Multi-user virtual environments, Part II

RELease 1.0, August 3, 1994 by Jerry Michalski

Phil clicks on the doorway and passes into the garden that Zoe has designed. A few messages flash across the bottom of the screen as the software retrieves and assembles the different graphic elements of the scene. The garden's description is on Zoe's computer, but it points to components that exist elsewhere. The chairs and fence are generic. Phil's machine generates them locally and puts them in the right places. The gazebo is Zoe's special design, and transfers quickly. The nighttime sky looks lovely. It's an hour-old snapshot straight from an observatory. Next year she'll use a video backdrop of a daytime autumn scene. Zoe's friend Juan designed the pond, with fish bred by E1-Fish; it's on its way from his machine.

Zoe has licensed special texture maps from a third party. Those arrive, and the scene is rendered within a couple of seconds. But wait. The birds she selected appear to be very popular: Their server is bogged down...now it's crashed. Looks like there will be generic birds in this scene.

In large volumes, activities like this one generate transactions that strain every part of a networked computing and communication system. If you look at these activities as a curious experiment in fictional 3D spaces, they may not seem important.

But it's possible that the flat, document/desktop-metaphor world we are familiar with so far is a warmup exercise for more complex and useful virtual worlds to come that will find common use in business. Perhaps our future computer file systems will be abstract, semantically arranged dataspaces or resemble our real-world offices; our business meetings will be held in virtual meeting rooms with avatars sitting in software armchairs; or our interactive tvs will have virtual-world interfaces.

Whether or not any of these interfaces becomes commonplace, their underlying infrastructures are precursors of architectures to come. To support these highly distributed and dynamic 3D environments, developers have to rethink client/server architectures and devise ways to create and share objects that exist in many places and will be viewed from many perspectives. To make these spaces friendly and useful enough that people are drawn to them out of more than mere curiosity will require advances in user-interface design and navigational aids. They may develop visual ways to communicate context and meaning. To make the whole thing a viable business proposition will require solid transaction underpinnings and global scalability.

Last month we described some virtual spaces, most of which run on centralized servers. We began with Apple's QuickTime VR, which allows on person to move through space defined on a single host computer, and progressed to systems that support multiple people on the same server or several linked servers, such as Virtual Universe and Kesmai's Aries.

This issue of Release 1.0 takes another step in the same direction: Most of the companies we cover are building virtual places that can hold large numbers of people and are hosted on many distributed servers. In some cases, the participants can create their own objects or places to share; in others, the environment is a context for shared activities.

Thorny (but fun) issues remain

Exploring virtuality leads developers to address a series of interesting questions, many of which are simultaneously practical and philosophical. For example, what kind of space should they build? Euclidean? Cartesian?(1) Hyperbolic (negatively curved)? Toroidal or ring-shaped? A Dyson Sphere? Conceptual or semantic? Some combination?

Individuals could have their own views on the space. For the venturesome or adventurous, a 3D world could have wormholes that teleport occupants to other spots, or doorways to worlds with other spatial models, where different rules of physics apply.

A compelling reason for agreeing on a Cartesian 3D space is that it gives people a way to wander about and meet other people. If there's no agreement on where the village square is, people can't stroll there and run into each other. Of course, a 3D virtual world will likely have real-estate battles, just as the real world does, despite the fact that virtual spaces can be collapsed inside each other. If people's sense of orientation depends on there being a point or axis of origin, then space nearest those foci will be dear. (Who gets to own 0,0,0?) There are only so many pixels to work with on a given screen.

Non-3D space is not immune from land-rush phenomena: Witness the current race to register domain names on the Internet, a race that is reminiscent of but less restrictive than those for vanity license plates or 800 numbers. (Who will register AAA Internet Services?)

Developers of virtual environments also have more tactical considerations, such as whether the space should be 2D, 3D or somewhere in between. Two-and-a-half dimensional space could have flat characters in a 3D space, or vice-versa. Avoiding full 3D will probably reduce the processing load on client computers.

Spatial integrity: A place to hang your hat