You are here

What Makes Wikis Work?

Montréal.
The next session here at WikiSym 2007 begins with a short paper by Sarah Guth from the University of Padua, on wikis in education. She's done some work using social software in teaching environments, and discusses the question of whether such teaching should take place in public or non-public social software environments. Using (public) wikis enables collective authoring (which enables critical reading and responsible writing); raises issues of individual and collective ownership (challenging conventional Western epistemologies of individual intellectual property); and highlights content as ego-less, time-less, and never finished while enabling continuous development. Publishing online also empowers students, and history and discussion functions focus on writing as process, not product.

So, the suggestion is that students in such wiki-based models may have a better responsibility, a better sense of collective ownership of content, and a better sense of knowledge sharing. Sarah tested these propositions by working with English as second language learners in Padua, working respectively in a semi-public wiki (readable by all, but writable only by students) and in a public wiki. This also meant that the communities in each wiki were substantially different, of course, and that content was created in the second case not only by students, but had mainly been created by other community members and was simply added to by students. There was also further difference in content in that the semi-public wiki contained opinions and facts, not facts only.

Results of such trials were that students found that the public wiki promoted collaboration beyond the classroom; increased their sense of responsibility and promoted more accurate writing; and that the knowledge sharing gave students a sense of empowerment. Drawbacks were that there was pressure felt from writing on someone else's wiki; frustration from having one's own contents edited by unknown users; a lack of confidence to edit contents and structure; and a frustration at following someone else's guidelines. Many challenges are present on both wikis: difficulty intervening on peers' texts, frustration at editing by others, difficulties in learning to negotiate, and a need for guidance; on the public wiki, however, quality improved, a greater sense of knowledge sharing evolved, and user empowerment was felt.

A public wiki requires a larger userbase, possibly collaboration with other universities around the world; guidelines must be written and edited by contributors; page creation and editing may change depending on context; and wiki development should be seen as a component of a collaboration project, not as the whole of the project itself. For Sarah, there is now a hope to create a larger base of students around the world participating in these projects, and to create a growing repository of material written by and for students.

The next speaker is Camille Roth, whose interest is in identifying the determinants for what makes wiki communities viable. There is now a large number of wiki communities. Some such communities operate effectively and successfully, others do not - why is this so? Camille qualitatively examined a dozen arbitrarily chosen wikis, and quantitatively studied some 7649 MediaWiki-based wikis. Such wikis necessarily always rely on relatively similar software which allows for collaborative editing and offers page edit histories, and often also offer account creation and talk pages; many wikis also employ some form of open source-inspired content licencing. The topological structure of content in wikis generally exhibits structural features similar to World Wide Web networks overall. Additionally, there are semantic and organisational features: the language(s) used, the explicit and implicit policies employed (such as Wikipedia's NPOV) which have evolved themselves from original ideas through collaborative wiki processes, and the technical policies on who can contribute, how interaction and social structure are organised, and what new technical features are introduced.

This also translates into demographic diversity across wikis, in terms of population size, content size, and activity (the number of edits). Ultimately, wikis are two things: a group of users, and a set of pages, with attendent population and content dynamics. Population dynamics can be generally distinguished into an initial bootstrapping periods, user incentives (altruistic, socially concerned, and selfish), and enrolment and leadership periods (with implicit and explicit role distributions that replace more egalitarian models at earlier stages ). Population very mildly appears to correlate with content size, which may suggest that distinct demographic paths are in place for different wikis; the post-bootstrapping phase may be very different, and there may be a link with governance models here: such models may determine success or failure. What's needed here is to identify distinct development paths for wikis, and link such paths with structural and governance features.

This leads us neatly on a panel on what determines the success (or otherwise) of wikis, which is chaired by Stewart Mader and involves Mark Bernstein, Dirk Riehle, Anne Goldenberg, Ted Ernst, Francois Beauregard. Mark Bernstein from hypertext publisher Eastgate begins by noting his own experience in starting wiki and other communities; Dirk Riehle from SAP notes past experience in using a wiki for supporting knowledge management in software development which benefitted from having a well-defined scope for the project and building on existing social networks; Anne Goldenberg is a postgraduate student and wiki researcher who developed a wiki for the art community which tied into graffiti artists' practices of building on and overwriting one another's works, and highlights the question of acknowledging other participants' contributions; Ted Ernst notes questions of generational change amongst wiki administrators; and Francois Beauregard notes especially the role of wikis for knowledge sharing within organisations, and the organisational dynamics which result from such processes - wiki use depends on organisational culture, but also creates organisational culture.

So, Stewart asks, how much pre-existing organisational culture must exist, and how much can it be created? In large companies like SAP, Dirk suggests, it must be created from the bottom up (sometimes against the resistance of managers); this may also lead to an abundance of individual wikis, however, and this overload is linked to the pre-existing organisational structure which has divided corporate populations into a multitude of hierarchical teams. Francois adds that in organisations, the question of safety is also important - people must feel safe to participate even if the quality of content isn't fully formed. Ted suggests that a clearly felt need for wikis to be used provides a further key driver, and Anne adds that in her context as well as in others, giving others access to one's own work can be a daunting step (especially if it also includes giving non-experts the ability to edit content) - negotiating authorship is crucial here, and this is also a political process of negotiating, overcoming, or re-forming hierarchies of contributors.

Stewart's next question is about statements which suggest that wiki work erodes expert authority; Dirk links this to older questions around knowledge management tools. Wikis work best with people who are enthusiastic about what they are doing and see a benefit from the collaboration, he says; to some extent this can be enforced through performance measurement as well, however (where wiki contribution becomes a factor in individual performance measurement in the corporation). Anne adds that in more open wikis there may also be an economic factor: if expertise is given away for free in a wiki, this is seen to undermine experts' ability to derive financial rewards from their knowledge (the same is true also in the educational environment, where using wikis is sometimes seen as undermining the teacher's special role in the classroom). Stewart's own suggestion is to experiment with putting just a little knowledge out on the wiki and exploring what happens - often an eye-opening experience.

A question from the floor further highlights that educational dimension of wikis as breaking open the enclosure of knowledge in academic frameworks, and in response Anne notes the Open Archive movement, which aims to do much the same (this links also with our AoIR keynote by John Willinsky a few days ago, of course). What's necessary here may be to ensure that students as well as academics understand the conditions under which knowledge is produced and disseminated - to understand the extent to which any form of publication can be trusted, and how their contents came to be. Stewart suggests that there will be evolutionary changes to the publication and peer review system over time, and that these will emerge to wider recognition once a critical mass of people have begun such alternative uses. In the process, distinctions between scholarly and other publishing are likely to blur further.

A further question from the floor also highlights the growing need for wiki reputation systems which trace contributors' track records and measure their standing in the wider community (and even across individual wiki, blog, and other social software spaces). Some such systems now exist, Anne says; at the same time, Francois suggests, such ratings are double-edged swords as they may also discourage new contributors. Stewart notes that the need for such systems differs between different uses - in organisations, there may be less need than in fully open wikis on the World Wide Web.

The next comment notes the experience of wiki-writing a book chapter, and says that all contributors did gain standing and status from their participation; it describes knowledge as a service, and notes the possibility of strong positive payoff from providing that service for free. Anne notes that in cooperating, it's clear at the end which contributor made what contribution, where in collaborating, the outcome is more important than the individual contributions. So, in a site like eBay, for example, contributrs necessarily need to be identified by name; in Wikipedia, this is less crucial. Another commenter notes his own work on a contributor recommendation system, and directs our attention back to questions of what community strategies work in wikis - 'make people feel safe', 'bug them to contribute', 'have a need to participate' appear to be some such strategies. Stewart asks a similar question by way of summing up.

Ted answers by saying that there are any number of wikis on many topics already out there, and encourages people to use those rather than start their own (except for where such wikis need to exist in a corporate enclosure). Dirk points out the need for a clear scope, mission, or vision again; how starting a wiki can be done successfully is more complex, and he highlights especially the need for a wiki evangelist who promotes and guides the wiki especially in its initial phase (and who needs to understand that their role will change down the track). There's also a need for internal reading cultures (which may also involve aggregating content and offering it as RSS feeds), help pages, pre-seeding wikis with some initial content. Against Ted's comment, Anne adds that multiple wikis on the same subject may be beneficial for content and idea diversity, and says that the difficulties may not be where people think they are - not in the functional editing process, but in the conceptual understanding of knowledge in a collaborative wiki context.

Francois suggests that in organisations, a good idea is to start with simple and valuable content which is not available anywhere else and draws users to it, and warns of overstructuring the content - in a wiki, content structures are a form of constraint. Dirk notes that wikis are bottom-up tools which are self-organising, and therefore must be allowed to self-organise; Mark highlights the role of content as the ultimate determinant of wiki success, and by comparison downplays the role of hierarchy of content and contributors. He also notes the importance of the knowledge that lies between the pages. Stewart, finally, suggests that starting with a small group is a good idea, and that this group should include some key drivers as well as some less experienced actors willing to learn and work together; from here, the attraction of the model can spread as new users will be attracted by the early successes. He also notes that blank pages are a major turnoff - even if content is as yet poor, it's better to enter some outline of the intended content into a page rather than leave it blank. Further, as groups of contributors get going, a barnraising session (getting people together in the same physical space and collaboratively seed the wiki with content) can be instrumental in creating community and community protocols. He also notes the crucial importance of having a shared purpose, and points us to his Website Wikipatterns.com as a repository of ideas.

Technorati : , , , , , ,
Del.icio.us : , , , , , ,