Signals in the Noise: Decoding Unexpected Engagement Patterns on Twitter Journal: PACMHCI Volume: 9 7 CSCW476 11 DOI: 10.1145/3757657 CCS: Applied computing Sociology

Yulin Yu, Houming Chen, Daniel Romero, Paramveer S. Dhillon · 2025

Read Signals in the Noise: Decoding Unexpected Engagement Patterns on Twitter Journal: PACMHCI Volume: 9 7 CSCW476 11 DOI: 10.1145/3757657 CCS: Applied computing Sociology by Yulin Yu, Houming Chen, Daniel Romero, Paramveer S. Dhillon (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-08128.txt.

Introduction

Why do some posts collect likes while similar ones draw replies instead? Yulin Yu and 3 co-authors decode mismatched engagement on Twitter in a 2025 social media study. Likes, retweets, and comments form a signalling system inside the attention economy, and the mix a post receives carries meaning beyond its total popularity.

An unexpectedness quotient quantifies how far each engagement type deviates from predictions. Analysis of over six hundred thousand tweets reveals distinct patterns behind the surprises, interpreted through signaling theory and attention economy ideas. Brands, campaigns, and researchers who read only total counts miss this structure; knowing which signal a post over-delivers helps explain what audiences actually valued about it.

Catalog record

CREATORYulin Yu, Houming Chen, Daniel Romero, Paramveer S. Dhillon
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE100 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-08128

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Signals in the Noise: Decoding Unexpected Engagement Patterns on Twitter Journal: PACMHCI Volume: 9 7 CSCW476 11 DOI: 10.1145/3757657 CCS: Applied computing Sociology Yulin Yu, Houming Chen, Daniel Romero, Paramveer S. Dhillon Abstract . Social media platforms offer users multiple ways to engage with content—likes, retweets, and comments—creating a complex signaling system within the attention economy. While previous research has examined factors driving overall engagement, less is known about why certain tweets receive unexpectedly high levels of one type of engagement relative to others. Drawing on Signaling Theory and Attention Economy Theory, we investigate these unexpected engagement patterns on Twitter 1 1 1 The social media platform has been renamed to ‘X,’ however our study data is from the time-period prior to the name change. , developing an “unexpectedness quotient” to quantify deviations from predicted engagement levels. Our analysis of over 600,000 tweets reveals distinct patterns in how content characteristics influence unexpected engagement. News, politics, and business tweets receive more retweets and comments than expected, suggesting users prioritize sharing and discussing informational content. In contrast, games and sports-related topics garner unexpected likes and comments, indicating higher emotional investment in these domains. The relationship between content attributes and engagement types follows clear patterns: subjective tweets attract more likes while objective tweets receive more retweets, and longer, complex tweets with URLs unexpectedly receive more retweets. These findings demonstrate how users employ different engagement types as signals of varying strength based on content characteristics, and how certain content types more effectively compete for attention in the social media ecosystem. Our results offer valuable insights for content creators optimizing engagement strategies, platform designers facilitating meaningful interactions, and researchers studying online social behavior. † † cc-license: by 1. Introduction Social media platforms have become central arenas for information exchange, social interaction, and content dissemination ( boyd et al. 2010 ; Zhang et al. 2018 ; Burke et al. 2010 ) . The unique affordances of these platforms, such as likes, retweets, and comments, have created a complex ecosystem of user engagement ( Meier et al. 2014 ; Scissors et al. 2016 ; Gilbert 2013 ) . While previous studies have analyzed overall engagement, we examine unexpected engagement, defined as situations in which content receives disproportionately high levels of one specific interaction type (likes, retweets, or comments) relative to the others. For example, a tweet providing in-depth political analysis might receive an unexpectedly high number of retweets compared to likes or comments, as users prioritize information dissemination over expressing agreement ( Gonzalez-Bailon et al. 2010 ; Zhang and Counts 2015 ; Semaan et al. 2014 ) . Conversely, controversial opinions often generate an unusually high volume of comments relative to likes or retweets, as users engage in debate without endorsing the message ( Zhang et al. 2018 ; Grevet et al. 2014 ; Semaan et al. 2015 ) . Similar patterns emerge with emotional content, where personal narratives tend to accumulate a disproportionate number of likes compared to other forms of engagement, as users express support without feeling compelled to share or discuss further ( Levordashka et al. 2016 ; Burke and Kraut 2016 ; Luo and Hancock 2020 ) . A substantial body of research has examined social media engagement mechanisms, revealing common motivations and patterns behind user behaviors across platforms ( Levordashka et al. 2016 ; boyd et al. 2010 ; Meier et al. 2014 ) . While several studies have analyzed engagement metrics collectively ( Tenenboim and Cohen 2015 ; Aldous et al. 2019 ; Pancer and Poole 2016 ) , they have primarily focused on predicting overall popularity or engagement volumes rather than examining the relationships between different engagement types. Previous research has identified key content features driving engagement: topic emerges as a primary factor, with domains like music, sports, and news generating distinctive engagement profiles ( Meier et al. 2014 ; Romero et al. 2011 ; Gonzalez-Bailon et al. 2010 ) ; emotional valence significantly affects interaction patterns ( Meier et al. 2014 ; Berger and Milkman 2012 ; Stieglitz and Dang-Xuan 2013 ; Tan et al. 2014 ) ; and textual complexity shapes user responses through factors like readability, URL inclusion, and content length ( boyd et al. 2010 ; Meier et al. 2014 ; Suh et al. 2010 ; Tan et al. 2014 ) . Our work extends these findings by examining how these content characteristics drive unexpected deviations in engagement composition rather than just volume. Unlike previous studies that treat engagement as a relatively uniform construct with varying intensities, we demonstrate how different content features systematically elicit specific types of unexpected engagement, revealing distinct signaling behaviors and attention allocation patterns that challenge existing engagement models. While these studies have provided valuable insights into how individual features drive overall engagement, they have not explored how these factors might contribute to asymmetric interaction patterns—situations where content receives unexpectedly high levels of one type of engagement relative to others. This gap is particularly significant as social media platforms increasingly serve as crucial channels for information dissemination and public discourse. To bridge this gap and provide deeper theoretical insights, we draw on two complementary frameworks: Signaling Theory and Attention Economy Theory. Signaling Theory and Attention Economy Theory together provide a comprehensive framework for analyzing unexpected engagement patterns. Originally developed to explain animal communication through observable actions ( Donath 2007 ; Connelly et al. 2011 ) , Signaling Theory illuminates how users leverage different engagement mechanisms as signals with varying strength and meaning. In social media contexts, likes function as low-cost signals of approval, retweets represent stronger endorsement as users share content with their networks ( Sharma et al. 2012 ; Bernstein et al. 2013 ) , and comments—requiring the most cognitive effort—demonstrate a willingness to invest time and thought, constituting the strongest form of engagement ( Gilbert 2013 ; Burke et al. 2010 ) . Complementing this, Attention Economy Theory posits that in information-rich environments, the ability to attract and maintain attention becomes a critical success factor ( Davenport and Beck 2001 ; Webster 2014 ) . On platforms like Twitter, where users navigate constant information overload, content must compete effectively for limited attention resources to avoid being lost in the endless scroll ( Mahmud et al. 2013 ; Wang et al. 2014 ) . Our study makes significant theoretical contributions by synthesizing these frameworks to develop a novel methodological approach—the “unexpectedness quotient”—that quantifies deviations from expected engagement patterns. This interdisciplinary approach bridges insights from communication studies, psychology, and information science ( Bernstein et al. 2013 ; Burke et al. 2010 ) , enabling us to examine not just overall engagement volume but the specific composition of engagement types. While previous research has explored general patterns, our study focuses on understanding and predicting unexpected deviations in engagement ratios, providing new insights into the nuanced ways users interact with social media content ( Scissors et al. 2016 ) . This approach allows us to identify content characteristics that systematically drive specific types of unexpected engagement, revealing patterns of user behavior that challenge conventional engagement models. Building on this novel methodological approach, we specifically seek to answer the following research questions: • RQ1: How can we quantify and measure unexpected engagement patterns across different

Read the complete text