Social Media Research: Observations
Summary: Network theory and network graphs present static, not dynamic social networks. Communication motivates social systems and social software sites. Graphs and topologies of these social networks requires better social software and social media research. The field is still needs to view users as motivated actors — not black boxes they are on graphs.
Social Software Dynamics
The science of network analysis has opened up a fascinating discussion about the structure and architecture of networked systems. We're all familiar by now with graphs comprising nodes and links, hubs and clusters, as we also recognize terms like "connectors," "diffusion," "scale-free," and the "power law." But it's easy to be seduced by the simplifying benefits of visual representations into thinking that they explain to us What's going on.
The first such chart was apparently a graph of bridges in a small Prussian city in which the question of whether one could walk the city without crossing the same bridge twice became a long-standing popular puzzle. It was a map of walks along paths made possible by connecting bridges.
In our current fascination with mapping network topologies, our emphasis has shifted from the dynamics that produce a network to the graph that captures its topology.
Let's assume that a city's inhabitants are engaged in solving a puzzle by means of walking across its bridges. Our topologies only show the totality of all possible paths. They show us the general. To capture the walker, and his particular walk, we need dynamics. People walk for a reason, namely to get somewhere. The paths they take can tell us something about the choices they face, and how they respond to them.
In the case of social systems, network dynamics are as important as their topologies, for it's in dynamics that we find the social practices we describe as emergent behavior. And if we are to understand social behavior, we need to supplement the network-centric perspective with a social systems perspective. Social systems provide us with an appreciation for the uniquely communicative and social interpretation of network constraints. Our interpretations illustrate the value we see in networking among people and within communities (organizations, groups, associations, etc.).
A dynamic model of social networks would show not just links (bridges) and nodes (islands). It would be capable of rendering networks at work.
- What motivation does a popular member (hub, connector) have to link out to those members whose links to him have made him seem popular? We know the network theorist's explanation for how he became popular: preferential attachment. But do connectors experience a social obligation to reciprocate incoming links? Does it cost them (is this a zero sum network?) to link out? Do we need to examine directionality of links? Does quantity of links equate to popularity or is it a byproduct of network growth? Do connectors suffer from excessive demands for communication? When is a popular person an authority, a local expert, a "go-to" person within a mediated community?
- Are all links equal but some links are more equal than others? Network topologies show nodes and links as equal: a link is a link and the same goes for a node. In social systems terms, though, quality of interaction and communication is highest among links of the first degree (these are our friends). Similarly, a social connector cannot possibly realize the value of her popularity among all linked individuals equally. In simple terms, it would seem that the more popular a member is, the more impersonal her interactions will be, or the more selective they will be. So all the links to a connector cannot possibly mean that she has equal interaction with each person (node) connected to her.
- Communication is discontinuous, but links appear as permanent. A simple triad is usually constructed of three nodes and three links (Granovetter's "forbidden triad" is two links among three nodes, or a chain). But if we were to map communication in practice, what we would see would look more like pairs switching on and off and exchanging position. The graph's display of three links masks the simple fact that we don't maintain continuous conversation with two people simultaneously. We take turns.
- Network topologies suffer from a lack of differentiation. If we were to map actual communication among social networks, we would probably see subnetworks lighting up during the course of different kinds of interaction. Among a group of friends, for example, subnetworks exist according to how members get together, for what kinds of activities, with what intensities and frequency, and so on. Network topologies ought to show the existence of functional subnets. Perhaps they could be constructed out of layers, each representing a subnet found to exist among members interested in some shared pursuit.
- Tie strength is not rendered in a synchronic view of a network. Social interactions occur over time, and with greater frequency among some ties than among others. A more accurate visualization of networks would highlight their dynamics and process. Our visualization would probably look more like a pattern of chains through which communication flows according to the subnetwork that's been put into play. Social interactions require diachronic modeling.
It would be exciting to see network analysis applied to mediated social networks. We will have to go far beyond existing topologies, however, if visual representations are to help us understand social process and the role played by social and group dynamics in the formation of networks.
If we built our models on shap shots over time we could capture some of the value that social networks have in maintaining routines. If we built them out of layers, we could distinguish members' relations based not just on who, but on for what purpose.
Diachronic modeling of social network topologies, constructed out of layers each representing a functional subnetwork, would seem to be a more accurate way to represent what happens in real social systems.