Approaches to Solving the Capture Problem

Traditionally, DR literature has emphasized that devising the right schema, i.e., one that captures the right information and structures it correctly, is the way to solve the problems of DRR usage. Yet designers’ resistance to DR capture exists regardless of what schema is used. Solving the capture problem will require research on more than schema design.

One direction taken by researchers working on solving the capture problem is to try to reduce the intrusiveness of DR capture, either by reducing the work of DR capture or reducing its disruptiveness in design or both. The MIKROPLIS [39, 44] and gIBIS [8] hypertext systems reduced the work of managing DR by providing extensive support for browsing, modification, and retrieval. This, by itself, however, was not enough. The cognitive overhead of DRR capture remained daunting.

One approach to reducing the cognitive overhead of capture is to use the strategy of differential description, in which designers only need to describe how the rationale for the current project differs from other rationale. One way to do this uses domain-oriented issue bases in PHI [18]. These contain rationale commonly used in projects in a given domain, including commonly raised issues, positions and arguments. Designers need to add only the missing information, including their decisions on the issues.

There are other ways in which differential description might be implemented. One would be by using rationale-annotated cases of similar projects, such as those provided by the ARCHIE system [29]. Another way might be to use design patterns annotated with rationale.

Of course, differential description only works for domains where previous design work has been done and where someone has built collections of issue-based discussion, precedent cases, or design patterns. By definition, this approach is not useful for unprecedented problems. It should also be noted here that Rittel’s theory of wicked problems, which led to the first DR method, included the notion that design problems are “essentially unique,” and thus not easily solved by looking to precedents [54].

A number of researchers have explored ways of capturing DR without use of any schema, either because schemas are too labor intensive to use or because they interfere cognitively with capture. For example, Shipman and his collaborators from Xerox PARC built “spatial hypertext systems” [61] that enable informal input of information in a 2D space and then infer the structure of that information from its spatial arrangement, work inspired in part by gIBIS’s graphical representation of IBIS structure. Reeves [52] also created a system that uses a schema-free approach to capture. With his system designers write their rationale as textual notes in the graphical representation of a physical artifact in a CAD system. The design history of the artifact then becomes the means by which rationale is structured. A different schema-free and completely nonintrusive approach is used by Myers et al. [47]. They add semantic information to a CAD system’s symbol library and then infer the DR from the designer’s use of the system. This approach, however, does not produce argumentation as such.

The idea of abandoning use of an explicit schema is controversial in the DR field. On one side of the debate, there are MacLean et al. [38] arguing for intrusive, schema-based approach to DR capture. At the opposite end are Shipman and Marshall [57, 58, 59] arguing for nonintrusive approaches that abandon use of schemas.