Normative behaviour in Wikipedia

Back to Programme

Goldspink, Chris; Edmonds, Bruce; Gilbert, Nigel (University of Surrey)

1 Centre for Research in Social Simulation, Department of Sociology, University of Surrey. Guildford, GU1 7XH, UK..

2 Centre for Policy Modelling, Manchester Metropolitan University, Aytoun Bldg., Aytoun Street, Manchester, M1 3GH. UK.

3 Centre for Research in Social Simulation, Department of Sociology, University of Surrey. Guildford, GU1 7XH, UK.

Email address of corresponding author: c.goldspink@surrey.ac.uk

Extended Abstract

This paper provides an account of research methods being employed in order to understand normative self-regulation mechanisms in the Wikipedia and in Second Life. The research is being undertaken for the project "Emergence In the Loop: simulating the two way dynamics of norm innovation", (EMIL) a three-year EU funded project (Sixth Framework Programme - Information Society and Technologies - Citizens and Governance in the Knowledge Based Society). Both qualitative and quantitative methods are employed in a fine grained analysis of behaviours and speech acts. The aim is to collect data which reveals the mechanisms involved in 'at a distance' normative coordination, particularly that associated with computer mediation. This research has as its target an understanding of the origins of social order in computer mediated environments and how different aspects of the technical environment interact with different governance structures to change social activity within online communities. The project has the potential, however, to contribute to a more fundamental understanding of the relationship between social structure and agency. The empirical analysis being undertaken through the Wikipedia and Second Life cases is designed to provide a point of grounding for the development of one or more computer simulations.

Wikipedia

Wikipedia has a clearly defined task - the production of a quality on-line encyclopaedia (Sanger 2005). Task coordination, production and social influence takes place by means of editing activity and communication about that editing activity. Editing activity is the primary means for creating, amending and maintaining the content of the Wikipedia. What can be done and how it is done is influenced by the motivations and intent of participating individuals (agency), the goal of the Wikipedia, the social structure (norms, rules and influence of community), and the constraints of the technology. Our focus is on the self-organising and self-regulating (governance) phenomena. Insight into this can be gained by examining the Discussion pages which accompany many of the articles as well as editing behaviour within or on the article.

Any coordinated (social) activity happens primarily in and through language. From this perspective institutions can be viewed as networks of commitments established, maintained and modified, in and through linguistic exchange (von_Krogh and Roos 1995; Rocha 1996). In the case of Wikipedia this is language exchanged asynchronously, (somewhat) anonymously and by text on the Discussion pages associated with an article.

In this study we selected a sample of 19 Discussion pages associated with articles flagged as controversial and 11 identified as of outstanding quality (Featured articles). These were coded at utterance level to reveal patterns of illocutionary influence. The results were also compared to an analysis of editing behaviour. To capture the subtle elements of linguistic influence the Virtual Response Mode (VRM) coding system was employed (Stiles 1992). Analysis was both qualitative and quantitative, with data being coded using the open source package WEFT-QDA and then converted to a form which would support statistical, including multi-level analysis (Snijders and Bosker 2004).

As a consequence of the preliminary analysis it was decided that an experiment would help answer some questions which could not be resolved through access to the Wikipedia community directly. This experiment was designed to examine the influence of pre-existing dialogue on subsequent editing behaviour and the influence of social identity (Smith-Lovin 2003; Smith et al. 2007) on responsiveness to norms.

The Second Life Case Study

Second Life (SL) is a 3-D virtual world built and owned by its Residents. Beginning in 2003 it currently attracts approximately 10,000 'residents' from around the world. As an open virtual world, Second Life supports a wide range of activity. Unlike Wikipedia it provides an opportunity for participants to identify goals and to self-organise around a wide variety of goals and tasks.

Each user interacts with others and the persistent objects in the virtual world through a virtual representation of themselves: an "avatar". Users can create environments which others may visit and interact within. Access to these can be controlled or open to all comers. Frequently it is the style and content of an environment which is a key factor in influencing the kind of visitor that may frequent it. In other cases access is controlled by membership of a group, by payment or by active policing by staff members. Such locations may have any theme or purpose that the owners wish (with a very few legal restrictions), such as: shopping Malls, clubs for users of different sexual orientations, brothels, role playing environments, parks and class rooms. Groups (which are the SL equivalent of discussion forum or mailing list) are a second way in which contact with users with similar interests may be made and maintained. Each individual has a "friends" list which aids regular contact between users. There is an SL currency that is convertible to the US$ on an exchange and a growing economy of virtual goods and services. Thus the means to create and grow new societies and social groupings are present here.

Those who form the groupings can (within certain constraints) define what can be done within them and how things should be done - including defining, suggesting and enforcing rules of conduct. Unlike Wikipedia, SL involves participants in synchronous communication. Until recently this was text based with the opportunity for exchanges to be broadcast to all those nearby or directed at specific individuals. The environment now also incorporates voice communication. Also in contrast to Wikipedia users interact in a simulated three dimensional space containing a variety of artefacts and can themselves create those artefacts. Since actors take on a visual presence through avatars which they can select and modify, there is a much a higher level of 'social presence' (Biocca 2003) compared to Wikipedia. Avatars are named but users may adopt several avatars (and hence identities) and the persons real world identity is hidden from those who interact with the user. Thus the actions of a single avatar may create consequences and a reputation for that avatar, but this may not impact upon the real life of the user. Simply changing avatars is easy, but will involve a cost to the user through the loss of non-tranferable artefacts. More importantly it would also entail the loss of the social network developed (friends, lovers and professional colleagues).

As Bainbridge (2007) has argued, that of this type of artificial environment is particularly suitable for the conduct of social scientific enquiry due to its openness. It is quite possible to recruit participants for involvement in situated or experimental research in much the same manner as would be possible in the real world. This recruitment is not constrained by geography and may involve potentially large numbers at low cost when compared to conventional research.

For our purposes, SL is attractive as it necessarily involves participants in new social situations: ones where they may participate in co-creating a novel environment. SL also allows a lot of social experimentation, with participants free to experiment with different identities, affiliations and behaviours with relatively low long term social costs. It is therefore, a social environment in which it may be easier to identify the need for norm innovation or the adaptation of norms imported from other environments outside of the virtual world.

EMIL has a presence in SL through the Centre for Policy Modelling at Manchester University. We were therefore well placed to collect data in this environment. Our procedure for Second Life involved two parallel forms of data collection. The first elicited information from 'Sim Managers' (people who have established and maintain large environments in SL) and thus helped to get a picture of how norms were developed and enforced over time. The second recruited people interested in normative behaviour in SL and collected information on 'normative events' which they encountered in their normal participation within SL and thus help to give a more detailed "snapshot" of deviation from norms and the reaction of others.

Some 25 participants in SL volunteered to participate in this research. Each collected data relating to one or more 'normative events' encountered while they went about their business within SL over a specified period (3 months). A normative event was defined as an event involving two or more people in which one person explicitly or implicitly expects or demands of the other/s compliance with some explicit or implicit standard, norm or rule. The recruit need not have been an active protagonist but must have been personal witness.

Participants were asked to document the event as closely as possible to the time of its occurrence. They prepared a brief narrative account in response to specific prompts. This data was again analysed both qualitatively and quantitatively. Transcripts of the communicative exchange were analysed in much the same way as the Wikipedia Discussion pages. The narrative accounts of participant/observers were approached using grounded theory (Strauss and Corbin 1997).

This paper provides an account of the methods and of the initial findings of the research, focusing in particular on lessons relevant to the design and conduct, and the fitness of the methodology for the purpose. It will be of interest to researchers of social behaviour, particularly in computer mediated environments. It will be of particular relevance to researchers interested in the interaction between social and technical artefacts and those interested in the effectiveness of alternative governance systems within open source or Web 2 environments. The work also provides an example of methods suited for use within established methodological frameworks such as activity theory (Kaptelinin and Nardi 1997; Engestrom et al. 1999), structuration theory (Orlikowski 1998) and its derivatives (Widjada and Balbo 2005). While the focus of our research was on norms the methods are relevant to an interest in emergent social structures more generally and to social research where the dialectic between micro and macro phenomena is of interest.

References

Austin, J. (1962). How to do things with words. Oxford, Oxford University Press.

Bainbridge, W. S. (2007). "The Scientific Research Potential of Virtual Worlds." Science 317: 472-476.

Biocca, F. (2003). "Towards a more robust theory and measure of social presence: review and suggested criteria." Presence 12(5): 456-480.

Engestrom, Y., et al. (1999). Perspectives on Activity Theory. New York, Cambridge University Press.

Habermas, J. (1976). "Some Distinctions in Universal Pragmatics: A working paper." Theory and Society 3(2): 12.

Kaptelinin, V. and B. Nardi (1997). Activity Theory: Basic Concepts and Applications. CHI 97.

Orlikowski, W. Y., JoAnne (1998) "Genre Systems: Structuring Interaction through Communicative Norms

" Volume, DOI:

Rocha, L. M. (1996). "Eigenbehavior and Symbols." Systems Research 13(3): 371-384.

Sanger, L. (2005) "The Early History of Nupedia and Wikipedia: A Memoir." Slashdot Volume, DOI:

Searle, J. R. (1969). Speech Act: An Essay in the Philosophy of Language. Cambridge, Cambridge University Press.

Smith-Lovin, L. (2003). Self, Identity, and interaction in an ecology of Identities.

Smith, J. R., et al. (2007). "Social Identity and the attitude-behaviour Relationship: Effects of anonymity and accontability." European Journal of Social Psychology 37: 239-257.

Snijders, A. B. and R. J. Bosker (2004). Multi-level analysis: An introduction to basic and advanced multilevel modeling. London, Sage.

Stiles, W. B. (1992). Describing Talk: A Taxonomy of Verbal Response Modes, Sage.

Strauss, A. and J. M. Corbin (1997). Grounded Theory in Practice. London, Sage.

von_Krogh, G. and J. Roos (1995). Organizational Epistemology. London, St Martins Press.

Widjada, I. and S. Balbo (2005). Structuration of Activity: A view on Human Activity. OZCHI 2005. Canberra, Australia.

Download Paper

Back to Programme