The obvious response is the question, “Must it add to the business?” Knowledge for knowledge’s sake is not a bad thing, and there is always the possibility that someone else will find a way to apply the results of basic research. In this case, however, I think that we can do a bit better than that.
To the best of my knowledge, which is limited, applications of network analysis to business usually come down to overlaying network diagrams on maps of the business units into which an organization or market is divided. It is then possible to quickly identify actors in bridging positions who may either facilitate or become bottlenecks to communication between the units in question. Alternatively, the overlay may reveal an absence of communication between units that is hampering business development. So far, so good. But suppose the question is how to staff a project team with the goal of creating something new, which entails creating relationships that do not yet exist either in the formal organization chart or in the network diagram. Choosing the members of a team is now a matter of art, in which the director of a project thinks of who might be good for a certain role, assesses the level of talent or skill that they bring to the table, checks their availability and then reaches out to them (or to their bosses, who must sign off on their new engagement).
But let’s be more concrete. When I was working for the Japanese agency that once employed me, my clients were often large international companies with, however, only a small presence in Japan. When agencies pitched for their business, they were shown reels of exceptional work produced by the agency’s stars. Then, having chosen an agency, they find that the stars are not assigned to their team, except, perhaps, in a distant advisory role. If they ask why, the answer is clear. If their budget is only a tenth of what their Japanese competitor is spending and the stars are already committed to other projects, they must work with teams composed of second or third-tier veterans or newcomers who have not yet achieved a stellar reputation. The work that they produce is often second-rate.
Asked by foreign executives what they should do if they find themselves in this situation, I suggested that they look at the newcomer awards in advertising annuals and request that one or more of these rising stars be assigned to their teams. Now, having studied network analysis, I can offer another possibility. Identify successful teams, i.e., those whose work is judged worthy of appearing in an annual, then look for individuals who appear as members of several different successful teams over a span of years, who have not yet won a newcomer award. Since the usual practice in Japanese agencies is to circulate new employees among several teams to broaden their experience, these are likely to be highly competent people whose careers are on the verge of taking off.
The NetLogo Team Assembly model suggests another potentially useful approach. The model is designed to relate creativity to one variable, the relative proportions of incumbents (people who have worked together before) and newcomers (people who have not worked together before) assigned to teams. The underlying theory, which seems quite plausible to me as far as it goes, is that either too many incumbents or too many newcomers will reduce creativity. The same people doing the same things will fall into ruts, while people who are thrown together for the first time may never get their act together. There will be, however, be some ratio (or range of ratios) of incumbents to newcomers that maximizes creativity. Knowing that ratio would be good for business.
You know, and I know, that this model is too simple. But as George Box famously said, while all models are wrong, some are more useful than others. Here we can think about what we would have to add to the model to make it more realistic. A great leap forward to a perfect solution is unlikely; but it still might be possible to make some progress step by step, by considering factors not yet included in the model.