Previous | Table of Contents | Next |
4. LINKS AMONG SYSTEM REPRESENTATIONS
A basic tenet of this project is that human-computer interaction design must embrace system usability, and system usability emerges from programmatically linking the end-user characterizations that drive user interface design to all phases of the product development including data definition and system modeling. It is not only the design of the user interface that requires bridging representational gaps. As has been highlighted with most recent statements of object-oriented analysis and design (e.g., Jacobsons Object-Oriented Systems Engineering), as well as other system design methods, systematic and traceable transformations from analysis, to design, to implementation, and testing are hallmarks of mature software development. User-centered design generally extends these methods through its focus on representations of user-centered data.
Using customer data as the source for analysis, and then consciously addressing what elements of system behavior should be revealed to the user and what elements should not, helps make the resulting system usable and useful by linking user-centered design practice with a software development discipline. For example, most Object-Oriented methods use class representations as part of a design. In these diagrams the user-visible objects can be separated explicitly from objects only visible within the software. Mapping an object analysis model onto the user view may indicate the need for mediating classes in order to preserve the user view. These mediating classes define requirements on the system design in support of the user interface and overall system usability, much as was shown in the discussion of conceptual models earlier. In this manner a unified object model (itself a representation) that links into the end users view of the world and the implementors analysis and design view can be created.
5. THE PSYCHOLOGY OF THE DESIGNER: TOWARD USEFUL AND USABLE REPRESENTATIONS IN DESIGN
Thus far the focus of this chapter has been on how the design process supports the creation of a representational system to bridge the gap between user-centered data and a concrete design. Moreover, it has been suggested that creating a representational system requires the transformation coding and recoding of information, and this process has necessary conditions for success. This section first takes this notion one step further and suggests some reasons for design success and failure of best practice and of inadequate practice that have their analog in other areas of cognition. The section concludes with recommendations for how to match design practice with the cognitive nature of the design problem.
5.1. DECISION MAKING IN DESIGN: WHY TRADITIONAL DESIGNERS HAVE BAD INTUITIONS ABOUT USABILITY
In his recent book Landauer (Landauer, The Trouble with Computers, 1995, p. 216) looked at the trouble with computers and considered the issue of why they are so difficult to use. Part of the problem may lie in the design practices of the people traditionally responsible for the bulk of the programs written. Landauer suggests some of the problem is that traditional designers take a software and hardware system point of view as they strive to decide how a program should work. Unfortunately, the user of the program is not interested in how the program works, but rather in how well it will help them do a particular task. This difference results in a mismatch in both knowledge and method between the designer and the end user.
The notion that traditional designers have bad intuitions about usability can be analyzed in view of the model for design discussed thus far. Recall the three key steps identified in the creation of the representational system: selection of the aspects of the represented world to be modeled, selection of the aspects of the representing world to do the modeling, and development of the correspondences between them. In this view, the designer is making decisions in the course of forming representations. Research on human judgment shows interesting patterns in how people make decisions. It suggests there are systematic biases in their intuitions resulting from the heuristics they use in everyday situations. Here the term heuristic is used descriptively to mean the rules of thumb guiding human judgment. Experts as well as novices are subject to these biases (Tversky and Kahneman, 1974). In effect, Bad intuitions may be thought of as the misapplication of a rule of thumb. Tversky and Kahneman have studied a variety of biases, in which individuals are shown to have ignored useful normative models of probability in their decision making. One example that might be germane is the availability bias. This bias is seen when people use the ease of recall of examples of events as an indicator of the likelihood of that event. Factors that normally enhance recall such as salience and familiarity come into play in the estimation of probability. In the case of system design, the essentially different bodies of knowledge possessed by traditional designers vs. their end users is a powerful source of misinformation. (You didnt know that alt-tab brings up a view of processes currently running? I use that all the time.) When the traditional designer taps into the only data they have for decision making, they use what is familiar or salient and therefore inaccurately estimate the skills, knowledge, goals, and intentions of the end user. While not all design decisions relate to inappropriate sampling of data, the point is that informing the design process with the appropriate data is central to design success. When taken to its extreme the failure to use end user data fallaciously equates expertise in the development of programs with expertise in the task domain. Finally, most fundamental may be the problem that traditional designers do not usually have access to disconfirming evidence for their own theories of user behavior.
In summary, there are critical pieces of information missing in the representational system of the traditional design scenario and key process steps missing from the traditional design process. User-centered design prescribes developing a representational system that includes (1) an adequate statement of the represented world and (2) a robust set of mediating abstractions to guide the design process. Iterations in user-centered design provide repeated refinement and validation, ensuring design judgments are guided by reasonable samples of the end user population.
Previous | Table of Contents | Next |