Perlin, K., & Fox, D. (1993). Pad: An Alternative Approach to the Computer Interface. In Proceedings of Computer Graphics (SIGGRAPH 93) New York, NY: ACM Press, pp. 57-64.

By Laura Slaughter

The authors of this paper describe an interface model called 'Pad' that allows users to organize objects on an infinite two-dimensional information plane. The Pad interface employs a spatial metaphor providing a surface that uses "portals" to navigate rather than the currently popular "desktop" model (windows, icons, menus, pointers). The paper states that the main advantage of this metaphor is that it will tap into human geographical and spatial ways of thinking. Another benefit is that, because it is one large surface, it simulates the real-world working environment and allows for objects to be moved from primary to secondary attention. Thus, it is a better model for capturing peripheral activity.

Pad is meant for a wide variety of applications. To illustrate their ideas, the authors furnished examples for a daily/monthly calendar, a hierarchical text editor, a painting program, a novel with bifurcated reading paths and video browsing. In addition, the use of Pad is not restricted for one primary user, the Pad space may be shared by many users for collaborative work. The authors proposed a distributed Pad system, where objects placed on the two-dimensional surface can exist on remote machines. Example cooperative applications listed by the authors include a community map with city cultural events and a shared spreadsheet.

The goal of the paper was not just to introduce the basic idea for an alternative interface but also to specify how the Pad model will 'look and feel." A description of the details necessary for implementation of Pad were outlined by the authors. The Pad Surface contains Pad Objects that can be displayed as either a graphic or a portal. Every graphic has an address on the Pad Surface consisting of a location x,y and a scale z. Portals are graphics with one additional address called Look-on (L). Portal views are continually changing and portals can look onto and into other portals. A "root portal" is the screen view with Pad Objects rendered on it. The use of portals allows semantic zooming by magnifying Pad Objects according to an expose event. The expose event determines the appearance, in terms of size, for the object receiving the event. Another use of portals is to "filter" Pad objects. The two filters used as examples were 1) a bar chart filter that changes tabular data into a bar chart and 2) a control modifier that allows users to select options for tools used in an application.

This paper makes several important contributions to the field of HCI. The most significant is that it proposes an alternative model for the user interface that has not been previously put forth. The ideas for Pad also advance HCI research, since they are built on previous work involving systems that allow better use of peripheral information. The authors assert that the Pad interface will be useful for collaborative/cooperative applications and that it will be less disorienting for users than jumping in hypertext. However, they do not have any evidence that these statements are true. By implementing the Pad model, an entirely new set of research questions involving zoomable user interfaces (multiscale interfaces) can be explored. HCI researchers can now begin to understand how users perform tasks in this type of information space. They can also learn what tasks this interface model might be used for, or what tasks it can't be used for, as well as methods/tools that will improve the interface model. In terms of computer science in general, Pad presents the challenge of actually building a working zoomable system. Moreover, it encourages further research on zooming algorithms. For psychologists, the Pad system can be used to learn how people use mental maps to navigate. How do people make decisions about where to go? How do they remember where they have been? For information scientists, Pad provides a new technique that can be used for organizing data, visualizing search result sets, and browsing information.

1. The most troubling aspect of this paper was trying to learn the system structure. If I had never seen Pad++, I would find it very difficult to imagine how this system works. The description of the system structure could be simplified. For example, the idea of portals can be confusing for the reader. Portals are displayed because a Pad Object creates them, yet "a pad object can look different when seen through different portals." The terminology used should be clarified. Isn't the Pad Surface just a "root portal."?

2. There may have been space limitations but I thought that the literature review could be improved. I was interested in reading more details about the other zoomable applications mentioned (Bad Windows, etc.). Also, did the authors do a literature review that indicates how people navigate spatially? This information might have helped them design the Pad model.

3. The authors didn't elaborate on the types of collaborative work they expect Pad will be appropriate for. Maybe they could have used the time-space matrix (page 481) in Shneiderman's Designing the User Interface. People might use Pad to work on asynchronous tasks (at the same place, different times or different places, different times) or on synchronous tasks (same time, different places).