This paper describes a sand-based ambient display as an experimental interface for peripheral communication of slowly emerging patterns of data. We describe a working networked prototype of the display that allows awareness of group activity over time. Our goal is to investigate physical and ambient interfaces that form representations of evolving processes rather than temporal descriptions of data.
Ambient media, tangible user interface, group awareness, persistent media, process visualization
Our environment is filled with subtle sources of information that are continuously changing, flowing in and out of our attention. The dynamic qualities of such environmental cues provide a consistent awareness of the changes in our surroundings. As we start to inhabit private and isolated spaces (including our corners of cyberspace), changes such as time of day, weather, sounds, and the flow of traffic are not readily perceived. There are also drastic changes in our environment that take place over long periods of time: the shifts of economic wealth, the carving of canyons by a patient flow of water, the geological motions of the earth. The world is in constant motion and we remain subjective participants at a localized scale, unable to observe these processes from an objective distance. We can only perceive the results of such processes, not the inner workings.
Conversely, there is an over-abundance of data on our desktop and activity on the network; an abundance not readily perceived in the environmental space around us. The question then becomes: how does one integrate physical and digital spaces to provide an awareness of environmental or electronic activity, in a unified and aesthetically satisfying manner? The use of ambient and tangible media such as air, water, sound, and sand offer rich possibilities for conveying such cues in a subtle manner.[1] We will explore the essential qualities of sand as a dynamic tangible medium and present a sample application that demonstrates its appeal as a natural interface to digital information.
In our environment, sandscapes are naturally formed
as the tangible result of air or water interacting with sand.
The wind and sea act as global and patterned forces that form
large structures in sand, such as the soft shorelines, dunes,
and valleys. Such interactions occur on time-scales too long for
us to observe. Yet one can "read" the history of this
environment though its dynamic surface.
Sand thus serves as a natural medium for the visualization of emergent behavior. Emergence here refers to the subjective evaluation that overall patterns are revealed through highly localized changes in the environment. Such a rule-based systemic approach has been previously explored by many of the connectionist theorists in order to explain natural phenomena.[2]
As a medium of communication, sand inherently provides several affordances as it forms a physically textured surface with natural persistence. Sand is not temporally bound since it retains its shape and form continuously, acting as a reservoir of inscribed actions or digital bits, over time. In contrast, other ambient media like air, water or sound must be manipulated continuously in order to achieve a persistent environmental effect, that is easily perceived.[3] Sand can be displaced in several ways, some of which are explored later in this paper. Sand can be viewed as a collective whole while applying simple localized changes, creating emergent patterns over time. Since there is always a conservation of the amount of sand, the affects of a localized phenomenon extend to a larger region. The granular nature of sand inherently enforces natural abstraction of data, rather than explicit binary states. Such abstraction is useful for representing emergent forms in large-scale data. Furthermore, temporal patterns gradually emerge due to the persistence of sand, as current forces either reinforce or destroy the results of earlier forces. The surface of sand thus permits a four-dimensional display of information over space and time. The resulting sandscape is a temporal sculpture that encodes dynamic data through its shape, form, and texture.
Our goal was to demonstrate a working prototype for sand-based display of both spatially and temporally dynamic information. We now describe the hardware device and a networked application developed on this platform.
For our implementation, we chose an approach where sand would be manipulated via air. Therefore, we have developed a "sand-plotter" that moves an air-blower within the x-y plane of a sand-box (Figure 1). The air tube is mounted on a mechanism that is moved via motors on two orthogonal bars placed above the sand-box. Air is blown through a flexible tube by a computer-controlled air-compressor, creating patterns in the sand as the grains are displaced by the force of the air. The duration of time that the air is blown creates varying imprints on the sand: the longer the airflow the deeper the contour becomes. The motors and air-compressor are controlled via the parallel port of a networked PC. Specialized software was written to permit a high-level data interface to the sand-plotter.
People enjoy moving their fingers through sand-boxes in their home or office, leaving traces of mental states, thoughts, or memories on the sand. As others come by, they too leave residual patterns on the sand. The sand box thus becomes a reservoir of group activity, evolving over time. Its success as an interface arises from these simple and meditative interactions. The sand-box can be considered a miniature landscape that provides a visualization of gradual, yet complex, global changes emerging in the environment. We were inspired to create a dynamic sandscape that depicted accumulated traces of group activity.
Top is a UNIX-based program that periodically displays the processes that are active on the system (Figure 2). The utility sorts all active processes according to raw CPU usage. Unfortunately, with such a data representation on the screen, it is not possible to gain an understanding of large-scale activity of a group of people over time. Such a display also shows redundant information, whereas an abstract representation should provide a glimpse into the emergent patterns of user activity over time. We chose this sample application to demonstrate the possibilities afforded through representation via a sand-based medium.
A program was written that sporadically runs top on a selected server and gathers usage data for specific users. This data is averaged over time and scaled such that it represents user activity relative to other users on the host computer. We chose to represent each user along the y-axis of the sand plotter and the amount of their activity along the pseudo z-axis, through the application of air forces. The duration of air blown is determined by the total percentage of calculated usage for a specific user. The x-axis was used to depict the elapsed time of day. The usage data was plotted at desired intervals of time, creating time-varying hills and dips in the sand. The motors were calibrated such that it took 24 hours for the entire space to be manipulated. After each 24 hour period, the plotter restarted from the beginning, either reinforcing or destroying the previous day's patterns.
Over the duration of the application, there was a continuous evolution of emergent patterns in people's activity. This emergence was translated into a smooth and continuous shifting of sand hills and valleys. The history of the group was made tangible at the surface: most of us worked during the day, leaving large valleys during this time. Curiously enough, some people here at the lab work exclusively during the night. This behavior, which deviated from the "norm", was immediately salient through the corresponding disparities between different areas along the y-axis.
To serve as another example application, weather-related data could be represented through this output device. Dynamically changing temperature or humidity at desired locations could be plotted at various corresponding spatial coordinates in the sand. The emergent shape of the sandscape over time, would provide a sense of the overall weather patterns. Controlled lighting could be used to augment sand patterns by introducing additional spatial or temporal context. Light on the sand could be made darker in regions of night-time activity, providing a temporal reference for the emerging patterns. Furthermore, images or graphical outlines projected on the sand would help associate the patterns with a spatial or geographic mapping.
By using sandscapes to show evolving patterns of processes, we have created a consistent interface that uses the natural properties of sand as a functional medium. The sand plotter translated digital information about the group environment into persistent physical contours. However, due to the use of a focused air stream, only a small area of the sand could be manipulated at any given time. Future work will address this issue by developing new techniques to allow local and global manipulation of sand. We have demonstrated sandscapes created by the interaction between sand and air, yet experimentation with water or light would help establish new aesthetic and emergent patterns. We also seek to leverage the rich aesthetic traditions of Zen rock gardens and temporal Buddhist sand paintings. Continued research will explore the potential of sandscapes as an active input interface for foreground activity.
We would like to thank professor Hiroshi Ishii and Brygg Ullmer in the Tangible Media Group at the MIT Media Lab, for their guidance and inspiration towards this work.
1. Ishii, Hiroshi and Brygg Ullmer. Tangible Bits: Towards Seamless Interfaces between People, Bits and Atoms. Proceedings of CHI '97, ACM. March, 1997.
2. Resnick, Mitch. Turtles, Termites, and Traffic Jams. MIT Press. Cambridge, 1994.
3. Strong, Rob and Bill Gaver. Feather, Scent and Shaker: Supporting Simple Intimacy. Proceedings of CSCW '96. pp. 29-30. Nov. 1996.