Paradoxically, stablished academic Social Network Analysis, as it appears in well known works as Wasselman and Faust’s, is hardly useful for getting the kind of information usually demanded by social analysis consumers. As Duncan Watts wrote in “Six Degrees“:

Because purely structural and static measures of network structure can’t account for whatever action is takin place on the network, the mehods offer no systematic way to translate into meaningful statements about outcomes (…) The principles must be interpreted in light of what the particular organization is trying to achieve and the kinds of people who work there.(…) Without a corresponding theory of behaviour -of dynamics- a theory of network structure is essencially uninterpretable and, therefore, of little practical use.

So, when we started some years ago to look for a general social network model in order to support the implementation of analysis, we looked towards Social Dynamics. In this kind of approach, origined in Economics, actors take their decissions based on a combination of their inherent payoff and their social support in the surrounding cluster.

Under this approach Michael Chwe, continued Peyton Young works, studying how different network topologies change social behaviour and network resultants. According to Chwe, there are thresholds in neighbour quantification of preferences, what make profitable to change actors own behaviours. As this measure is concrete, not continuous, the social dynamics could describe tipping points and sudden and “unexpected” changes in it’s equilibriums.

Towards a social network analysis model

In November 2003, the Spanish economist professor Juan Urrutia proposed in a widely distributed brouchure (popular in Spain because it predicted March 2004 mobs), a subtle but radical variation on Social Dynamics’ aproach to social network dynamics. According to Urrutia’s approach to Pyton’s and Chwe works: actors would like to act in a a certain way, and they want their own actions to be socially accepted. Futhermore, they have “political” desires, they would prefer some network architectures (in which their behaviour will be accepted or even shared by their neighbours) to others. Acceptance is mesured by the individual through its knowledge of his neighbours thresholds. So, cluster topology matters and affects actor’s strategies.

In fact this is a key concept: actors in social networks have a propagation strategy. They want something to communicate, and be communicated, by their enviroment (from their deserve of a better wage to their political points of view). They propagate this info or not according to the propagation threshold of their enviroment. This enviroment is composed by all the other nodes he/she is linked to. In this enviroment -his/her cluster- there are a certain volume of friendly receivers. We could define this volume as the number of nodes contacting him who he knows will accept him if he propagates the message. So, actor’s propagation threshold could be defined as the minimun number of friendly-receivers he should have in his cluster in order to propagate the message.

Urrutia ended his brochure with a fertile suggestion: the difficulty to propagate the messages they want to is in the origin of actors’ transformation strategies. Actors change their links, transforming their clusters and reorganizing global network, in order to achieve new propagation thresholds.

Consequences

According to the classification used by percolation models, social links could be opened or closed. A link is opened when information can flow between two nodes, and closed when it can’t. Strong ties (as friendship or love) are used to generate open links: we know if our close friends or couple are friendly-receivers or not of almost any message we could send them. But weak ties, as those we have with many social acquitances, are used to be closed. We don’t know how they would answer if we show them our points of view. So, when external or enviromental facts make change the main message of actor’s propagation strategy, actor will “scan” his network, looking for transforming closed links into opened links. With this information at mind, active actors will redefine their own clusters, changing the local and sometimes even global equilibriums.

But opposite to propagation thresholds, network topology changes leave back wide amounts of public information, from Internet logs to acountancy assets. We can approximate actors transformation strategy by the public register of their social actions. And with this information by hand it will not be hard to modelize actors propagation strategy and to predict the new global equilibriums. But it will need some mathematical tools and hound abilities which we will discuss in coming posts…