Sandra González-Bailón: The self-organisation of political protest and communication networks

Notes from the research seminar The self-organisation of political protest and communication networks, organized by the Internet Interdisciplinary Institute‘s research group GADE (e-Governance: Administration and Electronic Democracy) within the framework of the UOC’s Master’s Degree in Political Analysis and the new Master’s Degree in e-Government and Administration, in Barcelona, Spain, on September 17th 2013.

Sandra González-Bailón, Professor at the Annenberg School for Communication (University of Pennsylvania)
The self-organisation of political protest and communication networks.

Cyberactivism: there has been an obvious drastic reduction of costs of participation. May this be the reason more people are willing to engage in politics? Will these movements last in time? How do they work? How do they grow or are managed?

There also seems to be a certain degree of contagion of social movements: information flows through networks and enables new protests that replicate previous ones. Indeed, social networking sites help in putting in contact different and isolated communities, which also makes it easier for mobilizations to spread.

Network theory, network analysis and big data are being very handy tools for analysing what is happening in the field of social movements.

Threshold models

Watts, D.J. & Dodds, P.S. (2010). “Threshold Models of Social Influence”. In P. Bearman & P. Hedström (Eds.) Handbook of Analytical Sociology. OUP

Threshold models measure the likelihood that someone will do something above a certain threshold or number of people have already decided to do or have already done something. E.g. if your threshold of buying a new smartphone is 20%, you will decide to buy your new smartphone once 20% of your friends/network has already bought it.

  • the shape of threshold distribution determines the global reach of
    cascades;

  • individual thresholds interact with the size of local networks;
  • critical mass depends on activating large number of low threshold actors that are well connected in the overall structure;
  • exposure to multiple sources can be more important than multiple exposures from the same source (complex contagion)

The Spanish indignados movement or 15M

The Spanish indignados movement is highly hierarchical (high average degree), as most of online networks are. And the people that are in the core of the network tend to interact with other nodes in the periphery of the network (very low level of assortativity).

Most of the users had medium threshold levels — neither pure leaders, nor pure followers. What can be seen is that, actually, users with lower threshold values used to tweet at the beginning of the demonstrations while users with higher thresholds used to tweet later in time (i.e. a demonstration of the threshold model). When analysing the information cascades, once again it can be evidenced that messages spread virally and very quickly.

Where are recruiters or influentials and spreaders? The k-shell decomposition helps us to tell the degree of centrality of certain individuals or Twitter users. What we see is that recruiters do not necessarily belong to the core of the network, but are randomly distributed along the networks. But when it comes to analysing the lenght of the cascades that they initiate, core users spark longer cascades. In other words, messages are not always initiated at the core, but longer chains of messages are.

Thus, the power of networks have a relative weight, but hierarchies still have much weight in the diffusion of messages.

González-Bailón, S., Borge-Holthoefer, J. & Moreno, Y. (2013). “Broadcasters and Hidden Influentials in Online Protest Diffusion. American Behavioral Scientist.

Four type of users according to centralily and comparing ratio of mentions received/sent and ratio of following/followers: influentials (high ratio of received/sent, low ratio of following/followers), broadcasters (low, high), hidden influentials (high, low), common users (low/, low). Influentials usually initiate longer cascades.

Related to the evolution of the movement, at least what data say is that the first anniversary in 2012 was less concurred in terms of people tweeting or participating through Twitter. On the other hand, centrality grew, which means that hierarchy grew too. Why? Maybe because the leaders were less able to mobilize other people, maybe because these leaders became stronger leaders along time. Again, cascades in 2012 grew less than in 2011, which means that the reach of the message was shorter. Thus, more hierarchy and less reach. This evidence goes against the motto of “horizontality” of network-based citizen movements. When hierarchies are measured with Gini-coefficients, it becomes obvious that unequal distribution grows in all categories (influentials, etc.).

One of the consequences of this evolutoin of the network is the cloaking of the network: influentials become less large and more central, and thus they centralize more the debate. And sooner or later the risk of not being able to manage such centrality in the path of communications end up in cloaking the network and making it much weaker.

Brokers are people that bridge separate networks. It can be seen that brokers have low levels of structural constraint and actually send tweets with aim at putting in contact these different networks (e.g. by means of sending tweets with hashtags that belong to the “vocabulary” of more than one network). They sent more messages, got more retweets (RT) and received more mentions.

The problem is that there are just few brokers, which, again, pose a problem of cloaking the communication between networks would they disappear.

Conclusions

  • They are not horizontal structures.
  • They are not stationary. They are dynamic and change in time.
  • They are not robust and fluid: they have structural “holes” that difficult the processes of diffusion.

Discussion

Q: Is Twitter a social networking platform? And how does this affect the analysis? González-Bailón: it certainly is more than that, as it has a broadcasting component. And this sure affects the analysis as it fosters centrality and hierarchy more than other SNSs such as Facebook. On the other hand, some users are actually collective or institutional users, which also affects the rule of the game.

Share: