Types of Networks

Networks, the focus of MeshForum is a term that holds many different meanings, generally context specific.

Much of the literature on Networks focuses on either a specific type of network, generally within a very specific context, or on properties and traits of all "networks". While important, this is not all that needs to be thought about when thinking about Networks, nor does it sum up all of what will be covered at MeshForum.

Very broadly "Network" can both be a noun and a verb. As a noun, "Network" is some system of "nodes" and "links". As a verb, "Network" generally means creating links between nodes.

However, a number of important distinctions are missed by these basic definitions. One key distinction, often glossed over, is how a "network" changes over time. Most networks are dynamic and change from moment to moment, however much of the research on networks deals with questions about a static snapshot of a particular network.

Another key distinction is between "infrastructure" style networks - where the nodes and the "links" are of a long duration (Examples of this would be a street map, power grid, many IP based computer networks) and "dynamic" networks where nodes and "links" are rapidly changing and highly variable (Examples of this would be many economic systems where customers, exchanges of value, and items being purchased change hands frequently and often; other examples might be technical "mesh" networks where nodes and links are established on the fly and as needed, or many social networks where friendships and communications with friends change frequently - consider a nightclub as an example).

When looking at analyzing networks it is also important to remember and realize that many times there are multiple networks (or at least many different ways of looking at and tracking the data) that overlap and interact. Consider the case of electrical distribution.

On the one hand, the physical infrastructure of the power grid is fairly static - most "nodes" are large, phyisical structures that stay in place for a very long duration (power plants, power lines, transformers, meters, etc). On the other hand, consumption and "customers" as well as the economic and legal structures "owning" each part of the power grid can and do change rapidly and frequently. Companies are bought and sold everyday, customers move, seasons change, usage differs, in short the "network" when extended away from the broad infrastructure of the power grid and into the network of customers and companies becomes highly dynamic and changing.

If, for example, you are analyzing such a network to look at how changing the legal environment (say regulating prices) might impact the system; or if you are looking at how the introduction of new technology such as powercells where consumers might generate power vs. consume power, you have to keep in mind that there are many related networks interacting. Thus, while looking at and anlyzing the longlasting infrastructure elements is important, you have to also consider how to capture, represent, and study rapidly changing aspects of the network.

In general, infrastucture type networks are longlasting and fairly stable, thus more prone to being easily analyzed and studied. For one, it is possible to map the network, identifying the nodes and the links between the nodes with some assurance that the result is a very close representation of something real. In dynamic networks, however, as the nodes and links between the nodes can change, disappear or be added very rapidly, any attempt to map the whole network is difficult and potentially nearly impossible.

The nature of "links" and what is meant by them is also different. An infrastucture style network can be thought of as really three elements, nodes, links, and information being passed along the links. In the case of a powergrid, information in the form of "power" could be thought of as starting from one node and moving across links to some other node. More generally, in an infrastructure style network, the "action" of nodes interacting can occur other than by forming/destroying links - links can in some manner of speaking be "used". In most infrastructure networks, links are usually multidirectional, thus "action" can occur from either node connected to the other node.

On the otherhand, in a dynamic network "links" might be thought of as being highly fragile, they could be "created" as the act of use and immediately destroyed. As well, even more so than with infrastructure style networks, "links" in a dynamic network can be unidirectional - connecting one node to a particular other node, but not therefore connecting back to the original node. Thus in a dynamic network the direction of "connection" is critical for resolving quesitons about paths between nodes.

Taking this rough model and applying it to "Networks" such as the ones thought about and discussed in the "Social Network" field begins to illustrate the challenges faced by any attempt to "map" such networks. As well, it may help separate out the multiple networks that are, in fact, being looked at when social network analysis is performed.

At MeshForum we will be exploring these questions and many more, focusing on specific case studies of certain types of "Networks" as well as on the general questions raised when studying networks.

Consider the "Networks" you and your organization are a part of? Are they infrastructure networks? Dynamic Networks? Or some of both? When you think about what a "link" and a "node" are in these networks, what do you mean by "connecting"? How are these Networks you are a part of used? What challenges have you overcome around these Networks? What would you like to understand and overcome in the future?