Tony Sidaway, “Re: We Need to Recognize That Advocating Is a Basic Right,” wikien-l, May 5, 2006, (visited on May 5, 2006). Yet, the difficulties of achieving consensus and this blurring of approaches to decision making prompted Zoney to declare that any pretense of consensus should be abandoned and the community should move toward a “fixed method of decision-making”:

¶59 My problem is that perpetuating the lie that decision-making on Wikipedia is by consensus, we don’t strictly adhere to any other decision-making form (e.g. majority voting). In consequence, decisions are “whatever people can get away with.” Of course if there’s an actual real consensus (general agreement) then there’s a valid reason for a decision not being challenged. But more often than not all it means is that influential individuals ensure they get their way and others give up (that isn’t forming consensus by the way), or else we have majority/mob rule. 72 Zoney, “Re: ‘Consensus’ and Decision Making on Wikipedia,” wikien-l, June 29, 2007, (visited on June 29, 2007).

¶60 In the resulting thread some conceded there are problems, but not as bad as feared. The discussion touched on many of the challenges discussed in this chapter. And some took a pragmatic “time will tell” approach: hopefully the right thing happens more often than not. Sidaway felt that ultimately, “Decisions are more likely closer to ‘whatever offends the least number of people.’ This is sometimes less than optimal—I could give my list of things I think are poor decisions and you could probably give yours. The result is that nobody is ecstatic but we have something we can move ahead with.” 73 Tony Sidaway, “Re: ‘Consensus’ and Decision Making on Wikipedia,” wikien-l, June 29, 2007, (visited on June 29, 2007). Likening emerging consensus to nailing jelly to the wall and keeping the stuff that sticks, he continued his pragmatic perspective: “If hundreds of people edit a piece of work in good faith over a long period, what changes least over time may be presumed to be there by consensus. However even the most apparently stable elements of a work may be deposed quite easily. The result may be a new consensus or, in other cases, a period of instability where the new version and the old version compete.” 74 Tony Sidaway, “Re: ‘Consensus’and Decision Making on Wikipedia,” wikien-l, June 27, 2007, (visited on June 27, 2007). In response to the discussion Zoney asked in frustration if there is then “any hope of having a fixed method of decision-making on Wikipedia, rather than a shambolic pretence of achieving consensus …?” Wikipedian Adrian responded simply, “No.” 75 Adrian, “Re: ‘Consensus’ and Decision Making on Wikipedia,” wikien-l, June 26, 2007, (visited on June 26, 2007). Marc Riddell responded a little more encouragingly, “Yes, there is hope; if we can put our individual egos and emotions aside — and start using our heads in a responsible way.” 76 Marc Riddell, “Re: ‘Consensus’ and Decision Making on Wikipedia,” wikien-l, June 26, 2007, (visited on June 26, 2007).

§8 Conclusion

¶61 Humans naturally look for means by which difficulties can be clearly dissected and neatly dispatched. Yet, given its reliance upon an assumption of good faith and a preference for consensus in its decision making, one can conclude that the Wikipedia community is relatively tolerant of the ambiguities inherent to collaborating on a world encyclopedia and rather trusting of human judgment over the long run. While Wikipedia must often address many of the conflicts present in the real world, and has plenty unique to its own mission and methods, I would never argue that Wikipedia has become the global institution of “adjustment and adjudication” that Wells foresaw. However, Wikipedia is a fascinating example of a historic method of community decision making in a new context. In particular, its openness — the lack of topical and temporal scope, the initial lack of facilitation, the turnover in membership, and anonymity — brings a new salience to the challenges of consensus practice.

¶62 These challenges are seen in the humorous — perhaps cynical — definition of consensus at the head of this chapter (i.e., the scope of consensus includes only those who share your opinion). Similarly, in response to the question “How many Wikipedians does it take to change a light bulb?” Wikipedian Durova answered “69”:

¶63 1 to propose the change; 5 to support; 1 to dispute whether the change is a needed process; 7 more to pile on from IRC [Internet Relay Chat] and join the dispute; 2 to open a request for comment; 37 to vote at the straw poll; 5 to say votes are evil; 1 to MFD it [propose it as “Miscellany For Deletion”]; 9 to object until the MFD gets speedily closed; 1 to mark the proposal historical. Afterward on AN [Administrators’ Noticeboard], all opposers claim the consensus favored darkness [i.e., no light bulb!]. 77 Bibliomaniac15, “How Many Wikipedians Does It Take to Screw in a Lightbulb?,” Wikipedia, July 11, 2009, ...

¶64 In another “WikiSpeak” entry consensus is defined as “one of the three states that can be reached at the end of a discussion after all parties have become thoroughly fed up with it; the alternatives are no consensus or for pity’s sake, I wish I’d never gotten involved in this.” 78 Wikipedia, “User:Malleus Fatuorum/WikiSpeak (oldid=224745874)”. It can be an altogether frustrating experience. And in some circumstances, such as irreconcilable differences between community members or external threat, this tolerance can be incapacitating. And so, Wikipedia, like other open content communities, is also characterized by an odd type of leadership at the highest level: the “benevolent dictator,” the subject of the next chapter.