We have explored the concept and architecture of the Pond, an information ecosystem metaphor used as an interface to information on the Internet. There are several crucial issues to the design of especially multi-user interface artifacts being explored in this application such as the choice of interaction metaphor, the exploration of shoulder-to-shoulder collaboration, the "purpose" and role of the application itself and the adaptability and usefulness of the alternative technologies being explored.
Firstly, let us examine the actual nature of the pond table concept. In the Web Planetarium, users had to select items (web pages) by direct choice, i.e. by clicking on links in 3D space. There was no implicit introduction of new elements by the underlying system. In some applications this turned out to be a problem because users with no prior experience were too shy to discover what had to be done in order to introduce a new web page. The Library Demonstrator on the other hand is based on the concept of the user providing a set of (text) data in a way very similar to how a conventional database search is initiated and the number of items being searched is more or less infinite. The results of this initial search can then, of course, be used for further searches in a more intuitive and user friendly way (i.e. by direct manipulation of 3D objects).
With the Cyber Pond we try to combine the above approaches with an emphasis on user interface details. There is a fixed number of entries that is potentially very large (the Amazon sales catalogue of music CD’s) but there is no guarantee a specific item will be found or arrived at during any one session. The choices made are opportunistic and direct and very much dependent of what other users are currently interested in and the links being pursued when the underlying system introduces new items is very much ad hoc.
An inspiration for the present Pond, apart from the organic life, nature water etc. metaphors already discussed, is somewhat surprisingly the mechanical metaphor of the music jukebox and casino style gambling roulettes. The purpose of these contraptions is more or less obvious to people (users?) still the machines themselves carry an aura of intrigue and mystique. Many people find watching the working of the mechanical mechanisms endlessly fascinating and the designs are seen as a work of art in themselves. In the jukebox example the mechanical noises it makes and associated delays in "processing" the coin and actually getting a record on to the turntable are seen as a natural part of the mechanism that brings a certain transparency and understanding of what the machine is currently doing.
In terms of how information is displayed to users as well as the actual form factor of the artifact one of the things we want to explore is the position of the display. In our experience a horizontal presentation surface where users can gather around have an invitational collaborative sense to it. In, for instance, a cinema or conventional overhead presentation there exists a large gulf between the presentation and the audience, there is a significant effort required to get involved with the activities surrounding the presentation. A horizontal display where users can stand or sit around offers easy affordance for participating even on such a lightweight level as simply pointing to details of the presentation.
A serious drawback with the large area touch screen technology we currently use is a limitation when it comes to detecting multiple simultaneous touches. Future true multi-user touch screens will alleviate this shortcoming. In the present set-up we try to work around the problem by only using the touch screen for functions that inherently are single user, i.e. for the triggering of music playback. It of course makes no sense to play several samples at the same time, the result would only be a meaningless cacophony of sound.
One lesser problem is that the use of a separate application for playing the music samples (i.e. Real audio playback) breaks the illusion of an integrated 3D environment. This problem has a simple remedy in that it will be pretty easy to integrate support for the media format (of ones choice) into the application software
The RF tag system being used in the set-up can play a set of different roles. One of the more fundamental questions is if the tags are to be viewed as holders of search criteria (e.g. "country music") tags or as containers of special music items (e.g. "the 1993 Seattle Grunge Christmas Album"). The first case implies more of search-oriented application and given that there is no explicit character/keyboard input available to the user whereby more detailed search criteria can be defined (as in the Library Demonstrator). This function of the tags is probably not the most appropriate use.
We have opted to focus on the second usage of the tags whereby tags contain explicit references to pieces of music. The reasons for this choice are twofold. Firstly, the current version of the Pond Table is a demonstrator and the audience will probably not spend much time with the unit (15 minutes maybe). Secondly, keeping the handling of a music tag similar to how a music CD or cassette tape is used will reduce the feeling of unfamiliarity based on the new concepts used in the Pond. Behind the scenes there will of course still be an ongoing implicit search based on the Amazon recommendation database.
Given the nature of the tags and other functionality available in the pond it is also possible to imagine more interesting usages such as putting reflectors on top of the tags and placing them in the pond to be tracked by the video tracking subsystem. In this case the tags could for example function as tracking targets for flocking pond items.
The pond is designed to be more fun and social than precise and analytical. Our working view of the current Cyber Pond application is that it is very similar to an old style jukebox, but with a twist (a cyber jukebox?). And just as a real world jukebox, a pond not being interacted with will have an interesting behavior, it is a machine that is a pleasure to look at and listen to.
Finally, it would of course be possible to build a browser plug-in for a conventional desktop application but a lot of the fun and mystique of the pond artifact would be lost.