Studying the effect of similarity in online task-focused interactions

Although the Internet provides powerful tools for social interactions, many tasks-for example, information-seeking-are undertaken as solitary activities. Information seekers are unaware of the invisible crowd traveling in parallel to their course through the information landscape. Social navigation systems attempt to make the invisible crowd visible, while social recommender systems try to introduce people directly. However, it is not clear whether users desire or will respond to social cues indicating the presence of other people when they are focused on a task. To investigate this issue, we created an online game-playing task and paired subjects to perform the task based on their responses to a short survey about demographics and interests. We studied how these factors influence task outcomes, the interaction process, and attitudes towards one's partner. We found that demographic similarity affected how people interact with each other, even though this information was not explicit, while similarities or differences in task-relevant interests did not. Our findings suggest guidelines for developing social recommender systems and show the need for further research into conditions that will help such systems succeed.