A Formal Model of Collaborative Audience Interfaces

Anecdotal experience with collaborative audience interfaces such as Cinematrix suggest that such systems perform surprisingly well: untrained audiences rapidly self-organize to play games and follow mazes by controlling motion on a shared screen using individual plastic paddles and camera-based color detectors. We are developing a formal model to study such interfaces. In this model, each audience member is represented by a mathematically defined agent. We define ideal agents as well as a variety of abnormal agents that simulate dropouts, randomness, time-delay, and malicious behavior. A set (audience) of such agents takes as input the state of a shared screen and generates responses that are then aggregated to control a global cursor. When the goal is to move the cursor in a circular trajectory, we were surprised to discover that global performance is improved in the presence of abnormal agents. This quantitative model is consistent with the anecdotal evidence and suggests design guidelines for group interfaces where users interact over the Internet.

[1]  Kenneth Y. Goldberg,et al.  Collaborative teleoperation via the Internet , 2000, Proceedings 2000 ICRA. Millennium Conference. IEEE International Conference on Robotics and Automation. Symposia Proceedings (Cat. No.00CH37065).

[2]  N. L. Johnson,et al.  Collective Problem Solving: Functionality beyond the Individual , 1998 .

[3]  Allison Druin,et al.  Single display groupware: a model for co-present collaboration , 1999, CHI '99.