Art, Science, and Magic: What really happens during User-Centered Design?

A fundamental aspect of UCD involves gathering data from representative users, including market intelligence, user characteristics, wants and needs, tasks, and feedback to design ideas. It is therefore tempting to conclude that because UCD is a data-driven process, it is a science. At the very least, if UCD is not a basic science, we can call it an applied science, or better yet, engineering.

But it is simultaneously hard to argue that the UCD process is devoid of art. That which lies between gathering data and generating a design solution is a process that more resembles art than science. Somehow, data are magically transformed into a design. The process seems all the more magical in that first, it is not always controllable or predictable. Design solutions can manifest themselves in a flash, or only after long periods of deliberation. Second, the same set of data in the hands of different designers can generate vastly different design solutions, varying greatly in quality and effectiveness. Third, some UCD practitioners seem to have a better knack of turning data into design than others. Given these characteristics, it is hard to deny that there is a strong element of art in UCD.

It seems that most members of our community concede that UCD has elements of both science and art. However, to date, the scientific aspects of UCD have been given more thorough treatment, with only scant attention paid to an examination of the artistic aspects (Righi, Isensee, & Pierce, 2001). We are consequently left with the dilemma of defining the bridge between the science and art of UCD. What is the essence of what happens in this gap? What is the magic?

The field of UCD has a rich knowledge base grounded in, among other disciplines, the cognitive sciences. Therefore, I looked to the cognitive science literature for a possible model for explaining this phenomenon. I found the concept of subsumption. To understand subsumption, we need to first look at the cognitive structure (loosely, "the mind"). Ausubel, Robbins, and Blake (1957) posited that the cognitive structure is organized in stable, hierarchical clusters. When new bits of information enter the cognitive structure, they become catalogued under these more stable clusters. Ausubel et al. called this process subsumption. Subsumption has been employed to create tools called advance organizers to facilitate acquisition of new knowledge.

How does subsumption explain the bridge between science and art? It is possible that what occurs in the space between data and design is the gelling of individual pieces of information into a larger, less specific entity. Through the processes of UCD, the individual data points (requirements, tasks, user characteristics, etc.) are gathered into a greater structure (the design solution).

Here's a simple example. Consider the following user requirements for a Chamber of Commerce web site for a small tourist town:

• "I like to research things to do while I'm in town."
• "I want to find out what types of restaurants and are available to determine if the range is affordable."
• "I need to find out what the climate is like so I know what to pack."
• "I like to check a calendar of events to determine if there's anything exciting going on when I plan to visit."
• "I want to find out what kinds of shops are available in town and nearby."

Now, imagine that the design team is meeting to mull over these requirements. They may begin by talking about how each of these can be implemented. For example, they consider the "calendar of events" requirement, and imagine building a monthly calendar that resembles a typical calendar, with cells representing each day of the month. The cells would contain brief descriptions of the various events. The brief descriptions would link to pages containing complete descriptions of the events. An arrow control would allow users to move from month to month.

The team may perform this activity for various requirements independent of one another. But before too long, the team must create a unifying user model to support all these individual solutions. After some brainstorming, they decide to implement a web site that resembles a newspaper. The newspaper would contain sections that bear the name of typical newspaper sections, such as News and Weather, along with more custom categories such as Dining Guide and Events Guide. Users could visit each section to get the information they need.

In this example, the newspaper metaphor is the subsumer, the higher-level entity that incorporates all the constituent requirements.

Note that a user's conceptual model does not necessarily need to be a metaphor. Metaphors have been demonstrated to be effective subsumers in computer learning scenarios (e.g., Mayer, 1981), and have been used extensively in software design. However, a model can be any structure that serves to pull together the individual requirements. Again using the Chamber of Commerce example, the team may brainstorm and test with users various categories of requirements, and represent these categories as links in a navigation bar. Each link brings the user to a category (set of pages) of information. The site is therefore organized into a conceptual model whose essence is communicated through its navigation bar.

Whether the subsumer is or isn't a metaphor, or whether the platform is the Web, a desktop application, a kiosk, or a handheld PDA, the challenge for the design team remains the same: To gather constituent requirements into a logical, unifying design solution.

In the Chamber of Commerce web site example, the team expressed a degree of creativity in coming up with a design solution. The requirements had to be considered, weighed one against the other, and a single solution robust enough to meet all these requirements had to be generated. But note that in this case, the team pulled a model from their library of available design solutions, and modified the solution to meet the given requirements. In other words, a newspaper metaphor is common on the Web; the idea was adapted to meet the needs of the new web site. The available design solution is akin to as existing subsumer. But consider the case of a design team generating a unique design solution. Consider some products that have been designed and developed in the past: The telephone. The Walkman. The operating system desktop. The World Wide Web. These design solutions seem to represent a higher order of creativity. It seems, therefore, that subsumption can result in the adaptation of existing subsumers, or the creation of new ones, the latter being an apparently more difficult, more rare result.

The need to gather pieces of the user experience into a unified entity is not new to UCD practitioners. As designers, we are taught that as a milestone in the UCD process, we need to design a user's conceptual model - effectively, a subsumer - that pulls together all of the elements of the user's experience with the system. Unfortunately, we have not been able to guarantee that this will happen effectively and efficiently. We haven't been able to ensure that we can generate the best possible design solution, much less deliver it on schedule and within budget. From a practical standpoint, therefore, it would be valuable to find ways to facilitate the subsumption process. How can we make sure that once we have done all the work to gather data, those data can be brought together into a strong, effective, design solution? My observations of and participation in many design projects have generated some thoughts about what may facilitate subsumption and hence, the design process:

• Multidisciplinary teamwork. Bringing together people of differing viewpoints to analyze and debate requirements and other user data can help generate a wider range of possible design solutions. Unfortunately, teams are not effective by default. Team members should consist of a mix of people who can work well on a team and benefit from varied perspectives. Further, teamwork should be taught and practiced as a skill, and facilitated and managed throughout the design process.
• Creativity. Creativity is a complicated, hard-to-define construct. But although hard to define, it is usually recognizable. We need to seek out creative people who can use data to generate strong design solutions. Further, we need to learn to cultivate both inherent individual and team creativity through various activities and techniques. All the data in the world will not generate a design solution by itself. The process needs creative people to make it work.
• Facility with data interpretation. User requirements and other pertinent data must be considered in light of its validity, reliability, power, and context. Training and experience can help designers understand and apply relevant user data to generating design solutions.
• Familiarity with existing solutions. Practitioners who spend a great deal of time investigating existing solutions are valuable team members. By building up a library of existing solutions, a team can better draw from and adapt that which has solved design solutions in other contexts.
• Experience with the UCD process. A practitioner needs to understand that creating a solution is not simply a matter of solving many individual problems, but pulling it all together into a robust, unifying design solution.

To conclude, the magic of the Design in User-Centered Design is poorly understood. Subsumption is a mechanism that may offer an explanation for how data are transformed into design. But more important than explaining how this happens, we need to learn to teach, facilitate, and manage this process among UCD practitioners. Perhaps a better understanding and application of subsumption can help UCD prove itself as a process by which effective design solutions are generated.

References

Ausubel, D.P., Robbins, L.C., & Blake, E., Jr. (1957). Retroactive inhibition and facilitation in the learning of school materials. Journal of Educational Psychology, 48, 334-343.

Mayer, R.E. (1981). Psychology of computer programming for novices (Tech. Rep. No. 81-2). Santa Barbara: University of California at Santa Barbara. (ERIC Document Reproduction Service No. ED 207 592).

Righi, C., Isensee, S., and Pierce, E. (2001, June). Then a Miracle Occurs: Translating Data into Design. Paper presented at annual meeting of the Usability Professionals' Association, Las Vegas, NV.

User-Centered Design Karel Vredenburg, Scott Isensee, Carol Righi Prentice Hall, 2002.

Carol Righi, Ph.D.
Biography
March 26, 2002

Email the editor with any questions or comments.