Similarity Between Design Domains - CiteSeerX

A Comparative Programme for Design Research Martin Stacey Computer and Information Sciences, De Montfort University

Claudia Eckert Engineering Design Centre, University of Cambridge

Christopher Earl Design and Innovation, The Open University

Louis L. Bucciarelli School of Engineering, Massachusetts Institute of Technology

P. John Clarkson Engineering Design Centre, University of Cambridge

Proceedings of the Design Research Society 2002 International Conference: Common Ground Brunel University, London, September 2002 As ‘designing’ is a diverse phenomenon, but design processes have many important features in common with some other design processes, we can gain insights into how and why designers do what they do by making cross-domain comparisons. In this paper we propose a research programme for design studies: systematising these insights by using comparisons between design processes to compile a catalogue of patterns of designing – sets of features of design processes linked by causal mechanisms, that in combination with each other give a wide range of design processes their distinctive forms. The catalogue of patterns should include patterns describing features at different levels, linked by different sorts of causal mechanisms, so that different theoretical views and scales of description should be integrated in a richer unified understanding of designing.

Design research: mapping a diverse phenomenon Designing is a diverse, extremely complex and enormously variable phenomenon, so diverse that it is not obvious whether it makes any sense to regard ‘it’ as one thing, distinguished from non-designing by more than superficial characteristics. What scope do universal theories of design really have? Perhaps the crucial determinants of the nature of design are neither universal nor peculiar to individual episodes, but are shared by types of designing. What types of designing? And what are the crucial determinants of the nature of design? We have no definitive answers to these questions to offer. What we propose in this paper is a research programme for developing a richer and broader understanding of designing, by examining and mapping the similarities and differences between design processes. The goal of this research programme is to develop a progressively more coherent body of theoretical understanding of design, by working upwards from individual case studies.

A Comparative Programme for Design Research

This includes both comparisons between apparently-similar processes within one industry and between apparently-dissimilar processes producing radically different products, for instance helicopters and sweaters. We find similarity and difference relationships that cut across conventional industry boundaries – understanding them gives insights that are beyond the scope of universal theories of design and unreachable through single-industry process models. While design is the focus of our interest, our local and incremental approach to developing theoretical understanding of complex human activities means that we are not fundamentally concerned with the boundaries between design and non-design; our project can and should encompass the similarities and differences between types of designing and other complex human activities such as the practice of science. In contrast top-down universal theory approaches to design necessarily make the difference between design and non-design a central issue (see Love, 2002); this may be a fundamental mistake.

Design as a function of technology and culture The nature of an artefact – both its physical structure and operating principles, and its intended purposes (see Kroes, 2002) – exerts a powerful causal influence on the process through which it is designed. And design processes are constrained by human cognitive capabilities; although some designers have exceptional abilities, the findings of cognitive psychology give us universal constraints on what is possible for designers. Between these poles, a variety of cultural factors influence how designing happens: At the social level, the business models of the companies involved in the process (see for instance Eckert and Demaid, 2001), and their structures and social organisation. At the individual level, the knowledge, skills and values participants get from their professional training and experiences. Only some of these factors are unique; important elements are shared by the members of particular groups (see Bucciarelli, 1994; Henderson, 1999). Among the very wide range of factors that influence what happens in a design process, some serve to increase diversity, others serve to increase uniformity. This richness creates a methodological problem for design research: the sheer scale and complexity of comprehensive descriptions or explanatory theories of how complex artefacts are designed. Design researchers are forced to focus on limited subsets of the phenomena in front of them (for an illustration of this, see the twenty analyses of the same data reported in Cross, Christiaans and Dorst, 1996). Even rich case studies of individual processes or working environments are necessarily partial; ethnographers deal with this by adopting as a methodological axiom the view that different ethnographic accounts of the same culture can be equally valid. Universal theories of designing focus only on some aspects of a few significant parts of the design creation process. And descriptions of design differ radically in the scale of the phenomena they describe, as well as in the coverage they claim and the theoretical and methodological assumptions they embody. For instance Smithers (1998) presented a theory of the types of knowledge designers use; Papamichael and Protzen (1993) and Gero and Kannengiesser’s (2002) offered theories of the types and sequences of description creation in design; many engineering methodologists have based their views of designing on the theory of technical systems, for instance Andreasen’s (1980) theory of domains. Many other researchers have presented views of design to highlight particular aspects of designing, for instance Taylor (1993) stressed the inherent complexity and parallel nature of designing; we (Eckert, Stacey and Clarkson, 2000) highlighted the role of sources of inspiration in idea generation.

M.K. Stacey, C.M. Eckert, C.F. Earl, L.L. Bucciarelli & P.J. Clarkson

In order to make comparisons between design processes, we need to reduce the complexity of the problem by focusing on a limited subset of the phenomenon of design, as well as a limited set of design processes. Thus the research we do and the results we produce will necessarily be piecemeal, but they should be cumulative. In order to integrate our findings both with other comparative studies and the insights generated by the wide range of approaches to the study to design, we need a way to formulate our results that facilitates combination, both to create richer composite analyses and to reveal conflicts. In the rest of this paper we discuss our experiences of transferring insights between design processes, and the meta-theoretical view our experiences suggest, of how to do this systematically to build a structured mosaic of theoretical understanding of design.

Comparative analysis: useful in practice This papers draws on empirical studies undertaken by the authors in a variety of different domains and industries. The first author undertook a broad and detailed study of design in the knitwear industry, focusing on communication (Eckert, 1997, 2001; Eckert and Stacey, 2000) and the role of inspiration (Eckert and Stacey, 2001), interviewing and observing over 80 designers and technicians in 25 companies in Britain, Germany and Italy (see Stacey and Eckert, 1999 for a methodological discussion of our approach). The broad range of companies studied make it possible to compare patterns of activities across both similar and dissimilar companies, to assess the causes of observed behaviour. In the knitwear industry everyone followed the same steps, so that the design process as described in a detailed flowchart model (Eckert, 1997) proved remarkably universal; however there were very large differences between companies in the effort put into the different stages and the amount of backtracking to earlier stages. Eckert and Demaid (2001) compared design processes in a variety of industries characterised by needing to meet very rigid delivery deadlines. They identified consistent patterns across industries determined by the business models of the companies. How they sold their products had a huge effect on patterns of communication in the design process as well as on risk taking. Some companies design their own ranges and sell them through shows. They do market research, but have no direct interaction with the buyers who make purchasing decisions for retailers. They are free in their design, but take a high risk. At the other extreme some companies work very closely with buyers for retailers, who give them briefs for designs and guarantee the purchase of a certain number of designs. These companies communicate with their buyers, but have no direct contact with the market. They undertake a large amount of rework to satisfy their buyers, but carry little risk until the relationship with their buyer become tenuous. More recently the first author has been studying the design of complex engineering products, focusing on change processes in helicopters (Eckert, Clarkson and Zanker, in press) and diesel engines; and on process planning and communication in the automotive industry (Eckert and Clarkson, 2002; Eckert, Clarkson and Stacey, 2001), interviewing between 12 and 25 engineers and engineering managers in each company. She has gained interesting insights into engineering processes from her understanding of knitwear design. A stark example is the similarity in communication behaviour between artistic knitwear designers and technical knitwear technicians, and conceptual designers and analytical designers of diesel engines. Knitwear designers’ greatest skill lies in gaining a tacit understanding of the space of contemporary fashion and placing their own designs within this context (Eckert and Stacey, 2001). They communicate their ideas highly efficiently amongst themselves by specifying changes to known examples of designs that are interpreted within the shared context of contemporary fashion (Eckert and Stacey, 2000). Arguing with verbal explanations referring 3


to rational criteria from within a tacit context represented largely in visuospatial terms is extremely difficult; therefore the strength of subjective belief often replaces rational arguments. This mode of communication does not work well with people who do not share the context; it often appears handwavy (quite literally) and unspecific. The designers come across as inarticulate and less knowledgeable then they really are. Several years after observing the discourse of knitwear designers, we studied change processes in diesel engines, and interviewed several designers who were involved in the conceptual design of a new fourcylinder engine. The head designer had to make fundamental design decisions weighing up trade-offs, and defining the fundamental design that would later be realised and tested in several thousand person-years of design effort. This individual has a tremendous tacit knowledge of the properties of diesel engine designs. He uses the company’s old engines as well as competitor engines both as inspiration – for reassurance rather than copying – and as communication aids. When he has different design options, he might decide to follow a similar path to his competitors trusting them to have gone through a careful analysis process. Although design decisions are backed up with analysis, he follows his subjective instincts. In a discussion of degree titles it emerged that the diesel engine conceptual designers often don’t have a degree and if so feel strongly that they should be bachelors of art rather then science, because they see themselves as artists. They describe their work as an art rather than an exact science. Their mannerisms are remarkably similar to the knitwear designers including the gestures they make and the subjective beliefs with which they argue. In consequence their more analytical colleagues experienced them as vague and found it difficult to question their design decisions. When the author pointed out the similarity to knitwear design and explained the difficulty of rational explanation in communication about objects with emergent spatial or behavioural properties, designers from both groups found this very enlightening and requested that this issue be included in a formal feedback presentation. The conceptual designers had never realised that their way of talking was perceived as unscientific and therefore had not provided the rational arguments that they could have constructed; and the analytical engineers had never realised that this way of talking resulted from the task the conceptual designers were doing rather than their personalities (the two are of course not unconnected). So we have found that cross-process comparisons, focusing on a few locally-relevant aspects of designing, have enabled us to identify and explicate important causal influences on what happens in design processes. However what we have not yet done is formulate the results of such comparisons in a form that facilitates the development of a systematic body of understanding about design.

Similarity between instances of designing Our understanding of the diverse range of human activities we label ‘designing’ is influenced by the similarities we perceive both between pairs of design activities and across large categories. Similarity between domains arises from common features of designs or processes. Different categories of features correspond to a spectrum of layers of design descriptions (Eckert and Earl, 2002). At one end of this spectrum lie the designers’ domain knowledge and the organisation of the process whilst at the other end lie the details of components and the production processes that produce them. In a central layer, the features of products, which include solution principles, function and layout, link 'upwards' to the knowledge and process layers and 'down' to production and component layers. Common features between domains can be examined in each of the layers separately. However, in identifying similarities between domains it will be necessary to look across layers. Just as common features (and distinguishing features) can be used to compare domains, so can common relationships between features.


Not all shared features will have the same significance, either in our perceptions or in theoretically grounded reasoning. Moreover distinguishing features may overwhelm the similarity due to shared features. One purpose of the research programme we propose here is to change people’s perceptions of similarities and differences by raising awareness of factors other than superficial product characteristics and conventional divisions between industries. People commonly perceive ‘design’ as having a unity beyond being the activity of creating a plan for an artefact to meet a practical need; and many commentators have argued that it is fundamentally characterised by a cycle of proposing a partial solution, evaluating the solution, reformulating the problem and proposing another solution to the revised problem (Asimow, 1962; see Cross, 2000). But we suspect that much of the perceived unity comes from design activities sharing many characteristics with some others; in order to make sense of the relationships between individual processes and the range of design activities, we need to consider other kinds of similarity relationships. Most intuitively, similarity classes can be defined where all elements in the class share a set of common features. In mathematical analyses of similarity, closures arising when no more elements can be added to a class without reducing the numbers of features they all share are the critical similarity classes. These similarity classes are related in a lattice structure (Ganter & Wille, 1999) which provides a 'map' of potential similarity relationships. Similarity can also be based on tolerance relations defined by shared features. Tolerance classes are maximal sets of mutually similar elements, that is each pair of elements shares at least one feature (Schreider, 1975; Zadeh, 1971). This tolerance similarity is not transitive so that if domain A is similar to B which is in turn similar to C then A may not be similar to C. Similarity classes themselves are the subject of a tolerance relation based on sharing a common domain. A weaker definition of similarity classes relaxes the condition for common features to exist between each pair of domains in a class, allowing a chain of connection. Two domains A and C can be similar through a domain B which shares features with A and (not necessarily the same features) with C. A stronger specification takes into account the number of shared features. An m-connection arises from m shared features and corresponding similarity requires that domains share at least m features. Classes defined by chains of m-connections correspond to the multidimensional structures of complex systems (Johnson, 1995). In measuring the strength of similarity is necessary to consider both shared and distinguishing features as well as their significance (Tversky, 1977). The 'diagnostic' or classificatory value of features is dependent on the domains being compared. The strength of similarity between domains can reinforce itself in this classificatory role. Similarity is then perceived as greater within the group. The significance of features is also dependent on the range of domains being compared. Features common to all domains in a range have little or no significance. However, adding more domains, which do not share these common features, can increase the significance of the original features.

Patterns of designing The characteristics that some instances of designing share with some others come in clusters – different instances of designing can have a lot in common because what they share includes powerful determinants of the form of the designing process. Clusters of consistently shared characteristics form patterns. If we observe such a cluster, we can hypothesise that the shared characteristics are linked by causal relationships, or are all symptoms of some as yet unrecognised underlying cause. We can test such a hypothesis in two ways: by looking at 5


other design processes to see if the presence of some of these attributes predicts the presence of other attributes; and by trying to construct and test theories of how the attributes are causally related. In our view, recognising and making sense of patterns of designing is crucial to developing a rich, multi-level understanding of design. [Note that we avoid the term design pattern; this refers to an abstractly-formulated solution to a recurring problem, together with a description of the type of problem it fits and the consequences of using it, an idea introduced into architecture by Christopher Alexander (Alexander et al., 1977) and widely adopted in software engineering (notably, Gamma et al., 1995). This notion has long been implicit in much engineering practice.] Patterns of designing can be detected at all the scales at which designing can be analysed and described – of time, number of participants, the portion of the whole artefact being considered, and the activities that are the units of analysis. Moreover, given sufficiently rich observations of design processes we can look for patterns comprising features at different levels of description. But identifying similarities that can be represented as patterns of designing is not trivial: it involves finding the right abstractions of observable phenomena; this requires imagination and reframing the design situations in different conceptual terms.

Causal stories: elements of theories of designing Even if the elements of a group of characteristics appear to be consistently present or consistently absent in a range of design processes, the group will be unpersuasive as a pattern of designing until we have a causal explanation for why these features are related. Thus the next step given a putative pattern is hypothesising one or more plausible causal stories for how the features share common causes, or are linked by a chain of cause and effect. The instances of designing under study can then be scrutinised for supporting or disconfirming evidence. Hypothesising causal relationships enables the generation of more focused questions about what is really going on in an episode of designing. But the processes of designing complex artefacts are immensely complex and variable, and the causal mechanisms that influence the form of designing may operate with different strength. Causal influences on some aspect of designing may collide, chains of causality may be blocked or modified by other influences. So we should ordinarily formulate our causal stories in terms of causal pressures rather than rigid determination. We may observe only part of a previously defined pattern in another design process, so that some characteristic of the process is predicted by the pattern but is absent. We should look for reasons why the causal mechanisms that according to the pattern should produce it are blocked, as well as for evidence that they are not operating. Either finding should enable us to refine our formulation of the pattern or suggest a new one. We retain an open mind on the question of whether a science of design is possible, or whether the study of design is necessarily a humanistic discipline (see Darke, 1979, for an articulation of this view). But the collection of patterns of designing has the form of scientific research – what we are after is understanding regularities and causal processes, by hypothesising and testing covering laws relating observable variables in the form of clusters of characteristics of design processes, and partial theories in the form of causal stories for why the elements of the pattern are related. Such covering laws and pieces of causal explanation formulated as patterns


of designing are local and fragmentary – pieces of a fuller theory subsuming all valid universal theories and accurate process models, that it may be impossible ever to complete.

A research programme in comparative design studies Because designing any complex artefact involves an extremely complex process, and design processes are influenced by a wide variety of phenomena at several levels, complete causal explanations of why everything happens and the way it happens are infeasible. But design processes are neither completely dissimilar from each other nor completely unpredictable. We can aim to develop an understanding of design that explains the nature and effect of major causal influences on design processes; and that predicts to a useful degree the form of design activities and the problems they will meet. The immensity of the task of developing a complete theory should not worry us, provided that our theory fragments not only function as free-standing explanations of interesting phenomena, but can also be combined to understand what happens when different interesting phenomena occur together. We propose that cataloguing patterns of designing should be a significant part of future empirical design research. Researchers should take opportunities to use previously documented patterns to gain insight into new design processes. At the same time they should take opportunities to test, corroborate, falsify and refine previously documented patterns as well as add new ones. In each case they should explicitly raise the question of how general are the phenomena they see in individual cases. Through an incremental process of critical evaluation by a diverse body of researchers – everyone who wishes to participate – we can incrementally develop a progressively greater understanding of similarities and differences between design situations, and the causal influences that give design processes their form. The view of design processes as varying in similarity to each other along different dimensions, and the research programme we outline here, is independent of the theoretical assumptions and foundations that underlie any particular approach to understanding what is going on in design. The comparative programme should encompass and orient a variety of different theoretical and methodological approaches to design studies. It should provide a basis for examining the scope of particular analyses and their relationships to similar analyses of other design processes; and for relating different theoretical perspectives on the same phenomena. We welcome the participation of researchers with different concerns and theoretical viewpoints. Patterns of causal influence can spring from technical, organisational, social and cognitive causes, and can act at many different scales, and a key part of the programme we propose is looking to see how different patterns describing different types of phenomena interact. Researchers tend to interpret phenomena that they see on one specific layer and attribute problems to the cause they are most familiar with. Sociologists tend to attribute problems to the interpersonal nature of design, psychologists to cognitive factors, business researchers to organisational processes. For instance Eckert (2001) described a variety of causes for breakdowns in the communication of design ideas between commercial knitwear designers and knitting machine technicians (and argued that the primary causes were the inherent difficulties of describing knitted structures in the available representations and the participants’ lack of understanding of the nature of their communication problems, while several secondary causes made resolving the problems harder). The two groups are socially as different as possible: young university trained women versus older men trained on the job with little formal education. Sociologist repeatedly commented that only when they heard this fact, they understood why the two groups don’t communicate. However we have also observed 7


communication difficulties between socially homogeneous engineers with different areas of expertise and mental representations – what Bucciarelli (1994) termed different object worlds (Eckert, Clarkson and Zanker, in press). Comparative studies that allow us to test causal hypotheses should enable us to assess the validity of different explanations.

Comparative analysis through observational studies While theoretical analyses and experimental studies of designing in more or less realistic scenarios can yield important insights, particularly into the cognitive processes involved in particular kinds of problem solving, the primary source of information in the comparative design research programme will be studies of actual commercial design processes through interviews and observations. Sociology offers many different methodological approaches for studying cultures. Ethnography (see for instance Agar, 1980; Hammersley and Atkinson, 1995) has been widely adopted as a way to understand work cultures to assess the real requirements for software systems (for instance Suchman, 1987; Viller and Sommerville, 2000; see Anderson, 1994; Button, 2000), and has been taken up by many design researchers (notably Bucciarelli, 1988, 1994) including ourselves (Eckert, 1997, 2001). Ethnographic methodology can be applied to developing analyses of design in terms of cognitive analyses of expertise (Ball and Ormerod, 2000) and knowledge-level analyses of the information used and transmitted by the participants in a design process (Stacey and Eckert, 1999), as well as sociologically-oriented analyses of the types of statement that members of design teams make to each other (Minneman, 1991) and the role of visual representations in structuring design processes (Henderson, 1999). In an ethnographic study the researchers joins in the daily life of the groups they are studying and attempt to learn the skills and perspectives of the actors, while remaining conscious of their role as an outsider. This dual perspective enables the ethnographer to make sense of how the participants in the culture themselves see what they do and why they do it. Ethnographic studies are traditionally very detailed studies of one culture, which do not make any claims to generality. The criterion for validity is that assertions ring true to the participants in the culture; some researchers have attempted formal validation exercises with their participants, this is not unproblematic (Hammersley and Atkinson, 1995). However the rich data and insights gained by ethnographers can allow us to interpret behaviour we see in other situations. For example Anderson (1994) pointed up the analogy between Levi-Strauss’s (1969) study of marriage and gift-giving among Amazonian Indians and the social processes involved in experts helping their less knowledgeable colleagues customise software (Mackay, 1990). While some ethnographers argue that researchers should enter a culture without prior hypotheses, in practice hypothetico-deductive reasoning and explicit hypothesis testing is an important part of ethnographic fieldwork (Hammersley and Atkinson, 1995); we have argued that this is an essential part of understanding how cultural factors and the nature of the artefact interact to determine the form of the design process (Stacey and Eckert, 1999). Comparisons between design domains can provide a rich source of hypotheses to test and refine in observational studies; their use should be enhanced by making testable hypotheses available in a systematic catalogue of patterns of designing. In most of our own empirical studies of engineering, applying ethnographic methods is infeasible; we are largely reliant on interviews (Eckert, 2002). The information gathered is less rich, and suffers if the interviews are away from the engineers’ normal workplaces. Nor can


we always take statements in interviews at face value, as ethnographers well know. However we find interview studies to be effective given appropriately selected informants, especially if assertions can be discussed with several people with different viewpoints. We are able to actively test hypotheses derived from cross-industry comparisons. We employ aspects of the ethnographic mindset: we recognise the importance of understanding the participants’ own perceptions of tasks, situations and cultures; and that interpretations are selective and partial, so that different accounts may be equally valid. However central to our approach is generating and testing more general and abstract causal processes, and using comparisons between situations to test hypotheses and rival explanations.

The return journey: practical applications of comparative design studies Our approach is pragmatic. It aims to provide useful insights into particular situations from the richness of the immense variety of other design processes. While we are intellectually interested in understanding the nature of design, our ultimate goal is to find ways to support and enhance the work of practising designers. In the development of computer support tools, methodologies and management procedures, it is crucial to know the scope of phenomena the techniques are designed to address. It is necessary to identify the root causes of problems. The catalogue of patterns we envisage should provide designers, managers and software developers facing local practical problems with a toolkit of concepts with which to make sense of design processes they want to improve. Our comparative design research programme is intended to enable the transmission of best practice between companies and between industries by enabling the recognition of which design processes share needs and problems, instead of relying merely on the similarities between products. We also believe that our comparative approach can yield benefits for design education in a variety of design disciplines, which try to develop their own models and teaching methods without benefiting from each other’s experiences. Many working designers we have met in knitwear design as well as engineering fail to understand the nature of their colleagues’ work – for instance knitwear designers typically fail to comprehend that knitting machine technicians do designing (Eckert, 2001) – with visible adverse consequences for design processes. We find that designers in ‘subjective’ fields like knitwear design lack understanding and respect for the technical, problem solving aspects of their own work, while designers in ‘technical’ fields like engineering lack understanding and respect for the holistic, perceptual or subjective thinking involved in design. All these designers would benefit from a greater understanding of the similarities and differences between different kinds of ‘artistic’ and ‘technical’ design. In the long term, this more sophisticated view among designers may influence public understanding of design. The recruitment into different design professions is influenced by the public’s perception of the different fields, for example many believe that textile design would be more creative then engineering. Stereotypes perpetuate old gender divisions and attract certain personalities to certain fields.

Acknowledgements Claudia Eckert’s contribution to this research has been supported by the EPSRC block grants to the Cambridge University Engineering Design Centre. This paper was written while Louis Bucciarelli was a visiting professor at the Delft University of Technology. Several members of the Cambridge EDC have participated in our empirical studies of engineering design 9


processes, particularly Tim Jarratt and Brendan O’Donovan. We are very grateful to all our informants in many companies who have given their time to interviews and observations.

References Agar, M. (1980). The Professional Stranger: An Informal Introduction to Ethnography. London: Academic Press. Anderson, R.J. (1994). Representations and Requirements: The Value of Ethnography in System Design. Human-Computer Interaction, 9, 151-182. Andreasen, M.M. (1980). Syntesemetoder på systemgrundlag. PhD thesis, Lunds Tekniska Högskola, Denmark. Alexander, C., Ishikawa, S. & Silverstein, M., with Jacobson, M., Fiksdahl-King, I. & Angel, S. (1977). A Pattern Language. New York: Oxford University Press. Asimow, M. (1962). Introduction to Design. Englewood Cliffs, NJ: Prentice-Hall. Ball, L. J. & Ormerod, T.C. (2000). Applying ethnography in the analysis and support of expertise in engineering design. Design Studies, 21, 403-421. Bucciarelli, L.L. (1988). An ethnographic perspective on engineering design. Design Studies, 9, 159-168. Bucciarelli, L.L. (1994). Designing Engineers. Cambridge, MA: MIT Press. Button, G. (2000). The ethnographic tradition and design. Design Studies, 21, 319-332. Cross, N.G. (2000). Engineering Design Methods: Strategies for Product Design, 3rd ed. Chichester, UK: Wiley. Cross, N.G., Christiaans, H.H.C.M. & Dorst, K. (eds.) (1996). Analysing Design Activity. Chichester, UK: John Wiley. Darke, J. (1979). The primary generator and the design process. Design Studies, 1, 36-44. Eckert, C.M. (1997). Intelligent Support for Knitwear Design. PhD thesis, Department of Design and Innovation, The Open University, Milton Keynes, UK. Eckert, C.M. (2001). The Communication Bottleneck in Knitwear Design: Analysis and Computing Solutions. Computer Supported Cooperative Work, 10, 29-74. Eckert, C.M. (2002). Status in Empirical Studies. Workshop on the Role of Empirical Studies in Understanding and Supporting Engineering Design, National Institute of Standards and Technology, Gaithersburg, MD. Eckert, C.M. & Clarkson, P.J. (2002). Connectivity as a key to supporting design. In J.S. Gero (ed.) Artificial Intelligence in Design ’02. Dordrecht, Netherlands: Kluwer Academic Publishers. Eckert, C.M., Clarkson, P.J. & Zanker, W. (in press). Change and Customisation in Complex Engineering Domains. Research in Engineering Design. Eckert, C.M., Clarkson, P.J. & Stacey, M.K. (2001). Information Flow in Engineering Design – Problems and their Causes. Proceedings of the 13th International Conference on Engineering Design: Design Management – Process and Information Issues. Glasgow, UK: Professional Engineering Publishing, pp. 43-50. Eckert, C.M. & Demaid, A. (2001). Classifying Design and Design Management in Seasonal Industries. International Journal of Innovation Management, 5, 401-425. Eckert, C.M. & Earl, C.F. (2002). Perspectives on Similarity in Design. Proceedings of DESIGN 2002, Dubrovnik, Croatia. Eckert, C.M. & Stacey, M.K. (2000). Sources of inspiration: A language of design. Design Studies, 21, 523-538. Eckert, C.M. & Stacey, M.K. (2001). Designing in the Context of Fashion – Designing the Fashion Context. In Designing In Context: Proceedings of the 5th Design Thinking Research Symposium. Delft, Netherlands: Delft University Press, pp. 113-129.


Eckert, C.M., Stacey, M.K. & Clarkson, P.J. (2000). Algorithms and Inspirations: Creative Reuse of Design Experience. Proceedings of the Greenwich 2000 Symposium: Digital Creativity. London: University of Greenwich, pp. 1-10. Gamma, E., Helm, R. Johnson, R. & Vlissides, J. (1995). Design Patterns. Reading, MA: Addison-Wesley. Ganter, A.B. & Wille, R. (1999). Formal Concept Analysis: Mathematical Foundations. Berlin: Springer. Gero, J.S. & Kannengiesser, U. (2002). The situated function-behaviour-structure framework. In J.S. Gero (ed.) Artificial Intelligence in Design ’02. Dordrecht, Netherlands: Kluwer Academic Publishers. Hammersley, M. & Atkinson, P. (1995). Ethnography: Principles in Practice. London: Routledge. Henderson, K. (1999). On Line And On Paper. Cambridge, MA: MIT Press. Johnson, J.H. (1995). A language of structure in the science of complexity. Complexity, 1, 22-29. Kroes, P. (2002). Design methodology and the nature of technical artefacts. Design Studies, 23, 287-302. Levi-Strauss, C. (1969). The Elementary Structures of Kinship. Boston: Beacon Press. Love, T. (2002). Constructing a coherent cross-disciplinary body of theory about designing and designs: some philosophical issues. Design Studies, 23, 345-361. Mackay, W. (1990). Patterns of sharing customisable software. Proceedings of Computer Supported Cooperative Work ’90. Los Angeles: ACM Press, pp. 209-221. Minneman, S.L. (1991). The Social Construction of a Technical Reality: Empirical Studies of Group Engineering Design Practice. PhD thesis, Department of Mechanical Engineering, Stanford University, Stanford, CA. Xerox Palo Alto Research Centre report SSL-91-22. Papamichael, K. & Protzen, J.-P. (1993). The Limits of Intelligence in Design. Proceedings of the Focus Symposium on Computer Assisted Building Design Systems, Fourth International Symposium on System Research, Informatics and Cybernetics, Baden Baden, Germany. Smithers, T. (1998). Towards a Knowledge Level Theory of Design Process. In J.S. Gero & F. Sudweeks (eds.) Artificial Intelligence in Design ’98. Dordrecht, Netherlands: Kluwer Academic Publishers. Schreider, J. (1975). Equality, resemblance and order. Moscow: Mir. Stacey, M. K. & Eckert, C.M. (1999). An Ethnographic Methodology for Design Process Analysis. Proceedings of the 12th International Conference on Engineering Design. Munich, Germany: Technical University of Munich, vol 3. pp. 1565-1570. Suchman, L.A. (1987). Plans and Situated Actions. Cambridge, UK: Cambridge University Press. Taylor, A.J. (1993). The Parallel Nature of Design. Journal of Engineering Design, 4, 141152. Tversky, A. (1977). Features of similarity. Psychological Review, 84, 327-352. Viller, S. & Sommerville, I. (2000). Ethnographically informed analysis for software engineers. International Journal of Human-Computer Studies, 53, 169-196. Zadeh, L.A. (1971). Similarity relations and fuzzy orderings. Information Sciences, 3, 177200.

11


Biographies Martin Stacey studied cognitive psychology before doing a PhD in Artificial Intelligence at the University of Aberdeen (1992) and postdoctoral research on intelligent computer support for design at the Open University. He is now a Senior Lecturer in Computer Science at De Montfort University. His major research interests are in the psychology of design, intelligent systems to support design process planning, and in human computer interaction aspects of design support systems; he teaches software development methods and human computer interaction. Claudia Eckert’s education encompassed mathematics, philosophy, computer science and artificial intelligence, as well as experience of commercial software development. She completed a PhD in Design Studies at the Open University in 1997 followed by an ESRCfunded postgraduate research project, based primarily on a very large scale ethnographic study of the knitwear design process. Her research on knitwear design has focused on process modelling, the communication of design ideas, the use of sources of inspiration in various aspects of design thinking, and on intelligent computer tools for designers. She is now a Senior Research Associate at the Cambridge University Engineering Design Centre, where she is investigating planning processes, design modification processes and communication in large scale engineering projects. Analysing design processes using both a range of conceptual paradigms and cross-domain comparisons is her core research interest. Chris Earl is Professor of Engineering Product Design in the Department of Design and Innovation at the Open University. His main research areas since joining the Open University in 2000 are planning complex design processes, generative design descriptions and shape representation for design. Previously he was at the University of Newcastle, Faculty of Engineering where he conducted interdisciplinary research on the design of engineer-to-order products and on planning their manufacture and assembly. He has degrees BA, MSc in Mathematics and PhD in Design. Louis Bucciarelli is Professor of Engineering and Technology Studies at MIT. He received his BS from Cornell (Mechanical Engineering, 1959) and his PhD from MIT (Aeronautics and Astronautics, 1966). He was Director of MIT’s Technology Studies Program and is a member of MIT's Program in Science, Technology and Society. His major research interests include the development of alternative energy and residential energy instrumentation systems, engineering education, the history of science, and ethnographic studies of the engineering design process. He is the author of Designing Engineers (MIT Press, 1995). John Clarkson took a BA (1984) and PhD (1988) in Engineering at the University of Cambridge, before seven years working in design consultancy with PA Consulting Group. He returned to the University of Cambridge in 1995, where he is now Reader in Engineering and Director of the Engineering Design Centre, and a Fellow of Trinity Hall. His major research interests include the development of design methodologies to address specific design issues; medical devices; inclusive design; and the use of intelligent systems to support design process management.

WORD COUNT, excluding title, authors, abstract, acknowledgements, references and biographies: 4816.