Category "Conference Proceedings"
Category "Peer Production"
June 21, 2005
Authority Models for Collaborative Authoring
Krowne, Aaron, and Anil Bazaz. “Authority Models for Collaborative Authoring.” HICSS 2004 Proceedings. Jan. 2004. http://br.endernet.org/~akrowne/my_papers/authority_models/authority_models.pdf. 20 June 2005.
Krowne and Bazaz conducted an empirical study to research which authority model works best in a Commons-Based Peer Production (CBPP) environment. The CBPP in question was PlanetMath, which is powered by the wiki-like Noosphere interface. They recruited 25 mathematics students and asked them to contribute to the project over a period of six days. Noosphere typically operates with a hybrid authority model that is a combination of the two models researched, and so it was easily altered to offer only one model to users during each three-day section of the study. The two authority models in question are defined as follows:
- owner-centric: assumes an owner of each entry. Other users may suggest changes, but only the owner can apply the changes, and only he can grant other users ‘edit’ access to the entry (Krowne, 7). This model assumes that the owner is the subject-matter expert, and that all other users will defer to that expertise. This model will initially produce highly cohesive product with lower coverage.
- free-form: allows anyone to edit any entry at any time. (Ex: Wikipedia) It assumes that all users are on the same level, and that expertise will be universally recognized and deferred to. This model may initially produce high coverage with lower cohesion.
The exit survey indicated that while “the free-form model is the most productive, the owner-centric model is preferred by users. However, a significant fraction of users depart from a preference for the owner-centric model (about a third), indicating that there is probably a place for supporting the free-form model as well” (IV). The authors suggest that if production (economic reality) is paramount over users feelings, then the free-form model is appropriate for a project (as demonstrated by corporate wikis). In environments where users are not obligated to contribute and thus their feelings are more important, the owner-centric model may be an appropriate default.
They also noted that “swift trust” did not develop over the course of the project. Their hypothesis had supposed it would as participants noted the quality of each others work, but instead production dropped from 30 instances in the first half of the study to 4 in the second half. They posit that since the theory of swift trust was not developed for a commons-based setting, perhaps it does not apply to it.