Improvising Design - The Book

 Boundary-Spanning Design: How Information Systems Evolve Through Improvisation


Scaling up: a group is not a coherent set of individuals, working together

  • If we extend the model in Figure 1.3, to imagine group design as the convergence between multiple perspectives on socially-constructed, shared problem and solution-spaces, we need to examine the processes that lead to a shared understanding of organizational problems and IS solutions. How does a group manage understandings of the problem-space, the solution-space and the tasks required to move from the problem-space to the solution-space? These processes are seldom explicitly accounted for in design, which may lead to many of the project overruns and requirements definition problems observed in organizational IS (Walz et al. 1993).  To understand the role of shared understanding in this process, we employ the lens of social cognition. Drawing on the social psychology and organizational sensemaking literatures, we can trace two separate but interrelated theories of social cognition: socially-shared cognition and distributed cognition.
    The study of the processes by which human beings individually and collectively interpret, bound and make sense of their world originated in the fields of cognitive and social psychology. Human beings act according to internal, cognitive structures, variously referred to as schemas (Bartlett 1932; Neisser 1976), personal constructs (Kelly 1955), scripts (Schank and Abelson 1977) or mental models (Gentner and Stevens 1983; Johnson-Laird 1983). Mental models are often viewed as distinct from schemas, personal constructs and scripts in that they are seen as dynamic entities, constituted from the interplay between the cognitive schemas which represent different knowledge domains (Finke et al. 1996). Mental models become more complex, abstract and organized with experience -- this attribute is significant in IS design, where experiential knowledge is valued because of an increased ability for abstraction (Vitalari and Dickson 1983). These concepts from the psychology literature converge, and are extended to the analysis of social behavior, in the notion of a "frame" (Goffman 1974; Tannen 1993). The framing concept operates at the intersection of a psychological-cognitive and a social-behavioral approach to human interaction (Ensink and Sauer 2003). People behave according to "structures of expectation" (Tannen 1993) that guide how they predict and interpret the behavior of others. Individuals provide conversational cues, on the basis of which hearers are able to place the communication within a specific context (MacLachlan and Reid 1994). But an individual cannot contribute to a discourse without displaying their view on the subject matter. Thus, communications are framed simultaneously within a specific, situational context and from an individual perspective (Ensink and Sauer 2003; MacLachlan and Reid 1994; Tannen 1993). Individual frames are not static, but subjected to change during communicative and social interaction (Boland and Tenkasi 1995a; Ensink and Sauer 2003; Eysenck and Keane 1990). Lanzara (1983) views IS design as a dynamic process of framing and reframing problem-situations by a transactive process where different actors negotiate “their perspectives, values and (even!) facts”. Employing a framing perspective allows us to conceptualize how similarities and differences in individual perspectives and understandings guide collective action.

  • Groups of people who regularly work together on shared tasks have been observed to develop a repertoire of shared frames (Davidson 2002; Fiol 1994; Weick 1987). Shared frames provide cognitive "shortcuts" that permit a group to share common interpretations of the organization without the need for complex explanations (Boland and Tenkasi 1995a; Orlikowski and Gash 1994; Weick 1987). For example, IT developers share a vocabulary that is often unintelligible to other workers, but which allows them to communicate and to coordinate their work, using shorthand terms such as “this is a blue screen error”. This type of perspective-sharing requires not only shared knowledge, but also a shared system of norms, behavior and expectations that indicates why a specific term is significant (Brown and Duguid 1991; Lave and Wenger 1991; MacLachlan and Reid 1994). Shared framing is therefore embedded within a local system of socio-cultural values that make sense of “how we do things here” (Cook and Brown 1999; Lave and Wenger 1991; MacLachlan and Reid 1994):
      " Knowledge and understanding (in both the cognitive and linguistic senses) do not result from formal operations on mental representations of an objectively existing world. Rather, they arise from the individual's committed participation in mutually oriented patterns of behavior that are embedded in a socially shared background of concerns, actions, and beliefs."  (Winograd and Flores 1986, page 78) .
  • The ‘seminal’ study of shared framing in the IS literature was performed by Orlikowski and Gash (1994), who analyzed the degree of congruence between the shared “technological frames” held by members of technical vs. non-technical stakeholder groups in the adoption of Lotus Notes. They performed a qualitative analysis of key themes and terminology used in stakeholder interviews, to categorize various domains of technology-application that were associated with the frames of reference employed by members of the two groups to interpret the new technology and how it was implemented. They concluded that conflicts and difficulties may arise in technical change initiatives where members of key stakeholder groups hold "technological frames" that are not sufficiently congruent. Frame congruence does not imply that frames are identical, but that they are related in structure (possessing common categories of frames) and content (with similar values in the common categories) (Orlikowski and Gash 1994). But defining shared content depends upon the way in which the framing concept is itself  defined: we need to examine what-is-shared, to understand the degree of frame congruence (Cannon-Bowers and Salas 2001). Relating this to the convergence model of design, what-is-shared might usefully be indicated by an understanding of (i) the problem-space, (ii) the solution-space, and (iii) the design goals and task-definitions that arise from a gap-analysis between elements (i) and (ii).
  • If we examine studies of how design is framed, we find that the few studies that exist would seem to indicate that design depends upon shared understanding. Members of IS design groups, “successful in sharing plans and goals, create an environment in which efficient communication can occur” (Flor and Hutchins 1991, page 54). Only one study examines changes in design framing over time. Davidson (2002) studied a group engaged in the specification and design of an organizational information system. Through a thematic analysis of her data, she categorized various frame "domains" that resulted in a specific focus, excluding some design elements or issues from consideration and including others. In other words, adoption of a specific frame domain provided a conceptual boundary, or filter, to group discourse. Davidson found that different frame domains became salient to the group at different points in the process, resulting in the adoption of a different strategy towards the IS design. At times when the business value of IT frame-domain dominated group discourse, this led to radical reconsideration of project requirements. At times when the IT delivery strategy frame-domain dominated group discourse, the group reverted to a more conservative definition of requirements, consistent with the perceived need to deliver a known product. Tensions between the assumptions underlying each of these frame domains led to much of the instability in IS design group members' understandings and agreement of the requirements for a new system. In terms of what-is-shared, the frame domains identified relate to the intersection of task-related (experiential) knowledge and knowledge related to attitudes and beliefs (c.f. Cannon-Bowers and Salas 2001). Changes in the group's dominant frame domain appeared to be triggered or accompanied by the adoption of a new group metaphor for the rationale behind the current design strategy. But we cannot assume shared frames just because group members share a similar culture (Krauss and Fussell 1991) and we cannot assume the existence of a shared culture among members of a recently-established boundary-spanning design group, as groups with new members have diverse cultural values (Lave and Wenger 1991; Moreland et al. 1996). Hence we need to explore what is shared in design. By employing the concept of frame congruence, we can explore the extent to which definitions of a local problem-space or solution-space are shared by various group members.

  • There is a trade-off between shared understanding and the exploration of design alternatives. Group members who share common perspectives communicate more effectively. But in groups that have a high degree of shared understanding, there may be less of a tendency to explore alternative courses of action (Flor and Hutchins 1991; Mennecke and Valacich 1998; Wilson and Canter 1993). As discussed above, members of a design team derive from multiple and intersecting social worlds. The definition of novel business processes, work-procedures and IS roles is a major challenge for boundary-spanning groups. In traditional work groups, there are experts on which the group may rely for guidance, whereas in the design of novel organizational information systems, perceptions of expertise are subjective and negotiated: there is a "symmetry of ignorance" (Rittel 1972b). A diversity of perspectives is necessary for a sufficiently wide number of design alternatives to be considered in boundary-spanning design, but this conflicts with the need for intersubjective (cognitively-shared) understandings of potential outcomes (Flor and Hutchins 1991). A wide spread of experience must be expected to cause problems of group cohesion and productivity (Krasner et al. 1987b).

    • " Humans derive their (social) essences from the communities in which they are located, and human communities are contingent on the development of shared (or intersubjectively acknowledged) symbols or languages." (Prus 1991, page 10).
  • There is also the issue of how much we may expect stakeholders from radically different knowledge domains, with different educational backgrounds, to understand organizational problems and potential solutions in common. Established workgroups develop an understanding of who knows what, that allows them to operate with an understanding of others' tasks and the division of collective work (Hutchins 1991; Moreland et al. 1996; Weick and Roberts 1993). But the coordination of organizational expertise in newly-established groups is complex and difficult, especially in groups that span organizational boundaries. People rarely know who knows what in large organizations (Carlile 2002; Cramton 2001; Pfeffer and Sutton 2000). In established work groups, there are experts on which the group may rely for guidance, who interpret such embedded knowledge. But in the design of novel organizational information systems, expertise is subjective and negotiated. All team members may possess an equally strong claim to both expertise and ignorance of relevant knowledge.
    The concept of distributed cognition (Hutchins 1991) provides an explanation of how collaborative groups manage the division of labor across multiple knowledge domains, where they do not share common expertise. Weick (1987) discusses how teams performing collaborative tasks require a requisite variety of perspectives, to detect all of the significant environmental factors affecting collective decisions. But this is balanced by the need for a homogeneity of culture, within which team members can trust and interpret information from other team members. A wide spread of experience must be expected to cause problems of group cohesion and productivity (Krasner et al. 1987a; Orlikowski and Gash 1994). Understanding within a collaborative design team is distributed: each individual can comprehend only a part of how the target system of human activities operates. Groups appear to coordinate work most effectively if individuals have overlapping task knowledge (Hutchins 1991; Laukkanen 1994; Weick and Roberts 1993). However, a core tenet of the theory of distributed cognition is that there is a lack of overall congruence between how individuals frame organizational work. Understanding is not so much shared between, as "stretched over" members of a cooperative group, in distributed collaborative work (Star 1989). This provides an alternative to the assumption of shared knowledge, but it also raises the problem of coordinating knowledge across multiple work-domains, when individuals only possess a partial knowledge of others' work. Collective understanding is located in the process of "heedful interrelating":  this process takes place between individuals, not the group as a whole (Weick and Roberts 1993). A study of software development teams performed by Faraj and Sproull (2000) indicated that the effective management of distributed cognition was more significant than domain expertise in ensuring team effectiveness.

In operationalizing the concept of distributed knowledge, we note that heedful interactions between individuals in collaborative work are mediated by “boundary objects” (Star 1989). These are physical artifacts, maps and models that represent an incomplete representation of the superset of domain knowledge across various actors in cooperative work. Individuals are able to collaborate with others through the areas of interdependency, or overlap in their knowledge of what needs to be done and why, by ascribing a shared meaning to a boundary object. Boundary objects provide a sufficiently vague (global) representation of domain knowledge that they can be adapted to individual, local needs and constraints. For example, the topographical map of the London Underground train system does not represent a detailed model of the locations and distances between stations. But it is sufficiently vague that it can be used to coordinate knowledge about what to do (“how do I get from here to there?”),  ease of task (“do I have to change trains to get there?”), and travel costs (“which stations are in which travel zone?”). So it can coordinate collaboration between London Underground train drivers, platform guards, experienced travelers, tourists, ticket sales staff and ticket collectors, even when those individuals do not share the same knowledge about the elements that comprise the London Underground train system. These physical representations or external products of human interaction often contain a shared understanding that is not possessed individually by the people who produced them (Hutchins 1991; Star 1989; Weick and Roberts 1993). Thus, "shared" understanding is often not explicit, but communicated through representations of work and its context, that represent implicit and partial "maps" of what needs to be achieved (Hutchins 1991; Norman 1991; Schmidt 1997; Star 1989; 1998). We may therefore examine external representations produced through collaborative work, to understand the sum of the group knowledge: the union of individual design-frames, as distinct from the intersections that represent shared frames.

arrow Bibliography

All content is copyright © 2009-2010, Susan Gasson, Improvising Design. No content may be reused, duplicated in any form, or disseminated except with the express permission of the author.

Panorama theme by Themocracy