Communications of the ACM

Creative acts typically take place within a larger context of knowledge and existing practice; creating tools that enable experimentation with that knowledge will facilitate the creative process.

Support collaboration. This includes the ability to have a timely and meaningful exchange of ideas, enable multidisciplinary teams to work together, and provide a mechanism for exposure of one's work. It should enable critiques and feedback, and possibly a venue for competition and reward. The kiosks that housed the Expo'92 systems were explicitly designed to promote collaboration and group experiences. The design was very successful—the average group size was approximately three people actively using the system together while others observed. This illustrates that collaboration and sharing of ideas and creations may come as a result of the physical setup and environment as well as through software support.

Support iteration. It is fairly well accepted in HCI and user-centered design that one never gets it right the first time when designing and building a system. Expect that same need for iteration by the end user. A tool must support the ability to throw away or save as desired. Some iterations are helpful to save as part of the learning/creating process while some iterations are best forgotten. In the Expo'92 system finger paintings could be saved and retrieved or replaced. Similarly, voice messages and video imagery could be redone until the guests were content with their recordings or images. It was always up to the users whether they thought what they created was worth saving and/or sharing with others.

Support and perhaps encourage instructive mistakes. Systems should support challenging tasks that will lead to "instructive wrong" answers, obviating distracting mistakes by building in aids like automatic string completion and logic that precludes irrelevant inconsistencies so that attention may be focused on the act of learning and/or creating. Our group's approach to system design and development relies heavily on extensive and frequent user testing to discover how to improve the user's ability to accomplish what they want to do. This includes crafting the interface to preclude errors that distract visitors from their goals. However, in observing visitors use the Explore Modern Art system, it became evident just how useful certain kinds of errors could be if they were instructive mistakes. In evaluations of the Odd One Out task shown in Figure 1, visitors talked aloud as they carried out the task. As they examined a series of trials one could clearly hear them develop a hypothesis. They would then happen on a trial in which the use of that hypothesis would lead them to choose incorrectly. They would regroup their thoughts and discover new things.

Support the domain-specific actions that must be done. Such activities need to be enabled in a way that they are easily discovered, usable, and appropriately functional. In Explore Modern Art one of the things visitors want to do, and indeed need to do for certain purposes, is to magnify an image. Figure 2 shows the system's magnification function. The subject of this photograph—raindrops on a screened window—is not immediately apparent on the system without this magnification capability. Even in the original photograph it takes some effort to grasp the grid structure of the screen on the window. Here visitors are empowered to discover such details. With independent control over the size of the magnification box as well as the extent of the magnification a visitor can slide the magnification box anywhere in the image without losing contextual information. Visitors and curators alike indicate that maintaining the context of the magnified area enables them to discover new things and relationships within the image.

In the Expo'92 system the various functions provided basic sets of tools that were necessary to create interesting and imaginative artifacts. Given the user population (international, all ages, little computer experience), the functionality was designed to be accessible yet sufficiently useful. The finger painting function provided a simple palette, a set of brushes, the ability to clear the whole painting or a single stroke, the ability to save the painting and work on it later, and finally, an animated playback of the strokes. This relatively primitive set of functions was sufficient to support some extremely creative paintings. Upon discovery of the animation, people capitalized on this: for example, many created hidden writing paintings, where the message was only visible upon animating the sequence of strokes.