Following my assignment from last week, I am continuing to work with soc.motss. The metaphor I am using for my display is a cocktail party. Motss is often described as a cocktail party by its participants - idle chatter that moves from topic to topic, groups forming and leaving over time.
The main display unit in my interface is a person. Each person is a simple outline shape with a coloured body and a cocktail glass in his or her hand.
The body colours represent something about the person - part of the colour is blue or pink, to represent the gender of the person. The other part is brown to white, to indicate how active they are in the newsgroup. I wanted to communicate something about the person, and a gradient fill seemed to be a way to fit in two colours. I like the way the gradient makes the information seem a bit fuzzy, but the mixing of activity and gender is a bit odd. It might seem unusual to communicate gender so up-front, but I believe gender is an important part of soc.motss discussion and should be revealed. I have committed the error of making gender either/or, when several regulars on motss deliberately reject those categories. Perhaps the interface should allow each person to choose their own base colour for their gender.
The cocktail glass indicates a person's discussion interests. The colour here is pure, a mixture of red, green, and blue to indicate a mixture of three interests. I'm not sure what the three interests should be, perhaps red for "flirting", green for "serious discussion", and blue for "flaming." The eye can't distinguish a subtle mixture of these components, but if a person has an extreme of one behavior or another it should be visible. The greatest problem is that if a person is a mix of, say, red and green he or she will appear yellow, which most people don't register as "red and green".
There is some room left for more subtlety in the display. The faces are disturbingly blank - I left them this way because I wanted the face to communicate something about the person's own identity, but I didn't know how to do this. A snapshot of the person's own face would be the right idea, but won't work graphically because it won't map into the otherwise cartoonish image. The body shape could be varied to communicate something else about the person: their physical shape, perhaps (but this would be contentious, and hard to make work cross-gender).
The newsgroup as a whole is viewed as the party, a collection of people scattered around chatting about topics. Each cluster of people is one conversation. Technically, a conversation should be more than just one thread - often several disconnected threads are really about the same topic. Individuals are represented as before - body colour distinguishes their gender and activity and the cocktail colour indicates their general conversational style. Some groups are homogeneous, say a group of people all roughly green indicating that that conversation is probably about something serious. Most groups are heterogeneous, where the thread itself is a combination of different types of conversation (a common occurrence on motss).
The party view indicates two problems with the display. First, the pink/brown gradient for women is hard to distinguish at small scale - a different colour than brown would probably be better. Second, identity is now a bit confused. In a real cocktail party each person can only be in one place, but in a newsgroup one person could be participating in many conversations. If the goal is a thread oriented display then there is no problem, just draw the same person in several places. But if the goal is to show the dynamic, the cocktail party visualization is a bit awkward.
I imagine the party view to be scale free - the user can zoom out for a overview of the entire newsgroup, or zoom in to find out what a particular conversation is about. For example, the image at this scale reveals that this group of people are discussing Clinton's predicament. The text could then be connected to a display for the actual posts.
This interface doesn't show the actual messages themselves. I believe that sort of reading is better served by traditional newsreaders. My goal here was more to communicate the dynamics of the group, the connections between people and their conversations.
I like the spatial metaphor, especially if a scale-free display is feasible so that one can view different levels of overview or detail. The use of colour here is less effective, but I think with some further effort it can be made to work.
Nelson Minar | Created: September 15, 1998 |
<nelson@media.mit.edu> | Updated: December 15, 1998 |