Ratings Use in an Online Discussion System: The Slashdot Case

Abstract

Large-scale, online communication systems allow many-to-many interactions among users, which can result in a variety of positive outcomes. However, the prevalence of information overload and problems caused by the loss of shared communication channels in text-based environments may create barriers to realizing the benefits of online interactions. Attempts to manage online communication systems in the past depended on techniques that cannot be applied to, or do not allow, large-scale interactions. Slashdot is a large-scale, long-lasting Web discussion community that uses a form of recommendation system to provide feedback about the quality of comments users post to the site. This dissertation examines this novel approach to organizing an online communication system in terms of how users employ the ratings provided by the system, whether comment ratings have an effect on how new users of the site participate, and how users making recommendations about content actually provide ratings. I find that users do employ ratings to change how they view content, but that there is some resistance that prevents them from doing so readily. To overcome this friction, I recommend dynamic changes based on the choices of other users who seem more willing to make interface changes based on comment ratings. I also find that new user participation on Slashdot is affected by feedback on the initial comment made by the new member, but that user observation is just as important in determining how the new member will participate in the future. Finally, I find that ratings are being sufficiently applied to comments, but that some comments are not receiving fair attention because of when or where they are posted within the online discussion. The overall conclusions of this work are that pre-rating content helps to relieve the pressure of attaining sufficient ratings on comments, that rating labels provide valuable feedback for customizing how users with different motivations may read comments, and that comments ratings positively affect user experiences in a large, online discussion system. The Slashdot case shows how the use of recommendations in an online discussion system creates organization that ameliorates the problems of information overload and loss of communication channels, while still allowing for large-scale, heterogeneous interactions.