Reliable hci design: Cognitive Engineering and Patterns Adam Stork and John Long

Reliable HCI Design:
Cognitive Engineering and Patterns

Adam Stork and John Long

Ergonomics and HCI Unit

University College London
26 Bedford Way
London WC1H 0AP, UK
[e-mail]

ABSTRACT

The aim of improving the quality of life using computer systems is one which all CHI attendees would probably support. While invention and innovation play a role in that improvement, so does the ability of Human-Computer Interaction (HCI) practitioners to reliably specify computer systems.

We, and our colleagues, have been considering how to improve our ability to reliably specify computer systems, or do ‘reliable HCI design’, under the flag of Cognitive Engineering (CE). One of our expectations is that high reliability will take a long time. In the meantime, we have applied some of our long-term considerations to current HCI design to improve the reliability of today’s HCI designs.

Design patterns are beginning to be identified for HCI design. Patterns appear to be aiming for higher reliability than current guidelines, so making them interesting to us. However, we can also outline steps, based on our long-term considerations, that might make patterns support more reliable HCI design.

Keywords

Reliable HCI Design, Interaction Design Patterns, Pattern Languages, Cognitive Engineering

LONG-TERM RELIABLE HCI DESIGN

Cognitive Engineering (CE) [4] has a view of the development of the discipline of Human-Computer Interaction (HCI) that emphasises the reliability of HCI design. The long-term aim of CE is to acquire HCI knowledge that has a high reliability.

Examples of current HCI knowledge are: existing computer systems; guidelines; methods; and scientific models. None of these examples, however supportive, could be called highly reliable.

We, and our colleagues, have made some progress towards this long-term aim, (for example, [9]), but it looks as though high reliability will take a very long time indeed.

The high reliability of CE relies on knowledge being validated for design. The current understanding of achieving this validation is that we need:

1. Rigorous terms for the concepts that HCI designers will employ. These terms should enable representation of all HCI ‘s concerns. For example, we include: separation of the ‘design problem’ from the ‘design solution’, ‘performance of the users and computers in the world of work’, ‘mental workload of the users’, and ‘the effort of designing the system’.

These terms, and the associated knowledge, need to be able to be applied to design situations. For example, we need a relationship between mental workload and what users do in a design situation.

2. Application of the knowledge to specific design with an assessment of whether the resulting solution solves the problem.

3. A view of the structure of the knowledge. For example, we distinguish substantive knowledge, like guidelines, from methodological knowledge, like design methods.

Current proposals for the structure of highly reliable substantive knowledge is that it is a ‘general relationship between specific design problems and their solutions’ [9]. For methodological knowledge, it is that it is based on formal methods.

4. Strategies for acquiring this knowledge. Current strategies include: analysing designs; analysing the doing of design; and the re-orientation of scientific models.

The above are numbered to enable later cross-referencing.

MORE RELIABLE HCI DESIGN TODAY

We, and our colleagues, have applied these concepts to current HCI knowledge to try to improve the reliability of today’s HCI designs.

• HCI guidelines have been improved by putting them into a template that identified which HCI concepts were not covered [3]

• HCI methods have been subjected to initial validation by assessing whether they have been properly applied and whether the resulting computer system meets the requirements [10].

• Scientific knowledge of planning [6] and co-ordination [7] has been made more oriented towards design and initially validated.

• Methods have been enhanced to cover the range of HCI concepts [8].

Some of the above have also met long-term aims as well as aims for today. Over time, we hope to understand how far reliable HCI design can be taken and how well this long-term understanding can give more reliable HCI design today, and vice-versa.

INTERACTION DESIGN PATTERNS

The concept of Design Patterns originated in Architecture with Alexander [1], who saw the need to identify ‘generic solutions’ to re-use design experience to support the user-centred design of spaces that enable living.

Gamma et al. [5] are the main exponents of Design Patterns in Software Engineering as a means of improving the re-usability of object-oriented software.

Patterns are gaining popularity in HCI, following a CHI’97 workshop, reported by Bayle et al. ‘Towards a Pattern Language for Interaction Design’ [2]. Bayle et al. define an Interaction Design Pattern (IDP) as a ‘pattern that describes a connection between a repeatedly encountered problem and a solution that has been proven in the field, across time and circumstance’.

IDPs are of interest to us because they appear to be starting on the route to reliable HCI design, particularly relative to current guidelines. IDPs aim to be HCI knowledge with:

1. A separation of problem from solution.

2. A focus on being proven in specific design.

3. A structure in the form of a standard template for pattern representation, giving: problem; force; solution; and comments.

4. A strategy for their acquisition through a ‘writers workshop’, a form of constructive peer criticism. There has also been discussion for software engineering patterns on the merits of analysing past designs vs. current design vs. designers.

The list numbers match those of the CE list, so that the reader can identify the common route for reliability.

MORE RELIABLE INTERACTION DESIGN PATTERNS

We consider that IDPs, like other current HCI knowledge, could apply some of the CE concepts to try to improve the reliability of today’s HCI designs.

Possibilities to consider are:

1. Take-up of the CE concepts that are more rigorous than those of IDPs. Or, at least, consideration of those concepts.

2. Asking that each application of an IDP is recorded, along with any evaluations of those applications. These applications might be understood as more specific IDPs (see next point).

3. Inclusion in the IDP template of: the generality of the IDP and its relationship with more general and more specific IDPs; and the expression of the performance of the design problem and its solution.

4. Inclusion in the IDP of the process of acquisition of the IDP. Consideration of implications of the CE acquisition strategies for IDP strategies.

The list numbers match the CE list again. Evidently, there are further possibilities to consider, but given the current state of long-term CE and the desire to keep IDPs useful for current HCI, the above seems like a sufficient start list.

CONCLUSIONS

We have outlined our aim of reliable HCI design and current progress towards this aim. This progress enabled us to propose some broad changes to IDPs to improve the reliability of today’s HCI designs. We have started to implement and assess those changes, but significant further research is required. The next step is to investigate whether the proposed possibilities for IDPs lead to improved reliability of today’s HCI designs. We also intend to continue to develop long-term CE and promote the development of HCI knowledge to try to improve the reliability of HCI design

REFERENCES

1. Alexander, C. (1979). The Timeless Way of Building. Oxford University Press.

2. Bayle, E. et al. (1997). Putting It All Together: Towards a Pattern Language For Interaction Design. CHI’97 Workshop. http://www.pliant.org/personal/

Tom_Erickson/Patterns.WrkShpRep.html.

3. Denley, I. et al. (1997). Usability Principles for Service Design. Computers, Communication and Usability: Design Issues, Research and Methods. P. Byerley, P. Barnard & J. May (eds). Elsevier.

4. Dowell, J. and Long, J. (1998). Conception of the Cognitive Engineering Design Problem. Ergonomics, 41(2).

5. Gamma, E., Helm, R., Johnson, R., and Vlissides, J. (1995). Design Patterns, Elements of Reusable Object-Oriented Software. Addison-Wesley Publishing Company.

6. Hill, B., Long, J., Smith, W., and Whitefield, A. (1995). A Model of Medical Reception—the Planning and Control of Multiple Task Work. Applied Cognitive Psychology. Vol 9.

7. Lambie, A., Stork, A., and Long, J. (1998). The Coordination Mechanism and Cooperative Work. Proc. of the 9th ECCE.

8. Stork, A. and Long, J. (1997). Structured Methods for Human Factors Research and Development. Proc. of HCI Intl. 1997.

9. Stork, A. and Long, J. (1998). Strategies for Developing Substantive Engineering Principles. HCI ‘98 Conference Companion.

10. Stork, A., Middlemass, J., and Long, J. (1995). Applying a Structured Method for Usability Engineering to Domestic Energy Management User Requirements: A Successful Case-Study. People and Computers X, Proc. of HCI ‘95.