Automation in Construction 9 Ž2000. 409–415 www.elsevier.comrlocaterautcon Collaborative design: what is it? Thomas Kvan ) Department of Architecture, UniÕersity of Hong Kong, Pokfulam Road, Hong Kong, China Abstract Collaborative activities are an important application of computer technology now that telecommunications infrastructure has been established to support it. There are many students in schools of architecture who are undertaking collaborative projects using the Internet and many practices who work together exchanging files and interacting on shared digital models. Software vendors are developing tools to support such collaboration. But what are we doing? What is the nature of collaboration and what are the implications for tools that support this work? q 2000 Elsevier Science B.V. All rights reserved. Keywords: Collaboration; CSCW; Computer-supported collaborative design 1. Introduction It has become accepted practice to use the term ‘‘collaborative systems’’ to describe the computer systems which support distal communication between designers. In the field of design, we are beginning to see the term computer-supported collaboratiÕe design used increasingly often. Although the term normally used is computer-supported collaborative work, a similar word, ‘‘co-operation’’, appears to be used interchangeably with ‘‘collaboration’’, as seen in the journal name ComputerSupported CooperatiÕe Work or the paper title Exploring communication in collaboratiÕe design: co-operatiÕe architectural modelling w19x. In discussions about computer systems for collaborative design, their behaviours, specifications and implementation, the most fundamental arguments appear to be encountered on the issues of interaction w16,23x. What is the interaction to be provided, both ) E-mail: tkvan@arch.hku.hk interaction with the systems and interactions between participants? Not only do the participants need to share data but, how are they going to communicate with one another? To illustrate the issues and confusions, look at the discussions about the role of video — how important is it, how does it contribute to the interaction between participants and how wide does bandwidth have to be to support it. These issues are raised when considering other aspects of collaborative systems — data exchange; participant interactions with digital models; audio links; among others. Cuff w5x has observed that participants in collaborative design ‘‘did not necessarily participate equally or collaboratively.’’ From our own experience, we can observe that simply working together or talking about the same subjects does not make the act collaborative w25x. What specifically makes these acts of collaboration or not? Is there any importance in the name applied to this field of study ŽCSCW., which is sometimes rendered ‘computer-supported collaborative work’ and at others, ‘computer-supported co-operative work’? 0926-5805r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 5 8 0 5 Ž 9 9 . 0 0 0 2 5 - 4 410 T. KÕanr Automation in Construction 9 (2000) 409–415 2. What’s in a word? What do we mean by ‘collaboration’? Let us consider a variety of situations to explore the realm of activities that fall within our understanding of collaboration. We team together in design settings to take advantage of what Steiner w24x calls process gain. Collaborative success can therefore be said to be achieved when we have accomplished something in a group which could not be accomplished by an individual. Shea and Guzzo w22x identify three facets of a task which determine the success of group effectiveness: task interdependence Žhow closely group members work together., outcome interdependence Žwhether, and how, group performance is rewarded., and potency Žmembers’ belief that the group can be effective.. To be successful, a collaborative project must establish a definition of the team, identify their outcomes, ensure there is a purpose of the collaboration and clarify the interdependencies of the members. If this is to be possible, are there limits to collaboration, either in time, place or size? Can we collaborate if we have too many or too few participants? While Steiner’s work suggests strongly that the maximum number of participants in an effective working group is four, others have found successful collaboration extending to many more. Sudweeks and Rafaeli w26x took part in an extensive and prolonged exchange with over one hundred scientists and concluded that collaboration was possible with large numbers. As noted by Abarbanel et al. w1x, the many thousands of Boeing engineers who worked on the 777 consider themselves collaborators. From these examples, it appears that there are no numerical limits to collaboration, nor physical, when taking place in computer-supported environments. 2.1. Collaboration and co-operation Part of the problem with the term ‘‘collaborative act’’ is that the activities that are undertaken in such acts may vary in intent and degrees of participation, yet be called the same thing. Is the crew of a ship w10x guiding it into port collaborating with the pilot who has come on board as it enters the harbour? Is their working together not one of co-operation? The pilot points out the hazards, the captain issues commands to the crew. Are they collaborating? The roots of the words are, of course, frustratingly similar but they do carry distinctions worth pursuing. Indeed, the Oxford English Dictionary defines collaborate as ‘‘to co-operate, especially in literary, artistic or scientific work’’, deriving from the Latin words col labore, to work along side one another. Collaboration can be thought of as joint problem solving. It means working with others with shared goals for which the team attempt to find solutions that are satisfying to all concerned. Co-operation, as the Oxford English Dictionary tells us, is ‘‘to work together, act in conjunction Ž . . . . to co-operate for Ž . . . . mutual benefit’’ from the Latin co operari, to work with or along side. The dictionary also tells us that co-operation is an older concept Žthe first instance dates from 1616. while collaboration appears in the English language only in 1860, perhaps suggesting that co-operation is a simpler concept than collaboration. The important distinction between the two words is in the creative aspect of working together. In this way, collaboration in design is a substantially different activity than the collaboration shown by ants. Design collaboration requires a higher sense of working together in order to achieve a holistic creative result. It is a far more demanding activity, more difficult to establish and sustain, than simply completing a project as a team. I suspect that we collaborate far less often than we pretend to. From their survey of literature in collaboration, Mattessich and Monsey w15x have defined the words more thoroughly and have drawn a third distinction, that of coordination. Ø Cooperation is characterized by informal relationships that exist without a commonly defined mission, structure or effort. Information is shared as needed and authority is retained by each organization so there is virtually no risk. Resources are separate as are rewards. Ø Coordination is characterized by more formal relationships and understanding of compatible missions. Some planning and division of roles are required, and communication channels are established. Authority still rests with the individual organization, but there is some increased risk to all participants. T. KÕanr Automation in Construction 9 (2000) 409–415 411 Resources are available to participants and rewards are mutually acknowledged. Ø Collaboration connotes a more durable and pervasive relationship. Collaborations being . . . Ža. full commitment to a common mission . . . Authority is determined by the collaborative structure. Risk is much greater . . . As Mattessich and Monsey emphasize, collaboration requires a greater commitment to a common goal than co-operation with an attendant increase in risk. For this to occur, the level of trust must be higher. 3. Designing The act of designing is, to many, an act that involves others. Many practitioners cannot imagine completing an architectural design and saying ‘‘That is mine, and mine alone.’’ For most, the process of designing includes other members of their profession and members of other professions. To some, this is an ideological position — for example, those who hold the belief that the only good design is participatory design. Work such as that by Christopher Alexander w2x has attempted to formalise this approach into a design methodology. On the other ideological side, the Howard Roark image persists w20x. The brazen hero, working in defiance of society or preconceived notions of design delivers to a client a design that should be accepted must be accepted if the client is to be saved from branded an ignominious ignoramus. Discounting the extremes of these two positions as being positions that seldom occur and, if they do, are beyond the realm of normal collaborative pro- Fig. 2. Loosely coupled design process. cess, let us ask ‘‘Is design collaborative or co-operative?’’ What criteria could we use to distinguish between the two? Typically, we think of design as a continuous close-coupled process ŽFig. 1. in which the participants work closely to realise a design. Here, the participants work intensely with one another, observing and understanding each other’s moves, the reasoning behind them and the intentions. At any stage of the design, the observer cannot identify a discrete contribution to the design product from one participant or the other. Experience tells us that much design is in fact loose-coupled ŽFig. 2., with each participant contributing what they can in different domains of expertise at moments when they have the knowledge appropriate to the situation. The participants have been engaged to work together because each has a particular expertise that can be contributed to the solution process. In this situation, we see two or more experts operating in their own domains on a shared problem. The design moves in discrete steps, perhaps not a linearly as set out in the very simplified diagram below, but still in ways that we can see a step and identify what has happened during that step. 3.1. Some attributes of design actiÕities Fig. 1. Close coupled design process. During the design process itself, what is happening? Loose-coupled or close-coupled, how are the participants working together. Design is often thought of as a process that occupies a continuous space of time. If you ask a design team what they were doing, 412 T. KÕanr Automation in Construction 9 (2000) 409–415 the participants typically will not think of a time when they are not designing. They will describe a complex series of decisions, threads which were picked up and dropped, tasks and events which occurred. Typically, they will describe intense and extended periods of time when they worked intensively together to solve the design problem. Challenged to identify discrete tasks, they may tell you about tasks that took hours, days, or weeks. Gero and McNeill w7x have shown that design is in fact a process that consists of a series of distinct events that occupy discrete and measurable periods of time. Most significantly, they have shown that the temporal span of these design events is remarkably short. In one design analysis, they recorded the majority of events taking less than 30 s; another in which the participants were more expert in their field, most event lengths were less than 15 s Žsee Fig. 3.. As the authors note: ‘‘These differences may reflect differences in expertise with the experts moving quickly through the design task or they may reflect differences . . . between the designers.’’ In their paper, Gero and McNeill have shown that design can be seen as a series of discrete activities and that the level of expertise affects the way work is done, being discernible in the temporal pattern of their work. As collaborators come together in design, we can assume that the nature of their activity does not change since collaboration still requires a designer to attend to design as an individual as well as collaborate. Collaboration is probably episodic and cyclical too. This means that design remains a series of discrete steps. Collaborators work together for Fig. 4. A model of design collaboration. moments, then divide up and go their separate ways w11x. The participants act as individual experts addressing design issues from their perspectives. Their expertise may change during a design session as their understanding is supplemented and they learn from their involvement. This analysis is in sympathy with a model of collaborative design ŽFig. 4. postulated in by this Fig. 3. Event lengths during design Žfrom Gero and McNeill w7x.. T. KÕanr Automation in Construction 9 (2000) 409–415 author elsewhere w13,27x, which suggest that collaborative design consists of parallel expert actions, each of short duration, bracketed by joint activity of negotiation and evaluation. Thus the design activity itself is discrete, individual and parallel, not intimately linked. The participants act as individual experts addressing design issues from their perspectives. Their expertise may change during a design session as their understanding is supplemented and they learn from their involvement. We suggest that the cycle time in this model is very small — probably in the realm of seconds, perhaps not even the 15 s that Gero and McNeill have found, although we are still exploring this data. In this model, we can see that the work is in fact co-operative in nature, and that collaboration occurs as negotiation and evaluation, activities observed as essential and present in computer supported collaborative design w17x. This view of the design process is supported further in the categorisation of design collaboration identified by Maher et al. w14x in their experiment in collaborative design, namely: Ø Mutual collaboration, in which the participants are ‘‘busy working with the other’’. Ø Exclusive collaboration, in which the participants ‘‘work on separate parts of the problem, negotiating occasionally by asking advice from the other’’. Ø Dictator collaboration, where the participants decide who is ‘‘in charge’’ and that person leads the process. Indeed, Maher et al. note that the ‘exclusive collaboration’ model is the most effective and the one in which they observed most productive results. Mutual collaboration led to no result at the end of a very busy exchange between the participants, whereas dictator collaboration came to a conclusion as soon as the leader made up her mind. Design collaboration therefore requires a higher sense of working together in order to achieve a creative result than co-operative design. It is a far more demanding activity, more difficult to establish and sustain, than simply completing a project as a team. It should be noted that collaboration does not imply capitulation by individual members nor does it imply decisions by consensus Ža common mistaken assumption, see, for example, Ref. w3x.. But if we look further at collaboration, we see one more issue — compromise. Compromising suggests an expedi- 413 ent settlement that only partially satisfies those involved. It does not dig into underlying problems but rather goes for a superficial arrangement. This indeed is borne out by Dorst w6x with his observation that designers practice ‘‘satisficing’’ very often. In Cross and Cross w4x for example, collaborative designers reach design decisions which are not the best solution but which are adequate. This is not a pejorative to be dismissed, for the basis for the satisfied solution goes beyond superficiality and is often a truly innovative solution. 4. Implications for computer systems Whether we are co-operating or collaborating, we can be designing, but our expectations for the design environment changes if we think we are doing one or the other. A loose-coupled design process requires a very much different set of tools and conditions to be successful than a close-coupled one. Collaboration requires more than machinery and systems to occur. To what extent the systems are necessary for collaboration has yet to be addressed — many working in this field have their vision clouded by the issues of collaboration itself, failing to recognise the broader and far more important issues of systems for cooperative work. This over-riding desire to support collaboration leads to high-bandwidth demands to support close visual linking in order to replicate co-location w8,9,21x. Now that we have over a decade of experiential data, it is becoming apparent that these simulations of co-location do not lead to better work product outcomes w28x. Experimental data suggests that this conclusion is true w12,18,27x. To be provocative, I suggest that most of the time when people think they are working collaboratively they are actually co-operating and, even more important, compromising. And most of the time that is exactly what they should do. Collaboration is time consuming and requires relationship building and is suited to very particular problems that require such close coupling of the design process and its participants. It would be inappropriate to collaborate to accomplish most design tasks, in the strictest meaning of the word. In short, working together, even effectively, is not necessarily collaboration, nor should it be. 414 T. KÕanr Automation in Construction 9 (2000) 409–415 From this review, we see that collaboration is a deeper, more personal synergistic process and the term should be used selectively. Perhaps we should refer to our field as ‘‘co-operative design’’, recognising that the design process itself is one of negotiation, agreement, compromise, satisficing in order to achieve success. We might even talk about ‘‘compromised design’’ at the risk it might imply the wrong thing. I suspect again that we collaborate far less often than we pretend to. We focus on collaboration, failing to distinguish that act from co-operation, diminishing the usefulness of this looser coupling. Most design projects bring teams into a relationship which fits Mattessich and Monsey’s definition of co-operation or co-ordination more closely than it fits their definition of collaboration. Computer systems which support co-operative design can be loose-coupled components which allow for different systems and are reliant on a variety of communication channels. In a pursuit of systems for collaborative work, we risk over-emphasising particular features which may not be needed. For example, synchronicity may be unnecessary for much co-operative work. In place of striving for closer coupling of systems and tasks, we should perhaps pay greater attention to work processes which are needed to create the environment for successful co-operation or collaboration. We cannot expect to create such an environment simply with hardware and software tools. An over-reliance on tools can only lead to less than cost effective use of our resources. References w1x R.M. Abarbanel, E. Brechner, W. McNeeley, Fly thru the Boeing 777, in: M.L. Maher, J.S. Gero, F. Sudweeks ŽEds.., Preprints Formal Aspects of Collaborative CAD, Key Centre of Design Computing, Department of Architectural and Design Science, University of Sydney, Sydney, 1997, pp. 3–9. w2x C. Alexander, A Pattern Language which Generates MultiService Centers, Center for Environmental Structure, Berkeley, CA, 1968. w3x L. Caneparo, Coordinative virtual space for architectural design, in: M. Tan, R. Teh ŽEds.., The Global Design Studio, CAAD Futures ’95, Centre for Advanced Studies in Architecture, National University of Singapore, Singapore, 1995, pp. 739–748. w4x N. Cross, A.C. Cross, Observations of teamwork and social processes in design, Des. Stud. 16 Ž1995. 143–170. w5x D. Cuff, Architecture: The Story of Practice, MIT Press, Cambridge, MA, 1991. w6x K. Dorst, The design problem and its structure, in: N. Cross, H. Christiaans, K. Dorst ŽEds.., Analysing Design Activity, Wiley, Chichester, 1996, pp. 17–34. w7x J.S. Gero, T. McNeill, An approach to the analysis of design protocols, Des. Stud. 19 Ž1988. 21–61. w8x S. Harrison, S.A. Bly, S. Anderson, S.L. Minneman, The media space, in: K.E. Finn, A.J. Sellen, S.B. Wilbur ŽEds.., Video-Mediated Communication, Lawrence Erlbaum Associates, Mahwah, NJ, 1997, pp. 273–300. w9x S. Harrison, S.L. Minneman, Studying collaborative design to build design tools, in: M. Tan, R. Teh ŽEds.., The Global Design Studio, CAAD Futures ’95, Centre for Advanced Studies in Architecture, National University of Singapore, Singapore, 1995, pp. 687–697. w10x E. Hutchins, The technology of team navigation, in: J. Galegher, R.E. Kraut, C. Egido ŽEds.., Intellectual Teamwork: Social and Technological Foundations of Cooperative Work, Lawrence Erlbaum Associates, Hillsdale, NJ, 1990, pp. 191–220. w11x D. Klahr, preview of paper, to appear in Bull. Jpn. Cognit. Sci. Soc., personal e-mail received 18 September, 1998. w12x T. Kvan, A. Vera, R. West, Expert and situated actions in collaborative design, in: P. Siriruchatapong, Z. Lin, J.-P. Barthes ŽEds.., Proceedings of 2nd International Workshop on CSCW in Design, 2nd International Workshop on CSCW in Design, International Academic Publishers, Beijing, 1997, pp. 400–405. w13x T. Kvan, R. West, A. Vera, Tools for a virtual design community, in: M.L. Maher, J.S. Gero, F. Sudweeks ŽEds.., Preprints Formal Aspects of Collaborative CAD, Key Centre of Design Computing, Department of Architectural and Design Science, University of Sydney, Sydney, 1997, pp. 109– 123. w14x M.L. Maher, A. Cicognani, S.J. Simoff, An experimental study of computer mediated collaborative design, whttp:rrwww.arch.usyd.edu.aurkcdcrcmcdrpaperx, Int. J. Des. Comput. 1, 06 June 1998. w15x P.W. Mattessich, B.R. Monsey, Collaboration: What Makes It Work, Amherst H. Wilder Foundation, St. Paul, MN, 1992. w16x W.J. Mitchell, CAD as a social process, in: M. Tan, R. Teh ŽEds.., The Global Design Studio, CAAD Futures ’95, Centre for Advanced Studies in Architecture, National University of Singapore, Singapore, 1995, pp. 7–10. w17x M. Morozumi, M. Takahashi, R. Naka, N. Kawasumi, R. Homma, W.J. Mitchell, S. Yamaguchi, K. Iki, The levels of communications achieved through network in an international collaborative design project: an analysis of VDS ’96 project carried out between Kumamoto University, MIT and Kyoto Institute of Technology, in: Y.-T. Liu, J.-Y. Tsou, J.-Hou ŽEds.., Proceedings of the 2nd Conference on Computer Aided Architectural Design Research in Asia, CAADRIA ’97, Hu’s Publisher, National Chiao Tung University, Hsinchu, Taiwan, 17–19 April, 1997, pp. 143–152. w18x J.S. Olson, G.M. Olson, D. Meader, Face-to-face group work compared to remote group work with and without video, in: T. KÕanr Automation in Construction 9 (2000) 409–415 w19x w20x w21x w22x w23x w24x K.E. Finn, A.J. Sellen, S.B. Wilbur ŽEds.., Video-Mediated Communication, Lawrence Erlbaum Associates, Mahwah, NJ, 1997, pp. 157–172. C. Peng, Exploring communication in collaborative design: co-operative architectural modelling, Des. Stud. 15 Ž1. Ž1994. 19–44. A. Rand, The Fountainhead, The Bobbs-Merrill, Indianapolis, IN, 1943. R. Seward, D. Diaper, C. Sanger, The pod: a purpose-built environment to support group working, in: D. Diaper, C. Sanger ŽEds.., CSCW in Practice: An Introduction an Case Studies, Springer-Verlag, London, 1993, pp. 151–162. G.P. Shea, R.A. Guzzo, Group effectiveness: what really matters, Sloan Manage. Rev. 28 Ž3. Ž1987. 25–31. D. Shelden, S. Bharwani, W.J. Mitchell, J. Williams, Requirements for virtual design review, Architectural Res. Q. 1 Ž1995. 2. I.D. Steiner, Group Process and Productivity, Academic Press, New York, 1972. 415 w25x F. Sudweeks, M. Allbritton, Working together apart: communication and collaboration in a networked group, in: C.D. Keen, C. Urquhart, J. Lamp ŽEds.., Proceedings of the 7th Australasian Conference of Information Systems, Vol. 2, ACIS96, Department of Computer Science, University of Tasmania, 1996, pp. 701–712. w26x F. Sudweeks, S. Rafaeli, How do you get a hundred strangers to agree: computer mediated communication and collaboration, in: T.M. Harrison, T.D. Stephen ŽEds.., Computer Networking and Scholarship in the 21st Century University, SUNY Press, 1996, pp. 115–136. w27x A. Vera, T. Kvan, R. West, S. Lai, Expertise and Collaborative Design, CHI ’98 Conference Proceedings, CHI ’98, ACM, Los Angeles, 1998, pp. 503–510. w28x S. Whittaker, B. O’Conaill, The role of vision in face-to-face and mediated communication, in: K.E. Finn, A.J. Sellen, S.B. Wilbur ŽEds.., Video-Mediated Communication, Lawrence Erlbaum Associates, Mahwah, NJ, 1997, pp. 23–49.