TY - JOUR
AU - Sutcliffe,, Alistair
AB - Abstract This paper follows the scenarios and task models debate by reviewing the contributions of task modelling and scenario based approaches from a cognitive perspective. A framework of cognitive affordances is introduced to discuss the merits and limitations of each approach. An extension of the modelling theme, generic task models, is proposed to augment the contribution of knowledge reuse to the design process. The paper concludes by discussing how scenario based design might complement task analysis and reuse of task based knowledge. 1 Introduction This paper follows the debate started by Diaper's (2002) paper on scenario-based design and Carroll's (2002) response to it, with additional commentaries by Benyon and Macorlay (2002), Paterno (2002) and Carey (2002) . It is not my intention to elaborate the task model-scenarios debate directly, as this has already been conducted ably by the preceding authors. Instead, my objective is to broaden the debate by introducing reuse of HCI knowledge as a third paradigm for supporting the design process. At first sight, reuse may seem to bear little relationship to task modelling and scenarios, which both support de novo, bespoke-style development; however, Carroll has reused scenarios when they are bundled with claims in his task-artefact theory ( Carroll and Rosson, 1992 ), so I shall explore the relationship between specific (scenario-based) and generalised (task-based) knowledge in the design process. The title of this paper refers to a short paper ( Sutcliffe, 1989 ) I published in the first issue of Interacting with Computers , when I speculated on the convergence and possible co-habitation of task analysis, exemplifying the HCI approach to design, and systems analysis representing the software engineering tradition. It seems timely to reflect on that debate, since one of the implicit tensions in the Carroll–Diaper discussion turned on views of design traditions that could be located in Human–Computer Interaction or in the Interactive Systems Design community whose contributions can be found in DIS, SIGGRAPH and AIGA conferences. My paper is structured in two main sections. First, I join the task analysis-scenario analysis debate, but I will investigate the contributions both approaches can make to the design process from the viewpoint of cognitive reasoning processes. The subsequent section broadens the debate to review contributions to reuse of HCI knowledge that have escaped so far: claims ( Carroll and Rosson, 1992 ), design rationale ( MacLean et al., 1991 ), patterns ( Borchers, 2001 ) and generic task models ( Sutcliffe, 2002a ). The paper will conclude by reviewing the potential for co-habitation and competition between the different design approaches. 2 Roles of scenarios and task models in design One of the key distinctions between scenarios and any model is that the former are grounded examples of specific experience, whereas models are more abstract representations of phenomena in the real world. Unfortunately, the term ‘scenario’ has been abused in the literature and a large number of definitions exist (see Rolland et al., 1998 ). Indeed, much of the scenario literature, especially in the software engineering tradition ( Kaindl, 1995 ), is in fact describing event-sequence traces through state transition models. In object-oriented design it becomes difficult to distinguish between use cases, alternative paths through use cases, and scenarios, which are just another path through a use case ( Cockburn, 2001; Graham, 1996; Jacobson et al., 1992 ). However, this debate over the scenario/modelling boundary is a side track, since Carroll quite correctly uses scenarios as grounded narratives that describe specific examples of interaction. To understand the relative merits of scenarios and task models we need to investigate their roles in the design process and their cognitive affordances. 2.1 Scenario contributions Scenarios have several roles in design, although Carroll was emphasising one particular role as a ‘cognitive prosthesis’ or an example to stimulate the designer's imagination. A critical axis of the Carroll–Diaper debate turned on one's view of the designer's responsibility and freedom of action. In the task-analysis, model-based engineering tradition, the philosophy is to make design representations, design process and decisions explicit. The designer is a follower of methods, either as a novice in ‘cook book’ style or as an expert when methodological knowledge has been internalised. The scenario-based approach from Carroll's viewpoint exists to stimulate designers' creative imagination. Scenarios and other techniques such as claims are lightweight instruments that guide thought and support reasoning in the design process ( Carroll, 2002 ). Diaper correctly notes that this approach can lead to errors. A scenario, or even a set of scenarios, does not explicitly guide a designer towards a correct model of the required system. An extreme scenario might bias reasoning towards exceptional and rare events, or towards the viewpoint of an unrepresentative stakeholder. These biases are an acknowledged weakness of scenarios; however, Carroll's riposte is to trust designers as knowledgeable, responsible people who are capable of recognising such biases and dealing with them productively. Indeed, some propose scenarios that are deliberately exceptional to provoke constructive thought ( Djajadiningrat et al., 2000 ). Although scenarios are useful as cognitive probes for design, this is not their only role. Scenarios arguably are the starting point for all modelling and design, and contribute in several parts of the design process (see Fig. 1 ). Diaper pointed out that action lists in task models are essentially the same as event sequences in scenarios. This is quite correct and not surprising, as scenarios are examples of specific user behaviour which are collected in the early stages of requirements analysis. TAKD ( Diaper, 1999; Diaper and Johnson, 1989 ) explicitly refers to the process of generalisation from specific action sequences (i.e. scenarios) to a general task model that represents typical behaviour of a group of users. The process of generalisation inevitably loses detail and the analyst has to make judgements about when unusual or exceptional behaviours are omitted, or explicitly incorporate them in task models as branches in action sequences. Hence one criticism can be levelled at task analysis, that it inevitably omits detail which may be vital, while scenarios might be able to gather such detail but at the price of effort in capturing and analysing a ‘necessary and sufficient’ set of scenarios. Fig. 1 Open in new tabDownload slide Use of scenarios in different phases of the design process. Fig. 1 Open in new tabDownload slide Use of scenarios in different phases of the design process. One productive juxtaposition of scenarios and models is to use scenarios as test data to validate design models. This approach has been actively researched in the Inquiry Cycle ( Potts et al., 1994 ), which recommended using scenarios as specific contexts to test the utility and acceptability of system output. By questioning the relevance of system output to a set of stakeholders and their tasks described in a scenario, the analyst can discover obstacles to achieving system requirements. Input events can be derived from scenarios to test validation routines and other functional requirements. Obstacle analysis has since been refined into a formal process for discovering the achievability of system goals with respect to a set of environmental states, taken from scenarios ( Van Lamsweerde and Letier, 2000 ). Scenarios, therefore, can fulfil useful roles either as test data, as a stimulant to reasoning in validating system requirements, or by providing data for formal model checking. HCI uses scenarios in a similar manner in usability evaluation, although the role of scenarios is not articulated so clearly. Nevertheless, task or test scripts in evaluation methods ( Monk and Wright, 1993; Sutcliffe, 2000a ) are scenarios. Carroll also recognised the validation role for scenarios in the task-artefact cycle in which an implemented artefact (that embodies a task model) is evaluated, leading to design improvements and, by a process of claims analysis, to new HCI knowledge (see Fig. 2 ). Fig. 2 Open in new tabDownload slide Task-artefact cycle. Fig. 2 Open in new tabDownload slide Task-artefact cycle. The results of evaluations are recorded in usage scenarios that describe the problem that motivated a general design principle, called a claim, with trade-offs expressed as upsides and downsides ( Carroll, 2000; Sutcliffe and Carroll, 1999 ). Design rationale ( MacLean and McKerlie, 1995 ) is a closely related means of expressing generalisable knowledge accumulated during iterative design. Carroll has articulated several different roles for scenarios in the design process including as envisionment for design exploration, requirements elicitation and validation ( Carroll, 1995 ). Other roles are usage scenarios, which illustrate problems; initiating or visioning scenarios, which stimulate design of a new artefact; and projected use scenarios, which describe future use of an artefact that has been designed ( Sutcliffe and Carroll, 1998 ). 2.2 Task analysis roles Task models, in common with other conceptual models such as class structure diagrams, activity sequence or state transition diagrams in UML, provide an abstract view of the real world with a set of semantics motivated by the modelling goal. Hence class diagrams represent the inheritance structure of system objects that model the data-oriented view of a system, whereas state transition diagrams represent activity-oriented specification. Task models fit within the modelling genre that represents intent (goals) and activity (procedures or action sequences). Task modelling has been extended to cover data-oriented views via Domain Knowledge Structures in TKS (Task Knowledge Structures: Johnson et al., 1988 ), or has been adapted to fit within the object-oriented paradigm in MAD (Methode Analytique et Description: Rodriquez and Scapin, 1997 ). Others have extended task models to explicitly represented collaborative systems and multi-agent distributed tasks ( Van Der Veer et al., 1996; Van Der Veer and VanWelie, 2000 ). However, most task models (e.g. HTA: Annett, 1996 ; TAKD: Diaper and Johnson, 1989 ) focus on goal decomposition hierarchies with action sequence descriptions at lower levels of detail. A prime role of task models has been to represent the problem space when reasoning about functional allocation. Although the functional allocation literature acknowledges the need for task analysis, it rarely represents task models explicitly (see Dearden et al., 2000 ). Instead, scenarios may be used to motivate functional allocation decisions, within task-related frameworks such as IDAS (Information, Decision, Action, Supervision: Wright et al., 2000 ). However, task models can be used to partition activity during functional allocation ( Sutcliffe, 1995b; Vicente, 2000 ), and this remains one of their major roles. One criticism of task analysis is that is does not capture the richness of interaction that occurs in the real world compared with scenario narratives that concentrate on contextual description (e.g. Kuutti, 1995; Kyng, 1995 ). However, task models have been extended to describe information requirements implied in tasks ( Sutcliffe, 1997 ), and the role of work artefacts in ecological interface design ( Vicente, 2000 ). Another omission is the lack of explicit representation of communication between agents engaged in collaborative tasks, although this is partially specified in GTA ( Van Der Veer et al., 1996 ) and is dealt with more explicitly in coupling analysis, which provides a discourse analysis framework that can be incorporated into a task workload analysis ( Sutcliffe, 2000b, 2002b ). Finally, task models may be criticised for not representing the relationships between agents, activity and organisational structures, although these concepts are described in socio-technical system design frameworks such as ORDIT ( Eason et al. 1996 ); while more comprehensive modelling languages can be found in the i* requirements engineering method that analyses the dependencies between agents, tasks, goals and resources ( Mylopoulos et al., 1999; Yu, 1993 ). So task analysis might be accused of having a narrow scope of phenomena that it can model, and scenarios might be able to represent a wider range of phenomena, but they do so in an ad hoc manner and leave the responsibility of generalisation to the analyst. Task analysis needs to deal with system environment modelling, but more complexity is the penalty of extending the scope of modelling to the system's context. 2.3 Cognitive affordances Although the previous debate has covered the roles of scenarios and task models, it may be instructive to enquire a little deeper into how scenarios and task models may function in cognitive terms. Scenarios use language and concepts that are readily accessible to users and domain experts, whereas tasks and other conceptual models are expressed in a specialised language that users have to learn. Because scenarios invoke specific memory schema they help to recruit specific knowledge. This helps tune our critical faculties, since detail tends to provide more subject matter to detect inconsistencies and errors when we reason about models and specifications. In contrast, models are harder to comprehend because they represent abstract generalisations. While people form categorial abstractions naturally ( Rosch et al., 1976 ), we are less efficient at forming categories of concepts and functions ( Hampton, 1988 ). Unfortunately, formation of conceptual-functional categories is a necessary part of the generalisation process, so users can find reasoning about even simple conceptual models, such as data flow diagrams, difficult ( Sutcliffe and Maiden, 1992 ). Once learned, models become memory schema that represent abstract concepts removed from everyday experience, so their effectiveness depends on how well connected they are to more specialised memory schema representing scenario-based knowledge. The importance of the connection becomes clear when we try to validate models. Without any connection to specialised knowledge I can accept the validity of the general concept simply because it has a wide scope of meaning. For example, I might accept the proposition that <all birds can fly> as a true type definition of the class <birds> in the absence of more specific knowledge of penguins, kiwis, rheas, ostriches and dodos. Models therefore need to be integrated examples and scenarios and, furthermore, cannot exist profitably without them; indeed, human categorial memory is probably an integration of abstract models and specific examples ( Lakoff and Johnson, 1999 ). While scenarios might be effective in grounding reasoning, their downsides lie in reasoning biases and partial mental model formation, two problems Diaper was alluding to in his criticism of Carroll's position. Confirmation bias is a well known weakness of human reasoning ( Johnson-Laird and Wason, 1983 ). We tend to seek only positive evidence to support hypotheses, so scenarios can be dangerous in supplying us with minimal evidence to confirm our beliefs. While problem statement scenarios and anti-use cases ( Alexander, 2002 ) can counteract confirmation bias we need to be wary of this downside. Another potential pathology is encysting, more usually described by the saying “cannot see the wood for the trees”. Since scenarios are detailed they can bias people away from the big picture of important design issues towards obsession with unnecessary detail. Models exist to counteract this pathology. Partial mental model formation is another weakness when we test hypotheses ( Simon, 1973 ). Scenarios can encourage this pathology by reassuring us that we have covered all aspects of the problem with a small number of scenarios. This exposes the Achilles heel of scenario-based reasoning: it is difficult, if not nearly impossible, to be confident that a necessary and sufficient set of scenarios has been gathered to escape from the partial mental model problem. 3 Generalising task models and reusing HCI knowledge Having examined the relative merits of scenario and task analysis approaches, I will now turn to the question of reuse. Reuse is well established in software engineering in the form of architectural components and code ( Basili and Weiss, 1984; Frakes, 1995 ). In HCI, knowledge has been reused for many years as guidelines, heuristics, principles or golden rules ( ISO, 1997; Nielsen, 1993; Shneiderman, 1998; Sutcliffe, 1995a ). More recently, the patterns movement has sought to encapsulate such advice in a structured format of problem context, solutions, examples, and forces (e.g. Borchers, 2001 ). Carroll's task-artefact theory (see Fig. 2 ) places reuse in the pattern genre in the more sophisticated perspective of a process by which reusable knowledge is recruited, reused and refined over generations of artefacts. Task analysis creates task models for applications de novo, even though many tasks share common properties. If components can be reused in software engineering it seems to be reasonable to expect tasks, and hence the user interfaces which support them, to be reusable as well. The potential for reuse at the interaction level has been recognised in the form of interactors ( Duke and Harrison, 1994 ), which sought to formally model frequently occurring interactive problems, such as undo, redo ( Dix and Mancini, 1998 ), as reusable components. Reusable generic tasks have also been proposed in HCI. Taxonomies have been described by Zhou and Feiner (1998) , who focus on tasks involved with visualisation of information; and Wehrend and Lewis (1990) , whose taxonomy covers generic information processing tasks with mapping to suitable visualisations. Generalised task models have also been created in the knowledge engineering literature ( Breuker and Van Der Velde, 1994 ). These studies demonstrate that the regularities of task structures can be captured and reused; the question for HCI is what utility such general models might have. 3.1 Generalised task models General models inevitably omit detail. This devalues their reuse when considerable detail has to be added during a specialisation process. If additional reusable knowledge could be attached to generalised task models, then their utility could be improved; furthermore, the model would provide a context for interpreting reusable knowledge. The Domain Theory ( Sutcliffe, 2002a ) takes this approach by positing a library of generalised task models accompanied by heuristics and guidelines for functional allocation trade-offs, highlighting problems typically encountered when defining task support requirements, and suggesting generic functional requirements. To illustrate the concept of generalised tasks, diagnostic tasks are a class of similar problems that involve identifying the reason for a potential failure or malady in an agent, proposing a cure for the problem, and then taking action to effect the remedy. The essence of the problem is the same for radically different task contexts, from diagnosis of illness in medicine to diagnostic fault-finding in electronic equipment. Even though the detailed procedures for diagnosis are domain-dependent, the problem space for functional allocation decisions can be reused. For example, the sub-goal of finding the problem causes can be automated in more deterministic domains such as fault-finding in machines, whereas most attempts to automate less predictable domains (e.g. medical diagnosis) have failed. The reusable design advice is to ‘evaluate the extent of automation according to the dependability and predictability of the domain knowledge’. This can be refined to pose questions about the accuracy of observations about the faulty agent and the predictive power of the diagnostic rules. An example of the diagnosis generalised task model derived from a fault-finding task described by Rasmussen (1986) is illustrated in Fig. 3 . This is a complex generalised task that has two major sub-structures: causal analysis and repair. The generalised task specialises to models for diagnosis of problems with living things, designed artefacts and complex systems. Fig. 3 Open in new tabDownload slide Goal hierarchy for the Diagnosis task. Fig. 3 Open in new tabDownload slide Goal hierarchy for the Diagnosis task. Diagnosis is closely associated with gathering information, either by communicating with people (symptoms in medical diagnosis) or by manipulating designed artefacts to find out why they have failed. Information requirements are signs and symptoms for determining courses and appropriate treatments. In deterministic cases when diagnostic knowledge can be captured and modelled as formal rules, the whole task can be automated as an expert system (e.g. electronic equipment fault-finding and repair). However, most diagnostic tasks are cooperative in nature, and communication is necessary first for gathering signs and symptoms of the problem and asking follow-up questions, and then for testing that the repair has worked. In the latter, the computer supplies information on the faulty system, lists of potential faults, signs and symptoms of problems, and lists of possible causes, with repair strategies for fault types. Reusable knowledge is structured into problem alerts and generic requirements: Requirements problems : Gathering information for diagnosis when the signs might not be immediately accessible. Structuring facts to create a model of the problem. Functional allocation: in well known determinist domains, diagnosis can be completely automated as an expert system, e.g. faulting-finding in car engines from electronic tests. In less well known domains only partial support for the users' task is possible by pre-processing information, e.g. screening out hypotheses that do not match with known observations. Generic information requirements : Components; sub-assemblies of the artefact, machine or agent being considered. Fault/failure history for the object/artefact instance. Typical faults found with the class of objects/artefacts. Heuristics or rules for forming diagnoses from the observed evidence; differential diagnosis hints. List of treatments coupled to diagnosed causes. Generic functional requirements : Pre-processors to sort and rank symptoms and observed problems in order of severity, importance, time, etc. Questioning of checklist dialogues to ensure full capture of diagnostic information. System model simulations to locate problems and support diagnostic reasoning. Expert system inference to diagnose possible causes from observed signs. See Sutcliffe, 2002a for further details. The relationship between families of generalised tasks is illustrated in Fig. 4 . Analysis/modelling, Diagnosis and Validation/testing are all involved in problem solving which is, in turn, a precursor of Decision-making. Navigation is not closely connected with other tasks, although navigation in its conceptual sense is necessary for Information Retrieval when wayfinding through large information spaces. Fig. 4 Open in new tabDownload slide Relationship between families of generalised tasks. Fig. 4 Open in new tabDownload slide Relationship between families of generalised tasks. One group of tasks is associated with various forms of problem solving, ranging from discovering the reasons for observed problems to designing new objects. Another group involves planning and ordering future activity in a system, scheduling and planning; while a third group concerns processing or finding information. Physical tasks are not described because their structures and procedures are closely associated with details of machines and objects in the environment. For instance, driving a car or riding a bicycle depends on details of the artefact's controls, so abstracting generalised physical tasks does not promise much reuse payoff. However, the cognitive activity associated with physical tasks can be generalised. For instance, driving a car involves Navigation and Planning generalised tasks. 3.2 Reusing generalised task models Generalised task models provide an initial conceptual model of activity in the domain. Reuse commences by identifying the generalised tasks that are implicit within a domain. In some cases, such as diagnosis and fault finding, the mapping between the concrete and abstract model is fairly obvious; however, for cases where the mapping is not apparent, synonym tables and decision tables/trees are provided to help identification ( Sutcliffe, 2002a ). Once the generalised task models have been identified, additional detail is captured to identify sub-classes in a family of generalised tasks, and the requirements problems associated with the model. The added value in reusing generalised tasks lies in accelerating the initial scope stage of task analysis and in the advice and trade-off knowledge attached to the models. Our experience to date indicates that generalised tasks can be difficult to identify in new applications ( Sutcliffe, 1997 ), which agrees with the common finding that abstract models are not easy for most people to identify and comprehend ( Guindon, 1987 ). Once the abstract model has been identified, then the process becomes much easier and reusable advice can prevent software designers making repetitive errors. However, generalised task models, as well as models created by conventional task analysis, are still abstractions which do not relate well to people's everyday experience. This limitation led to investigation into how generalised models and more specific knowledge in claims and scenarios might be integrated. 3.3 Integrating scenarios, claims and generic models Claims offer more targeted delivery of HCI knowledge by virtue of the task-artefact cycle that provides a context. There are different utility trade-offs for knowledge reuse between general models (e.g. walkthroughs: Wharton et al., 1994 ; generalised task models: Sutcliffe, 2002a; Wehrend and Lewis, 1990 ) and richer yet domain-specific models ( Timmer and Long, 1997 ). More general models can target a wider range of domains but at a penalty of delivering less knowledge, while the converse is true for domain-specific models. Since claims have a domain-specific anchor in the artefact context, if they could be integrated with models, then insight into the general to domain-specific trade-offs may be gained. To solve this problem we ( Sutcliffe and Carroll, 1998 ) associated claims with generic domain models. Claims are linked to object system models ( Sutcliffe, 2002a ) if they concern problems of transaction processing and real world interaction; for instance, many human factors problems are associated with monitoring tasks. Other examples are claims for task support functionality in hiring/loans, sales order processing, and logistics/distribution application classes. If claims pertained to a particular task then they are indexed to generalised tasks that match their original context; for instance, claims based on a medical expert system for diagnosis or television fault diagnosis would be indexed to the diagnostic generalised task. Claims have upsides and downsides that present usability arguments as trade-offs in psychologically based design rationale. Claims are generated by basing a design on task analysis and influences from theory, evaluating the usability of artefacts then extracting the claim and iteratively improving the design. The artefact becomes an example that instantiates the claim, while the claim and links to theory provide justification about why the claim should deliver usability, with supplementary information on dependencies and design trade-offs. It is interesting to note that Carroll has acknowledged the role of task analysis in task-artefact theory, although his more recent writings have demoted the influence of task analysis in design. Claims are situated in a context not only by a scenario of use and the artefact that helps designers understand how to apply usability arguments, but also by a generalised task model that provides an application context for future reuse, hence overcoming some of the limitations of domain-specific knowledge. The extended knowledge representation schema for claims ( Sutcliffe and Carroll, 1998, 1999 ) and an example of a claim are illustrated in Figs. 5 and 6 . Fig. 6 Open in new tabDownload slide Example of a claim (after Carroll et al., 1992 ). Fig. 6 Open in new tabDownload slide Example of a claim (after Carroll et al., 1992 ). Fig. 5 Open in new tabDownload slide Schema for representing claims-related knowledge, illustrating relationships between scenarios, claims and generalised tasks. Fig. 5 Open in new tabDownload slide Schema for representing claims-related knowledge, illustrating relationships between scenarios, claims and generalised tasks. By associating claims in this manner the designer can have the best of both worlds. Claims with their associated artefacts and scenarios provide grounded examples of design advice and a specific context, which can be readily assimilated. Generalised models, on the other hand, enable the scope of applicability to be evaluated beyond the narrow context delineated by the scenario and artefact. Furthermore, claims can be associated with interaction patterns to relate HCI knowledge to software engineering use cases. An interaction pattern for the generalised task of information retrieval is illustrated in Fig. 7 with associated claims based on research into information-seeking behaviour ( Sutcliffe and Ennis, 1998 ) and information visualisation ( Sutcliffe and Patel, 1996 ). Fig. 7 Open in new tabDownload slide Interaction pattern for the Information Retrieval generalised task. Claims are attached to either the whole task or specific components depending on the original task/artefact context. Fig. 7 Open in new tabDownload slide Interaction pattern for the Information Retrieval generalised task. Claims are attached to either the whole task or specific components depending on the original task/artefact context. The first three claims in Fig. 7 describe trade-offs for design features which help query formulation by providing libraries of pre-formed queries, conceptual maps of the database or an active thesaurus with suggests alternative query terms. The second claim, conceptual maps, was derived from research on information display artefacts ( Sutcliffe and Patel, 1996 ) and is described in more detail in Fig. 8 . Fig. 8 Open in new tabDownload slide Visual conceptual maps claim. Fig. 8 Open in new tabDownload slide Visual conceptual maps claim. This view of claims is similar to the schema of patterns which recommend that design advice is presented in the context of a motivating problem, and with an example of its application ( Borchers, 2001 ). Although patterns do have a clause that indicates the range of problems the design advice can be applied to, this scoping is ad hoc. Advocates of patterns proposed relationships between individual patterns constructed into a hypertext-like pattern network or language ( Alexander et al., 1977 ) to set the context. Unfortunately, pattern languages are also ad hoc and tend to be incomplete. 4 Integrating contributions So far I have argued for a third development paradigm coexisting alongside task analysis as advocated by Diaper and scenario-based design as proposed by Carroll. Furthermore, I have suggested we might be able to have the best of both worlds by integrating specific grounded knowledge contained with scenarios, and more generalised knowledge in claims and generic task models. Unfortunately, there are obstacles in the way of using multiple representations, even though many advocate them in HCI and software engineering ( Mylopoulos et al., 1999; Paterno, 1999 ). In reality it is difficult to get practitioners to accept complex representations, and even simple ones get misused and customised to individuals' needs. Take MacLean's ( MacLean et al., 1991 ) QOC variant of design rationale, for example. This is a simple representation of a design question, alternative solutions and evaluation criteria for the solutions. However, QOC has been difficult to introduce into new communities of practice ( MacLean and McKerlie, 1995 ), and similar problems have been encountered with the gIBIS ( Conklin et al., 1988 ) version of design rationale ( Sutcliffe and Ryan, 1997 ), ( Buckingham Shum, 1996 ). Task analysis as a method or notation has not been readily adopted in practice ( Bellotti, 1988; Diaper 1999 ) apart from the human factors safety engineering community where it is the notational lingua franca (e.g. Hierarchical Task Analysis: Annett, 1996 ). Carroll in his more recent work has simplified claims ( Carroll, 2000; Rosson and Carroll, 2001 ), abandoning complex formatting. Claims are presented as simple design principles, in association with a motivating scenario and occasionally an artefact. In terms of process, Carroll advocates a more creative view of design, with scenarios playing roles of thought prostheses and challenges of experience for design. Hence modelling and reuse are marginalised, which I argue sends improving HCI back into the evolutionary crawl of craft-level knowledge, following Dowell and Long's (1998) metaphor. So is there a synthesis for model analytic and creative exploration approaches to design? A partial answer is acknowledging the ‘horses for courses’ argument. A differentiation between formal model-analytic and creative exploratory design approaches will always be necessary for applications which are more safety critical on one hand and those oriented to entertainment, education, and general commerce on the other. A more satisfactory answer is to set the approaches in the context of where interactive systems design is going in the future. Systems are becoming more intelligent, ubiquitous, collaborative and pervasive. Design is extending beyond the interface to artefacts, products and information appliances ( Norman, 1999 ). Carroll's design approach with scenarios has agility and adaptability on its side, but can it scale? In Carroll's exploration of educational technology the scenarios hardly explore the problems of shared awareness, access control to shared artefacts, and synchronisation of interaction, all acknowledged problems in CSCW for which considerable reusable knowledge already exists ( Olson and Olson, 2002 ). Diaper's criticism of the scope bias in scenarios makes this point. However, task modelling has hardly matured to deal with the complexities of mobile and ubiquitous computing, so a pessimistic conclusion is that neither model-based nor scenario-based approaches to interactive system design are able to equip designers with the necessary tools to deal with the complexities of rapidly evolving technology. However, I am not encouraging HCI to abrogate its responsibilities as an engineering discipline in the face of technological advance. The synthesis between the three paradigms becomes clearer when the nature of methodological interventions in the design process is examined. Methods, guidelines, principles and models are rarely used explicitly by expert designers ( Guindon and Curtis, 1988 ). Novices might use them ab initio, but design knowledge soon becomes internalised as the designer's skill. I argue that reusable knowledge in the form of generic task models, claims, and design rationale should become part of the skill-set of interactive system designers. Task analysis can contribute by providing a knowledge representation schema for reusable models and a process for specialising more generic models. Scenarios, as Carroll suggests, support the design process at run time as probes to test assumptions and stimulate creation. So both approaches have their contributions to make, albeit in different phases of design education and practice. Scenarios can stimulate thought, but knowledge can only be reused effectively in a generalised form as task models, claims, principles and guidelines. The interesting research question is when explicit recording of tasks (or other models) is necessary for design introspection and passing on for reuse. My final reflection concerns the contribution of theory, which has been absent from the preceding debate. Without theory, scenarios, task models, principles and artefacts are only reflections of best practice. Unless we advance basic understanding by developing interaction theory and the synthesis of psychological theory with interactive system design, advance in design quality and efficiency for the complexities of collaborative, ubiquitous systems will only improve at the pace of an indecisive tortoise. References Alexander, 2002 Alexander I. , Initial industrial experience of misuse cases in trade-off analysis Proceedings of RE-02 IEEE Joint International Conference on Requirements Engineering, Essen 2002 IEEE Computer Society Press , Los Alamitos, CA OpenURL Placeholder Text WorldCat Alexander et al., 1977 Alexander C. Ishikawa S. Silverstein M. , A Pattern Language 1977 Oxford University Press , Oxford Annett, 1996 Annett J. , Recent developments in hierarchical task analysis Robertson S.A. Contemporary Ergonomics 1996 Taylor Francis , London OpenURL Placeholder Text WorldCat Basili and Weiss, 1984 Basili V.R. Weiss D.M. , A methodology for collecting valid software engineering data , IEEE Transactions on Software Engineering 10 ( 6 ) 1984 ) 728 – 738 Google Scholar Crossref Search ADS WorldCat Bellotti, 1988 Bellotti V. , Implications of current design practice for the use of HCI techniques Jones D.M. Winder R.L. People and Computers, IV: HCI-88 1988 Cambridge University Press , Cambridge 79 – 96 OpenURL Placeholder Text WorldCat Benyon and Macaulay, 2002 Benyon D. Macaulay C. , Scenarios and the HCI-SE design problems , Interacting with Computers 14 ( 4 ) 2002 ) 397 – 405 Google Scholar Crossref Search ADS WorldCat Borchers, 2001 Borchers J. , A Pattern Approach to Interaction Design 2001 Chichester , Wiley Breuker and Van Der Velde, 1994 Breuker J. Van Der Velde W. , CommonKADS Library for Expertise Modelling 1994 IOS Press , Amsterdam Buckingham Shum, 1996 Buckingham Shum S. , Analyzing the usability of a design rationale notation Moran T.P. Carroll J.M. Design Rationale: Concepts, Techniques and Use 1996 Lawrence Erlbaum Associates , Hillsdale, NJ 185 – 215 OpenURL Placeholder Text WorldCat Carey, 2002 Carey T. , Commentary on ‘Scenarios and task analysis’ by Dan Diaper , Interacting with Computer 14 ( 4 ) 2002 ) 411 – 412 Google Scholar Crossref Search ADS WorldCat Carroll, 1995 Carroll J.M. Scenario-Based Design: Envisioning Work and Technology in System Development 1995 Wiley , New York Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Carroll, 2000 Carroll J.M. , Making Use: Scenario-based Design of Human–Computer interactions 2000 MIT Press , Cambridge, MA Carroll, 2002 Carroll J.M. , Making use is more than a matter of task analysis , Interacting with Computers 14 ( 5 ) 2002 ) 619 – 627 Google Scholar Crossref Search ADS WorldCat Carroll and Rosson, 1992 Carroll J.M. Rosson M.B. , Getting around the task-artifact framework: How to make claims and design by scenario , ACM Transactions on Information Systems 10 ( 2 ) 1992 ) 181 – 212 Google Scholar Crossref Search ADS WorldCat Carroll et al., 1992 Carroll J.M. Singley M.K. Rosson M.B. , Integrating theory development with design evaluation , Behaviour and Information Technology 11 ( 1992 ) 247 – 255 Google Scholar Crossref Search ADS WorldCat Cockburn, 2001 Cockburn A. , Writing Effective use Cases 2001 Addison-Wesley , Boston, MA Conklin et al., 1988 Conklin J. Begeman M.L. , GIBIS: a hypertext tool for exploratory policy discussion , ACM Transactions on Office Information Systems 64 ( 1988 ) 303 – 331 Google Scholar Crossref Search ADS WorldCat Dearden et al., 2000 Dearden A. Harrison M. Wright P. , Allocation of function: scenarios, context and the economies of effort , International Journal of Human–Computer Studies 52 ( 2 ) 2000 ) 289 – 318 Google Scholar Crossref Search ADS WorldCat Diaper, 1999 Diaper D. , Task analysis for knowledge descriptions (TAKD): a requiem for a method , Behaviour and Information Technology 20 ( 3 ) 1999 ) 199 – 212 Google Scholar Crossref Search ADS WorldCat Diaper, 2002 Diaper D. , Scenarios and task analysis , Interacting with Computers 14 ( 4 ) 2002 ) 379 – 395 Google Scholar Crossref Search ADS WorldCat Diaper and Johnson, 1989 Diaper D. Johnson P. , Task analysis for knowledge descriptions: theory and application in training Long J. Whitefield A. Cognitive Ergonomics for Human–Computer Interaction 1989 Cambridge University Press , Cambridge OpenURL Placeholder Text WorldCat Dix and Mancini, 1998 Dix A. Mancini R. , Specifying history and backtracking mechanism Palanque P. Paterno F. Formal Methods in Human–Computer Interaction 1998 Springer , London 1 – 23 OpenURL Placeholder Text WorldCat Djajadiningrat et al., 2000 Djajadiningrat J.P. Gaver W.W. Frens J.W. , Interaction relabelling and extreme characters: Methods for exploring aesthetic interactions Boyarski D. Kellogg W.A. Conference Proceedings: DIS2000 Designing Interactive Systems: Processes, Practices Methods and Techniques 2000 ACM Press , New York 66 – 71 OpenURL Placeholder Text WorldCat Dowell and Long, 1998 Dowell J. Long J.L. , A conception of the cognitive engineering design problem , Ergonomics 41 ( 2 ) 1998 ) 126 – 139 Google Scholar Crossref Search ADS WorldCat Duke and Harrison, 1994 Duke D.J. Harrison M.D. , Unifying views of interactors Proceedings: Workshop on Advanced Visual Interfaces, AVI'94, Bari, Italy 1994 ACM Press , New York OpenURL Placeholder Text WorldCat Eason et al., 1996 Eason K.D. Harker S.D.P. Olphert C.W. , Representing socio-technical systems options in the development of new forms of work organisation , European Journal of Work and Organisational Psychology 5 ( 3 ) 1996 ) 399 – 420 Google Scholar Crossref Search ADS WorldCat Frakes, 1995 Frakes W.B. Kent A. Williams J.G. Software Reuse 1995 Marcel Dekker , New York 179 – 184 Graham, 1996 Graham L. , Task scripts, use cases and scenarios in object oriented analysis , Object-Oriented Systems 3 ( 1996 ) 123 – 142 OpenURL Placeholder Text WorldCat Guindon, 1987 Guindon, R., 1987. A model of cognitive processes in software design: An analysis of breakdowns in early design activities by individuals. MCC Technical Report STP-283-87. Austin, TX: Microelectronics and Computer Technology Corporation. Guindon and Curtis, 1988 Guindon R. Curtis B. , Control of cognitive processes during software design: What tools are needed Soloway E. Frye D. Sheppard S.B. Human Factors in Computing Systems: CHI 88 Conference Proceedings 1988 ACM Press , New York 263 – 269 OpenURL Placeholder Text WorldCat Hampton, 1988 Hampton J.A. , Disjunction in natural categories , Memory and Cognition 16 ( 1988 ) 579 – 591 Google Scholar Crossref Search ADS PubMed WorldCat ISO, 1997 ISO, 1997. ISO 9241: Ergonomic requirements for office systems with visual display terminals (VDTs). International Standards Organisation. Jacobson et al., 1992 Jacobson I. Christerson M. Jonsson P. Overgaard G. , Object-oriented Software Engineering: A Use-case Driven Approach 1992 Addison-Wesley , Reading, MA Johnson et al., 1988 Johnson P. Johnson H. Waddington R. Shouls R. Jones D.M. Winder R. Proceedings: HCI'88 Proceedings: HCI'88 1988 Cambridge University Press , Cambridge 35 – 61 OpenURL Placeholder Text WorldCat Johnson-Laird and Wason, 1983 Johnson-Laird P.N. Wason P.C. , Thinking: Readings in Cognitive Science 1983 Cambridge University Press , Cambridge Kaindl, 1995 Kaindl H. , An integration of scenarios with their purposes in task modelling Olson G.M. Schuon S. Designing Interactive Systems: DIS 95 Conference Proceedings, Ann Arbor, MI 1995 ACM Press , New York 227 – 235 OpenURL Placeholder Text WorldCat Kuutti, 1995 Kuutti K. , Workprocess: scenarios as a preliminary vocabulary Carroll J.M. Scenario Based Design 1995 Wiley , New York OpenURL Placeholder Text WorldCat Kyng, 1995 Kyng M. , Creating contexts for design Carroll J.M. Scenario Based Design 1995 Wiley , New York 85 – 108 OpenURL Placeholder Text WorldCat Lakoff and Johnson, 1999 Lakoff G. Johnson M. , Philosophy in the Flesh: The Embodied Mind and its Challenge to Western Thought 1999 Basic Books , New York MacLean and McKerlie, 1995 MacLean, A., McKerlie, D., 1995. Design space analysis and user-representations (Technical Report EPC-1995-102). Cambridge: Xerox Research Centre Europe. MacLean et al., 1991 MacLean A. Young R.M. Bellotti V. Moran T.P. , Questions, options and criteria: Elements of design space analysis , Human–Computer Interaction 6 ( 3/4 ) 1991 ) 201 – 250 Google Scholar Crossref Search ADS WorldCat Monk and Wright, 1993 Monk A.G. Wright P. , Improving Your Human–computer Interface: A Practical Technique 1993 Prentice Hall , Englewood Cliffs, NJ Mylopoulos et al., 1999 Mylopoulos J. Chung L. Yu E. , From object-oriented to goal-oriented requirements analysis , Communications of the ACM 42 ( 1 ) 1999 ) 31 – 37 Google Scholar Crossref Search ADS WorldCat Nielsen, 1993 Nielsen J. , Usability Engineering 1993 Academic Press , London Norman, 1999 Norman D.A. , The Invisible Computer: Why Good Products Can Fail, the Personal Computer is so Complex, and Information Appliances are the Solution 1999 MIT Press , Cambridge, MA Olson and Olson, 2002 Olson G.M. Olson J.S. , Groupware and computer supported cooperative work Jacko J.A. Sears A. The Human–Computer Interaction Handbook 2002 Lawrence Erlbaum Associates , Mahwah, NJ 583 – 595 OpenURL Placeholder Text WorldCat Paterno, 1999 Paterno F. , Model-based Design and Evaluation of Interactive Applications 1999 Springer , Berlin Paterno, 2002 Paterno F. , Commentary on scenarios and task analysis by Dan Diaper , Interacting with Computers 14 ( 4 ) 2002 ) 407 – 409 Google Scholar Crossref Search ADS WorldCat Potts et al., 1994 Potts C. Takahashi K. Anton A.I. , Inquiry-based requirements analysis , IEEE Software 11 ( 2 ) 1994 ) 21 – 32 Google Scholar Crossref Search ADS WorldCat Rasmussen, 1986 Rasmussen J. , Information Processing in Human–Computer Interaction: An Approach to Cognitive Engineering 1986 Amsterdam , North Holland Rodriquez and Scapin, 1997 Rodriquez F.G. Scapin D. , Editing MAD* task description for specifying user interfaces at both semantic and presentation levels Harrison M.D. Torres J.C. Conference Proceedings: DSVIS 97: Design, Specification and Verification of Interactive Systems, Granada Spain 1997 Springer , Berlin 193 – 208 OpenURL Placeholder Text WorldCat Rolland et al., 1998 Rolland C. Achour C.B. Cauvet C. Ralyte J. Sutcliffe A.G. Maiden N.A.M. et al. , A proposal for a scenario classification framework , Requirements Engineering 3 ( 1 ) 1998 ) 23 – 47 Google Scholar Crossref Search ADS WorldCat Rosch et al., 1976 Rosch E. Mervis C.B. Gray W. Johnson D. Boyes-Braem P. , Basic objects in natural categories , Cognitive Psychology 7 ( 1976 ) 573 – 605 Google Scholar Crossref Search ADS WorldCat Rosson and Carroll, 2001 Rosson M.B. Carroll J.M. , Usability Engineering: Scenario-based Development of Human–Computer Interaction 2001 Morgan Kaufmann , San Francisco Shneiderman, 1998 Shneiderman B. , Designing the user Interface: Strategies for Effective Human–Computer Interaction third ed. 1998 Addison-Wesley/Longman Simon, 1973 Simon H.A. , The structure of ill-structured problems , Artificial Intelligence 4 ( 1973 ) 181 – 201 Google Scholar Crossref Search ADS WorldCat Sutcliffe, 1989 Sutcliffe A.G. , Task analysis, system analysis and design: Symbiosis or synthesis , Interacting with Computers 1 ( 1 ) 1989 ) 6 – 12 Google Scholar Crossref Search ADS WorldCat Sutcliffe, 1995a Sutcliffe A.G. , Human–Computer Interface Design second ed 1995 Macmillan , London Sutcliffe, 1995b Sutcliffe A.G. , Task Related Information Analysis Benyon D. Palanque P. Proceedings: IFIP 13.2/2.7 Workshop on Methods and Design Architectures for User Interface Design 1995 OpenURL Placeholder Text WorldCat Sutcliffe, 1997 Sutcliffe A.G. , Task-related information analysis , International Journal of Human–Computer Studies 47 ( 2 ) 1997 ) 223 – 257 Google Scholar Crossref Search ADS WorldCat Sutcliffe, 2000a Sutcliffe A.G. , Bridging the communications gap: developing a lingua franca for software developers and users Actes du XVIIIe Congres: INFORSID, Lyon 2000 Inforsid , Toulouse OpenURL Placeholder Text WorldCat Sutcliffe, 2000b Sutcliffe A.G. , On the effective use and reuse of HCI knowledge , ACM Transactions on Computer–Human Interaction 7 ( 2 ) 2000 ) 197 – 221 Google Scholar Crossref Search ADS WorldCat Sutcliffe, 2002a Sutcliffe A.G. , The Domain Theory: Patterns for Knowledge and Software Reuse 2002 Lawrence Erlbaum Associates , Mahwah, NJ Sutcliffe, 2002b Sutcliffe A.G. , Modelling collaboration in loosely coupled inter-organisational relationships Systems Engineering for Business Process Change: Collected Papers from the EPSRC Research Programme 2002 Springer , London OpenURL Placeholder Text WorldCat Sutcliffe and Carroll, 1998 Sutcliffe A.G. Carroll J.M. , Generalizing claims and reuse of HCI knowledge Johnson H. Nigay L. Roast C. People and Computers XIII; Proceedings: BCS-HCI Conference, Sheffield 1–4 September 1998 Springer , Berlin 159 – 176 OpenURL Placeholder Text WorldCat Sutcliffe and Carroll, 1999 Sutcliffe A.G. Carroll J.M. , Designing claims for reuse in interactive systems design , International Journal of Human–Computer Studies 50 ( 3 ) 1999 ) 213 – 241 Google Scholar Crossref Search ADS WorldCat Sutcliffe and Ennis, 1998 Sutcliffe A.G. Ennis M. , Towards a cognitive theory of information retrieval , Interacting with Computers 10 ( 3 ) 1998 ) 321 – 351 Google Scholar Crossref Search ADS WorldCat Sutcliffe and Maiden, 1992 Sutcliffe A.G. Maiden N.A.M. , Analysing the novice analyst: cognitive models in software engineering , International Journal of Man–Machine Studies 36 ( 5 ) 1992 ) 719 – 740 Google Scholar Crossref Search ADS WorldCat Sutcliffe and Patel, 1996 Sutcliffe A.G. Patel U. , 3D or not 3D: Is it nobler in the mind? Sasse M.A. Cunningham R.J. Winder R.L. People and Computers XI. Proceedings: HCI-96, London August 1996 Springer , London 79 – 94 OpenURL Placeholder Text WorldCat Sutcliffe and Ryan, 1997 Sutcliffe A.G. Ryan M. , Assessing the usability and efficiency of design rationale Howard S. Hammond J. Lindgaard G. Human Computer Interaction INTERACT-97 1997 IFIP/Chapman and Hall 148 – 155 OpenURL Placeholder Text WorldCat Timmer and Long, 1997 Timmer P. Long J. , Separating user knowledge of domain and device: a framework Thimbleby H. O'Conaill B. Thomas p People and Computers XII: Proceedings of the HCI 97 Conference 1997 379 – 396 OpenURL Placeholder Text WorldCat Van Der Veer and VanWelie, 2000 Van Der Veer G.C. VanWelie M. , Task-based groupware design: Putting theory into practice Boyarski D. Kellogg W.A. Conference Proceedings: DIS2000 Designing Interactive Systems: Processes, Practices Methods and Techniques, New York 2000 ACM Press , New York 326 – 337 OpenURL Placeholder Text WorldCat Van Der Veer et al., 1996 Van Der Veer G.C. Lenting B.F. Bergevoet B.A.J. , GTA: Groupware Task Analysis: Modeling complexity , Acta Psychologica 91 ( 1996 ) 297 – 322 Google Scholar Crossref Search ADS WorldCat Van Lamsweerde and Letier, 2000 Van Lamsweerde A. Letier E. , Handling obstacles in goal-oriented requirements engineering , IEEE Transactions on Software Engineering 26 ( 10 ) 2000 ) 978 – 1005 Google Scholar Crossref Search ADS WorldCat Vicente, 2000 Vicente K.J. , HCI in the global knowledge-based economy: designing to support worker adaptation , ACM Transactions on Computer–Human Interaction 7 ( 2 ) 2000 ) 263 – 280 Google Scholar Crossref Search ADS WorldCat Wehrend and Lewis, 1990 Wehrend R. Lewis C. , Proceedings: First IEEE Conference on Visualization: Visualization 90 Proceedings: First IEEE Conference on Visualization: Visualization 90 1990 IEEE Computer Society Press , Los Alamitos, CA OpenURL Placeholder Text WorldCat Wharton et al., 1994 Wharton C. Reiman J. Lewis C. Polson P. , The cognitive walkthrough method: a practitioners guide Nielsen J. Mack R.L. Usability Inspection Methods 1994 Wiley , New York 105 – 140 OpenURL Placeholder Text WorldCat Wright et al., 2000 Wright P. Dearden A. Fields R. , Function allocation: A perspective from studies of work practice , International Journal of Human–Computer Studies 52 ( 2 ) 2000 ) 335 – 355 Google Scholar Crossref Search ADS WorldCat Yu, 1993 Yu E.S.K. , Modelling organisations for information systems requirements engineering Fickas S. Finkelstein A.C.W. Proceedings: First International Symposium on Requirements Engineering—RE'93, San Diego, CA 1993 IEEE Computer Society Press , Los Alamitos, CA 34 – 41 OpenURL Placeholder Text WorldCat Zhou and Feiner, 1998 Zhou M.X. Feiner S.K. , Visual task characterization for automated visual discourse synthesis Karat C.M. Lund A. Coutaz J. Karat J. Human Factors in Computing Systems: CHI 98 Conference Proceedings, Los Angeles 1998 ACM Press , New York 392 – 399 OpenURL Placeholder Text WorldCat © 2003 Elsevier B.V. All rights reserved.
TI - Symbiosis and synergy? scenarios, task analysis and reuse of HCI knowledge
JO - Interacting with Computers
DO - 10.1016/S0953-5438(03)00002-X
DA - 2003-04-01
UR - https://www.deepdyve.com/lp/oxford-university-press/symbiosis-and-synergy-scenarios-task-analysis-and-reuse-of-hci-pNDEPTzTZ0
SP - 245
EP - 263
VL - 15
IS - 2
DP - DeepDyve
ER -