Graphical Conversation

Assignment 5 | 3.14

The assignment was to look closely at two graphical communicative environments. I chose these two:

• Active Worlds (3D)
• VZones (2D)

How would you design a better graphical environment?

I looked at Chat Circles and how it could be improved. Chat Circles avoids many of the pitfalls and false expectations of systems such as Active Worlds. Instead of trying to provide realistically rendered worlds with human-like avatars and gestures, Chat Circles is tightly focused on how to make the graphical elements represent information about the communication process. Like the best examples in Edward Tufte's books, it is deceptively simple yet illustrates principles of data density, enforcement of visual comparisons, and use of small multiples.

Some possible improvements:

• The circles could be transformed into other shapes depending on the content of the message. For example, when the user writes an angry message, the circle could turn into a spiky word bubble, borrowing a page from the comic book vocabulary. Gradually, over time, the bubble would transform back into a circle. This could be done automatically through automatic text analysis similar to the one performed in Comic Chat, with the option for the user to override and specify directly the shape of the balloon by choosing from a palette.
• Going further, I would try using more animated elements. Master animators like Chuck Jones (The Dot and The Line) have shown that one can convery emotions by using the expressive power of animated primitive shapes. Using the previous example, the angry message could cause the circle to tremble, shake, and twist violently, until it gradually calmed down to its original form. Or, in a more cartoon-like fashion, the circle could grow a tiny volcano with puffs of steam, or launch tiny missiles at the target circle of the angry message. Such animations must be done in a clean, minimal way that they do not create too much of a distraction for the users and the rest of the interface: they should provide just enough animation to make the interface more fun and dynamic.
• Fisheye views could be applied to the Conversation Landscape for browsing particular threads of conversation. Currently the text for only one particular message is displayed when the mouse is rolled over the corresponding horizontal segment. The horizontal segments could be replaced the actual messages, with the text in very small font. When mouse rollover occurs over a message, its text is enlarged to be readable. Messages close to that specific message are also made large enough to be readable, but smaller than the rolled-over message. Such a fisheye view interface would allow a user to read a specific message in the context of adjacent messages, to get both the detail and the context more quickly.
• In terms of use of space for different purposes, private areas could be created for conversations not recorded by the Conversation Landscape system. (This is also mentioned in the reading.) The private area itself could be a large circle, one with a different coloring or line type to indicate that it represents the boundaries of the private conversation. The participants would still be represented by their respective circles within the large circular wall. Other users would not be able to enter this wall: they would make a dent in the wall if they collided with it, but would bounce off the wall, and the wall would return to its circular shape. (Golan Levin implemented a cellular structure somewhat like this for a class last term.) The owner of the private circle would be the one who could allow outsiders to join the circle.

Required readings:

• Rossney, R. 1996. Metaworlds. Wired 4.06 (June 1996).
• Kurlander, D., T. Skelly and D. Salesin. 1996. Comic Chat In Proceedings of ACM SIGGRAPH.
• Viegas, F. and J. Donath. 1999. Chat Circles. In Proceedings of CHI 99
• Eisner, Will. Comics and Sequential Art. Selections from chapter 5.

James Jung-Hoon Seo | 3.13.2000