David Millen

Energy Mapping: Incorporating Emotion into the Design of Collaboration Tools

Biography
David R Millen is a group manager in the Collaborative User Experience group at IBM T J Watson Research in Cambridge, MA. His group studies the social and technological implications of on-line communities and large-scale collaboration. Through field study and prototype applications, his group explores how to create and support distributed teams and on-line communities. Many of these applications include interactive visualizations to enhance the user experience and help make large amounts of information accessible to the community. Prior to joining IBM, David worked at AT&T Labs, where he explored how new technologies changed employee work activities, organizational roles, and patterns of communication.

Position Paper
Energy is an important but invisible variable mediating many human-human interactions. We all have the sense of wanting to spend more time with those people who seem to exude a positive energy. Recently, social scientists (Cross and Parker, 2004) using methods of Social Network Analysis (SNA) have provided evidence that people identified by others as high energy were also associated with higher levels of performance.

In this paper, we investigate the energy patterns within a large organization from a social network perspective. Our goals are twofold. First, we want to understand how this approach can offer insights into social and organizational behavior. And second, we want to understand whether the social network perspective is useful in informing interaction design.

When we talk of incorporating emotion into design we are appealing to the need to acknowledge a set of powerful yet invisible elements which seem, sometimes irrationally, to influence not only our attitudes but also our behavior. This paper suggests that individual “energy” is one of those invisible elements which can have a real, measurable impact on how we collaborate with other people and the outcome of that collaboration.

The implications for organizational design are clear—seek opportunities to bring those people who create energy into roles and responsibilities that have strong performance outcomes. Based on our experience, we believe that SNA methods provide a powerful way to quickly assess individual, group and overall organizational relationships. We have observed how this kind of organizational insight can be used to facilitate better decisions for restructuring (or fine tuning) the organization, and to identify areas for training and development.

We also believe that social network methods are important for interaction design. One intriguing idea might be to annotate communication tools such as email, with energy or heat maps of the organization. This idea is not so far fetched. Some work is already underway. Email has been mined for non-affective communication patterns to provide a “spectroscopy” of an organization (Tyler, Wilkinson, & Huberman, 2003). Others have mined such patterns to explore novel interfaces for email (Fisher & Dourish, 2004; Whittaker, Jones, & Terveen, 2000). It is important to note that the raw communication data, for example link structures, are often available as metadata and relatively easy to mine.

Socio-emotional network information, on the other hand, is less readily available. User ratings of the affective dimensions of both people and content are becoming increasingly common. For example, moderators of Slashdot forums (Slashdot.org) rate postings on a five point scale, which also includes a verbal label such as: funny or overrated. Buyers on ebay (www.ebay.com) rate sellers on a positive, neutral, negative scale, along with free form comments. This reputation system is one attempt to measure (and monitor) trust among buyers and sellers. Not surprisingly, HCI researchers have shown great interest in understanding the measurement and user experience of both sites (Lampe & Resnick, 2004; Resnick, Zeckhauser, Swanson, & Lockwood, 2002).

Several important questions remain for further research. Further work should be done to ensure the validity and reliability of online ratings of affect? It would be potentially useful to discover behavioral markers of socio-emotional relationships that would allow machine mining of these relationships? For example, can an organizational spectroscopy of socio-emotional measures be mined from everyday tool such as email or chat? And always, there needs to be serious discussion about the ethical and privacy implications of the measurement and use of these kinds of data.

Back to Social Computing Symposium 2005