Crowds, both real and virtual
Familiar Stranger
The familiar stranger phenomenon discussed by Milgram is striking. One would assume that encountering a person or set of persons on a very regular basis would be cause some sort of interaction. Yet this is not the case: a different type of relationship is formed, one that reinforces the lack of interaction.
Were I not to have experienced this phenomenon firsthand, I would not believe it. Falling into this type of relationship is surprisingly easy. When I would commute to and from the city for work, I formed this familiar stranger with a number of people at my railroad station. These were people that stood in the same place day after day, at the same time, and (if they did at all) spoke to the same small group of people.
More interestingly, I have found that this type of relationship pervades almost all social situations, even those that would seem to be too intimate to allow this sort of interaction. For example, working at the Media Lab, I encounter people with whom I never interact, and whom I have never been formally introduced to. I do, however, have this familiar stranger relationship with these people, and were I to see them outside of the context of MIT, I find myself a little more at ease.
The fact that I feel more comfortable when around familiar strangers is curious; I don't know these people besides their physical appearance, and perhaps some small part of their daily schedule. Yet, though they play a very small role in my community, they are members. Perhaps this social position is more important than it may seem at first glance.
Part of what makes a situation psychologically comfortable is the familiarity one has with their surroundings. As such, these familiar strangers may play the role of "familiar surroundings" in such a way that, though they are people and not items in a room, the lack of interaction casts them into a role that is somewhere in between the two, and therefore allows me to incorporate them into my view of my surroundings.
This logic would dictate that their presence automatically makes me more familiar with my surroundings, and puts me more at ease, knowing that there are known objects around me. This role as peripheral objects is directly at odds with the close personal relationship that Milgram describes when some ill circumstance befalls a familiar stranger. Instead of becoming more attached personally to a stranger, as in Milgram's example of an old woman who collapsed on a street in Brooklyn to whose aid another woman, for whom this old woman was a familiar stranger, came, a familiar stranger can be cast into a less personal role of an object.
I would seem then, that the familiar stranger is given more breadth in terms of social role because of their familiarity.
It is interesting to consider Milgram's final words: "...Perceptual processing of a person takes considerably less time than social processing....If the temporal relations were reversed...a quite different phenomenon would result: We would typically talk with people whom we did not have time to visually perceive." Certainly, Milgram was not considering online communities when he wrote these words. Newsgroups, as per our discussions over the past few weeks, seem to exhibit just this reversal of temporal relations: we don't have any perceptual processing of a person available to us, and we have an extended period of time to interact socially.
Yet, in my experience, there still exists the notion of a familiar stranger. In reading newsgroups (and not participating in the discussions) I have developed the familiar stranger relationship with certain members of the newsgroup's community. The phenomenon of the familiar stranger must therefore be much more deeply rooted in social interaction than Milgram suspected, and is a fundamental part of a community.
Visual Who
Visual Who does a good job at displaying group membership. However it makes the assumption that the groups displayed are both equal in weighting (given some scheme for weighing the groups, such as importance or public vs. private) and mutually independent (therefore the graph it displays is multidimensional, with groups representing axes). Furthermore, its displays make salient over all patterns in group membership, but befuddle any attempt to look at the membership microscopically (this of course is not a design flaw, but a feature!)
These three supplemental aspects of the visualization presented by Visual Who provide interesting extensions to the project.
Group Weighting
Visual Who (as described in Judith Donath's paper) does not represent heterogeneous groups. While the mailing lists used as an example can be assumed (at least initially) to be on equal footing with respect to one another, given any reasonable rating scheme (group membership, of course, can be viewed as such a weighting scheme, with more populated lists getting greater weights) that is secondary to group membership. This would include ratings like mood, activity, physical proximity of members, etc. that are tangential to group membership measurements (as well as the secondary interaction of people who are on similar lists, described later in the paper).
Providing such supplemental measurements would allow users to visualize the relationships between the groups, the members, group membership and these supplemental measurements. This might be accomplished via use of rotation or size of the members and group labels--determined mathematically as the conjunction of these measurements from different groups for a particular person, in the same way that location and acceleration are determined by the interaction of the multiple springs pulling on a single person. Fonts could also be used to portray some of this information, but using this requires careful planning, as part of the elegance of Visual Who derives from the fact that the people and groups combine on screen to create a flowing sea of text. Introducing font changes can make the visualization less coherent, as well as less readable.
Alternatively, the display can be expanded into four dimensions (x, y, and time being the dimensions already present). While such a display can make visualizing the graph on a 2D monitor more difficult, if coupled with 3D immersive technology, such that the graph is shown in true 3D, then the visualization can be much more powerful, and display much more information.
Public vs. Private
The rating of groups as public vs. communal vs. private is in some ways similar to a measurement scheme as discussed above, and may be portrayed as such, however it seems that this particular measurement is important and special enough that it should be given it's own visualization within Visual Who. This can be addressed in at least two ways: Turning people and groups on or off depending on their visibility, and augmenting other attributes based on the public or private nature of the people or those attributes.
Turning users and groups on and off over time provides an intriguing augmentation to Visual Who. Similar to the display of users logging in and out over the course of the day, people state of public or privateness changes over time (both during the day, and over longer periods of time, presumably for different reasons, which is another discussion).
A time lapse graph of people's movement to and from public and private over the course of a day (or other time period) would be very informative about people's patterns of socializing. (This information about moving from public to private and vice versa also moves to and from public and private knowledge, so in effect, this is a recursive measurement--I don't know how to deal with this, but it's an interesting aspect to think about).
Integrating the publicness of information is more subtle. While it can be done by weighting the influence of a measurement based on it's publicness, this portrayal of data would likely be lost on the user. Some way of making this meta-measurement more salient is needed, although I am at a loss to suggest one. Once found, this augmentation to Visual Who should add yet another dimension to the visualization.Macro vs. Micro Representation
To accomplish this, the first augmentation to Visual Who should be a zoom feature. This would allow users to zoom in on heavily populated areas to see more clearly the interaction of the people in that area. Alternatively, turning groups of people on and off would also make the graph less dense (but I am assuming this capability is already available).
Fundamentally, however, Visual Who runs into a wall; The representation is meant to show inter-group interaction, however, diving deeper into the visualization, users may want to view intra-group interaction. Assuming that Visual Who works also for this type of visualization (which may or may not be the case), there would have to be a smooth transition to such a visualization, so that a user could move from macro- to microscopic views of the data (and back) fluidly, while allowing the user to maintain his view of his place within the data. This would add a number of interesting problems to the task of visualization, not the least of which would be maintaining a view of the macroscopic group interaction while concentrating on a microscopic view.
Solutions to this difficulty are not readily apparent, however one place to look would be the work on Scale-Free Images done at the Lab. While I have not studied the visualizations developed for that project fully, much of their display deals with moving from macro- to microscopic views of their data.
page last updated