social networks

Readings:
Garton et. al: How would you depict these relations? You may also want to consider how the individuals are represented. Explore you ideas by sketching a small network - perhaps of people you have been in contact with in the last day or two, or your family, or characters in a novel, etc.

Wellman and Frank: What aspects of their analytical model do you think are most affected by these changes? One of the big limits on support is time and attention - a person with a huge network cannot be supportive of everyone, all the time. How would you reframe their analysis to take into account levels of need and levels of ability to provide support? Your answer may be in the form of a short essay, a re-working of their model, or a sketch (similar to problem 2).

Would a more nuanced approach to ties, as with Granovetter's discussion of weak ties and their function in linking otherwise unassociated networks, affect this model?  If so, how?


Visualising ties and relations
       
        In considering ways to better visualise the content, direction and strength of the "relations" and "ties" as described by Garton Istarted by thinking about ways of emphasizing the direction ofresource exchange. I chose a coloured light analogy, with the colourrepresenting the content or type of resource. Then light shines from a source of a resource and shadow starts to indicate resource use or importance.
       
        I quickly abandoned the light metaphor as it is too strong an association to play with. Changing its rules means playing with physical laws that would obviously seem out of place or unnatural to an observer. i like the effect of a show however, and started to see it as a trail or motion line towards another.
     
        This suggested to me an indication of importance. I found my eye easily followed the tail through a person (circle) to the person they are tied with. Length of tail then gives an indication of importance to that person of the resources provided by another. I though I could also include colour as an indicator of content. More content also increases the number of lines, this make the tail darker overall, which corresponds with strength of relations.
     
        The next step was to include information on the history of interactionand balance of exchange of resources (I prefer this interpretation to that of straight ties). I settled on a simple wave form between point. The wave is split, with each persons provision of resources on one side. The frequency of the peaks represents the frequency of interaction, the amplitude is meant to suggest amount, importance or intimacy (depending on the resource provided). By overlaying wave forms of several colours (content again), an over all impression of exchange can be obtained. Imbalance can quickly be identified in the asymmetry of a curve. From a distance a darker link suggests a stronger tie/resource exchange, and can be caused by higher frequency, greater amplitude or more relations.
       
        Finally I would like to extend this diagram with a temporal element. Position of circles (people) is based on recent interaction levels. Wave forms describe long term averages of interaction level.  Many tails starts to become confusing, so these would appear and fade (or cycle) to indicate importance to each individual of the resources provided by others, over time.
       
        I chose not to include identity information to the circles representing each person as this mixes the viewer's interpretation of this identity with the interaction statistics presented, and I hoped that the latter would be interpreted for patterns on their own.

       
A multi-level model and online networks
       
        Wellman and Franks multi-level model relies heavily on characteristics which are often not fully communicated in online networks such as gender and age. In the case of ties formed solely or mostly via computer mediated communication their analysis will fail to adjust to lower weights for these factors.
       
        The authors do touch on some of the issues effecting online networks in their conclusion. They describe the shift from "a solidarity to network", with ties now functioning as "dyads and small clusters" rather than the "densely-knit groups" that have historically dominated social networks. They also recognize that people are now more likely to "move through partial specialized involvements with multiple sets of network members". This is particularly apparent online in the case of an individual posting a question to a discussion forum to obtain information with no previous or subsequent interaction with the group or responding alter. Even when a user becomes a regular poster to a forum, a broadcast-like tie is established - something which Wellman and Frank's model and the models proposed by the other authors discussed read this week definitely does not consider.
       
        The lack of permanence in "fragmentary and loosely-coupled" modern interaction is also a key factor, as is the privacy of many online interactions. Interaction conducted "tucked away in private homes or telecommunications" is much less visible to a group, and therefore less likely to influence others in a network.
       
As Donath asserts:
"One of the big limits on support is time and attention - a person with a huge network cannot be supportive of everyone, all the time."

        I think this is particularly true of online interaction, where the cost associated with initially establishing a tie or request is low, and so the number of ties formed can be great. To emphasize this "ability to support" component, I suggest a visualization that assumes some 'total support' level, that is then distributed between the ties established.
I have used a simple container and flow analogy here. I would also consider whether being supported adds to ones ability to support. A conservation law of sorts.
       
        So that a container would be 'topped-up' by the support of others. Having chosen this approach, I am still undecided a but this metaphor, and question its underlying assumptions.

       
Slightly too low friend of a friend count?
       
        Newman's calculation of numbers of "friends of friends" provides improved accuracy in prediction over previous methods by accounting for not only the mean number of ties your friends have, but also the "distribution" from some friends with only a few ties to some with many and the "clustering coefficient"  which accounts for a reduction in the total count of friends of friends due to connections amongst themselves. In his calculations, Newman makes the simplifying assumptions that the chance of your friends knowing each other is high, and that cases in which two of your friends know someone who is a stranger to you, and don't know each other are rare. In a more nuanced consideration of weak ties, which, as Granovetter discusses are often provide links to completely unassociated groups, this assumption may not be the case for a percentage of your friends ties.This would push the friend of friend count back up.