A recent post addressed the social media metrics debate among SEMs. Missing from these conversations, as far as I can tell, is any mention of metrics used in network theory proper. These measures have been used by researchers studying real world networks, including the internet. It might be useful to build a bridge between these metrics and the ones that SEMs propose.

First, a quick overview of network (graph) theory and its methodologies. A network is made up of nodes and links (also called edges):

The metrics used to make sense of a network use these basic components to make predictions about the network. The metrics vary based on whether the network is directed (links go in one direction; for example, I link to you, but you don’t link back) or undirected (links are always both ways). The web is a good example of a directed network. A system of interstates, on the other hand, is an undirected network.

Centrality measures determine how important a node is to the network. This information can be useful in answering question like, What is the shortest path across the network?, and, Which nodes are most likely to receive information traveling across a network? A node’s centrality is called its “prestige” in a directed network. Pagerank is Google’s approximation of network prestige.

The individual centrality measures that result in prestige are degree centrality (the number of links a node has, interpreted as how much risk a node stands of to catch what flows across a network), closeness centrality (a node’s relative distance from the “action” in a network), and betweenness centrality (how important a node is in the flow of information across a network). These measures, along with study of the power laws that govern networks, are applied to information diffusion, prestige, community structure, growth, and robustness of biological and digital networks.

Now that these terms are clear, let’s translate network theory to social media communities. Let’s say you are brainstorming a social media strategy for a client. Because there is a thriving blog community in their field, you recommend a blog. First, you might make up simple guidelines to help you define what defines a relevant blog in the community, such as a certain number of posts on a given topic. After coming up with a map of the community, you can determine who the key players are in the community by using centrality measures. Which blogs are most likely to be stopping points to users browsing the community? You can recommend to your client which bloggers to seek interaction with, including linking, using knowledge of the structure of the network.