Análise da Rede Social da Rede Educação Viva
Abaixo encontram um excerto do artigo Activism and the New Science, publicado no nosso Medium. Nesse artigo exploramos a ferramenta de análise de redes sociais para caracterizar, analisar e suportar o trabalho de redes intencionais.
Pinging the Network
The digital communication channels of the network (a google group, mailing list and facebook page) were used to send the links to the forms together with an explanation of the mapping initiative taking place. I call this step “pinging the network”. In IT networks “PING” is used to measure the reachability of a host computer to a specific node on the network. This email invitation to participate in the mapping worked like a PING as, in principle, only active “nodes” will respond to it and the time it takes to respond is also an informal way of measuring the nodes’ response time (this response time was not registered, however).
By the end of February 2017, out of the 65 recipients of the mailing list, 37 people had responded to the “PING”, along with 22 projects.
Following a step of cleaning up and structuring the data, this stakeholder Kumu map was created. This map has two element types (person and project) and each node type has 3 dimensions (besides the usual contact details and picture). For persons these dimensions represent:
1- Geographical location
2- Professional occupation
3- Other personal interests
For projects the three dimensions represent:
1- Target group (pre-school children, young people, adults, etc)
2- Project inspiration (Montessori, Waldorf, etc)
3- Status (seeding, active, growing)
Here is a snapshot of the map. Existing formal relationships between people and projects where also represented. There is no specific meaning in the location of the elements: they are placed in a random position.
Social Network Analysis
Phase 2 of this project consisted of trying to reconstruct the social network linking the elements of the current ecosystem. So, once the stakeholder map was completed, a second round of querying was emailed, this time only for those people who responded to the first “ping” and were properly mapped on the Kumu stakeholder map.
The purpose of this second online survey was to create the personal ties (or links / edges) between the people on the map and weigh these ties with a figure from “0” to “3”. You would rate a tie with “0″ if you don’t know this person at all and with a “3” if you have already worked together sometime in the past.
From the 36 people who were invited to participate in this second phase, 28 responded. Although Kumu also provides a powerful social network analysis toolset, I have chosen to use Gephi for this second analysis phase. The resulting network is shown in the picture below.
In the image above, the nodes are sized according to its betweenness centrality (a measure of that person’s role as a broker or bottleneck in the network) and colored according to its degree centrality (the number of connections that person has). The thickness of the lines between people, represent the strength of the relationship as assessed by each person individually.
For a detailed description of what the metrics represent and what can be done with this kind of social network analysis, I strongly recommend this presentation prepared by Jeff Mohr, one of the founders of Kumu.
For me, it is important to underline here both the objectivity of the mathematical metrics and the subjectivity of its interpretation. Obviously, the metrics are strongly related to the quality of the data gathering process, the proper representativity of the graph (not all the people replied to the second phase request, for example) and the inherent subjectivity of the tie strength ranking. Having said that, common sense dictates that any interpretation derived from this graph should be taken with a grain of salt.
Recall again my “main question” which is whether social network analysis can be used to diagnose the resilience and action potential of horizontal, decentralized grass-root community initiatives. What recommendations can be made on how to improve the conditions for emergence of aligned and purposeful action within members of the network?
To characterize the resilience of the network I propose the use of a number of metrics:
– Network density: measures how many of all the theoretically possible ties are actually present. 100% means all nodes are connected to all nodes. The above network has a density of 26%.
– Network diameter: is the longest of the shortest path between any two nodes and, together with the density it suggests how closely knit the network is. Our network has a diameter of 3. That means, any two persons on the network are linked by a maximum of 3 intermediaries.
– Maximum value of betweenness centrality: betweenness centrality of a specific node counts how many times the node is on the shortest path between any two nodes. Elements with high betweenness usually play the broker or bottleneck role in a social network. In our network, the highest betweenness centrality is 214. This means that, on all the shortest paths between members, there is a member which lies in 214 of them.
– Distribution of eigenvector centrality: eigenvector centrality measures how well connected a node is by measuring it’s links to neighboring highly connected nodes. This measure suggests what elements will have more influencing power over the network. By looking at the distribution of eigenvector centrality one can speculate about how news, information and initiatives spreads throughout the network. Eigenvector centrality distribution shows no clear signs of what I call the “one man show” effect but I am still unsure what the ideal distribution would look like. Any suggestions? Comment below!
Stress testing the network: resilience
Resilience is the capacity of a system to maintain its function, after suffering from an external disruption.
One interesting study that can now be performed, is to see how the network changes in response to a disturbance. One of the most common ones, the burnout of the network broker, the element with the highest betweenness. This is a common occurrence in grass-root community projects. What happens to the metrics when this person needs to take some time off or disappears?
This can be done by simply deleting the node from the graph and calculating the metrics again. Note that the broker function is still there, but it is now assumed by a different member with a different relationship pattern.
These figures suggest that there is room for improvement as far as minimizing the impact in betweenness of a missing node. In real life, of course, the ability of the network to adapt to such events depends significantly on the culture, history and purpose of the network and cannot merely be determined by mathematical metrics. Nevertheless, looking at the network structure and to a few metrics, may inform us as to which people to connect to during the next networking events so as to increase the network density for example and motivate the elements to start joint initiatives that will strengthen the weaker ties and create new ones.
It may also point us towards areas of innovation and new ideas. These are usually introduced in the network by outsiders or lone-wolves that are located in the periphery of the central core: as the saying goes, “innovation happens at the fringes”.
As it becomes largely accepted that the network paradigm is by far a more accurate model of how work gets done in organisations, social network science is becoming more popular and useful.
In this article I hoped to prove that a very simple stakeholder and network graph can enrich a team’s discussion around the topic of how we organize our relationships around a common purpose and how can we make ourselves more resilient. It also provokes some personal reflexion about what role we play in the network, whether we know it or not and how we can take care of this common asset which is the web of ties that weaves us all together.
Network science is only one of many other tools encompassed by complexity science and systems thinking that can be put to the service of the change makers community.
Advances in computational technology allow us to access wonderful analysis tools that, if used with a good dose of humility and common sense, can provide great insights and support to solving wicked social systems problems. Because now most scientific and technical information is readily available online for free, it has never been so easy to make use of this information and put it to use for the advancement of civilization. If there is any hope for us to solve mankind’s super-wicked problems, like climate change, inequality, terrorism, etc., I am sure social complexity science and systems thinking will play a defining role in looking for some insights.
To conclude, I wish to acknowledge the members of REV for their support, interest and resilience.