Contact Consortium Logo

BayCHI Newsletter for April 1997

March Meeting Report

by Stephen Attaway (

This abstract provided courtesy of Bay CHI. All rights reserved.


Interacting and Designing in Virtual Worlds on the Internet

A panel discussion with:

The new medium of multiple-user virtual worlds offers thought-provoking challenges to those of us involved with computer-human interaction. What form does the "interface" take when the computer is generating a face for the human behind the mask, or avatar? How can emotions be expressed? How does cyberspace represent itself as a 3D model? The panelists have worked out some preliminary answers to these questions in the marketplace. They are also keenly interested in the social implications of their work.

Bruce Damer led off by pointing out that multi-user virtual worlds are moving the net beyond the document-centric model imposed the World Wide Web in '93. He thought these virtual worlds might prove to be as significant a development as Mosaic was.

He pointed out that designing a virtual world requires a number of different specialists, including cultural anthropologists and even a primate gesture specialist. He then gave an overview of some of the worlds his consortium works on. It literally was an overview, since the worlds are laid out on a 3D map where the user can zoom in on the area of interest. Bruce led us through:

All these environments run on 28.8 modems on Pentiums. Bruce pointed out that his medium is not driven by corporate labs or business plans but driven out of people's homes - he sees his consortium as putting to tools into people's hands so that everyone can be their own computer/human interface designer.

Bruce concluded with a story to illustrate where this medium may be going: A new baby, Marquis, was born into a family that spent a lot of time in Alpha World, one of his consortium's online environments. The baby watched the family interact every day with the places and avatars of this world. Finally, at age seven months, Marquis got out of his crib, crawled over and got into Alpha world by steadily pressing the spacebar. Marquis started sucking - and suddenly all kinds of terrible sounds began emerging from his mother's avatar in the world. Everyone active in the world noticed and gathered around, asking his mother's avatar what was wrong - at which point Marquis started screaming . . .

Steve De Paula (Onlive) spoke next. He began with a question: should the panelists be at a computer game conference instead of a user interface conference? He pointed out that the panelists were trying to accomplish something substantive - building virtual communities online. Steve's preferred medium uses audio chat emanating from avatars in virtual space, not text chat.

Steve pointed out that to build a community you must have people who communicate. The most natural emulation style remains the human voice. Yet a number of problems must be solved to emulate a group conversation. With several people talking at once, there are multiple inputs to organize. Steve's group uses a spatial metaphor where the people whose avatars are closest are loudest. To hear people better, you simply walk over to them.

Steve's group has based its work on research into social spaces, particularly the cocktail party setting. In cocktail parties groups of 3-6 tend to congregate and form a conversational unit. Steve has created an online virtual cocktail party, with the goal of making the user believe s/he is actually there. This is done through a consistent spatial metaphor as well as some "realist" conventions. For example, the user cannot leave his or her "body".

Steve sees many uses for such a space: distance learning as well as online virtual communities. He pointed out that people in 3D worlds tend to congregate in tribes, not in nuclear groups of 2 people. He thinks this is because the rich 3D environment actually puts us into a tribe sensibility, into a rich social space. This space requires all the social support of actual communities, even police. AOL warned Steve's group that the problems of the real world leak into the virtual world, and that he would have problems with pedophiles long before he thought he would.

Reid Hoffman (Fujitsu) has been running a virtual world in Japan since 1990. Reid's group started from the viewpoint of social engineering. How can you approach these worlds as an HI designer? The key is to remember you are designing a shared space - where people both represent themselves and understand that there are other people there.

A fundamental problem for the user is: how do I know someone is talking to me? Reid's group uses a 3rd person perspective with dialogue balloons and a point-and-click interface. Clicking on another person's avatar causes your avatar to move over to and face the other, so that it is reasonably clear who you are addressing with your dialogue balloon.

A second problem is how to design in the right social distance, how to set a balance between intimacy and anonymity. How do you communicate to the user how much is being revealed about himself or herself? Steve's group at Fujitsu did some research on people's relationship to their own avatars. They asked, "Are you your Avatar?" 50% said yes, 50% no. They then asked, "Is your Avatar the same as you in the existing world or you in your dream world?" Again, 50% yes, 50% no - but not the same 50%! The responses fell into quadrants! So some people interacting in a virtual world are working personas in a dream space, while others have the experience of talking to others through a screen - potentially on opposite sides of the same interaction.

Rob Rutherford (Black Sun) talked about the emerging relationships between robots and humans online. Many users have told him, "I am bothered when I am chatting with a robot and don't know it's a robot." Some users have suggested robots be made to wear a distinguishing badge. Rob himself, however, likes to adopt robot avatars, and even uses street signs or tables as avatars. He enjoys the resulting ambiguous interactions between people representing themselves as robots, robots representing themselves as people, and people wondering what is "real".

Kirk Parsons (Avatar Designer) talked about the technical challenges of avatar design, most of which have to do with constraints imposed by the modeling software chosen. Kirk gave three things a good Avatar should do: Motion is a lot more important in achieving these three things than a high resolution static presentation. Kirk prefers to generate a number of low-res versions of a very high-res "master". He uses different animation techniques depending on what he is modeling. For gross body movements, like walking, he uses key frame; for facial expression, morphing. He also uses inverse kinimetics and dynamics.

Since Kirk makes a lot of Avatars, he is interested in design standards, creating a shared library of parts and techniques for creative re-use and eventually independent modeling/authoring tools.

Kirk concluded with some useful definitions: an Avatar is a visual representation of a person. A Bot is a program simulating a person. An Agent is a Bot capable of executing a task.

Back to Home Page Hand

Back to Consortium Home Page

Please send any comments to our Webmaster

© 1997 Contact Consortium, All Rights Reserved.