Issue
Article
Vol.28 No.1, January 1996
Article
Issue
Issue |
Article |
Vol.28 No.1, January 1996 |
Article |
Issue |
DIS was the first in what is hoped to be series of conferences focusing on the design process of interactive systems. The conference was started by a group from the CHI community who wanted to see more focus on design and had three cooperating sponsors: IFIP Working Group 13.2, the Association for Software Design, and ACM SIGCHI. Attendance was approximately 125 from USA, Europe and Japan. Many new conferences go through an identity crisis and DIS is no exception. The main objective was to shape the vision of what design is and how it can be improved. The papers did not all directly reflect the design theme, but that is probably a `chicken and egg' problem of needing papers to define what design is before a conception of design exists. Anyway, there is my personal reflection on some of the papers and sessions.
The opening plenary presentation by Henry Petroski set out an interesting perspective of engineering design. Taking the humble paper clip and pencil, Petroski showed how design is often sub-optimal, a response to constraints and economics of the real world and evolves over time. How the design compromises and variations for something as apparently simple as a paper clip, scale up to complex software systems is a key issue, however, Petroski's presentation stopped short of covering complex artifacts. The main lessons were taking constraints into account and learning from design histories. In civil engineering there is a 30 year cycle of design evolution to extinction (e.g. a type of bridge design is pushed to its limits then another design paradigm takes over).
Wednesday's sessions covered evolutionary design, experience and requirements, and organization concerns. The session titles were often misleading and came in for criticism by discussants. The discussant role worked well throughout the conference and developed many interesting perspectives on the papers. On the first session, Tom Moran illustrated how design varied by its scale and intent. As discussant, Tom compared the quite different views of design presented in the first three papers. In the first of these Jack Carroll presented a framework within which others can then design for themselves in an electronic village which gives people a means of interacting and authoring material for a local Web community. This differed from the design view in the paper by Fischer, Nakakoji and Ostwald which described designing an environment which allows people to design classes of applications-domain oriented design environments. The third paper discussed design as accommodation to constraints of existing technology as Catalina Danis talked on how to fit speech technology into work contexts.
This session was followed by an `experience' session reporting use of the ADEPT toolkit for task oriented design by Stephanie Wilson, QMW, integration of techniques for requirements and analysis by myself and customization of HCI guidelines within the commercial design process from Scott Henninger.
Communication and group working in design were featured in an `organizational concerns session', with papers Volker Wulf (Univ of Bonn) looking at organization models of the design process and papers from NYNEX (Alison Lee) and Colorado (Jonathan Ostwald) describing annotation tools for facilitating user-designer feedback, and how `suggesting' agents can help the design process by providing intelligent assistance with mixed initiative user system interaction. In summarizing this session, John Karat pointed out that "failure in communication" seemed to be being raised as a central problem in design.
Thursday started with more on collaborative aspects of design in multi-disciplinary teams. The papers covered 1) reflections on applying HCI methods in software design from Jonas Löwgren (one of the few papers reflecting the Scandinavian experience of design), 2) a neat framework for analyzing how design issues connect across different parts of the organization and different components of the designed artifact from Catherine Burns in Toronto, and 3) a comparative study of industrial and engineering design from William Holmes from Coventry, UK.
The next session, design in context, fell into the misnomer category. Anne Rose, Univ. of Maryland gave a report on ethnographic analysis for redesign of a criminal records system and found, as many others have, that ethnographers give little practical advice for designers/analysts in our community. Another Colorado paper followed on extending scientific visualization modelling tools by DM-macro like programming. This session finished with Keith Butler from Boeing showing how UIDEs really work when you coupled them with a reuse library for the application functionality.
The afternoon panel brought education into the picture with Jack Carroll, Gerhard Fischer, Jim Haines and Elliot Soloway giving different pictures of what design for learning should be about and how traditional design conceptions for teaching/instruction need a radical overhaul. The panel also raised issues of whether learning is separate from doing, a theme dear to Gerhard's heart and present in many Colorado papers. Can learning be seamless with doing via learning centred design? However, as Ruven Brooks pointed out, you don't want the pilot of jet aircraft to be learning on the job. Interesting issues about how design and learning interact in creativity remain, as well as how overtly we should design to encourage user learning as well as just helping users do their job.
The final session was mistitled CSCW, in reality it was about empirical analysis of design meetings using speech act techniques for conversation/content analysis by Antonio Maia, and gesture analysis by Mathilde Bekker. Wendy Kellogg was an amusing discussant questioning the relevance of the papers in providing real insight for improving design; a fair point, but nevertheless interesting scientific insight can help build new design conceptions as Bekker argued in preserving gesture communication within CSCW tools.
Friday started with "design environments". The first (Inference (Grizzly) Bear) was in the UIDE/programming by example tradition and attempted to automate design by inferring dialogue from user interaction patterns. This was followed by two more Colorado papers on themes of reusing analogically matched designs (Kurt Schneider), and providing explanations embedded within operation of drawing tools could help user learning (Chris DiGiano).
The Formal methods session was misnamed or not depending on which side of the Atlantic you came from. From the European view Greg Abowd's talk was the only formal method, demonstrating how formal notations coupled with a FSM prover engine can actually deliver scaled up analysis of dialogues. The other papers were by Gerrit van der Veer looking at evolution and fit between HCI design methods; and an object-oriented method for Interactive systems from Birgit Kneer.
The final paper session, on which I did the discussant slot, gave three different perspectives on scenarios. Colin Potts used schematic scenarios and goals to improve requirements, rather like test data for programmes, Colin proposed a method using test situations for elaborating user requirements. Following this, Herman Kiandl's paper described a hypertext tool which linked goals to early descriptions of user behavior and system functions, while in Traci Royer's paper scenarios were dialogue walkthrough for new versions of an existing system and used by the design team to debug problems. Traci's paper was the one of the few from "real" industry, which reflects the ongoing problem of getting industry people into `academic' conferences. Of course other companies were well represented, IBM, Apple, NYNEX, Boeing to name but a few, but these people are inherited from CHI.
The final plenary shifted focus to design in a different domain, in this case mega-design of cities by Ernesto Arias from Colorado. Ernesto started by giving a vision of what design-in-the large really means: try the new capital city of Nigeria for size. The main focus for design on this scale becomes setting the parameters and framework within which others will do smaller scale design. Layers of analysis and modelling are necessary to choose optimal land use, zoning, placing key buildings, streets etc. Ernesto then shifted to simulation environments and engaging people in the design process by providing game-like kits for end users (in his case people in a Denver neighborhood) to try out design for redeveloping their area. This led to considerable debate about the merits of top down versus bottom up design and user participation. Ernesto seemed to be not too happy with the mega-design projects of the past and more in tune with his collaborative design projects. Nothing like learning by experience.
In summary DIS was a stimulating and well organized event (credits Allan Maclean, Gary Olson and Sue Schon). It may not have produced exactly what the organizers expected but, as it turned out in review meetings, they didn't have a precise picture of what they wanted either. The debate over what DIS is continues. Using plenary speakers to give a perspective of design from other communities worked well, and is probably the best way to recruit such experience rather than trying to make the conference appeal to people from graphics, engineering, architecture, etc. My reflections of what was missing are that not much appeared about supporting the creative aspect of design; indeed, what is the nature of creativity. Also design for work context seemed to take a back seat; we need to know more about how to design systems to help users achieve their work goals, and explore the dimension of design for task fit to enabling new ways of using technology.
DIS will probably continue with a conference in the Netherlands in August 1997, so watch for this event.
Alistair Sutcliffe (A.G.Sutcliffe@city.ac.uk)
Issue |
Article |
Vol.28 No.1, January 1996 |
Article |
Issue |