Network Analysis


Creating Knowledge:
Network Structure and Innovation

Steve Borgatti, Carroll School of Management, Boston College



All knowledge is socially constructed, but some more than others.

In some cases, an individual interacts with a number of others who may be completely unaware of what problem he is trying to solve, and then, with the knowledge gained, the individual goes off by himself and synthesizes a solution.

In other cases, the new knowledge is co-created by interacting individuals who are bouncing ideas off each other and actively integrating their different perspectives.

These two kinds of knowledge creation are supported by different kinds of network structures.

Individual Creativity

To maximize individual creativity, a person needs access to a diversity of skills and expertise. The relationships between the knowledge builder and the resources they draw on do not have to unusually close. They shouldn't be enemies or competitors (more on them later), but friendly acquaintances will do fine. All parties need to have some skill at communicating across disciplines.

The more diverse people a person can call on, the better the opportunities for knowledge creation. Since individuals are limited in the number of relationships they can maintain, efficiency is important. A person who has many colleagues drawn from one discipline and/or social circle will not access as much diversity of ideas as a person who is connected to the same number of people drawn from different disciplines, departments and social circles. People who interact daily come to know many of the same things, and are in that sense informationally redundant. In contrast, people who do not interact will often know many things that the other does not know.

The property of having ties to people who are not in the same social circles with each other is called betweenness or "structural holes". A person rich in structural holes has many ties, and the people they are tied to are not tied to each other.

It's important to realize that in a small group, it is difficult for many people to have personal networks rich in structural holes. For this to happen the network has to be fairly diffuse.

Interactive Creativity

Interactive creativity also calls for heterogeneity -- it is the successful synthesis of different perspectives that creates something new. But because the interaction in this context is more intense and more important, the relationship between the people needs to be very good. In particular, they need to be able to understand each other well. This tends to mean that the participants are fundamentally similar in language and background concepts. It also means that affective elements like simply liking each other are helpful, as are good social skills.

Radical versus Incremental Innovation

People need access to a diversity of skills and knowledge in order to innovate. This argues for being as well connected as possible. If we want everyone in a group to be in a position to innovate, this will mean a very dense network in which everyone is connected to almost everyone.

This is great for incremental improvements within a well-established paradigm, but tends to stifle radical innovation. Michael Polanyi wrote the following about one of his contributions to physics:

"I would never have conceived my theory, let alone have made a great effort to verify it, if I had been more familiar with major developments in physics that were taking place. Moreover, my initial ignorance of the powerful, false objections that were raised against my ideas protected those ideas from being nipped in the bud."

In other words, if radical innovators are too well connected to the network, they can get swamped by the prevailing wisdom. As a result, radical innovation is facilitated by sparser and clumpier networks -- as in a skunk works.



The answer to the question 'what should my organization's network look like to enable innovation?' depends on the kind of innovation.

Type of Network to Facilitate Innovation


Type of Innovation

    Incremental Radical
Individual large diffuse sparse
Interactive dense clumpy


2005 Steve Borgatti Hit Counter