You are here

Looking under the Hood of Wikipedia

After some drama getting here (note to self: Qantas may be in trouble at the moment, but avoid Scandinavian Airlines like the plague), I'm now in Copenhagen, and we're about to start the programme proper of the ninth annual Association of Internet Researchers conference. Although the presentation by Dr. Hala-Seuss which runs in a parallel session was very tempting, I'm starting the day in a session on wikis. Timme Bisgaard Munk is the first presenter, presenting on his study of the Danish Wikipedia.

What makes people contribute to Wikipedia? Classic theories of user motivation do not seem to explain its success particularly well, so Timme surveyed a number of Danish Wikipedia administrators to get a better sense of their motives; this, of course, biases the results towards the admin class on the site - it provides only limited information about 'normal' users. Traditional motives listed for participation include ideological concerns, altruism, the drive for social status and belonging, the joy of writing and participation, and a few others; broadly, there is a distinction between mono- and multi-causal explanations. The latter describe Wikipedia as a case of peer production, a dynamic learning community, a system for contributions by users in different roles, a social movement, or a virtually unknowable community motivated by a myriad different factors.

Timme suggests that the mono-causal explanation is too simple; he also says that Wikipedia is not a clear example of the peer production model (he compares it with open source software development, where job outcomes can be a major driver for participation - in my view, this vastly overestimates the role that potential job outcomes play in open source participation motivations, though), or of a learning community (as this implies that all users strive for advancement to admin roles, he suggests). He suggests that the explanation that different roles drive participation is the appropriate one, including a drive to achieve status, be part of a community, pursue political agendas, develop and display an online identity, etc.

There are a number of variations here: one, people may gradually become more serious in their contributions; two, the role people play on the site may determine their motivations; and three, there is a link to the assignments people take on on the site. The optimal strategy for contribution then becomes writing articles on narrow subjects for few readers, becoming niche experts; in this view, indifference (i.e. a lack of complaints) can be interpreted as passive positive recognition. This also means that everyone can then feel that they have made a successful contribution. This also both motivates and demotivates, however; the result for all is undifferentiated recognition.

Next up is Hichang Cho, whose focus is also on motivations for wiki participation. What factors influence knowledge contribution behaviours? Suggested factors include motivational factors (both extrinsic and intrinsic rewards), relational and social factors (reciprocal obligations, community membership and belonging, and the role of social networks and social capital), normative factors (that is, subjective norms), and efficacy (knowledge self-efficacy). Hichang suggests a relation between these, with motivational factors influencing social and relational factors, which in turn exercise behavioural control and finally determine behaviour itself.

Hichang conducted a survey of some 223 Wikipedians in the US and Singapore, and from this tested the links between motivational factors and behaviours. Interestingly, recognition and membership were found to be insignificant as motivating factors; the rest were (sometimes strongly) significant. Additionally, intrinsic rewards were seen to be more important than extrinsic rewards; and reciprocity and community membership were important as social and normative factors. Social and relational factors need to be taken into account in understanding Wikipedia processes, then. That said, the present study looked especially at fairly active Wikipedia members and did not examine anticipated general reciprocity, meta-knowledge, and a number of other variable, so more work is to be done.

Up next is Rut Jesus, whose focus is on the study of Wikipedia articles rather than contibutors. This is a non-trivial undertaking, since wiki articles are famously multi-authored and palimpsestic; at the same time, several tools for the analysis of such content are not becoming available. These include HistFlow, which visualises the edit history of wiki pages and over time tracks the fate of edits made (that is, tracks the changing authorship of articles); WikiChanges (which simply tracks the number of edits made in any one time period, especially showing peak editing periods that may be driven by external events), and of course standard internal Wikipedia statistics (number of edits, number of editors, recency of changes, etc.). Additionally, the article's talk page also provides information on the tacit knowledge which has influenced the development of the wiki page, and an examination of contributions to the talk page therefore also provides useful information.

Other tools are WikiDashboard, which tracks the activities of individual editors; WikiScanner, which tracks IP addresses and can be used to uncover self-interested edits by people with a stake in the topic of the page; the well-known WikiTrust, which colorises the entry text according to how thoroughly passages have been edited; and tools which create animations of edits of specific pages over time, chromograms which provide a visualisation of editing practices, and biclique networks which track which editors are editing what pages. Many such tools are open source, versatile, Web-based, and available for any article; some, however, remain at a conceptual stage and are presented only through examples of their output.

Finally we move on to Thanomwong Poorisat, whose interest is in the perceived credibility of collaborative Websites. Collaborative Websites tend to lack traditional gatekeepers, of course, and instead rely on the collective intelligence of their participating communities; this opens up significant pitfalls if the content evaluation processes within such communities are flawed or ineffective.

Wikis can be seen to generate a kind of emergent credibility, especially where they openly draw on external sources; clearly identified interactive features may play a considerable role in influencing such perceptions of credibility. The relative prevalence of pseudonymity and anonymity also affect credibility perceptions. The present study surveyed users (university students) of both wikis and non-collaborative Websites to gauge their perceptions of such sites.

What the researchers found was that sites with collaborative cues were perceived to be more credible; however, such perceived credibility was unrelated to the memory of collaborative cues (meaning, I think, that respondents did not remember specific cues, but were simply generally aware of their existence on collaborative sites).

Technorati : , , , , , , : , , , , , ,


Thanks for this update. I'm especially interested on the papers that are user surveys. Do you know if the conference will publish the papers on their website?

Sure - if presenters have uploaded their full papers to the conference submissions system, they will be made available shortly on the Association of Internet Researchers Website...