Footprints : interaction history for digital objects

Abstract

Digital information has no history. When we interact with physical objects, we are able to read the traces left by past interactions with the object. These traces, sometimes called "wear," form a basis for the interaction history of the object. In the physical world, we make use of interaction history to help come up with solutions and guidance. This is not possible in the digital realm, because the traces are missing. This dissertation describes a theoretical framework for talking about interaction history. This framework is related to work in anthropology, ethnomethodology, architecture, and urban planning. The framework describes a space of possible history-rich digital systems and gives properties which can be used to analyze existing systems. The space consists of six properties: proxemic/distemic, active/passive, rate/form of change, degree of permeation, personal/social, and kind of information. We also present an implementation of these ideas in a system called Footprints, a toolset for aiding information foraging on the World Wide Web. Our tools assume that users know what they want but that they need help finding it and help understanding - putting in context - what they have found. Footprints is a social navigation system, designed to show that information from past users can help direct present problem-solvers. We present results from informal use of the tools over the last two years, and from formal surveys and experiments on a controlled task. These experiments showed that people could achieve the same or better results with significantly less effort by using our tools.