Opinion Dynamics in Networks with Antagonistic Relationships

31 concepts18 papers2 proposals

Traditional models of social influence operate on a simple assumption: influence is a positive force. If a friend adopts a new technology or belief, you are more likely to follow suit. However, real-world social structures are "signed," containing both positive edges (trust, friendship) and negative edges (distrust, enmity). In these environments, influence does not just propagate; it refracts, flips, and sometimes backfires.

The Mechanics of Antagonism

In a signed network, the relationship between two individuals determines how an opinion travels. While positive edges encourage consensus, negative edges trigger what Cisneros-Velarde et al. describe as a "repulsive effect." Instead of moving toward a neighbor’s view, an individual may move toward the opposite extreme.

This behavior is often modeled as a "state-flipping" mechanism. As Shi et al. demonstrate, when a node interacts with a negative neighbor, it may perceive the neighbor's state and intentionally adopt the inverse. Rawal et al. leverage this logic in their COSiNe model, where a node adopts the opposite of a neighbor’s opinion across a negative edge. This creates a "boomerang effect": an attempt to influence an enemy can result in a reactionary shift that strengthens their original opposition, as noted by Li et al..

From Influence to Opinion Maximization

Most research in social networks focuses on Influence Maximization—the binary problem of getting as many people as possible to adopt a product. However, Gionis et al. argue that in polarized environments, we should instead focus on Opinion Maximization: shifting the average sentiment of the population.

This shift in perspective changes who we should target. In a purely positive network, you target the most connected "influencers." But in a signed network, Gionis et al. show that the most effective strategy is often to target "anchor nodes"—those with the most extreme negative internal beliefs. Converting these individuals provides a greater net gain for the network's average opinion because it removes the strongest downward pull on the collective equilibrium.

Conversely, targeting widely distrusted nodes can be disastrous. Xu et al. explain that if a "negative influencer" (someone many people distrust) adopts a desired opinion, the rest of the network may reactively shift in the opposite direction.

The Topology of Negative Influence

In signed networks, a node's importance is not just a function of how many people they know, but the sign of those connections. Eger demonstrates that being strongly disliked by an out-group grants an agent as much power to shape the final distribution of opinions as being liked by an in-group.

This leads to a counterintuitive property: global influence can be negative. Sun et al. show that in signed versions of the Friedkin-Johnsen model, increasing one person's internal belief can paradoxically decrease the overall average opinion of the group. This happens because the network structure flips the sign of the transmitted influence as it passes through antagonistic links.

Emergent Polarization and Blocking

The presence of negative ties fundamentally changes the long-term stability of a network. While standard models usually lead to consensus, Eger shows that out-group discrimination explains persistent disagreement and stable polarization. Cisneros-Velarde et al. further prove that while two antagonistic factions will polarize, three or more groups can lead to never-ending fluctuations in beliefs.

Negative influence also acts as a barrier to information spread. In the Competitive Linear Threshold model proposed by He et al., negative influence can dominate positive influence, effectively blocking a competitor's message. Di Ianni et al. describe nodes becoming "refractory"—ceasing to be influenced entirely if the negative pressure from their enemies exceeds their threshold for adoption.

Ultimately, ignoring these negative links leads to significant inaccuracies. Chen et al. find that treating negative relationships as positive or simply removing them degrades the predictive power of diffusion models. As Li et al. suggest, distrust acts like a prism: it doesn't just block information, it refracts it into opposing viewpoints, ensuring that even a positive message will inevitably generate negative spreaders in a polarized society.

Key Foundational Work

Influence diffusion dynamics and influence maximization in social networks with friend and foe relationships (2013) — contributed 2 concept elements to this direction.
Influence Blocking Maximization in Social Networks under the Competitive Linear Threshold Model (2012) — contributed 2 concept elements to this direction.
Opinion Maximization in Social Networks (2013) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_negative_social_influence_44965806,
  title = {negative social influence},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/negative-social-influence/rdr_44965806f0fa43bfba7885c10b90f6c9},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_negative_social_influence_44965806_v1,
  title = {negative social influence},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/negative-social-influence/rdr_44965806f0fa43bfba7885c10b90f6c9/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake