Wikipedia has become a prime example of really existing mass cooperation with low barriers to entry; perhaps the most important example. This is why it is important for researchers to map its operations. I want to focus on situations where governance is most apparent: in situations of conflict, when direct negotiation fails, insults fly and people appeal to higher authorities for justice.

In a recent series of interviews on peer governance in Wikipedia posted by Vasilis Kostakis, Michel Bauwens argued that the victory of deletionists over inclusionists imposed an artificial scarcity, which breeds boundaries and hence power structures. Now, the inclusionist vs. deletionist battle is useful for defining the specificity of Wikipedia: it’s true that what distinguishes an online encyclopedia written by amateurs of varying competencies from a traditional expert-based encyclopedia is the absence of space constraints. Another consequence of this freedom and absence of editorial oversight may be a lack of discernment in distinguishing the trivial from the profound, a collapsing of what is historically momentous and what is “just for the fans” – for example the English Wikipedia entry on “democracy” has 9,550 words; the entry on the TV series “Lost” has 11,380 words. So, inclusion or deletion is a central philosophical question regarding what Wikipedia should be in terms of content. But when it comes to issues of power, I would argue that notability battles are a symptom of a larger issue, which is that conflictuality is endemic in the project.

Why is there so much fighting on Wikipedia? The answer could be framed in structural terms: governance in peer projects depends on the diffusion of decision-making. When combined with fuzzy guidelines such as “notability” and “consensus” the result is innumerable decision-makers with their own take on the rules. Or it could be an institutional problem: there is no Constitution to spell out important roles and processes such as the exact powers of the charismatic founder or recall mechanisms for abusive authorities. Then again conflicts could arise because of the nature of the technology: the speed of the transition between writing and publishing does not give much of an opportunity for people to cool off. Or they could result from the nature of an encyclopedia, which is based on assertions as to what truth is, rather than on opinions, which can vary: such assertions do not leave much space for others.

All of these answers are contributing factors. But the simplest and most convincing explanation is that on Wikipedia, completely different people with completely different opinions are attempting to work together; let’s not forget that a lot of content disputes originate not in whether to delete or keep an article, but in what articles actually say. To explore the issue, I will draw on a recently published article which offers a remarkable analysis of conflicts in the French Wikipedia, whose title could be translated as “The negotiation of points of view: a social map of conflicts and quarrels in the French Wikipedia” by Auray et al. The article combines two distinct approaches: social network analysis, with its focus on quantitative maps of social relationships, and pragmatic sociology, with its emphasis on interactions between people understood as sequences of actions.

I have argued elsewhere that the development of autonomous online scientific expertise has evolved since the origins of the Internet when it was based on the proven expertise of hackers, to the present situation where the competence of contributors to open projects such as Wikipedia is impossible to ascertain. In a similar vein, Auray et al. preface their analysis with the observation that the rise of Wikipedia parallels an evolution of our relationship to “scientific truth” as manifested by the reopening of scientific controversies and the questioning of traditional authorities.

The population of Wikipedia editors comprises a great mix of people: scientists, interested amateurs, consumers, advertising agencies, industry spokespersons. All these people with diverse interests and competencies must negotiate together, which can give rise to conflicts. Negotiation has previously been analysed as an essential part of the work of open communities such as free software projects. When people start fighting, they have two avenues: they can try the mediation of a public discussion, but the heterogeneous nature of participants and remnants of previous confrontation can result in new disputes erupting, governed by the logic of honour, of not backing down. A second avenue is arbitration: the appearance before a tribunal where claims or accusations are made, evidence is presented, and witnesses can testify. Editors who contribute a lot tend to make lots of accusations. This is because as participants become more and more involved in the encyclopedia, they become more and more familiar with its rules, and seek more and more to apply them to enable the project’s functioning, as a form of personal engagement which is also a moral initiative.

Auray et al. identify several factors which contribute to conflictuality, such as the number of participants, the location of disputes, and the identity choices of participants.

The larger the number of contributors, the more likely discussion is; the threshold number seems to be eight. When there are more than ten participants, discussion increasingly moves to the talk pages of users, and is more likely to degenerate into insults. A surefire indicator of fights are references to policy pages. These can be statistically measured: research by Kriplean and Beschastnikh has shown that pages with more than 250 posts had 51% of the links towards policy pages.

There are two main types of articles where conflicts erupt: first, the usual suspects are topics with burning current affairs value involving inter-ethnic or inter-faith conflicts; second, “scientific” categories with low academic legitimacy such as homeopathy and chiropraxy are strong conflict zones. Suspected “sock-puppetry” (fake identity) is also a source of conflict; an attenuated version of this being the lack of regard for people who have not registered on the site and instead just use an IP address: more than half of the text inserted by “IPs” is deleted, and they are more likely to be present in semi-protected articles which is where disputes and insults typically occur. IPs are also more likely to insult others, so there are suspicions that IPs are registereds users who use “socks” to engage in insulting behaviour which they would not dare to do under their registered identities.

In conclusion, the authors ask what happens when conflicts arising from the negotiation of points of views cannot be resolved or pacified into reasoned discussion? In this case there occurs what they call a “re-singularisation”, where, after the discussion being generalised, taken to the higher level of public mediation, it reverts to one person being blamed. This finding of a personal “fault” by troublemakers or persecutors can be described as a key “managerial strategy” on Wikipedia (others may be tempted to call it a major flaw).

This analysis has the benefit of explaining the often-found fact that the proportion of policy and regulatory discussion in relation to mainspace content is rising – as a function of the process of acculturation into the project of its custodians, the pre-admins and admins. It also points to the paradox that since the project’s development relies in part on the constant entry of enthusiastic “newbies”, the subsequent herding of these novice autonomous content providers by administrators along normative policy lines cannot but generate resentment and the feeling of injustice, in the shape of participants who feel they have been ill treated, or even humiliated, by registered editors and admins. Unfairness can be hard to evaluate, as both sides in disputes invariably feel they are in the right, so a structural example will best illustrate the issue: creators of articles set its tone. Because of a “first-mover advantage”, the initial text of an article tends to survive longer and suffer less modification than later contributions to the same article. It is to be expected that article creators who maintain an interest in the article would put it on their watch list and, despite the project’s injunctions, would experience feelings – if not of ownership – at least of heightened sensitivity and possible unhappiness if someone attempts to “improve” their baby. The problem is compounded when editors dispose of administrative tools. Further, if these kinds of situations involve existing friendship cliques, there is an increased likelihood of abuses of admin authority. In this sense, conflict would appear to be an inevitable by-product of the Wikipedia development model.