Simulation Creates Realistic Social Networks to Study Disinformation Campaigns

The race to understand online misinformation has reached a critical milestone with the development of a groundbreaking simulation framework by researchers at the University of Murcia. The team, led by Alejandro Buitrago López, Alberto Ortega Pastor, and David Montoro Aguilera, has created a sophisticated system that generates realistic synthetic social networks to study how false information spreads online.

Their innovation addresses a persistent challenge in misinformation research: the ethical and practical limitations of studying real-world social networks. Rather than mining actual user data, the framework creates artificial yet remarkably realistic online environments populated by computer-generated agents with distinct personality traits.

“This represents a significant step forward in our ability to understand the dynamics of online information spread,” said an independent researcher familiar with the work who wasn’t authorized to speak publicly about it. “Until now, we’ve been limited by privacy concerns and data access when studying real platforms.”

The system’s power lies in its sophisticated agent design. Each simulated user possesses demographic-based personality traits and follows predictable behavioral patterns governed by finite-state automata—essentially, rule-based systems that determine how agents will react to various stimuli. This approach allows for both realism and interpretability, a crucial balance for scientific research.

Perhaps most impressive is the content generation system. Powered by large language models similar to those behind ChatGPT and Google’s Gemini, the framework produces contextually appropriate social media posts for each agent. These posts reflect the agent’s established profile and “remembered” experiences, creating an authentic simulation of online discourse.

The researchers have also implemented what they call a “red module” based on DISARM (Disinformation Analysis and Risk Management) workflows. This component allows for the simulation of coordinated disinformation campaigns, enabling researchers to study how malicious actors can manipulate online conversations and test potential countermeasures.

Rigorous validation confirms the framework’s realism across multiple dimensions. Network analysis shows the synthetic platforms exhibit the same “small-world” properties found in real social networks—where most users can be connected through surprisingly few intermediate connections. The simulation also successfully reproduces phenomena like homophily, where users tend to connect with others similar to themselves.

In experiments, the team generated networks with over 10,000 agents, producing more than 7.7 million posts. The simulated users displayed realistic behaviors, including increased emoji usage over time and tendencies to engage with abusive content in predictable ways. Interestingly, abusive agents naturally gravitated toward central positions in the network—a pattern observed in real-world platforms that has concerning implications for online harm.

The system includes a Mastodon-based visualization layer that allows researchers to observe agent interactions in real-time and validate the simulation’s accuracy. This feature transforms abstract data into an intuitive interface that resembles familiar social media platforms.

Industry experts suggest this framework could have far-reaching applications beyond academic research. Social media companies might use similar approaches to test new content moderation policies before deployment. Government agencies could employ such simulations to better understand the dynamics of foreign influence operations and develop more effective countermeasures.

Despite its innovations, the research team acknowledges limitations in the current implementation, particularly around scalability and content evaluation. They’re continuing to refine the system to support even larger simulations and more sophisticated agent behaviors.

The development comes at a critical time, as concerns about online misinformation have intensified globally. With major elections approaching in several countries, tools that help understand and potentially counter coordinated disinformation campaigns could prove invaluable to maintaining information integrity.

The research team’s work represents a significant contribution to the growing field of computational social science, where computer simulations help unravel complex social phenomena that would be difficult or impossible to study through traditional methods.