Adam Stork and John Long
Ergonomics and HCI Unit University College London
26 Bedford Way London WC1H 0AP
Accepted for Publication: 1994 . In Proceedings of the International Working Conference on Home-Oriented Informatics, Telematics and Automation, Amager, Denmark.
Abstract
This paper seeks to contribute to more effective design practice in the longer term both for Human Computer Interaction in general and for Home Informatics in particular. The ultimate goal of more effective design practice is held to be practice supported by ‘general design principles’ (Dowell and Long 89), which are conceptualised, operationalised, tested, and generalised. The value of such a goal can be seen in the ‘hard’ Engineering disciplines, such as Electrical or Mechanical Engineering, which have general design principles that are better conceptualised, operationalised, tested, and generalised than those in other design disciplines, such as the emergent discipline of Human Computer Interaction. These Engineering disciplines can offer earlier success and a better guarantee in the solution of design problems.
The paper first argues that a pre-requisite of, and so one potential contribution towards, the development of general design principles for Human Computer Interaction would be to develop operationalisations of specific design problems and specific design solutions that are better conceptualised and operationalised than existing such operationalisations. The initial conceptualisation and operationalisation of a specific design problem is then described.
The specific design problem, object of the case study reported here, was one of the failure to maintain desired comfort and costs by a heating system of a typical home. The resulting specific design problem specification is considered to be better conceptualised and operationalised than other specifications of such design problems. It expresses the domain of application and interactive worksystem as they relate to the design problem of supporting the planning and control of comfort and costs of heating in the home.
The paper concludes that the conceptualisation and operationalisation raise no undue concerns for the aim of developing general design principles in the longer term and demonstrates some modest progress towards this aim. Further research towards achieving this aim is identified.
Keywords
Energy Management Systems, Engineering, Home Informatics, Human Computer Interaction, Planning and Control
Introduction: Scope and Aims of this Paper
Technology continues to penetrate the home (Home Informatics). Even the home, however, is a challenge for the discipline of Human Computer Interaction (HCI), an emerging discipline which has demonstrated only limited success in other more developed technological areas. This paper describes work that aims to improve the success of HCI in the longer term for Home Informatics in particular, but also for HCI in general.
The paper describes work that is part of an overall effort to improve HCI. The research can be summarised by describing the ultimate goal of HCI knowledge to be practice supported by ‘general design principles’ (Dowell and Long 89), which are conceptualised, operationalised, tested, and generalised to offer an earlier and a better guarantee of success in application. A strategy is described for the development of general design principles for which a pre-requisite is the development of operationalisations of specific HCI design problems and specific HCI design solutions that are better conceptualised and operationalised than existing such operationalisations.
The work can be summarised as the development of a conception of the general HCI design problem and the operationalisation of that conception for a specific HCI design problem. This operationalisation is to be better conceptualised and operationalised than existing such operationalisations.
The achievements towards, and implications for, the overall research consequent on this detailed work are then considered.
Engineering: The Likely Future of Human Computer Interaction
Dowell and Long have made proposals concerning the nature of the discipline of HCI (Dowell and Long 89; Long and Dowell 89). Firstly, they have characterised HCI as a design discipline rather than as a scientific discipline: HCI knowledge supports HCI practice which is to provide solutions to general HCI design problems. Secondly, they have identified from this statement a means of assessing the effectiveness of HCI knowledge and used this means to argue for certain properties of that knowledge for it to be effective. These properties can be used to assess the current HCI discipline and propose a more effective (a possible future) discipline.
The effectiveness of HCI knowledge is dependent on its ability to support HCI practice to solve general HCI design problems. Dowell and Long identify four necessary properties of effective HCI knowledge: it must be conceptualisable; operationalisable; testable; and generalisable.
The current state of the HCI discipline is characterised as that of a ‘Craft’ discipline. Knowledge is currently implicit and informal, consisting of ‘heuristics’; and practice is that of ‘implement and test’ (and iterate). Heuristics are poorly, if at all, conceptualised (often only through example practice), which leads to them being difficult to operationalise, therefore unlikely to be testable, and so unlikely to be generalisable. This characterisation represents an understanding as to the current state of the poor guarantee of success offered by current HCI knowledge and practice.
The alternative, more effective, HCI discipline is characterised as that of an ‘Engineering’ discipline. Knowledge would need to be conceptualised, with explicit, complete, consistent, and formal definitions, to be operationalisable, testable, and generalisable. Dowell and Long call this knowledge ‘engineering principles’, and practice would be that of ‘specify then implement’. HCI engineering principles would offer a high guarantee of success of application in practice, similar to that currently enjoyed by the existing (‘hard’) Engineering disciplines like Electrical or Mechanical Engineering.
A Strategy for Developing Human Computer Interaction Engineering
The development of HCI engineering principles requires the development of knowledge to support practice. HCI practice is the provision of general design solutions to general design problems, which are general over types of user, types of computer, and types of domain of application. HCI knowledge contains general relationships between general HCI design problems and general HCI design solutions.
It is suggested that one possible means of developing this HCI knowledge is to identify general relationships between general design problems and solutions from specific relationships between specific design problems and solutions, where a specific design problem and solution is defined as being particular to a user, a computer system, and a domain of application. Such a development requires the conceptualisation and operationalisation of specific HCI design problems and solutions from a conception of the complete general HCI design problem and solution which is general over all types of user, all types of computer, and all types of domain of application, hereafter referred to as ‘a conception of the general HCI design problem and solution’.
A conception is understood to be a set of concepts, which are abstractions on a class of objects based on their common aspects, and their relations. Conceptualisation is the process of generating a conception. Operationalisation is the process of instancing a conception to produce an operationalisation. An operationalisation of a conception is a set of less abstract concepts (related to the concepts in the conception) that reference observables in the ‘real’ world. It is convenient to introduce the concept of ‘a set of metrics’ and the process of ‘metrication’. Metrication is the process of instancing an operationalisation (in the limit) to produce a set of metrics. Metrics quantify the less abstract concepts of the operationalisation in an observable relation with the ‘real’ world.
For the knowledge to support engineering principles, the operationalisation of the specific design problems and solutions needs to be explicit and formal, where formal is understood as a representation that has defined rules of syntax and semantics, and is therefore understandable by some people for some purpose. Formality requires the metrication of the operationalisations of the conceptions of the specific HCI design problem and solution.
It is proposed that tractable re-design problems be selected, then described as ‘user requirements’, and finally design solutions to these problems developed using best current HCI practice (craft ‘heuristics’, including evaluation, applied by an HCI designer). These design problems and solutions will be operationalised according to the conception of the specific HCI design problem. It is considered easier to operationalise installed design solutions than to operationalise design problems. The selection of re-design problems, instead of design problems, enables the operationalisation (according to the conception of the specific HCI design solution) of the existing (installed) design ‘solution’ (to an earlier, probably implicit, design problem) to guide the operationalisation of the re-design problem. The selection of a tractable re-design problem ensures: that the differences between the operationalisations of the current design ‘solution’ and the tractable re-design problem be minimal; and that a design solution exists.
This approach is the one adopted here. The current state of the present work is that conceptions of the general HCI design problem and solution have been identified and instantiated to conceptions of the specific HCI design problem and solution. An initial tractable re-design problem has been described as ‘user requirements’. The current design ‘solution’ has been operationalised, according to the conception of the specific HCI design solution, and this operationalisation has guided the operationalisation of the initial tractable re-design problem, according to the conception of the specific HCI design problem.
The user requirements for this initial tractable re-design problem were intended to be operationalisable according to the conception of the specific design problem in order to prevent concentration, for the moment, on the relationship between the user requirements and the operationalisation of the specific design problem.
The strategy adopted to perform the operationalisations of the conceptions of the specific HCI design problem and solution was to develop an explicit operationalisation as a step towards an explicit and formal set of metrics. The current work, as presented here, describes the explicit operationalisations of the conceptions of the specific HCI design problem and solution with the specific tractable re-design problem and current specific design ‘solution’.
Conceptions of the General Human Computer Interaction Engineering Design Problem and Solution
Dowell and Long (89) have proposed a conception of the general HCI design problem and this conception is presented here with some additional detail. No attempt is made to indicate in all cases where their work ends and the present begins to achieve an uninterrupted exposition of the concepts. Readers interested in the differences, however, are directed to read the original paper by Dowell and Long.
Dowell and Long state the general HCI design problem informally as ‘the design of interactive worksystems for performance’. They propose a more precise description as follows (slightly amended for typographical considerations):
‘the design of behaviours constituting a worksystem {S} whose actual performance (Pa) conforms with some desired performance (Pd). And to design {S} would require the design of human behaviours {U} interacting with computer behaviours {C}. Hence conception of the general design problem of an engineering discipline of HCI is expressed as:
Specify then implement {U} and {C}, such that
{U} interacting with {C} = {S}Pa=Pd
where Pd=fn(Qd, Kd)
Qd expresses the desired quality of the products of work within the given domain of application;
Kd expresses acceptable (i.e., desired) costs incurred by the worksystem, i.e. by both human and computer.’
This statement expresses Dowell and Long’s distinction between the behavioural system (that is the interactive worksystem) that performs work, and the world of work (that is the domain of application, within which the work is performed). It is inferred from their work that Pa is a function of the actual quality of the products of work within a particular domain of application (Qa) and the actual costs incurred by a particular worksystem (Ka).
Furthermore, an expression of the general HCI design solution is inferred from their work as:
‘a specification and implementation of {U} and {C}, such that
{U} interacting with {C} = {S}Pa=Pd
where Pd=fn(Qd, Kd)
and Pa=fn(Qa, Ka)’
It follows from these expressions that a statement of the general HCI design problem requires a statement for the desired performance of the desired interactive worksystem whereas a statement of the general HCI design solution requires a statement of the actual performance of the actual interactive worksystem.
The components of performance are now considered, following Dowell and Long, for desired performance and actual performance. Important occurrences of the concepts are highlighted to aid their identification and permit an informal assessment that the later operationalisation indeed operationalises these concepts.
A Conception of Desired Performance
The desired performance, Pd, is conceptualised as a function of the desired quality of the products of work, Qd, within the domain of application and the acceptable or desired costs, Kd, incurred by the interactive worksystem.
The interactive worksystem boundary criteria allow assertion of the behavioural system which constitutes the interactive worksystem, that system ‘whose purpose is to achieve and satisfy … common goal[s]’. The domain boundary critieria allow assertion of the world of work which constitutes the domain of application, that world of work which is determined by the requirement to express these common goals.
A Conception of Actual Performance
Actual performance, Pa, is conceptualised as a function of the actual quality of the products of work, Qa, within the given domain of application and the current or actual costs, Ka, incurred by the worksystem.
The interactive worksystem boundary criteria and domain boundary criteria are the same as for the conception of desired performance.
A Conception of Desired Quality
Dowell and Long conceptualise the world of work as consisting of objects that have attributes that have a set of possible states (defining their affordance for change). The desired quality of the products of work to be achieved by the interactive worksystem are conceptualised as transformations of states of attributes of objects that are desirable, called product goals. These objects and their attributes are conceptualised as abstract or physical, and related or unrelated. The transformations described by a product goal can be identified for each attribute, and these transformations are termed task goals.
Dowell and Long describe the difference between abstract and physical attributes of objects as: ‘abstract attributes of objects are attributes of information and knowledge’; and ‘physical attributes of objects are attributes of energy and matter’. They also conceive that ‘different attributes of an object emerge at different levels within an hierarchy of levels of complexity’ and in general abstract attributes emerge at a higher level than physical attributes. Similarly, ‘objects are described at different levels of specification commensurate with their levels of complexity’. Furthermore, attributes of objects are related to attributes of other objects both across and within levels of complexity.
A Conception of Actual Quality
The actual quality of the products of work achieved by the interactive worksystem are conceptualised as similar to desired quality, with transformations of states of attributes of objects that are achieved, called product achieved goals, and transformations for each attribute, called task achieved goals.
A Conception of Desired Costs
Dowell and Long conceptualise the interactive worksystem (the behavioural system) as ‘human and computer behaviours together performing work’. They distinguish human behaviour as purposeful from computer behaviour as purposive. They claim that human behaviours correspond with the transformation of objects in a domain and that an expression of them must ‘at least be expressed at a level commensurate with the level of description of the transformation of objects in the domain’. These statements would appear to hold for computer and interactive worksystem behaviours.
These behaviours can be abstract or physical . Abstract behaviours ‘are generally the acquisition, storage, and transformation of information. They represent and process information at least concerning: domain objects and their attributes, attribute relations and attribute states, and the transformations required by goals’. Physical behaviours express abstract behaviours and are ‘related in an hierarchy of behaviour types’.
Dowell and Long conceptualise the user as having cognitive, conative, and affective behaviours. ‘The cognitive aspects of the user are those of knowing, reasoning and remembering, etc.; the conative aspects are those of acting, trying and persevering, etc.; and the affective aspects are those of being patient, caring, and assuring, etc.’
Dowell and Long conceptualise humans and computers as ‘having (separable) structures that support their (separable) behaviours’. Furthermore, ‘Human structures may be physical (neural, bio-mechanical, and physiological) or mental (representational schemes and processes)’. Similarly, computer structures may be physical or abstract.
Dowell and Long claim that ‘work performed by interactive worksystems always incurs resource costs’. They identify resource costs as behavioural or structural and associated with the human or the computer (separable). These costs can be further associated with abstract (mental) and physical behaviours or structures. Examples of resource costs related to the human are: physical workload for human physical behavioural costs; mental workload for human abstract (mental) behavioural costs; physical development and deterioration for human physical structural costs; and mental development and deterioration for human abstract (mental) structural costs. Examples of resource costs related to the computer are: energy emission and consumption for computer physical behavioural costs; software and functional resource (transaction and access resources) usage for computer abstract behavioural costs; system (hardware) development and degradation for computer physical structural costs; and software and functional development (and degradation) for computer abstract structural costs.
The desired costs are conceptualised as the necessary resource costs of the interactive worksystem to achieve the desired quality.
A Conception of Actual Costs
The actual costs are conceptualised as the actual resource costs of the interactive worksystem to achieve the actual quality.
Conceptions of the Specific Human Computer Interaction Engineering Design Problem and Solution
The conceptions of the specific HCI design problem and solution are operationalised from the conceptions of the general HCI design problem and solution. The specific HCI design problem and solution are particular to a user, a computer system, and a domain of application (and by definition).
The (specific) desired performance is conceptualised as a function of the desired quality of the products of work within a particular domain of application and the desired costs incurred by a particular interactive worksystem consisting of a particular user and a particular computer system.
The (specific) actual performance is conceptualised as a function of the actual quality of the products of work within a particular domain of application and the actual costs incurred by a particular interactive worksystem consisting of a particular user and a particular computer system.
An Operationalisation of a Specific HCI Design Problem
The following tractable specific HCI (re-)design problem was selected as the first specific HCI design problem to be operationalised according to the conception of the specific HCI design problem developed above. This operationalisation was guided by first operationalising the current (inadequate) design ‘solution’ consistent with the strategy adopted for this research. The following is a (shortened) statement of the user requirements for the specific HCI (re-)design problem.
‘if A stays at home to work and leaves after 8 o’clock in the morning then the house is too cold until the gas-powered central heating is turned back on. If he expects to be at home for a short time then he often uses the one-hour boost facility on the heating controller to turn the heating back on, which can result in him being too cold if he is at home for longer than expected. The current gas bill is acceptable and an increase could be tolerated although a decrease would be desirable.’
The particular user is A, the particular computer system (or device) is the energy management heating system, and the particular domain of application is the planning and control of the comfort of A in the home of A. The concepts are highlighted to aid identification and permit an informal assessment that this operationalisation in fact operationalises the earlier concepts.
The Specific Actual Performance
The specific actual performance is operationalised as the union of the specific actual quality and the specific actual costs. The interactive worksystem boundary criteria are operationalised by the requirement that the constituents of the interactive worksystem have the common goals of the current (level of) achievement and satisfaction of the planning and control of the comfort of A in the home of A using the radiators. The domain boundary criteria are operationalised by the requirement that the constituents of the domain of application express the current (level of) achievement and satisfaction of these common goals.
The Specific Desired Performance
The specific desired performance is operationalised as the union of the specific desired quality and the specific desired costs. The interactive worksystem boundary criteria are operationalised by the requirement that the constituents of the interactive worksystem have the common goals of the desired achievement and satisfaction of the planning and control of the comfort of A in the home of A using the radiators. The domain boundary criteria are operationalised by the requirement that the constituents of the domain of application express the desired achievement and satisfaction of these common goals.
The Specific Actual Quality
The specific actual domain of application has two main physical objects: A and the house. A has a physical attribute of temperature and an abstract attribute of comfort. The attribute of comfort is related to the attribute of temperature having a range of acceptable temperatures (between 35.75˚C and 37.5˚C).
The second physical object is the house, which has physical objects that are the rooms. The rooms have a physical attribute of their temperature and physical objects of the radiators. The radiators have a physical attribute of their temperature. The temperature of the room is related to the temperature of A (an approximately linear relationship) and the temperature of the radiators (related through convection, u-value of the room, etc.). The temperatures of the radiators are controlled by the heating system in the interactive worksystem.
The current states of the temperatures of the radiators result in the state of the comfort attribute of A being ‘not comfortable’ at some times. This state is a task achieved goal and defines the product achieved goal of the actual quality by interpretation of the relationships between this attribute and the other attributes in the actual domain of application.
The Specific Desired Quality
In this case, the specific desired quality is very similar to the specific actual quality of the current inadequate ‘solution’ described above. The difference is that the specific desired quality has a task goal to maintain the state of A’s comfort attribute as ‘comfortable’ instead of a task achieved goal of ‘not comfortable’. This task goal results in the product goals being different from the product achieved goals.
The Specific Actual Costs
There are two main sub-systems in the interactive worksystem: the user (A) and the heating system (a common combination boiler system with a simple two-period time controller). The heating system has the following interacting physical behaviours: receive control of the temperatures (at times) of the radiators; receive press of advance button; and receive press of a one-hour boost button. The user has the following interacting physical behaviours: perform control of the temperatures (at times) of the radiators; perform press of advance button; and perform press of one-hour boost button. The non-interacting physical behaviours include, as examples: for the heating system, fire-up the main burners from the pilot light; and for the user, move to (and from) the location of the heating controller. A further non-interacting physical behaviour of the heating system, and an example of a behaviour that corresponds with the transformation of the attributes of objects in the domain of application, is flow of hot-water to and from the radiators.
The physical structures can be derived from the physical behaviours, for example the heating controller has a physical structure of a one-hour boost button and the user has a physical structure of a body.
The abstract behaviours of the heating controller are the execution of an algorithm on some data structures, where the algorithm and data structures are the abstract structures of the heating controller. The algorithm and data structures can be described in pseudo-code as follows:
‘boostTime = off // Counter decrements with time. At zero is ‘off’.
task 1 – constant
currentTime // Maintains the current time.
heatingState = domain.radiatorOnOffState
if (currentTime > 6.40am & currentTime < 7.20am | boostTime != off)
heatingState = on
else
heatingState = off
task2 – on boostButton.pressed
if (boostTime = off)
boostTime = 1hour
else
boostTime = off’
The cognitive abstract (mental) behaviours of the user are the operation of mental processes on mental representations, where the mental processes and the mental representation are the cognitive abstract (mental) structures. The mental processes and mental representation can be described in pseudo-code as follows:
‘comfort = domain.A.comfort
leaving = leavingPlan(comfort).leaveWithinHalfHour?
if (comfort = uncomfortable.tooCold & !leaving)
moveToController
pressBoostButton
returnFromController’
The conative and affective abstract (mental) behaviours and structures are not operationalised here.
The physical behavioural costs of the heating system can be operationalised by the energy consumption and release due to the behaviours of the heating system, for example: potential energy is stored and dissipated when the boost button is pressed and released. The abstract behavioural costs of the heating system can be operationalised by the transaction and access resource usage incurred in performing the algorithm and data structures given above, for example: the frequency of access of the timer.
.
The Specific Desired Costs
The specific desired costs are similar to the specific actual costs of the current inadequate ‘solution’ described above. The differences, in this case, are due to statements in the user requirements and interpretations of the user requirements. The ‘current gas bill is acceptable and an increase could be tolerated although a decrease would be desirable’ (from the original problem statement) suggests that the operationalisation of the physical structural costs of the heating system should be within a range that allows for this desirable decrease or acceptable increase in gas usage. It is assumed that the heating system can be upgraded and, therefore, the operationalisation of the physical structural costs of the heating system should also be within a range that allows for a different installation and maintenance price.
It is assumed that the other costs should either remain the same, and be operationalised in the same manner—for example the user physical structural costs—, or decrease if possible, and be operationalised to be within a range—for example the user physical behavioural costs.
Further Research
The operationalisation of the planning and control specific HCI design problem needs to be developed further, both in terms of the range of concepts operationalised and the operationalisation to the level of metrics. A rigorous demonstration that the operationalisation indeed operationalises the concepts needs to be constructed. A design solution has been developed using best current HCI practice, and this solution will be installed and evaluated soon. The conceptualisation of the specific HCI design solution will allow this design solution to be operationalised. The conceptualisation and operationalisation of this initial specific HCI design problem and solution is expected to enable specific relationships to be identified. Further specific HCI design problems and solutions will then be addressed and general relationships, as putative general design principles, identified where possible. The overall research strategy will then be assessed and pursued or amended.
Conclusion
The current operationalisation of the specific HCI design problem can, informally, be said: to be developed from the conception of the general HCI design problem and to have failed to identify properties in the conception of the general HCI design problem that could not be operationalised. There arise, therefore, no negative implications for the conception of the general HCI design problem. Hence, there are no negative implications for the strategy for the development of general HCI design principles.
The validation of the conception and operationalisation of general HCI design problems, and the strategy for the development of general HCI design principles, will be achieved if putative general HCI design principles are produced and validated.
The current operationalisation, however, can be considered to be better conceptualised and operationalised than other operationalisations of the general HCI design problem. The very production of the operationalisation demonstrates that this conception is better than other conceptions of general HCI design problems (and HCI knowledge) that have not produced operationalisations, for example Storrs’ conception of HCI (89). The authors are not aware of other operationalisations of general HCI design problems based on this conception (or indeed other conceptions) of the general HCI design problem.
The operationalisation also serves as a interesting example that contains explicit, partly structured, but informal HCI design knowledge of a planning and control Home Informatics design problem that could be used by Craft HCI practitioners.
Overall, this work is proceeding according to a well-developed view of the requirements for a discipline of HCI for Home Informatics. The work aims to offer long term benefits towards the development of general engineering design principles for HCI and shows progress towards that aim. Such engineering general HCI design principles would offer a better and earlier guarantee, one that is significantly greater than the current (low) guarantee of existing HCI Craft-based knowledge. This improvement will benefit the challenging area for HCI of introducing technology into the home.
Acknowledgements
The research associated with this paper was carried out under a SERC CASE studentship sponsored by Schlumberger Industries. Views expressed in the paper are those of the authors and should not necessarily be attributed to SERC or Schlumberger Industries.
References
Dowell J. & Long J. B. (1989). Towards a conception for an engineering discipline of human factors. Ergonomics, November 1989, 32(11), pp. 1513-1535.
Long J.B. & Dowell J. (1989). Conceptions for the discipline of HCI: Craft, Applied Science and Engineering. In: A. Sutcliffe & L. Macaulay (eds). Proceedings of the Fifth Conference of the BCS HCI SIG, Nottingham, England, 5-8 September 1989, Cambridge: Cambridge University Press. pp. 9-32.
Storrs, G (1989). A conceptual model of human-computer interaction? Behaviour and Information Technology, 5(4), pp. 323-334.