Whether you work in academia, business, education, the fine arts, construction, or just about any other field, chances are you have been a member of some sort of team charged with solving a problem or to create something new. Within the last couple of years, team formation has been looked at through the lens of network theory, and there are some new and interesting findings.
A group out of Northwestern University (NU), led by Prof. Luis Amaral, has looked at how effective creative teams are formed. Their data sets included looking at the teams that have created Broadway musicals for the past century and the publication records of the top journals in the areas of social psychology, economics, ecology and astronomy, each over the last half century. I'll just summarize their findings here, but the full paper (published in one of the top science journals Science, April 29, 2005) is available online; select the PDF file for the article entitled "Team Assembly Mechanisms Determine Collaboration Network Structure and Team Performance."
The model used in the analysis distinguishes between veterans who have been involved in creative collaborations before and rookies, who are about to see their names appear in print for the first time. What is interesting as far as what leads to success (for plays success is getting to Broadway, and in publications it is to be published in a top journal) is two parameters: the fraction of the team that is composed of veterans (or incumbents, as used in the paper), and the propensity of the veterans to use their connections within the field's network and select agents they have collaborated with in the past. What the research shows is a phase transition, from one regime where you find a large cluster connecting a substantial fraction of agents in a particular field, and another regime where there are large number of isolated clusters of agents. Veterans tend to make up the large clustered, well-connected portion of the network (and are more likely to be hubs within a network), and rookies tend to be more isolated since they have not yet had the time nor experience to become embedded within the larger network (tend to be in the periphery of a network). The size of successful teams tends to vary within the scientific fields studied, but for the Broadway case, the team size averages seven members of a team. It is also important to realize that regardless of the team size, teams tend to be embedded in a larger network because of the fact that veterans on the team tend to know others within the field that may be collaborating elsewhere.
The network formation of the model used by the NU group does indicate a scale-free architecture to the larger network, where hubs are formed due to the rookies' desire to 'make a name for themselves' and wanting to be associated with better known veterans in the field. This is called preferential attachment in network language. For teams that do not make getting veterans to join the team a priority, success is less likely and the network these teams belong to are much more idolated from the rest of the field. However, as the desire to include veterans on the team increases, teams are more successful, and in addition, coalesce into a well-connected single cluster, where links between separate teams exist because of the veterans' links to past collaborators who belong to other teams. This phase transition shown in the data and predicted by the network model is clear evidence for what has been called the "invisible college" proposed by other researchers since the 1960s (and a Wikipedia article about the Royal Society claims Robert Boyle used this term as far back as 1646). The invisible college is the web of social and professional contacts that link, say, scientists across universities.
The main, generalized conclusion from the formal application of network theory to a social system such as creative teams is that to be successful a 'dream team' needs to be built with a majority of veterans who have not necessarily worked together before. Staying with the same collaborators over and over again can actually hurt the creative process and performance of the team (i.e. you need some new blood from time to time, and this can take the form of a rookie or another veteran from the larger network who you may not know very well, to keep the creative juices flowing). In other words, change is good for the creative process to work well.
Much of this seems intuitive. Many people change jobs after a few years because they become bored and simply need a change to spice up their career. Over time, members of teams within a company may find it more difficult to stay productive if no change ever takes place, regardless of past success and creativity. It is rare to find teams of performers who are successful over long periods of time because they have trouble finding something new to keep their audience interested. Normally an athletic team consisting mostly of rookies and other young players, and few veterans, will not find success; rather, it takes a mix of established players (veterans) and some new blood to be successful. And each year or every other year, teams may want to get rid of some of the younger talent and bring in some different young talent or another veteran who would be new to that team, in order to remain successful. This is precisely what the Chicago Bulls did in their dynasty years with Michael Jordan as the main hub. It is truly interesting that applying network theory to team formation shows this is the scenario needed to have the highest chance for success. It pays to dip into the larger network and make some changes in team structure in order to remain fresh and creative.
It may be worthwhile to contrast this with what Howard Gardner writes about for creative individuals who make groundbreaking discoveries. As brilliant as Albert Einstein was, and his nearly individual burst of creativity in physics from 1905 to about 1925, he was largely unproductive in the second half of his career because he isolated himself from the rest of the scientific establishment (i.e. network) and rarely collaborated.