Small worlds and networks for collective intelligence
At Swarm we’re really interested in how you can create networks that actually work, networks that succeed in mobilising a group of individuals or organisations to solve a particular challenge. Collaboration works but successful collaborations owe that success to many factors – team dynamics, experience of collaborators, shared sense of mission and values and structure.
In this context, Brian Uzzi and Jarret Spiro’s ‘Collaboration and Creativity’ has a lot of great insight. In their paper Uzzi and Spiro are concerned with how we create better, more successful ideas through collaboration and to explore this concern they looked at Broadway musical hits between 1945 and 1989.
As a tall man not particularly suited to theatre seating and with musical tastes that don’t quite extend to Cats or Les Miserables, I can’t think of anything worse than sitting through over 40 years of musicals. But the findings are definitely worth exploring in a bit more detail.
Small world networks
For Uzzi and Spiro, creativity is the product of “a social system of actors that amplify or stifle one another’s creativity”. To illustrate the idea they introduce the notion of the small world network.
A small world is fairly self-explanatory – it is a group of people who know each other and who are locally clustered. A small world network exists when these groups are connected to other, similar small worlds through common members.
This closeness and cohesion shapes behaviour in the system: it allows ideas to circulate rapidly across the network as a whole, jumping from small world to small along shared connections. Ideas passing through small world networks are not just transmitted more rapidly but also accepted more readily, introduced as they are by a credible member of a pre-existing group. So in practice, Emily overhears an interesting idea put forward by Tom on her leadership course. The next week, she shares it with Swarm and because it’s interesting (and because it’s Emily), Swarm apply it in our own work.
The authors use a measure, called Q, to assess whether a network is ‘small worldly’ or not. Networks where there is long path length (that is the amount of people, on average, it takes to connect two random members of the same network) and low clustering (measured by the cluster coefficient) have low Q, while those with short path length and high clustering have high Q. For a more detailed look at the equation used, check out the original paper here.
When Uzzi and Spiro applied this formula to the world of Broadway musicals they discovered an interesting result. When Q is too low (long path length, low clustering), the musicals flopped (as judged by the amount of positive reviews, rave reviews and financial success). And when Q is too high… they flopped too. At the low end of Q, the authors contend, networks were too splintered to properly circulate good ideas and the lack of group cohesion meant that even if a good idea was introduced its acceptance could be difficult. When Q is too high, on the other hand, the ideas circulating are more likely to be stale and too easily adopted by people who know each other too well.
As Uzzi and Spiro put it “Any successful production is likely to be a combination of convention and innovative material – material that extends conventions by showing them in a new form or mode of presentation.”
Small worlds and structuring networks:
Contrary to what some have written, Uzzi and Spiro’s work does not have direct implications for the optimum way of structuring a team. As, Jordan Ellenberg, writing in Slate points out, their work does not show that hit musicals were more likely to have teams made up of collaborators who had worked with each other in the past as well as a few fresh faces.
The real insight is around how to structure networks: who to involve, how to connect them and what types of collaborative interactions should be developed and trialed.
As we build collaborative ventures which not only engage partners to operate the venture but also engage stakeholders (in the form of members, customers, beneficiaries, etc.) to help shape and drive the venture, we are effectively creating networks. Two of the collaborative ventures we’ve been working on – Good For Nothing and the Wild Network – are already large and growing. They are networks that depend on innovation, creativity and a shared sense of mission to sustain themselves and to thrive and so considering Uzzi and Spiro’s findings when evolving the structure – and infrastructure – of these ventures is valuable.
At Good For Nothing, for example, we use a chapter model to expand. Chapters are run by leaders who take the Good For Nothing concept and apply it within their local communities, effectively creating small worlds that are joined together – a small world network. We ran a two day event to blood in new chapter leaders in 2013 – a way of effectively bringing up the cluster coefficient and the Q. However, with chapters all over the UK and all over the world, we need new ways to create meaningful connections and means of interaction across the network to help good ideas spread and feed the initiatives of local chapters. It’s a task that we’ll spend a lot of time thinking about this year.
New questions to answer:
There are other questions that emerge from the research. What is the impact, for example, of the strength of the collaboration? In the world of Broadway Musicals, collaboration means working together for months, sometimes years to develop and deliver a show. When we’re talking about collaborating around issues of say, food security or green energy, simply having conversations and getting to know one another might not be a like for like substitute. If that is the case then how can we structure networks that allow participants to work more meaningfully together for concentrated and/or extended periods of time. Some of this will happen organically – a campaigner may work on a wind energy campaign and then go on to consult for a green energy fund, etc. but what other opportunities can we create to aid these types of interaction, especially when collaborators are geographically dispersed?
Another question relates to how the focus of the network changes the network dynamics. A Broadway Musical involves a diverse set of individuals coming together to apply various skills at the service of creating a successful production. Solving large social and environmental issues involves a similarly diverse cast of characters but the array of motivations as well as outputs is also diverse, much more so than in the case of Broadway. How can this be adjusted for?
Uzzi and Spiro’s insights into small worlds are valuable for all of those looking to create platforms, networks and tools that help drive collectively intelligent solutions and these new questions help to frame new constraints to look out for. We will continue to play around with these ideas and let you know what we find out.