Network Analysis


Facilitating Knowledge Flows

Steve Borgatti
Carroll School of Management, Boston College



What are the conditions that facilitate knowledge transfer? There are two kinds of issues to consider: relational and structural. Relational refers to qualities of the relationship between two people who might share information with each other. Structural refers to the pattern or structure of relationships across the entire network.

Relational Conditions

A key relational concept is multiplexity. Multiplexity refers to the extent to which one kind of tie between two people is accompanied by another kind of tie between the same two people. For example, two people who trust each other might also share information with each other, lend money to each other, and so on.

What are the relational conditions that facilitate knowledge sharing between two people? In particular, when I have a particular problem, what allows me to seek information from you? Perhaps the most important factor is my knowing that you have expertise in the relevant area.

Another factor is already being acquainted with you, or knowing that we have mutual acquaintances who can make the introduction -- most people hate to make cold calls.

Access is also a factor, especially in a sprawling multinational corporation where people may be widely separated geographically and in very different time zones. Even if we are in the same office, it may be that you are much sought after for advice, and are always too busy to see me. Or, there may be social and cultural barriers that inhibit our interaction, such as being in different, possibly competing, departments, or you may be much higher status than I am.

Another set of issues are trust, psychological safety and dependence. In seeking information and help from you, I will have to reveal, to some extent, what I'm doing and how much (or little) I know. I need to trust that you will not misuse this information. I also need to know that you will not want to much from in return for the help. Finally, I need to feel that you will not make me feel stupid for revealing my ignorance.

Structural Conditions

The structure of social network affects how rapidly information flows from one end of the network to the other. Ultimately, the speed of information flow is a function of path lengths. When the length of the shortest path between a pair of nodes is high, it will take a long time for information to flow from one to the other. Networks with high average path lengths take longer to transmit information to all members. In turn, the average path length in a network is a function of a number of structural factors. I consider each in turn.

Density. The density of a network is the number of ties it has. In general, the more ties the network has, the shorter the path lengths, and the fast information moves. However, there are costs to having too much density. First, it may mean that people are spending more time interacting and maintaining relationships than working. Second, high density implies lots of different pathways. The more pathways, the greater the potential for hearing the same information many times, and worse, hearing different versions of it each time.

Centralization. Networks that revolve around a single very well connected node tend to be very efficient transmitters of information. Nodes are at most just 2 links away from each other. These networks are great for spreading best practices. The danger in these networks is that the network is highly dependent on one person. If that person leaves, there will be disruption for some time. And if the person gets things wrong, or is pursuing their own agenda, they can hurt a network.

Centralization also creates a property known as searchability. It means that it when information is distributed across the network, people are able to locate who's got it with relatively little trouble. In the case of a highly centralized network, the center of the network is a natural go-to person who can direct people to the right person.

Core/periphery. An extension of the concept of centralization, core/periphery networks revolve around a set of central nodes (not just one) who are well-connected with each other, and also with the periphery. Peripheral nodes in contrast are connected to the core, but not to each other. Core/periphery networks contrast with "clumpy" networks, which consist of two or more subgroups that are well-connectd within group but weakly connected across groups -- like a collection of islands. If we compare networks with the same density, core/periphery networks have shorter average path lengths than clumpy networks. Tie for tie, they are more efficient spreaders of knowledge. However, because they have a dominant core, what is spread is what the core wants to spread.

In sum, dense, core/periphery networks are very efficient at spreading knowledge. The other side of coin, however, is that they are not good at innovation because it is too easy for the conventional wisdom to swamp new ideas.


Promoting knowledge sharing is a matter of (a) creating the relational conditions that facilitate interpersonal transfers, and (b) creating the structural conditions that facilitate diffusion.

2005 Steve Borgatti Hit Counter