Overview



    RATIONALE & AIMS

    Designers in industrial and commercial contexts have to balance a desire for innovation against constraints of limited resources. Re-use of ideas and solutions from previous projects can cut costs and design time; also innovative ideas that were ahead of their time can be reconsidered as circumstances change. However, inappropriate re-use can actively stifle innovative design. This project, funded by the ESRC Cognitive Engineering initiative, has investigated re-use in highly innovative design environments, and has developed Desperado, a tool to support this process. The research approach was to collect data, using a novel mixture of naturalistic and controlled studies, concerning the ways in which information is re-used by expert designers working on major engineering and commercial projects.

    ACTIVITIES

    Three kinds of study were carried out. The first adopted a 'cognitive ethnography' approach (see Ball & Ormerod, in press), in which we carried out longitudinal studies of information documentation, retrieval and re-use in four major industrial design settings. The second involved using design episodes gathered in the ethnographic studies as materials for a card sort task, in which designers sorted design activities into natural categories for documentation and retrieval (see, Ormerod, Rummer, & Ball, in press). The third involved analysing protocols of designers tackling short but realistic design briefs. In these studies, we compared the design processes of both commercial and academic designers, working without support, with Desperado as an encoding tool only, and with Desperado as a complete encoding and retrieval environment.

    On the basis of the first two sets of studies, we developed Desperado (see Ormerod et al, 1999; Ball et al, 1999). Two versions were developed: Desperado I, a standalone Macintosh application written in C++, that accesses an object-oriented database, and Desperado II, a web-based interface for a Filemaker Pro™ database, that runs on PC platforms. The system uses a prioritisation interpreter that acts on behalf of the designer as an encoding and retrieval agent, which makes decisions about what labels to supply during the encoding process and what episodes to suggest for retrieval during ongoing design. At the same time the designer is stepped through the elicitation of a limited design rationale. In this way, Desperado embodies an agency metaphor based around the notion of a 'surrogate manager'.

    OUTCOMES

    Evaluation of Desperado is currently in progress, but early results are encouraging. In a study of Desperado in encoding mode only, we have found that the start-up times for using the system are short (around 10 minutes), and the amount of permanent documentation increases whilst the amount of writing overall decreases. More importantly, designers using Desperado appear to make as much progress in tackling the brief as the control group, while generating more solution options and critically evaluating them more systematically than in the control condition. Interestingly, we see a large amount of re-use of ideas from previous episodes, even though the system is in encoding mode only, because designers use the prioritised labels supplied by the system as guides to interesting options and important criteria under which to evaluate their ongoing work.

    Last revision: 15-6-99
    Comments welcome.
    J.Mariani@lancaster.ac.uk