Online network organization of Barcelona en Comú, an emergent movement-party

Pablo Aragón, Helena Gallego, David Laniado, Yana Volkovich, Andreas KaltenbrunnerComputational 2017 · DOI

Abstract

The emerging grassroots party Barcelona en Comú won the 2015 Barcelona City Council election. This candidacy was devised by activists involved in the Spanish 15M movement to transform citizen outrage into political change. On the one hand, the 15M movement was based on a decentralized structure. On the other hand, political science literature postulates that parties develop oligarchical leadership structures. This tension motivates to examine whether Barcelona en Comú preserved a decentralized structure or adopted a conventional centralized organization. In this study we develop a computational methodology to characterize the online network organization of every party in the election campaign on Twitter. Results on the network of retweets reveal that, while traditional parties are organized in a single cluster, for Barcelona en Comú two well-defined groups co-exist: a centralized cluster led by the candidate and party accounts, and a decentralized cluster with the movement activists. Furthermore, results on the network of replies also shows a dual structure: a cluster around the candidate receiving the largest attention from other parties, and another with the movement activists exhibiting a higher predisposition to dialogue with other parties.

Introduction: The Paradox of the Movement-Party

Political science has long grappled with the "Iron Law of Oligarchy," which posits that complex organizations inevitably trend toward centralized, elite-led structures. This presents a significant challenge for modern social movements, such as Spain’s 15M ("Indignados"), which are founded on principles of horizontal, decentralized participation. When these movements transition into formal political parties—becoming "movement-parties"—do they sacrifice their democratic roots for institutional efficiency?

This study investigates this tension by analyzing the online organizational structure of Barcelona en Comú (BeC) during the 2015 Barcelona City Council election. By developing a computational methodology to map Twitter interactions, the authors examine whether BeC preserved its decentralized movement structure or adopted a conventional centralized party organization.

Methodology: Mapping the Political DNA

The researchers analyzed two distinct types of Twitter networks: Retweets, representing endorsement and information diffusion, and Replies, representing dialogue and interaction. To ensure the stability of their findings, they employed several innovative techniques:

  • N-Louvain Method: To address the stochastic nature of standard community detection, the authors ran the Louvain algorithm 100 times. Nodes were only assigned to a cluster if they appeared there in at least 95% of the runs, effectively filtering out noise and identifying stable political cores.
  • Network Metrics: Rather than relying solely on size, the study used the Gini coefficient of in-degree distribution to measure hierarchy. The authors argue that Gini is more robust for heavy-tailed social graphs than standard centralization metrics. They also utilized $k$-core decomposition to identify the "inner core" of communities and assess social resilience.

Dataset and Scope

The study analyzed a dataset of 507,597 tweets collected between May 1 and May 26, 2015. The data collection focused on seven primary political parties contesting the election, seeding the network from official party and candidate accounts.

Political parties and their corresponding Twitter accounts analyzed in the study.

The temporal distribution of the dataset shows significant spikes in activity corresponding to key campaign events, such as televised debates and the election day itself.

Distribution of the number of tweets in the dataset over time, highlighting peaks during the debate and election.

Key Findings: The Dual Structure of BeC

The most striking result of the analysis is the unique organizational DNA of Barcelona en Comú compared to traditional political entities.

1. Centralization vs. Decentralization

While traditional parties (such as PSC, PP, and CiU) appeared as single, centralized clusters, BeC manifested as a hybrid of two distinct but overlapping groups:

  • BeC-party: A centralized cluster focused on official accounts and candidates (notably @adacolau), exhibiting high inequality (Gini = 0.995).
  • BeC-movement: A decentralized cluster of grassroots activists with significantly lower inequality (Gini = 0.811) and high internal connectivity.

2. Resilience and "Small-World" Dynamics

The BeC-movement cluster proved to be the most resilient in the study. With a maximum coreness ($k_{max}$) of 17, its members were more tightly interconnected and less dependent on central leadership than supporters of any other party. Its clustering coefficient (0.208) was nearly double that of traditional parties like CiU (0.114), indicating a robust, "small-world" activist base.

3. Interaction Patterns

In the reply networks, the candidate account @adacolau served as a primary target for attention from opposing parties. Conversely, the BeC-movement cluster demonstrated a higher predisposition for horizontal dialogue and cross-party engagement, particularly among parties favoring a Catalan independence referendum.

Conclusion and Future Directions

The findings suggest that movement-parties can maintain a dual organizational structure that balances institutional campaigning with grassroots activism. However, the authors note that Twitter data provides only a partial view of political organization. Future research should explore longitudinal data to determine if these "movement" clusters eventually dissolve into traditional party structures over multiple election cycles, a process known as institutionalization. Applying this N-Louvain framework to other international movement-parties, such as Podemos or the Five Star Movement, could further validate whether this dual structure is a defining characteristic of 21st-century digital politics.

Summary updated

Abstract

The emerging grassroots party Barcelona en Comú won the 2015 Barcelona City Council election. This candidacy was devised by activists involved in the Spanish 15M movement to transform citizen outrage into political change. On the one hand, the 15M movement was based on a decentralized structure. On the other hand, political science literature postulates that parties develop oligarchical leadership structures. This tension motivates to examine whether Barcelona en Comú preserved a decentralized structure or adopted a conventional centralized organization. In this study we develop a computational methodology to characterize the online network organization of every party in the election campaign on Twitter. Results on the network of retweets reveal that, while traditional parties are organized in a single cluster, for Barcelona en Comú two well-defined groups co-exist: a centralized cluster led by the candidate and party accounts, and a decentralized cluster with the movement activists. Furthermore, results on the network of replies also shows a dual structure: a cluster around the candidate receiving the largest attention from other parties, and another with the movement activists exhibiting a higher predisposition to dialogue with other parties.
BibTeX
@inproceedings{aragon2017online,
  author = {Pablo Aragón and Helena Gallego and David Laniado and Yana Volkovich and Andreas Kaltenbrunner},
  title = {{Online network organization of Barcelona en Comú, an emergent movement-party}},
  booktitle = {Computational},
  year = {2017},
  doi = {10.1186/s40649-017-0044-4},
}
Figure