Performance and perceptions of human computation games for image tagging

Applications that use games to harness human intelligence to perform various computational tasks are known as human computation games (HCGs). Most HCGs are collaborative in nature where players to cooperate to score points. Competitive versions, where players work against each other have been argued to address shortcomings of collaborative HCGs. However, there is yet little work done in understanding how different HCG genres affect players' perceptions and performance. In this paper, we focus on image tagging HCGs, where games are used to generate keywords for images. Three versions were created: collaborative HCG, competitive HCG and a control application for manual tagging. The applications were evaluated to examine the quality of the tags generated and users' perceptions of these genres. Results suggest while participants reported liking the collaborative and competitive HCGs over the control application, those using the latter seemed to generate better quality of tags.