Social Network Analysis within Agile Teams


It is possible to apply techniques borrowed from social network analysis) to software development teams. Once revealed, social networks can be actively or passively stimulated for the benefit of team formation and cohesion. Agile principles incorporate social network stimulation on an almost subliminal level; this is one of the reasons why agile principles work.

All organizations have a formal network of relationships between their employees, which defines their interaction. These formal networks really aren't networks, they're hierarchies of knowledge and power. Explicitly the hierarchy is one of power; implicitly this is manipulated by access to information. Information is passed on or not passed on selectively.

Think about it: do you tell your boss everything? Do you really think he tells you everything? Legitimate power, based on rank, is expressed in the formal network. However, it is naive to think that people interact solely on this level. The informal network, or social network, can alter the dynamics of power by allowing different routes of access to information.

Organizations develop a multitude of social networks over time. These may run to the thousands in large companies, generating networks of complexity orders of magnitude higher than the org chart. The structure and connectivity of these informal networks are highly significant in determining the flow of information and influence within organisations.

It is possible to apply techniques borrowed from social network analysis (SNA) to software development teams. Once revealed, social networks can be actively or passively stimulated for the benefit of team formation and cohesion. Agile principles incorporate social network stimulation on an almost subliminal level; this is one of the reasons why Agile works.

Typically, non-Agile organisations pay scant attention to team dynamics, and operate at the level of the individual and their position and role in the defining hierarchy. Conversely, approaching team dynamics at the levels of the informal networks within a group underpins Agile and Scrum practices. Informal networks are by their nature organic and self-organising. It is not possible for management to explicitly engineer these networks or overtly manipulate them to achieve objectives, but their formation, evolution and complexity can be stimulated by both active and passive techniques.

This article discusses current concepts in SNA and how these apply to Agile software development. We then describe some novel methods for social network stimulation, such as ‘speed dating.' 

Social Network Analysis
To clarify: in the current context, the term ‘social network' does not refer to Facebook, MySpace, LinkedIn or similar tools for online networking. The interactions on these networks are not representative of reality because they are limited by the structure of the sites.

Social networks are the networks that arise through informal connections between people, regardless of their position in the hierarchy of the organisation. For instance, in a team, or on a larger scale within an organisation, a social network could (and often does) consist of the people who take their kids to school and get to work earlier, the people who go out and play tennis, or the people who smoke.

In recent years, SNA has been accepted as a serious research discipline. Quantifying organic, complex relationships is challenging endeavour, both mathematically and conceptually. Social networks require a conceptual shift from the realm of the complicated to that of complexity.

Like any complex system, a social network is more than the sum of its individual parts and their attributes. The Cognitive Edge network (formerly the Cynefin Centre for organisational complexity), founded by Dave Snowden, and of which the author is a member and certified practitioner, is one group taking on the challenge of practically applying SNA and complexity to everyday team function. There is distinction here between social complexity and organisational complexity: organisations represent just one of many social networks that individuals form.

In every organisation some people are key influencers of decision-making. Known as the core group, these are the people who most strongly influence decisions within the organisation. Even though they may not be high in the formal hierarchy, their opinions are crucial to any action taken by the group - and the organisation. The core group is the prototypical social network in any organisation. The core individual's influence can be overt, as in people who become known as ‘fixers' or ‘movers' in a team, or subtler.

To take an example, the generals do not run the military, it's not run by the officers, the sergeants run it. That network of sergeants, or the head nurses in a hospital, is the real core group. They are the ones who get things done, and who drive organisational change. Identifying core people and core networks is a fundamental part of social network analysis. Organisational change is rarely successful when done as management mandate. To succeed, it has to be implemented at the social network level, but needs to be supported at the core group level.

Team members' attributes and behaviour differ according to their position in the informal network. Individuals with more connections are often exposed to more, and more diverse information. For instance, everyone hears the progress update at the meeting, but the people who share a smoke or a ride home swap the gossip. Highly connected individuals can be more influential, and may be more influenced by others. The roles adopted by individuals in the network will to a large extent depend on their personality, unless otherwise influenced by factors that force them to give up ingrained behaviour patterns.

SNA in Groups
Performing SNA in groups of any size can be entertaining and informative. Mapping the social connections generates curiosity and appeals to basic human needs for connection and feedback. There are a number of ways in which this can be done. The network can be presented as a 2D visualisation [fig. 1] derived from the numbers and strengths of connections between each node. Connection strength and reciprocity are represented by line thickness and arrowhead direction, respectively.


Fig. 1 A graphical visualisation of the social interaction of a software development team


Recently, researchers like Benjamin Waber, a doctoral student at MIT, have experimented with live, real-time social network mapping. Waber's ‘reality mining' technique uses RFID equipped badges to log the location and duration of interactions between employees. The author has been discussing with colleagues at Nokia the possibility of using mobile phones for a similar purpose. The core group and the structure of the informal network become apparent very rapidly. Different patterns of interaction reveal different types of team, for instance the ‘pulsing star', indicating a very creative team - they spread out, and then regroup to share information. Waber can improve the density of informal networks by spotting weak areas and engineering connections between individuals, in a process dubbed ‘network tightening', which is his intuitive approach to social network stimulation.

Note that the types of connections mapped by Waber, in terms of who meets and communicates with whom, are non-emotive, unlike a questionnaire asking about levels of trust. The networks mapped are of neutral interaction and information exchange, and do not ask for subjective ratings of individuals by other individuals.

Mapping the social network of an organisation can have drawbacks. For instance, it can engender resentment among managers when it becomes clear they aren't the hub, or even a hub, of the network. Snowden and others recall incidents where management deliberately marginalised strongly connected individuals whose position in the hierarchy did not, they felt, entitle them to that role. For such ‘political' reasons, people will frequently rate their relationships with their seniors more positively than is in fact the case, even in confidential surveys.

Privacy is also a concern - to what extent can or should a firm track employees' communications and activities? The questions and information necessary to understand and track real "social" interactions are very sensitive. In addition, experience has proven that many people can guess the persons represented by the nodes of a SNA chart with a high degree of accuracy even when the names are removed.

What does the social network of an Agile team look like? Should all actors be reciprocally connected to all others, or would gaps in the network be a natural and necessary result of team diversity? In a ‘complete' network all available information would flow freely between nodes and all team members would have equal influence.

Even in dedicated Agile groups, this state of team dynamics is rarely observed; there will always be stronger and weaker connections. For instance, a programming pair will tend to share more information than each would with management, on a daily basis. There will be someone who contributes less in meetings, someone who takes on the core role and gets things done for others. 


Cognitive Edge:

Art Kleiner, Who really matters - the Core Group theory of power, privilege and success, London, 2003

Richard Corriere, Life Zones - How to win in the game of life, New York 1986

About the author

StickyMinds is a TechWell community.

Through conferences, training, consulting, and online resources, TechWell helps you develop and deliver great software every day.