The Exploration of Solution Options in Design:
A ‘Naturalistic Decision Making’ Perspective
Linden J. Ball
Nicola J. Lambell
Sue E. Reed
Fraser J.M. Reid
This paper examines how a ‘naturalistic decision making’ (NDM) approach to understanding cognitive processing in complex, real-world environments may facilitate an improved theoretical interpretation of solution-generation and solution-evaluation in design. In reviewing research that has investigated solution exploration in design, we contrast studies of individual designers with studies of team-based activity. We propose that both individual and collaborative design is based around the application of a dominant satisficing approach to solution exploration, and we discuss how the satisficing that is seen in design situations mirrors decision-making strategies observed in a wide range of NDM contexts. We argue for a view of satisficing as being adaptive and efficient in ill-defined and dynamic contexts such a
s professional design. We conclude with an overview of the prospects for an NDM account of design, focusing particularly on avenues for future studies of the solution-exploration processes that underpin design expertise in collaborative contexts. D
esign activity is often viewed as a form of problem solving (e.g., Goel and Pirolli 1992, Smith and Brown 1993, Ball, Evans and Dennis 1997). Researchers espousing this position typically emphasise the complex and ill-defined character of design tasks and the need to manage such complexity through processes of problem structuring (including the clarification of requirements and constraints), sub-problem identification, and the exploration of solution ideas through abstraction hierarchies that take the designer toward an implementable level of detail. Other researchers (e.g., Akin and Lin 1995, Ullman, Herling and Sinton 1996) have characterised design more as a decision-making endeavour, and focus on the processes associated with the generation, evaluation and selection of solution options. In reality, of course, design can be conceptualised as involving both problem-solving and decision-making components, with neither view being incompatible with the other. Klein (1998), for
第五届cctv舞蹈大赛Linden Ball, Nicola Lambell, Sue Reed and Fraser Reid
example, argues that the distinctions between problem solving and decision making tend to blur in natural settings, and he proposes that although some prefer to treat problem solving as a subclass of decision making (invoked when a new course of action needs to be formulated) others prefer to see decision making as a subclass of problem solving (invoked when several courses of action need to be compared).
The apparently symbiotic relationship between the problem-solving and decision-making processes that arise in naturalistic settings is also reflected in recent developments in the fields of problem solving and decision research. In particular, these two formerly disparate areas of psychological concern are moving ever closer together in terms of their fundamental aims (i.e., to understand cognitive processing in complex, real-world environments), their associated research methods (including increased use of field-based techniques such as ethnographic observation), and the theoretical models that are being proposed to explain underlying representations, processes and strategies. Such integration is certainly to be welcomed, as it is sure to pave the way toward the formulation of improved psychological accounts of domain-based expertise as it is played out in situated and contextualised settings such as real-world design.
Much of this new generation of problem-solving and decision research has been spearheaded by tho
se working within the so-called ‘naturalistic decision making’ (NDM) tradition that emerged in the early 1990s (for collections of key papers see Klein, Orasanu, Calderwood and Zsambok 1993, and Zsambok and Klein 1995). At least three themes have tended tend to dominate the research agendas of NDM investigators. Firstly, there is the ambition to understand decision making that is embedded in large, dynamic situations that involve many of the following characteristics: ill-structured tasks; information that is incomplete, ambiguous, or changing; goals that are shifting, ill-defined, or competing; decisions that occur in multiple event-feedback loops; constraints on time; high stakes; and a requirement to balance personal choices against organisational norms and goals. Second, there is a central concern with understanding the role that expertise plays in situated activity. For example, how do people use their knowledge and prior experience in coping with complex, real-world decision tasks? Third, there is the desire to understand the influence of multiple co-participants on the decision-making process. It seems clear that the NDM approach should have much to offer in relation to an understanding of design cognition since design seems to be a prime example of the kind of complex real-world activity that the approach is geared-up to addressing.
In light of the potential relevance of the NDM perspective to the design domain, we pursue three primary objectives in the present paper. First, we present a review (in section 1) of key findings concerning solution-generation and solution-evaluation activities in
The Exploration of Solution Options in Design:
A ‘Natural Decision Making’ Perspective design that derive from a range of studies that cut across a variety of design domains. One particular interest that we have in relation to this review is to compare descriptive findings concerning the nature of solution exploration in design with prescriptive accounts of idealised solution-development activity. Another theme motivating our review is to contrast studies of individual designers with those based on team design activity. Second, we examine how the NDM approach may facilitate a deeper conceptual understanding of findings concerning solution-exploration activities in design. Third, we overview the prospects for an NDM account of real-world design, focusing particularly on suggestions for future empirical research that might shed further light on the processes of solution generation and solution evaluation that underpin design expertise in collaborative contexts.ndm
1. Solution exploration in individual and collaborative design 1.1 The prescriptive position
The prescriptive position on solution exploration in design emphasises the importance of designers systematically generating and evaluating a range of solution alternatives to any problem or subproblem (e.g., Pahl and Beitz 1984, Pugh 1991, Cross 1994). The idea that exploring multiple de
sign alternatives has high utility derives primarily from economic considerations. In essence, since the financial impact on a company of a failed product can be high, it is seen as advisable for designers to strive to avoid poor design outcomes by selecting the best available design option from a range of alternatives at all stages of the design process. From a prescriptive standpoint it is often seen as especially important for designers to defer any premature commitment to initial design ideas in the upstream, conceptual phases of design activity, since a high-level design concept may itself subsume many weeks or months of more detailed design effort.
A further, noteworthy, aspect of such prescriptive models of effective design practice is that they rarely exhort designers to attempt to achieve optimal design solutions, since it is recognised that: (i) the space of possible solution concepts is extremely large (and therefore computationally intractable), and (ii) important design requirements (including aesthetic ones) may be subjective and non-quantifiable. In spite of this acknowledgement of the impossibility of design optimisation, a prescriptive emphasis still remains on the importance of identifying the best solution in a range of possibilities. Simon (1981), arguing from a psychological standpoint, can be viewed as concurring with the prescriptive design literature when he proposes that the pursuit of alternative design ideas is a central aspect of design activity. Simon also acknowledges that the solution-generation process is
not an optimising one because of the limitations of available information-processing capacity in human designers. Instead, he proposes that designers apply a ‘satisficing’ principle, by means of
Linden Ball, Nicola Lambell, Sue Reed and Fraser Reid
which they strategically search for any acceptable solution to a design problem and then stick with such a solution once it is identified.xjw
1.2 What individual designers do in practice
Anecdotal reports abound in the design literature concerning the nature of solution-generation in design, and it is often claimed that the pursuit of multiple alternatives is a common aspect of professional design practice. Although such anecdotal evidence is suggestive of the possible role of multiple-solution exploration in design, actual empirical research on this theme presents a more complex picture. One pioneering study (Kant 1985) revealed that expert algorithm designers rapidly developed a single ‘kernel’ idea which was progressively refined through levels of increasing detail. Similar evidence for an apparent lack of exploration of alternative design concepts coupled with ‘fixation’upon initial design ideas was reported in various studies during the 1980s and 1990s in relation to software engineering (e.g., Jeffries, Turner, Polson and Atwood 1981, Adelson and Solowa
y 1986, Guindon 1990), mechanical engineering (e.g., Ullman Dietterich and Stuaffer 1988), architecture (e.g., Rowe 1987), and electronic engineering (e.g., Ball, Evans and Dennis 1994, Ball, Maskill and Ormerod 1998). Taken as a whole, then, the empirical evidence seems to support a view of solution development in design that deviates from the prescriptive position that designers should explore multiple alternatives at all levels of design abstraction. Moreover, the aforementioned studies cut across a number of different design domains, problem types and levels of expertise (although the majority of studies focused on experienced designers), suggesting that minimal solution exploration is a generic feature of the natural practice of individual designers.
Cross (in press) argues that fixation on initial design solutions may not necessarily be a bad thing. For example, a number of studies of outstanding designers (e.g., Darke 1994, Cross and Clayburn Cross 1998) have indicated that they, too, exhibit a form of fixation on a guiding solution theme or principle, and can be tenacious in their pursuit of solution concepts that fit this theme. It is, of course, also possible that designers with different levels of expertise fixate upon initial solution ideas for different reasons (Ball et al. 1998). For example, experts may readily be able to re-invoke relevant solution ideas from their vast stores of experience-based knowledge, and may stick with these ideas because they know that they are good ones. Novices, on the other hand, may fixate upo
n the first solution idea that they come up simply because it is the only solution that they can generate given their limited knowledge-base, and they know that they are unlikely to be able to generate any viable alternatives.
One apparent flaw in this latter argument, however, is evidence that experts, like novices, are sometimes seen to generate initial concepts that turn out to have significant weaknesses and
The Exploration of Solution Options in Design:壮族姑娘
出嫁为什么要穿黑嫁衣?
A ‘Natural Decision Making’ Perspective shortcomings (Rowe 1987, Ullman et al. 1988, Ball et al. 1998). Interestingly, too, both experts and novices seem to be reluctant to abandon such failing solutions, and will tend to develop them into satisfactory ones by producing ‘multiple slightly improved versions’(Ball et al. 1998) or by applying ‘patches’ to rectify weaknesses (Ullman et al. 1988). It is possible, however, that poor initial solutions - and the subsequent patching that they necessitate - are relatively rare aspects of expert design that only arise in the context of highly unfamiliar and non-routine problems, where the expert’s knowledge-base is severely stretched. In such a context it would be hardly surprising to see the designer displaying behaviours more akin to a domain novice. Furthermore, although patching seems like an impoverished design strategy, it may
well be the most rational strategy to adopt in highly time- and cost-constrained contexts such as design, where solution abandonment in favour of a new idea would be wasteful of design efforts already made, and downstream activities might be just as likely to throw up unexpected problems with an alternative high-level solution concept anyway. It might actually be more cost-effective for the designer to stick with an idea that they believe can be made satisfactory, without recourse to the exploration of an entirely new concept.
mird-124
1.3 What collaborative design teams do in practice
Much real-world design work is situated in commercial contexts that involve a high degree of collaborative effort. Few studies, however, have investigated whether the solution-exploration behaviour manifested by individual designers generalises to team-based activity in company-based contexts. One study that stands out as an exception to the lack of research in this area is that conducted by Olson et al. (1996). This research focused on design teams working in early software-design meetings, and studied 10 such meetings relating to four different projects in two organisations. All communication was analysed to uncover the structure of design activity, including the generation and evaluation of alternatives. These analyses revealed a high degree of commonality across meetings in the sequential organisation and time distribution of behaviours. Of
particular interest were the observations that over 21.5% of design time was spent generating and clarifying alternatives (making solution exploration the largest category of activity), and that for 80% of design issues (i.e., design questions or problems) two or more alternatives were considered, with the average being 2.5 alternatives per issue. It appears, therefore, that Olson et al. have uncovered good evidence for moderately increased levels of alternative-generation activity in group design compared with individual design. It is noteworthy, however, that many alternatives appeared to be assessed in a partial or piecemeal manner. Indeed, Olson et al. observed that 33% of
Linden Ball, Nicola Lambell, Sue Reed and Fraser Reid
alternatives were not actually evaluated through the explicit application of any design criteria.
Olson et al.’s findings have been broadly supported in two recent studies of collaborative design reported by Turner and Cross (2000) and by Ball and Ormerod (2000a). Turner and Cross focused on five small-group design meetings that were primarily concerned with the informal review of design progress and associated design development. Using Olson et al.’s classification scheme for coding design activities, they too found that much more time was spent on the exposition of design alternatives than on any other activity. Turner and Cross propose that their teams appear to have ado
pted an ‘Inquiry Cycle’ (Potts and Catledge 1996), whereby ideas that are first developed in activity outside of meetings are then presented and discussed within the meetings, finally being refined after such discussion. Turner and Cross suggest that this activity cycle may account for the comparatively large amount of time spent on explaining alternatives in the meetings at the expense of evaluating them through the application of criteria (note that the time distribution of these activities occurred in a 5:1 ratio). This apparent lack of full and systematic option evaluation is resonant of Olson et al.’s findings that not all design alternatives were explicitly evaluated. In Turner and Cross’ study, however, what seems to be minimal solution evaluation may be an artefact of their time-based analysis of design activities, since it could be the case that the collaborative evaluation of design options through the application of criteria is a relatively quick process, whereas actually explaining options to co-designers simply takes more time to accomplish.
Ball and Ormerod (2000a) reported a study of a single, company-based design-review meeting that centred on the conceptual aspects of an industrial product. They employed a Questions, Options and Criteria (QOC) scheme to code design activity (see MacLean, Young, Bellotti and Moran 1991), where the Q, O and C elements are isomorphic with the Issue, Alternative and Criterion elements of the scheme adopted by Olson et al. (1996) and applied by Turner and Cross (2000). Ball and Ormer
od found that the mean number of options generated per question was 2.8 (which compares closely to the mean rate of 2.5 alternatives per issue identified by Olson et al.). In addition, over 95% of options were evaluated by the application of at least one evaluative criterion, with the average rate of application of criteria per question being 7.4 (and the range being between one and 17). These findings suggest reasonably high rates of alternative generation and evaluation by this collaborative design team.
工业产品生产许可证管理条例One further aspect of Ball and Ormerod’s findings was that most of the options that were generated (i.e., 22 out of 25) actually derived from the activity of the ‘Project Champion’ associated with the product being developed, rather than reflecting the input of novel design ideas from the rest of the design team. This finding was interpreted as
豫公网安备41160202000603
站长QQ:729038198
关于我们
投诉建议