In my last post I discussed some of the basic techniques of social network analysis (“SNA”). While SNA can be a powerful evaluative tool, the results of SNA should be analyzed closely to ensure that the correct conclusions are drawn from the particular measures and to determine cautionary measures that could be taken to preserve the value and efficiency of the network. For example, a node may have a high level of degree centrality (i.e., a large number of direct connections); however, the influence of the person occupying that node may be relatively limited if the connections are limited only to other nodes close by in his or her immediate cluster. Also, the importance of what appears to be a connector or hub really depends on how much information does in fact flow through that node. Another important feature of SNA is the way in which it can identify nodes that have drifted away from the central network. SNA usually identifies one or more nodes, or clusters of nodes, with relatively low centrality scores meaning that they are no more than peripheral members of the particular network. There may be a number of reasons for this such as personality factors, the nature of the activities performed by the persons occupying those nodes and problems with the flow of information within the network. However, peripheral nodes can be very valuable for the network either by virtue of the skills they represent or the connections that they themselves have to resources and information from outside of the network and it is therefore important for the company to find a way to create higher levels of connectivity with these peripheral nodes. As for nodes that serve as brokers (i.e., a high level of betweenness centrality), care must be taken to anticipate the consequences of a break down in the flow of information through that node because the occupant leaves the company or becomes disenchanted with management policies to the point where he or she abandons the role of connector. Finally, a very centralized network with a handful of hubs can be quite dynamic; however, there are clearly risks associated with the possibility that one central node will suddenly become disabled and bring the entire network to a complete standstill. Put another way, it does not take much for a “connector” to turn into a “bottleneck,” either because the person in the node is simply overworked and unable to push information or along or simply decides that it is in his or her interest to hoard information and dispense it selectively even though it slows down company initiatives.