A recent study published in Scientific Reports finds that more extreme or eccentric ideas tend to gather more “likes” on social media platforms. Conducted by researchers at Binghamton University, State University of New York, the study utilized data from online experiments and social media to analyze the relationship between the popularity of ideas and their eccentricity. The results showed a clear trend: ideas that deviate significantly from the norm are more likely to capture attention and engage interactions.
Social media’s influence on public opinion and social dynamics has been a focal point of research for some time. Previous studies have largely focused on extreme ideologies related to politics or religion, examining how these beliefs propagate through networks and influence behavior. However, this study broadens the scope to include less overtly controversial topics — like preferences in food or lifestyle — suggesting that even these can polarize opinions and attract significant attention online.
The motivation behind this research stemmed from a desire to understand how societal norms are influenced by the digital conversations that pervade our everyday lives. By focusing on eccentricity — defined not just in terms of radical political or religious views but any significant deviation from mainstream opinions — the researchers aimed to uncover the ways in which social media shapes public discourse.
To explore how eccentricity correlates with popularity on social media, the researchers analyzed data from two distinct sources: controlled online experiments and real-world social media interactions.
The first part of the data collection involved online experiments at a mid-sized U.S. university. Participants in these experiments were students from various academic majors, recruited to engage in tasks that mimic social media activity on a platform designed to resemble Twitter. The tasks varied in terms of required collaboration: one involved writing marketing taglines for laptops (high collaboration), and the other involved crafting short fictional stories (low collaboration). These tasks were chosen to stimulate creative idea generation and sharing within a controlled environment.
Participants were organized into networks, mirroring the social structure of typical social media platforms, where they could only see and interact with posts from their designated network peers. Each idea posted could be liked and commented on, akin to real social media dynamics. This setup allowed the researchers to control the environment and observe how ideas spread and receive feedback within a limited and known group of individuals.
The second source of data came from GAB, a social media platform known for its popularity among far-right users in the U.S. The researchers used a snowball sampling method to collect data, focusing on posts made between August 2016 and January 2021. This method involved collecting posts and connection information from a large subset of GAB users, ensuring a comprehensive capture of the network’s dynamics.
About 147,000 posts from approximately 3,000 users were analyzed. To handle the vast amount of textual data, posts were first cleaned of stop words, punctuation, and numbers, and then processed using the Doc2Vec method to convert text into numerical vectors that represent the semantic content of the posts. These vectors were then further analyzed to assess the eccentricity of ideas based on their deviation from the norm within the user’s social network.
The core of the methodology lay in defining and quantifying the eccentricity of ideas. Eccentricity was measured by the semantic distance between an idea and the ‘center’ of all other ideas within a user’s social neighborhood at the time the idea was posted. This center is calculated as the mean vector of all recent ideas within the neighborhood, providing a dynamic and evolving benchmark against which to measure eccentricity.
Additionally, the researchers introduced the concept of self-eccentricity to capture how a user’s ideas deviate from their own previous posts. This metric provided insights into how individuals’ opinions evolve over time relative to their own historical perspectives.
The researchers found that ideas which deviated more significantly from the norm consistently received more likes across both datasets. This pattern held true regardless of the nature of the task or the platform, highlighting a general trend in social media behavior.
In the controlled experiments, where participants engaged in creating content like laptop marketing taglines or short stories, those contributions that were more unusual compared to the average submissions of the group attracted more attention and interaction.
The analysis of data from GAB, a platform with a user base known for extreme views, further strengthened the study’s conclusions. The findings from GAB mirrored those from the controlled experiments. Posts that were more eccentric—further away from the semantic center of discussion topics commonly found on the platform—garnered more likes. This was consistent across various discussion threads and topics, ranging from politics to more general interest subjects.
Sriniwas Pandey, a lecturer at Binghamton University, highlighted that the study he co-led demonstrated the necessity of more eccentric opinions for maintaining popularity on social media.
“If you want to get attention about anything, not even a political or religious subject, you will get attention if you say something that’s kind of extreme,” Pandey explained. “The more you see others sharing similar opinions, the more likely it will start cementing your belief in that particular direction.”
The researchers also found significant variations in self-eccentricity among users, indicating that people’s ideas often become more extreme or diverge more from their past ideas over time. This trend was particularly evident in the data from GAB, where users had a history of postings that could be analyzed for such patterns.
The findings suggest that individuals undergo personal shifts that may make them more receptive to or creators of more eccentric ideas over time. This has significant implications for how we understand the spread and adoption of extreme views, indicating that individual trajectories can also contribute to the polarization observed on social media platforms.
“This study draws a picture of how extreme ideas and opinions may spontaneously arise in society,” the researchers wrote. “Everyone wants to gain social acceptance and become popular and influential. As being eccentric in opinions helps attract neighbors’ attention, people start expressing out-of-center opinions. And as most of the social neighbors do the same, it would be difficult to notice that one’s opinions are becoming more eccentric compared to others.”
“If people do not recognize the heat, there would be little feedback mechanisms to stop them from becoming more eccentric. These behavioral patterns form a cycle and may reinforce each other. These conclusions illustrate the need for further study of how to detect such spontaneous escalation dynamics in society, and if appropriate, how to implement effective interventions so that it will not cause undesirable negative impacts on our lives.”
The study, “Generation and influence of eccentric ideas on social networks,” was authored by Sriniwas Pandey, Yiding Cao, Yingjun Dong, Minjun Kim, Neil G. MacLaren, Shelley D. Dionne, Francis J. Yammarino, and Hiroki Sayama.
