Paul Dourish
Apple Computer, Inc.
Current address: Xerox Palo Alto Research Center, Palo Alto,
California.
dourish@parc.xerox.com
Position paper for the CHI'97 Workshop on Awareness in Collaborative Systems (Atlanta, Georgia, March 1997).
The provision of "awareness" information has become an active area of research in systems supporting collaborative activity. However, the primary focus of these investigations has been support for synchronous collaborative activity.
This position paper lays out some fundamental properties and motivations for the development of awareness services, and proposes extending the notion of awareness beyond the area of synchronous interaction. In particular, it relates traditional models of awareness both to asynchronous collaboration and to single-user interactive systems.
The Polyscope system developed by Alan Borning and Mike Travers [1991] at EuroPARC in 1990 was the first system to explicitly address the issue of how computer systems could support passive awareness within a workgroup. Polyscope was motivated by the centrality of awareness in use of the EuroPARC "RAVE" media space. In 1990/1, its successor, Portholes [Dourish and Bly, 1992], focussed on awareness in distributed working groups. Portholes provided not only the coordination benefits of awareness information, but also a means to foster a sense of community in groups who did not share physical space. Portholes-like systems have since been developed in other media space environments (e.g. UToronto's Postcards, UCalgary's Peepholes, Bellcore's Gallery, DEC's Argohalls and NYNEX Portholes).
Polyscope and Portholes provided awareness information independent of tasks. "Workspace awareness" has also been explored in shared workspaces such as collaborative editors. Studying shared text editing, Dourish and Bellotti [1992] show how sharing of information through a "shared feedback" approach allowed group members to flexibly and fluidly coordinate their work. Providing and visualising awareness information is an active CSCW research area [e.g. Gutwin et al, 1996].
I want to argue that we should take a wider perspective on awareness. In particular, I will discuss how awareness information can be used in asynchronous collaborative systems and even in single-user interactive systems. To do this, I will first consider the function of awareness information.
Dourish and Bellotti's study points to just the same sorts of coordination in distributed settings. Shared feedback renders others' work visible, and allows individuals to gauge the progress of their own work in relation that of the group.
The passive nature of this information is important. Information arises directly out of each person's activity, rather than having to be managed explicitly; awareness information does not need to be sought out. This is similar, perhaps, to the way in which information visualisation systems seek to move the load of interpreting information from the cognitive to the perceptual systems. In much the same way, collaborative awareness systems allow one to perceive the activities of others.
The motivations for asynchronous and synchronous awareness are the same. Portholes was inspired by the way in which individuals sharing a physical space use cues in that space to build an understanding of what's going on around them; and similarly, studies of copresent artifact-based collaboration show how people orient towards each other's action in managing their own activity. Asynchronously, we can see how people use cues for understanding activity they don't directly witness either in physical space (imagine walking into someone's office and seeing signs of recent occupancy) or through artifacts (such as marked up documents).
One way to support asynchronous as well as synchronous awareness information is to place it within a collaborative frame which does not, itself, distinguish between synchronous and asynchronous work. An example of this approach is the Prospero collaboration toolkit [Dourish, 1996], which characterises interaction as a repeated cycle of divergence and synchronisation. This cycle operates uniformally over timescales ranging from seconds (similar to "synchronous" approaches) to weeks (similar to "asynchronous").
Even without this, though, awareness information can be incorporated into purely asynchronous collaboration systems. A key research issue is the sorts of visualisations that can be effectively provided and exploited in these situations. One approach is to place the awareness information within the shared workspace itself. This approach is used by some synchronous awareness systems, but is especially effective in asynchronous systems for two reasons. First, the shared artifact is, essentially, the only shared space available to the different participants; and second, the key information required in asynchronous collaboration is information about activity to the artifact, not information about activity performed by others (in other words, the artifact is the focus).
One highly compelling example is Hill et al's [1992] "edit wear". Their metaphor is a document which "wears out" as it is used. One side of the document window contains indications of read and write activity. Any activity "wears away" a part of the document window, placing a mark in the margin. As marks accumulate, the window edge takes on a ragged appearance which gives an immediate visual cue to activity over the document or artifact.
My current research looks at systems which can give accounts of their own behaviour [Dourish, 1995]. While mental models research concerns the understandings which people develop of systems they use, we are looking at models which systems present themselves, explicitly and implicitly. Unlike explanation-based systems, the focus is not on accounts of the system which exist outside (as "commentaries" upon it), but ones which arise from the system itself in the course of its activity (and frequently are manifest in how the system does what it does). Feedback provides awareness of the system's response to the user's action; we are concerned with visualising aspects of the system's own behaviour.
Although this sort of single-user awareness is qualitatively different from the sorts of awareness support which we are familiar with in collaborative systems, it is drawn from precisely the same theoretical accounts of coordination behaviour. The theoretical underpinnings of this work lie in ethnomethodology [Button and Dourish, 1996]. Ethnomethodology's analytic frame is based on the accountability of human social action, which refers to the how ordinary social action is organised so as to be seen or read as meaningful to others. Similarly, I am looking at taking the information which systems give of their own operation, as part and parcel of what they do and how they do it, and making it meaningful in terms what what's going on beneath the surface. This is essentially a new architectural approach to the design of interactive systems, but which builds upon exactly the same mechanisms at work in traditional collaborative awareness systems.
The provision of awareness information isn't just a neat hack; it's theoretically motivated, and those motivations also point to new opportunities. Although awareness support has largely been explored in the context of synchronous collaborative systems, the same factors which motivate its support also argue for the support of awareness information in other areas. I have argued that it is time to explore awareness information more generally. In particular, I have discussed avenues for the exploration of awareness information in two other domains--in asynchronous collaboration, and in single-user interactive systems.
References
Biographical Information
Paul Dourish is a senior research scientist at Apple Research Labs in Cupertino, California. He is investigating software architectures for interactive systems which support fluid, improvised activity. Before coming to Apple, he worked at the Rank Xerox Research Centre (EuroPARC), where he worked on a variety of projects in collaboration and multimedia communication. In 1990-1, along with Sara Bly, he developed the Portholes distributed awareness service as part of the RAVE media space. His research interests include metalevel software architectures, collaborative work and the relationship between social and design sciences.
Notes
[1] In fact, I think the distinction between synchronous and asynchronous systems is essentially spurious, and simply an artifact of the architectural approaches we have become accustomed to using. Real work moves freeling between different degrees of synchrony and coordination.