Evaluation of 3 Ubicomp systems

Prototypes in the Wild: Lessons from Three Ubicomp Systems by Scott Carter and Jennifer Mankoff, in IEEE Pervasive Computing Journal, October-December 2005 (Vol. 4, No. 4) pp. 51-57. This paper is an account of three ubicomp systems evaluation at multiple design stages:

  • PALplates: to support office workers in doing everyday tasks by presenting key information and services at places of need, or locations where workers were most likely to need them.
  • A nutrition tracking system that uses inexpensive, low-impact sensing to collect data about what household members are purchasing and consuming and then uses simple yet persuasive techniques to suggest potential changes.
  • Hebb: a system which capture and convey shared interests, it senses group members’ interests via email analysis software and displays relationships between members on public and private displays to encourage conversation about those topics. The Hebb system includes interest sensors, presence sensors, and public and private displays

Why do I blog this? For each of these projects, the authors provide a description of how they evaluated them (mostly with paper prototypes first and field experiments with interactive prototypes afterwards). What is strikingly interesting is that their computer science perspective lead them to "a struggle to balance quality of evaluation and ease of prototyping" as they say. This paper is yet another element to attest that testing ubicomp applications in field settings is particularly important. I was also interested by the fact that they studied their prototypes at different stages of design.