Innovation networks, how to increase creativity by manipulating networks

This image comes to us from Fleming and Marx, 2006, via visual complexity.

This paper is a good example of applied network theory. It focuses more on the applications of network theory to management of creative resources rather than analyzing the network for its statistical properties. In this case, the researchers are trying to find out how network structure affects the innovation of individuals.

So the innovation structure of yore churned away in seclusive highly clustered groups controlled by technology firms. Inventors were hired, for instance, by Bell Labs, and the inventors in Bell Labs would only collaborate with other Bell Labs inventors. This is a very disconnected network with disjoint sets of actors. From the data in this paper, these situations are good for getting ideas distributed, accepted, and moving from concept to tangible product. But these situations are not good for fresh innovation.

Large connected networks are better for fresh innovation and foster creativity, perhaps due to the diversity on contacts. Highly clustered groups, such as the one boxed in the image above, are more prone to groupthink and other failings of tight-nit groups and *shudder* large meetings.

I'm a little unclear as to how they built the work, but it seems that connections are created by two people collaborating on a patent. The colors are the firm they worked for, and the color of the links indicate how old the links are.

If the innovation structure in the days of yore was tightly contained in firms, in this more recent age firms don't have the same strangle on inventors that they once had. People move from company to company and firms essentially compete for talent. This starts to create a 'small-world' network. This is the famous 'six degrees of separation' concept. Although the network can be very very large, every single node can reach every other node through only a small number of steps. Thus communication and information can spread very rapidly from one point to another. I think Jon Kleinberg explains this principle very well in this paper. But for those who don't dig stochastic simulations and optimization problems the authors of this paper explain it in a different way that I've never encountered:

A small world network can be visualized as a “caveman” social structure. Imagine an extended community inhabiting a series of caves which, although not close, are each within walking distance of one or more caves. In such a community, one would expect a dense network of relationships (or “ties,” in network parlance) within each cave, owing to daily interaction among those in closest proximity. Sociologists would describe the internal cave structure of dense, overlapping, and redundant relationships among occupants as a cohesive cluster. The upper left box in Figure 2 illustrates a very cohesive cluster of inventors, linked by overlapping collaborative ties.

In contrast to the multiple and redundant ties within caves, fewer ties — possibly none at all — would exist between caves, owing to less frequent interaction among occupants of different caves. If the caves were linked by a few connecting relationships, however, then together the collection of caves would constitute a “small world.” These bridging connections dramatically reduce social distance because each cave member can now reach a different cave member via a few indirect links. Anyone within a given cave would probably know the local “bridging” person who, in turn, would know a few members of the other caves. Individuals who bridge the caves create a small world; in the absence of such individuals, people remain isolated in their local caves and their worlds remain “large.”

Their suggestions for those managing inventors? Find out the 'gatekeepers' are by mapping your local network of collaboration. The gatekeepers would be the cavemen in a cave who have contacts at other caves. After you find who they are, they have to be offered incentives, but not adminstrative and business incentives since they have to be encouraged to continue to collaborate and innovate. At the same time the highly clustered group of innovators that you do maintain should be kept this way. The cluster allows for good ideas to actually see the light of day. This way the creativity and the productivity of your team is enhanced, and you figured it all out using applied network science.

~yargh.