CSCW-2010 Workshop on Collective Intelligence in Organizations: Towards a Research Agenda February 6, 2010. Savannah, GA, USA 1 Assessing Group Performance from Collective Behavior Daniel Olguín Olguín and Alex (Sandy) Pentland MIT Media Laboratory Human Dynamics Group {dolguin, pentland}@media.mit.edu Introduction The Human Dynamics research group at the MIT Media Laboratory has demonstrated that wearable technology can be used to characterize face-to-face interactions, measure individual and collective patterns of human behavior, and automatically map out a company's de facto organizational chart (Choudhury & Pentland, 2003; Pentland, 2006; Olguin-Olguin et al., 2009a, 2009b). This capability can be an extraordinary resource for studying group behavior, group performance and team formation processes. With that goal in mind we developed the Sociometric badges, wearable electronic sensors capable of detecting face-to-face interactions, conversations, body movement, and proximity to others (Olguin- Olguin, 2007). The Sociometric badges are capable of extracting speech features without recording the content of conversations in order to maintain privacy, and of wirelessly transferring data to a central server. We have used them in several organizations to capture face-to-face communication patterns and study the relationship between collective behavior and performance outcomes, such as productivity and job satisfaction (Olguin-Olguin et al., 2009a, 2009b; Wu et al., 2008). The design of the Sociometric badges was motivated by the fact that a large number of organizations already require employees to wear RFID name tags that identify them and grant them access to several locations and resources. These traditional RFID name tags are usually worn around the neck or Đlipped to the user’s ĐlothiŶg. With the rapid ŵiŶiaturizatioŶ of eleĐtroŶiĐs, it is Ŷoǁ possiďle to augment RFID badges with more sensors and computational power that allow us to capture human behavior without requiring any additional effort oŶ the user’s side. By capturing individual and collective patterns of human behavior with Sociometric badges and correlating these behaviors with individual and group performance, it is possible to identify successful vs. unsuccessful teams, high performing teams, and predict group outcomes. The added value for the users is the feedback that they can receive about their daily behaviors and interactions with others, and how these behaviors affect their individual and group performance. Collective intelligence is sometimes defined as the ability of a group to solve problems more effectively than any of its individual members (Heylighen, 1999). The term is also used to describe several web tools aimed at improving group performance, such as wikis, social networking sites, and other software programs that facilitate group collaboration. Sociometric badges are measurement tools that faĐilitate the studLJ of ĐolleĐtiǀe ďehaǀior aŶd help orgaŶizatioŶs ŵadžiŵize their groups’ collective intelligence through specialized software that analyzes behavioral patterns and generates automatic feedback reports and dynamic visualizations. We can design organizational interventions based on these measurements and feedback mechanisms.