Wikipedia in the anti-SOPA protests as a case study of direct, deliberative democracy in cyberspace

ABSTRACT This paper contributes to the discussion on deliberative, direct democracy and volunteer mobilization in the Internet era by analyzing the vote participation levels of Wikipedia volunteer editors (Wikipedians). On 18th January 2012 in the ‘first Internet strike’ against the American ‘Stop Online Piracy Act' legislation, over two thousand Wikipedians took part in the vote concerning whether their site should undertake a protest action, with vast majority expressing support for this action. However, the vote participants formed only a tiny fraction of the total number of Wikipedians who number in millions. Although Wikipedia can be seen as an open, democratic forum practicing deliberative, direct democracy, the process of voting on Wikipedia is significantly influenced by participation inequality, with a majority of the vote participation coming from a small group of most active contributors – an effective oligarchy. This paper discusses the intricate dynamics between Wikipedia egalitarian ethos and the creed to discuss project matters deliberately on one hand and the conspicuous lack of promotion and advertisement stemming from a rule against ‘canvassing’ and an overall skepticism regarding the status of majority votes. While voters' passivity and lack of interest play a major role, as expected, another factor emerges as a significant factor responsible for the low levels of participation: an inefficient information distribution system, as the vast majority of Wikipedians were not aware of the ongoing discussions and the vote itself until after their conclusion.


Introduction
This paper analyzes an incident in the decision-making process of the online encyclopedia Wikipedia, contributing to our understanding of the (e-)democratic and deliberative decision-making in transnational organizations. In studying the participation levels in the voting leading to the blackout of the website in early 2012 through the prism of theories of participation inequality and Michel's Iron Law of Oligarchy, the viability of the Wikipedia project as a possible platform of deliberative, direct democracy is critically considered. Given its size and resulting complexity, Wikipedia allows us to see how e-democratic decision-making and deliberation occur at a meso-level: in an online organization whose volunteer members outnumber populations of entire countries.
The specific caste studied is the largest ever 1 vote on Wikipedia, in the broad context of the 'first Internet strike' of 18 January 2012 against the SOPA ('Stop Online Piracy Act') legislation proposed in the USA. This protest action involving a large number of Internet sites and communities was organized by the anti-SOPA digital rights/free culture movement. The SOPA legislation was first discussed in the US Congress in October 2011, presented by its corporate supporters as a tool aiming to stop copyright infringement by foreign websites. In the opinion of its detractors, it was seen as threatening to disrupt the free expression on the Internet and framed as a threat to its (and Wikipedia's) very existence. Brought to the attention of Wikipedia community in December, in a little more than a month over two thousand individuals, primarily Wikipedia volunteers, took part in a series of votes on whether Wikipedia should take part in a protest action, eventually arriving at a clear consensus to join the said action (Konieczny, 2014).
Said volunteers (Wikipedians), who number in millions, are all allowed to participate in the site's governance, often through polls and votes where each participating volunteer is treated in an egalitarian way (Jemielniak, 2013, p. 47;Reagle, 2010, p. 117). The site's model of deliberative, democratic discourse was not imposed on Wikipedia from an outside agency, but established by the volunteers themselves in a still-ongoing process. Over the years a number of scholars (Black, Welser, DeGroot, & Cosley, 2008;Firer-Blaess, 2011;Hilbert, 2009;Jemielniak, 2013;Klemp, 2010;Konieczny, 2010;Kostakis, 2011;Noveck, 2009) have commented on the fact that Wikipedians are in effect making or reinventing the law; presented Wikipedia as an open, democratic forum practicing deliberative, direct democracy; and suggested that Wikipedia's unique model of decisionmaking can be used to improve existing models of participative, democratic discourse.
As I show below, while in theory Wikipedia can be indeed seen as an inclusive, open forum practicing deliberative, direct democracy, in practice it suffers from the common problem affecting majority of the e-democracy projects: a very low turnout among the eligible voters or discussants. While as expected lack of interest in the vote matter itself was one of the major reasons for abstaining from the vote, it was responsible for only about a third of abstentions among the most active editors. The activity of Wikipedia volunteers is shown to exemplify participation inequality, with majority of voter participation coming from a small group of most active (vested) contributors. While those editors are not consciously limiting the participation of others, the Wikipedia community has evolved rules against invasive advertising of its internal discussions (known on Wikipedia as 'canvassing'), even when such discussions are about issues of core importance to the project's governance (such as the SOPA legislation, which was framed by numerous supporters of the vote as endangering the very existence of Wikipedia). Wikipedians expect that editors interested in policy-making will regularly check centralized noticeboards listing current debates, and are averse to advertising them through mass mailings, site banner notices, and such. In practice, this means that only a tiny minority of Wikipediansprimarily, the most active oneshave a chance to become aware and participate in those discussions, while the vast majority is unintentionally but effectively excluded. Even among the most active editors, a quarter was not aware of the vote, a number which was certainly much higher among the less active volunteers; and among all respondents over two-thirds felt that discussions about issues such as SOPA are not adequately advertised to the community. In other words, a major reason for the low turnout in this particular vote, in addition to the expected lack of interest from the votersand likely reinforcing itwas an inefficient information distribution system, which I show to be a product of participation inequality and oligarchy formation. The said oligarchy formation can be understood as a process of demands about the vote organization, influenced by the Wikipedia encyclopedic ethos (in particular, desire to reduce canvassing and vote stacking), but de facto resulting in the empowerment of the oligarchical system of most active editors which led to a significantly restricted access to information, and a general lack of awareness of the vote, for majority of potential participants.

Decision-making process on Wikipedia
To most of the world, Wikipedia is just the world's biggest encyclopedia. Underneath that simple description lies the fact that Wikipedia is an organization run by millions of volunteers, with characteristics of a social movement, and with an innovative, self-created, and still evolving governance system (Konieczny, 2009(Konieczny, , 2010Jemielniak, 2013, p. 151;Reagle, 2010, p. 90). Jemielniak (2013, p. 259) in his book dedicated to the organization and governance of Wikipedia described it as a heterarchy, which he defines as a 'meritocratic adhocracy with a dispersed power structure'. Wikipedia has no governing body, official or otherwise, that tells editors what to do, or that is responsible for drafting policies. There are no 'Wikipedia employees', and the hundred or so of the Wikimedia Foundation employees, responsible for managing the legal aspects of the project, have only a limited impact on the decisionmaking process (Jemielniak, 2013, p. 149;Konieczny, 2010). The site's entire governance structure, managing millions of volunteers working on a similar number of content pages, has been created by its volunteers.
Officially, Wikipedia declares that its preferred decision-making process is a discourse that leads to a consensus, a practice often discussed in the context of a similar decisionmaking system found among the Quakers (Jemielniak, 2013, p. 62). Wikipedians stress that voting, when it takes place within the project, should be seen as polling, that is, a tool that aids the deliberative discourse, rather than supersedes it. Jemielniak (2013, p. 84) notes, however, that many rules on Wikipedia are idealized versions more suited to what the Wikipedia community wishes themselves to be rather than what it really is. In fact the direct democracy, practiced through Wikipedians participating in various votes, is a common occurrence in the project. Similarly Black et al. (2008) noted that much of what Wikipedians advocate with regard to their best practices in policy-making is in line with theories of democratic deliberation. Despite the Wikipedia rule stating that 'polling is not a substitute for a discussion', it often becomes exactly that, particularly in situations where the number of participating editors becomes large enough that it is not practical to expect them to communicate effectively through consensusseeking deliberative discourse. Voting can also help to arrive at decisions more quickly than the deliberative consensus decision-making style. It is also more friendly toward changing the status quo, and reduces the chance of an early minority galvanizing its hold on the discourse, which is not an uncommon problem on Wikipedia (Jemielniak, 2013, p. 18;Reagle, 2010, p. 111). Overall, voting is a common occurrence within the Wikipedia project.

E-democracy
Using terms ranging from 'direct democracy' to 'teledemocracy', various scholars since the times of Rousseau, Mill, and de Toqueville have advocated a plethora of participatory models of decision-making. In the traditional context, direct democracy has a venerable tradition, most famously exemplified in the form of the Athenian democracy. In recent decades, the advent of new information and communication technologies has revitalized it, with the promise of new media overcoming the limitations of the old, namely, allowing the decision-making process to be more effectively stretched to encompass millions of citizens. E-democracy, as this phenomenon has often been labeled, has led to numerous experiments, including on the national level, such as the decade of electronic voting in Estonia (Madise & Vinkel, 2014). Numerous countries are experimenting with variants of this approach; a more detailed discussion of various examples can be found in Lee, Kaiju, and Stokes (2011). E-democracy, however, is hardly limited to governmentsit has been incorporated into a number of citizen-and NGO-driven initiatives; among them, some have been based on the wiki technology, leading to the use of the term 'wiki government' (see Noveck, 2009;Tapscott & Williams, 2010, for numerous case studies).
Supporters of e-democracy have argued that citizens would be significantly more active given the opportunity to directly participate in the voting. Direct democracy has been often said to motivate participation by energizing citizens with political efficacy and a sense of civic duty, and the positive effects of direct democracy on voter participation and turnout rates are common arguments expressed by supporters of that model of decision-making (Madise & Vinkel, 2014;Morrell, 1999;Pateman, 1970, p. 104).
Of course, direct democracy is not without its critics, who point out that it can lead to problems such as an information overload, voter dissatisfaction, confusion, fatigue, declining turnout, and government paralysis (Kampen & Snijkers, 2003;Wilensky, 2002, p. 415). Related to this is a bias toward voters who are educated and have plenty of spare time, and the effects of digital divide. Objections to direct democracy also include its vulnerability toward the polarization of opinions, populism, demagoguery, and marginalization of minorities (Abromeit, 1998, p. 139). For an in-depth discussion of the pros and cons of direct democracy, see Butler & Ranney, 1994 (chapter 2) and Budge, 1996 (chapters 3 and 6).
One of the measures of deliberative democracy is voting participation, or turnout. Though not the only measure, it has the advantage of being easily operationalized, facilitating comparative studies. Turnout in direct democracy initiatives can differ significantly. Schiller (2011, pp. 21-22) notes that while turnout in European ballot voting on initiatives and referendums varies, about 50% of registered voters is an average. Still, some votes have attracted less than 5% of those eligible to take part in them (Reilly, 2010, p. 48;Schiller, 2011, p. 281). According to Schiller (2011, p. 151, 197), the major factors responsible for the turnout are the size of the community, the type of the voting procedure, and the issue at stake.
Early research on direct democracy in the context of the Internet focused on the digitization of existing democratic practices, the practical implications of which have so far been rather sparse and disappointing, with resulting calls for a more effective digital implementation of tools related to direct participation and citizen representation (Hilbert, 2009). In recent years, wikis in general and Wikipedia in particular have become a focus of several works analyzing its implications for direct and deliberative democracy (in particular, see Black et al., 2008;Firer-Blaess, 2011;Hilbert, 2009;Jemielniak, 2013;Klemp, 2010).
One of the earliest works that noticed Wikipedia's democratic and deliberative mechanism was that of Black et al. (2008), who noted that the decision-making process on Wikipedia, exhibiting many democratic elements, demonstrates a higher level of problem analysis and information dissemination than those found in most traditional settings. Hilbert (2009) argued that the Wikipedia project is one of new Web 2.0 sites that has the potential to break the long-standing democratic tradeoff between group size and depth of argument. Klemp (2010) noted that the Wikipedia model of online interactions may represent a useful supplement to traditional deliberative practices by promoting largescale inclusion and improved accuracy. He concludes that Wikipedia proves that online deliberation does not have to succumb to the threat of group polarization, and proposes that the Wikipedia model should be used as a basis of real-world state referenda and evaluation of elected representatives' and bureaucratic performance.
Firer-Blaess (2011) builds on Hansen, Berente, and Lyytinen's (2007) work, analyzing the Wikipedia process through the Habermasian model of rational discourse. Both conclude that Wikipedia meets that model. Firer-Blaess (2011) echoes Klemp (2010) in suggesting that the Wikipedia model can be used to improve a number of existing political discourse processes, such as legislation design and electoral process.
Wikipedia's model of governance has also been shown to suffer from a number of problems. Kostakis (2010), Flöck, Vrandečić, andSimperl (2011), andJemielniak (2013, p. 39) have discussed the danger of active and organized minorities prevailing over less active majorities, group polarization, the tyranny of structurelessness with hidden hierarchies, information overload, and coordination and communication problems. Another concern about Wikipedia governance that has attracted relatively little scholarly attention so far is the fact that despite the professed openness of the project, only a small number of Wikipedians take part in it. Jemielniak (2013, p. 138) has been the first to discuss examples of how and why the norms of Wikipedia community's culture have led to an inefficient information dissemination. He presents case studies of the tensions between the community of volunteers and the Wikimedia Foundation, with the community refusing to grant the Foundation the right to advertise project governance discussions in an efficient manner (such as through site-wide banners or mass emails), and at the same time, being unable to reach in a timely manner a consensus of what, when, and how to advertise. The result of this is that most core policy changes happen after debates that attracted only a few discussants. This was evident in the case of the SOPA discussion and vote analyzed here, where the Foundation request to advertise the vote more widely through a site banner was approved by the community only after weeks of delay, already after the vote was more than half-way over, just days before it was to end.

Participation inequality on Wikipedia and the Iron Law of Oligarchy
Studies of Wikipedia have shown it to be a good representation of the participation inequality. The power law, also known as the '1% rule' or the 'long tail', suggests that in most online communities a small proportion of activists creates the majority of content. This inequality has been commonly ascribed to many trends on the Internet, and confirmed to affect many aspects of participation online, including those of Wikipedia contributors (Morzy, 2012;Ortega, 2009). According to Wikipedia's own statistics as of May 2012 (available at http://stats.wikimedia.org/EN/TablesWikipediaEN.htm), the project had 16,785 million registered accounts, but only about a third (5.3 m) made one or more edits. Only about 115,000 (the average for December 2011 and January 2012) made at least one edit monthly, with about 40,000 making 5 or more edits; about 3500 making 100 or more edits; and 250, 1000 edits or more. It is those most active editors who have created most of Wikipedia content: 19% of editors (about 772,500) with 10 or more edits contribute 95.5% of all content, and 1.3% (about 52,300) with 300 or more edits, over 80% of content.
This participation inequality represents the emergence of an oligarchy. In his landmark study in 1915, Robert Michels has pointed out that all organizations are faced with problems of coordination that can be solved only by developing a bureaucratic oligarchy, incompatible with democracy (Leach, 2005;Michels, 1915, p. 28). This argument was founded on a crucial observation that everyday operations of an organization cannot be run by mass membership and the effective functioning of it requires the concentration of power in the hands of the few (Michels, 1915, p. 114). Although modern technology in particular, the Internethas allowed us to overcome a number of issues, others still remain as valid today as a century ago, namely, the technical indispensability of leadership, the tendency of the leaders to organize themselves, the gratitude of the led toward the leaders, and the general passivity of the masses (Michels, 1915, p. 240).
Leaders offer the advantage of swift decision-making, and their position is further strengthened by their skill in controlling information and managing relationships. Even in a group dedicated to democratic goals, leaders soon find ways to evade community control and become the 'most equal among equals', seeking to preserve and enhance their position within the organization. Ordinary members of the community, in turn, are often unwilling to take the risk and effort to enforce a change in leadership, and suffer from what Michels referred to as the psychological need to be led. Michels stresses that 'for democracy, the first appearance of professional leadership marks the beginning of the end' (Michels, 1915, p. 28).
Yet another key element of the oligarchical system of control is the fact that party elites have a near-monopoly on efficient communication. Michels notes how oligarchies try to control the mass media (specifically for his time, press), and how they try to create rules and restrictions on what can be said and by whom (Michels, 1915, pp. 149-151, 177). In many instances this means that the leaders can ensure that certain messages are well advertised, while others are left to languish in obscurity. However, in Wikipedia, something different seems to be happening.
Decisions about information flow are made by tiny groups of highly active editors who are likely to be intimately familiar with relevant rules and customs of the specific, low-visibility discussion forums, often attracting less than ten participants, where key decisions about whether a topic will or will not be advertised to the wider audience are made. Those dynamic, tiny oligarchies (a particular example of which is discussed later in the context of a 'MediaWiki_talk:Watchlist-details' discussion), whose ever-changing and rotating membership could even be described as adhocratic (Konieczny, 2010), are not enforcing limited information access in order to pursue their individual agendas or protect their positions of power (both aspects, although not inexistent on Wikipedia, are less relevant in its relatively flat hierarchy composed of volunteers than in traditional organizations where most leaders are part of a professionalized, salaried elite).
Instead, in the case of Wikipedia, the emerging oligarchy can be operationalized through participation inequality rather than a clear emergence of a conscious leadership group. A small faction of very active volunteers, who are aware of and who take part in the decision-making process of the site, dominate the said process, but they are operating in a bureaucratic environment that has evolved under a number of factors (in particular, ones relating to the organizational culture, such as encyclopedic ethos and resistance to canvassing, discussed in detail in subsequent sections). As shown in the 'MediaWiki_talk:Watchlist-details' case, this means that the site oligarchical organization has not reached the point (and perhaps never will) where it can consistently advertise or suppress a message. It may well do one or the other, but such a decision will be based on the input of a very tiny fraction of the Wikipedia's community, one, however, that is almost always composed of some of its most active editors.

Chronology of Wikipedia's decision-making regarding anti-SOPA protest action
Implications of SOPA for the Wikipedia project have been discussed by some Wikipedians as early as November. On 10 December, Jimbo Wales, who commands a significant following as the project's founder, posted on his English Wikipedia talk page a straw poll regarding what, if anything, should Wikipedia do. The ensuing five-day discussion attracted 508 participants, of whom a significant majority was in favor of taking some protest-like action, and resulted in the creation of a dedicated Wikipedia:SOPA initiative page (http://en.wikipedia.org/wiki/Wikipedia:SOPA_initiative). Wales continued to be highly supportive of the idea that Wikipedia should join other websites in a protest action. Staffers of the Wikimedia Foundation, a critic of SOPA, started another straw poll on 13 January, this time regarding what, if any, action should be taken by the Foundation to support the Wikipedia community if it desires to join the protests. That vote lasted till 16 January, attracting 1674 editors of whom approximately 90% supported joining the protest.
Subsequently the English Wikipedia (and a number of other Internet sites) became blacked out on 18 January. Within hours of the protests, the support for the SOPA legislation in the US Congress dropped.

Hypothesis and methodology
To test whether the most experienced editors dominated the vote, and thus, whether the power law was reflected in this decision-making process, the following hypothesis was formulated.
Experience hypothesis (H1): Experienced editors will make up a disproportionate number of voters.
Data were collected from publicly available Wikipedia pages. A total of 2097 editors were identified as having participated in the voting process, forming the studied population. Next, the Edit Counter tool (http://toolserver.org/~River/cgi-bin/count_edits?) provided data on the exact edit count and registration date for each editor in this study.
However, even within the group of most active Wikipedians, less than 10% of the Top 100 most active contributors took part in the vote. To gain a deeper understanding of why this was the case, I invited the first 100 nonparticipants from the list of most active Wikipedians active at that time to take part in a survey, where they were asked why they did not vote (they were allowed to state multiple reasons), and whether they think that the SOPA vote in particular, and similar issues in general, are well advertised to the Wikipedian community.

Findings
With regard to the number of edits, the overall distribution is presented in Table 1 and Figure 1.
With regard to the number of edits, not counting 11.4% anonymous editors (who voted without creating an account), 10.5% registered editors had fewer than 10 edits and 21% had fewer than 50 edits. On the other end of the spectrum, 77.7% of editors had over 10 edits and 40% had over 300 edits. If the voters distribution was to be similar to that of the editors throughout Wikipedia, based on the number of edits, only 20% of editors should have 10 or more edits, and only 1%, over 300.
Participation is also positively correlated with activity. The number of editors who make several edits per day, so at least over a 100 edits per month, is very small: a little over 3500. Out of that group 1264 participated in the votes, giving this group a turnout of 36%. The turnout among the editors who make at least 10 edits a month was only 5%, and within the group which makes at least 1 edit a month, 1%.
Moving beyond a simple edit count, criticized as insufficient by Jemielniak (2013, p. 40) and Konieczny (2014), 64.5% of editors who participated in the vote were defined as veteranseditors who have 50 or more edits project wide, including at least one to a Wikipedia policy page (or its discussion page) and at least one to any discussion page, have been registered for over a month, and have a user page.
With regard to the length of registration, the time data confirm that two groups are significantly overrepresented: the oldest and the newest editors. Around December 2010-January 2011, editors who have been registered for over 5 years formed 8.1% of all Wikipedia editors, but they represented over a third (37.1%) of the vote participants. At the same time, the group of editors registered for barely a month, usually only a tiny percentage of Wikipedia editors (around 0.1%), formed 8% of the voters in this vote. Another confirmation that experience played a role in attracting editors comes from the analysis of the administrator's participation. They were also overrepresented: while forming only 0.03% of Wikipedia editor population, they accounted for almost 7% of the voters. Finally, 71% of vote participants had a user pagea page on which they can voluntarily present information about themselves, compared to only 30% of the Wikipedia editors with one or more edit.
A logistic regression analysis was conducted to compare the population of editors who voted only in one vote (December or January) to those who voted in both. The following variables were tested: being an admin, and being a veteran editor (the latter being a composite of several other variables). A test of the full model against a constant-only model was statistically significant, indicating that the predictors as a set reliably distinguished between acceptors and decliners of the offer (χ 2 = 57.326, p < .000 with df = 42). Nagelkerke's R 2 of .04 indicated a weak but significant relationship between prediction and grouping. The Wald criterion demonstrated that veteran status made the most significant contribution to prediction (p = .000), followed by admin (p = .035). The XP(B) values indicate that a change in status to administrator increases the chances of voting by 50%, and being a veteran editor, doubles it. This confirms that one's activity is significantly related to one's chances of participating in the decision-making and deliberative process on Wikipedia.
With regard to the survey of non-voters, thirty volunteers out of the hundred surveyed returned their questionnaires. Approximately a quarter (26%) were not aware of the vote, a third (36%) were aware of it but where not interested, a fifth (20%) felt that too many people voted by the time they became aware of the vote for their contribution to matter, and 16% chose to not participate on principle of refusing to join a discussion that dealt with real-world politics which they felt Wikipedia should not be involved in (thus, echoing sentiments also found in Konieczny, 2014). Two-thirds of the editors who chose not to vote while being aware of the vote (64%) felt that the SOPA vote was advertised well enough, though as expected the quarter who were not aware of the vote strongly disagreed with that statement. In fact, 100% of the said quarter stated it was not advertised sufficiently. Out of all responders, 69% felt that such issues are, in general, not advertised well enough across the community.

Discussion
Who voted: the 80/20 community Regardless of the metric used, it seems clear that the vote was as of interest primarily to the most experienced Wikipedians; this means that the H1 experience hypothesis is supported. The top 1% of most active Wikipedia editors (with 300 or more edits) who create over 80% of the project content also accounted for 40% of the vote; and the top 20% group (with 10 or more edits) accounted for slightly over 77% of it. Thus, the Wikipedia community can be described as an '80/20 community'one where the community policies and actions are significantly shaped by the small group of its most active members.
Nonetheless the newbies, first-time editors, and anons were also overrepresented, accounting for about a fifth of the participantshardly an insignificant proportion. During the SOPA votes on Wikipedia, there were practically no calls from the experienced editors for the exclusion of the less active ones, or even the anons; instead, they were recognized as having the right to participate in the discussion on equal terms with the veteran members. This suggests that despite the existence of a power law-type, oligarchical participation inequality in voting, biased toward the experienced editors, Wikipedia is still dedicated to transparency, openness, and its egalitarian ethos. Its governance and decision-making process are not discriminatory, and it invites participation from less experienced members of the project. Yet with such a large pool of potentially eligible voters, in an atmosphere supportive of participation, why did so few choose to participate in the vote?
The size of the electorate On the surface, the participation of 2097 editors in the vote equals to an abysmally low 0.003% turnout from the population of 5.3 m of editors who made at least 1 edit. Yet taking the 5.3 m number at face value is problematic due to the potentially large number of 'fake editors'. We can discount tiny (<2000) numbers of 'bots'accounts operated by automated scripts and deceased volunteers. There are, however, no good estimates for editors who left the project ('emigrated'), who changed their username, and serials vandals/spammers who are known to create numerous fake accounts. The number of editors retiring is even more crucial; it has been high enough to stop the growth in Wikipedia's editor base since around 2007 (Halfaker, Geiger, Morgan, & Riedl, 2013). Based on the analysis of editing status of Wikipedia's most active contributors, I estimate that the number of retired editors may be as high as 30-40%. Regardless of exact numbers, it is clear that we cannot consider the number of all-time Wikipedia editors in the same fashion as we would for citizens of a given country or statetoo many of them have 'emigrated', or are 'criminal-like' vandals not interested in any civic issues.
Rather than looking at a total number of Wikipedia editors, it is more useful to consider their activity in a given time frame. Based on Wikimedia (2012) data, a number of 130,000 editors can be proposed as an estimate of active editor population (editors who made at least one edit) on Wikipedia in the period 10 December 2011-16 January 2012. This number allows us to control for retired, deceased, and retired editors, though unfortunately not that easily for the vandals. Accounting for 2% of vandal-only accounts, and assuming that majority of those accounts represents unique individuals, this gives us a more reasonable but still relatively low turnout of 1.6%.

Electorate's awareness of the vote
Most discussions of voting participation and turnout concern real-world elections about issues that are widely advertised in numerous media, and are often a topic of personal conversation between citizens. While we know that many citizens in democratic countries still remain poorly informed about the elections, and others purposefully chose to not exercise their right to vote, existing literature operates under the unwritten assumption that said citizens are at the very least aware that a vote is happening. This key assumption cannot be, however, taken for granted on Wikipedia. This 1.6% number would be acceptable if we assumed that an average Wikipedian was at least as aware of the SOPA discussion and vote as an average citizen in a democratic nation is about his or her upcoming election. But the first discussion and vote, the one on Wales' pages, did not receive any official advertisement; it was thus visible only to the approximately 2600 Wikipedians who are monitoring this page (Wikipedia-speak has 'watchlisted' it). Given that number, the turnout for the December vote becomes a more respectable 19%.
After the dedicated page for SOPA discussion was created, it was announced through the Wikipedia's weekly newspaper, Wikipedia Signpost, first on 19 December and then again on 16 January; the latter report was, however, posted only within hours of the vote concluding. The discussion was also featured on Wikipedia Centralized Discussion page for a month. Within about two days before the end of the vote, the discussion was also advertised throughout Wikipedia using a 'site notice'a banner-like advertisement visible to editors who use the 'watchlist' function. It was also announced numerous times on Wikipedia's Facebook feed, the Wikimedia Foundation's blog, which issued a press release, and it received coverage by outside Wikipedia, from citizen journalism sites to mainstream press including title such as The Independent and the Washington Post.
Signpost provides statistics about page viewsand its December and January reports were read, in those months, by about 500 individuals. The Centralized Discussion page received about 3000 views in this period. Readership of other pages is unfortunately nearly impossible to estimate. The previous estimates on the unequal participation of editors offer, however, a useful hint: only the most active members, those who make at least several edits each day, are likely involved enough in the project to read its newspaper, look at the list of centralized discussion, or look at their watchlist. The number of editors who make several edits per day, so at least over a 100 edits per month, is however very small, numbering only a little over 3500. Out of that groupwhich can be defined as 'editors likely informed about the vote' -1074 participated in the January vote, giving a turnout of 28%. That number, in par with the December one, is more in-line with a number examples of real-world direct democracy turnouts discussed in the literature (Reilly, 2010, p. 48;Schiller, 2011, p. 281).

Reasons for low turnout
Before drawing the final conclusions, I would like to account for Michels' argument about the passivity of the masses. The topic was framed by vote supporters, including some of the project's top elitesnamely, Wales and the Wikimedia Foundationas of core interest to the Wikipedia community, concerning its very survival. Nonetheless, we can see that even among the top 100 contributors who did not vote, about a third declared they, simply, had no interest in the issue.
At the same time, an overview of Wikipedia discussion pages related to the SOPA initiative shows that a number of editors, numbering at least several dozens, left messages on various Wikipedia discussion pages and Wikimedia Foundation blog comments, stating that despite considering themselves reasonably active, they were not aware of the discussionin which they would have liked to participateuntil after the fact. This is a testament to a simple fact of life: most editors did not pay attention to the newspaper articles and centralized notices; and the most visible note, the site banner, was used only for about two days. Even among the group of most active contributors, a dedicated core of volunteers who log in on a daily basis, only two-thirds were aware of the vote. A third was nota number that would certainly grow among the less active volunteers.
This can be taken as evidence of an emergence of oligarchy, centered around different levels of Wikipedians' activity, and relevant to two of Michels' key concepts. First, the information flow on Wikipedia is restricted, with the majority of the community not being effectively informed about the ongoing debates. Interestingly, this is not an outcome of an intentional leadership-enforced information monopoly, as Michels would have expected. As noted by Konieczny (2010), Reagle (2010, p. 129), andJemielniak (2013, p. 51), there is no conspiratorial 'cabal' pulling the strings on Wikipedia, though such a view persists among some disgruntled former Wikipedians (O'Neil, 2011). Rather, the comments of editors participating in the discussion about advertising the SOPA discussion and vote in mid-December (https://en.wikipedia.org/wiki/MediaWiki_ talk:Watchlist-details/Archive_5#SOPA_RfC) as well as the more general observations made by Jemielniak (2013) suggest that this is justified by the editors who see this information control as preventing unwanted spam and information overload, distracting the rest of the volunteers from their core mission of creating encyclopedic content.
At the linked MediaWiki_talk:Watchlist-details where decisions are made on whether certain discussions are to be announced to all registered editors through a short notice in their watchlist tool (which, it should be noted, is used only by more active editors anyways), only seven editors participated in a short, less than 400 words long discussion. The said discussion had four editors in rough support of adding a notice, and two againstone felt that it would be not welcomed by non-US editors who might not be interested in what that editor perceived as a US-only topic of interest, and another felt that the topic is sufficiently advertised through Centralized Discussions page; the seventh participant was an admin whose entire contribution was a court conclusion of 'no consensus', which in effect means rejection of the proposal. In the end, objections of just two editors (one of them a regular of the MediaWiki_talk:Watchlist-details page) and an arbitrary weighting of them by a single admin managed to derail the likely awareness of the vote and participation in it of hundreds of others. A discussion about advertising of a major debate concerning Wikipedia's future (not to mention topics of digital rights and civil liberties worldwide) which could have brought it to the attention of hundreds if not thousands of active editors was held in an arcane niche of the Wikipedia project, of which almost no participant of the main SOPA discussion was aware. Further, five of the seven participants of this discussion have had prior involvement (familiarity) with the discussion about topics eligibility for the watchlist inclusion, another indicator that it was a decision made by a de facto oligarchya tiny group of editors who are familiar with this particular tiny cogwheel of the big Wikipedia machine.
The concern about avoiding unnecessary notifications can be traced to one of the policies of Wikipedia, entitled Wikipedia:Canvassing, which states 'When notifying other editors of discussions, keep the number of notifications small, keep the message text neutral, and don't preselect recipients according to their established opinions' and: Indiscriminately sending announcements to editors can be disruptive for any number of reasons. If the editors are uninvolved, the message has the function of 'spam' and is disruptive to that user's experience. More importantly, recruiting too many editors to a [dispute resolution] can often make resolving the dispute impossible. (Wikipedia, 2013) This last sentence offers a partial explanation: Wikipedia model was originally built around consensual, deliberative discourse models, and discouraged voting. Majority of discussions on Wikipedia have been and are small affairs, with only a few discussants. In such an environment, it is easy to see why canvassing would get a bad press: in an average Wikipedia discussion with few active participants, an ability to advertise it on outside fora and thus generate an influx of 'single purpose' votes can be highly destabilizing. The sheer size of English Wikipedia and its contributors makes it a different environment from most other Internet communities, which are much smaller, usually numbering dozens to hundreds of active members. With millions of pages under constant development, decisions on Wikipedia are based on a super-majority of voters in a specific dispute and not on a majority of all project participants.
Voting, however, is necessary for large-scale discussions to proceed efficiently (Jemielniak, 2013, p. 63) and is better executed in the environment were all voters are aware of the vote, and thus can exercise their right to participate in it. The Wikipedia norms, however, evolved in such a way as to encourage deliberative discourse and discourage voting, and even in situations where all editors can be considered 'involved', the cultural norms against 'spamming' are very strong, acting as an efficient deterrent to mass notifications. It is likely for this very reason that seemingly obvious solutions which could have drastically increased the vote participationsuch as sending all registered editors an email or a talk page messagehave not been implemented.
Wikipedia editors assume instead that interested editors will know where to find information about such debates. As one of the survey respondents stated: 'The information is out there if you choose to invest the time to visit the Village Pumps and read The Signpost and Tech News (and first find out these things exist).' An unspoken collateral is that the vast majority of the Wikipedia volunteers are not active enough to justify that assumptions about their knowledge of the project governance. This takes us back to the Michels argument about the passivity of the masses: it is indeed clear that majority of Wikipedians chose not to participate in the deliberations and the vote. What is less clear is whether their decision is consciously based on their presumed lack of interest, or rather, is an outcome of an inefficient information distribution system about such discourse and votes on Wikipedia, failing to inform the editors of the very existence of the said deliberations.
Then there is also the group, about a fifth of the editors, who felt that too many people voted by the time they became aware of the vote for their contribution to matter (a collective action problem). The existence of this group can be explained using the freeriding model, a useful counterpart to Michels' sets of theories. Based on Olson's (1971) logic of collective action, it suggests that it is quite rational for many members of an organization to participate as little as possible, while benefiting from others' engagement, something that Olson noted was particularly common for organizations providing public goods (and Wikipedia clearly qualifies here). Here, we clearly see a group of editors who looked at the vote, decided that the outcome was guaranteed, and saw no need to participate themselves. Such members may even be supportive of the oligarchy (activity of others), as it absolves them from the need to take care of this themselves.
Overall, the reasons of editors who did not vote, or in other words, did not participate in the democratic process with a well-framed positive outcome, including why so much free riding rather than bandwagoning happened here, would be a worthwhile topic to pursue in a follow-up paper based around a new survey asking more relevant questions about reasons for non-participation.
Finally, the resistance to 'canvassing' can also be seen as a way to preserve the status quo, that is to say, the dominant position of the active minority, though in the highly transparent and officially very anti-hierarchical Wikipedia system, such an argument is rarely brought up seriously, and when it is, it tends to focus on a critique of the administrators rather than the top 1% most active editors (O'Neil, 2011). Jemielniak (2013, p. 48) writes that 'there is a strong resentment of administrators and their true of imagined abuses of power, and adminiship is perceived [ … ] as a proof of editors' inequality'. Administrators form an easily identifiable block, though one that only obscures the reality of Wikipedia's hidden oligarchy, composed of highly active editors, of whom administrators are only a small part, as the SOPA vote 1:9 ratio of administrators to veteran editors demonstrates.
Given that the anti-canvassing rule has evolved to prevent small groups from dominating niche discussions (for a specific case, see Jensen, 2012or numerous examples in Jemielniak, 2013, the question that should be asked at the end is whether Wikipedia community should implement a different set of rules for discussions of project governance level. If we conceive of Wikipedia with respect to some notion of political life, its primary activities, ones that the canvassing rules evolved to regulate, would invoke different conceptions of politics. After all the peer-production of an encyclopedia is more in line with Aristotle's or Dewy's notion of politics as concerning action producing ends internal to that field of activity. Meanwhile, the SOPA protest is a form of politics involving demands for interest recognition by participating in a field of activity external to Wikipedia, something that the project's rules were not intended to cover. In other words, elite/minority control in mass participatory projects may present a problem in cases where individual interests are involved (such as in content edit wars on Wikipedia), but not in more top-level political cases involving project governance. Inefficient advertising of a vote, rather than protecting it from a takeover by a minority with vested interests, may facilitate such a situation, particularly when members of said minority are in a better position to learn that a vote is taking place in the first place due to factors such as exclusive knowledge about information channels advertising the vote's very existence (one of the classic ways and oligarchy entrenches in power, according to Michels).

Conclusions
Several scholars suggested that the Wikipedia model of deliberative discourse is highly successful, and should be used as a model for future projects aiming to improve the spread of deliberative democracy (Black et al., 2008;Firer-Blaess, 2011;Hilbert, 2009;Klemp, 2010;Kostakis, 2011;Noveck, 2009). In this light, several important observations should be made, centered on the fact that Wikipedia's poor information access transformed the vote on SOPA into an exercise in oligarchical decision-making.
The analyzed vote on Wikipedia, the largest ever in the history of the project, was significantly influenced by experienced, veteran editors, although the participation of new and first-time voters was not insignificant. The data support a power model estimate, with about 80% of vote coming from the 20% of most active editors, though no intentional marginalization of inexperienced editors could be detected. The exact turnout of eligible voters on Wikipedia is difficult to estimate due to the fact that it is hard to even define the range of the eligible voting populationparticularly when we consider that the vote was in theory open to the entire world. Only a small fraction of eligible editors voted (less than 2000 out of millions registered), and even fewer people from the outside (about 300 unregistered voters) took part in the process. In practice, out of the group of editors who were active enough to be reasonably expected to have a chance to be aware of the very existence of the discussion and the vote, the turnout was about 15%. This, however, still puts it at a level lower than the roughly 50% of a turnout expected in well-functioning, traditional models. While it is not impossible that in better conditions the Wikipedia model could produce significantly higher turnouts, it should be stressed that as it functions currently it does not appear be able to do so.
It seems likely that the turnout was significantly undermined by the inefficient system of informing the eligible voters that such a vote is happening. As Michels noted in his Iron Law model, most organizations willdespite their best intentionsevolve oligarchies that will work against democratic norms. Wikipedia, for all its openness, efficient merger of deliberative discourse and direct democracy, praised in literature, and valued in the Wikipedia community itself, seems not to be immune to this process. There are clear indications that decision-making on Wikipedia can and is dominated by the oligarchy formed by the most active editors. This oligarchy, however, is barely conscious of its own existence, yet still it has succeeded in ensuring that its members will have significant control over the decision-making process, even when the said process is open to all editors, in a community that puts much value on the egalitarian ethos. This has not occurred through a conscious effort of the said oligarchy, or of any identifiable leaders, aiming to exclude participation. Nonetheless, the facts speak for themselves: the SOPA discussion was popularized because a Wikipedian 'celebrity', Jimbo Wales, decided to bring it to the community's attention, and was then dominated by the most active editors. However, this is a development contrary to Michels' view that oligarchies will assume efficient control of information channels, closely controlling what information flows, who generates it, and who reads it: in the case of Wikipedia, there is no one (or no group) in control of the information, but the resulting effect -inefficient communicationachieves a similar effect of poorly informed populace (of volunteers).
In big part, the development of oligarchy on Wikipedia, and the failure in its communications stem from the project's cultural norms, which state that the editors are expected to be interested enough in the project to monitor policy discussion pages on a regular basis, rather than being deliberately notified about them. Yet the numbers of editors who follow Wales' talk page, or other major discussion fora like the Signpost can be counted in tens of thousands at most. As few editors can live up to an idealized version of an informed editor, the policy on canvassing in practice ensures that only the most active editors can be reasonably expected to be aware of the ongoing decision-making processes. In essence, Wikipedia's deliberative, direct democracy, in theory open to all, is in practice hidden in the bowels of the project, where only the most active and dedicated volunteers venture.
Time, and further studies, will likely tell us if Wikipedia's system will continue evolving toward a more centralized oligarchical system, capable of controlling the information system, or more curiously, whether it will remain stuck in its present half-way state, where decisions are made by the most active editors, who are nonetheless not organized enough and not aware enough of their own existence as an emerging oligarchical decisionmaking group, to exercise control over what message the Wikipedia project wants to send out (either to its own members or to the outside world).
Contributing to the body of literature on democratic decision-making, this research therefore offers a cautious lesson regarding assumptions of voters (or discussants) turnout: in some cases, the low turnout is not due to the voters' lack of interest (passivity of the masses), but due to the failure of the vote organizers to inform the electorate of the vote's very existence. Further, such a failure may not be intentional, but a byproduct of the inefficient communication channels existing within a given organizational structure. Note 1. As of early 2016, however, several other votes have exceeded said threshold; for an up-to-date list see https://en.wikipedia.org/wiki/Wikipedia:Times_that_1000_or_more_Wikipedians_ supported_something.

Disclosure statement
No potential conflict of interest was reported by the author.

Notes on contributor
Piotr Konieczny is an Assistant Professor at the Department of Informational Sociology, Hanyang University. He has received his Ph.D. degree from the University of Pittsburgh. He is interested in the sociology of the Internet and social movements, in particular in topics such as wikistheir impact on individuals and organizations; decision-making processes and organizational structure of Wikipedia; patterns of behavior among Wikipedia contributors; relation between wikis and social movements; free culture movement; and teaching with new media. [email: piokon@post.pl]