Figure 1 - uploaded by Hai Liang
Content may be subject to copyright.
Examples of viral and broadcast diffusion cascades. Arrows indicate the flow of information. Step 1 users shared messages from the seed user while the information flowed from the seed user to the step 1 users. In plot A, black and gray nodes are users from different ideology groups. Sharing between black and gray nodes indicates a case of cross-ideological sharing (i.e., S→4, 2→5, and 7→8). The probability of cross-ideological sharing in plot A increases from 25% (step 1) to 33.3% (step 2) and 100% (step 3).
Source publication
Sharing cross-ideological messages on social media exposes people to political diversity and generates other benefits for society. This study argues that the diffusion patterns of political messages can influence the degree of selective sharing. Using a large-scale diffusion dataset from Twitter, this study found that messages that spread through m...
Contexts in source publication
Context 1
... second way in which messages can spread is through the viral model (person-to- person), which can be represented graphically as a diffusion tree ( Figure 1A). People can share the same messages from different users (intermediaries) in addition to the source account. ...
Context 2
... can share the same messages from different users (intermediaries) in addition to the source account. In Figure 1A, individual 2 is an intermediary between the seed user S and individual 5. A diffusion path is a chain of sharing actions. ...
Context 3
... Figure 1A shows, a message posted by seed user S could spread in four different paths: S→1; S→4; S→2→5; S→3→6; and S→3→7→8. ...
Context 5
... collection of diffusion paths is called a diffusion cascade, of which Figures 1A and 1B are two examples. A broadcast model means that all people share the message directly from the seed user ( Figure 1B). ...
Context 6
... collection of diffusion paths is called a diffusion cascade, of which Figures 1A and 1B are two examples. A broadcast model means that all people share the message directly from the seed user ( Figure 1B). On the other hand, diffusions following the viral model have many intermediaries. ...
Context 7
... the other hand, diffusions following the viral model have many intermediaries. The diffusion trees are composed of many person-to-person diffusion paths ( Figure 1A). A central difference between the figures is the cascade depth, which is the number of generations or steps in a diffusion tree. ...
Context 8
... length of the chain indicates how far the original message has spread. In Figure 1A, the cascade depth is three. Individuals 1-4 are the sharers at step 1, individuals 5-7 are the sharers at step 2, and individual 8 is the sharer at step 3. ...
Context 9
... Figure 1B, the cascade depth is one, which indicates a broadcast diffusion model. Theoretical models of information diffusion through interpersonal networks have generally been framed with analogies to contagion models of infectious diseases. ...
Context 10
... we converted a list of retweeters (Figure 2A Cross-ideological sharing was measured by comparing the ideological difference between adjacent participants in a diffusion tree. For example, if A (who leans left) retweets a message from B (who leans right), that qualifies as a case of cross-ideological sharing (see Figure 1). If retweeters are from the same ideology group, it is classified as a within-ideology case of sharing. ...
Context 11
... second way in which messages can spread is through the viral model (person-to- person), which can be represented graphically as a diffusion tree ( Figure 1A). People can share the same messages from different users (intermediaries) in addition to the source account. In Figure 1A, individual 2 is an intermediary between the seed user S and individual 5. A diffusion path is a chain of sharing actions. For example, S→2→5 is a diffusion path that indicates that individual 5 has shared a message posted by S through intermediary 2. A message can be diffused through different intermediaries to reach even more individuals through multiple ...
Context 12
... second way in which messages can spread is through the viral model (person-to- person), which can be represented graphically as a diffusion tree ( Figure 1A). People can share the same messages from different users (intermediaries) in addition to the source account. In Figure 1A, individual 2 is an intermediary between the seed user S and individual 5. A diffusion path is a chain of sharing actions. For example, S→2→5 is a diffusion path that indicates that individual 5 has shared a message posted by S through intermediary 2. A message can be diffused through different intermediaries to reach even more individuals through multiple ...
Context 13
... Figure 1B, the cascade depth is one, which indicates a broadcast diffusion model. Theoretical models of information diffusion through interpersonal networks have generally been framed with analogies to contagion models of infectious diseases. Messages are assumed to move through multiple steps from their sources in the manner of epidemics (e.g., Leskovec, Singh, & Kleinberg, 2006;Watts, 2002). Diffusion cascades with more steps indicate a higher probability of person-to-person contagion. Previous studies have found that person-to- person diffusion on social media platforms is more likely to occur between users who are already well connected in a community (Liang & Fu, 2016); users in dense communities are more likely to be homogenous with respect to political ideology or other attributes ( Barberá et al., 2015;Conover et al., 2011) and thus are less likely to share opposing messages. Therefore, cascade depth is negatively associated with the probability of cross-ideological sharing. This argument is also consistent with previous studies on mass media and cross-ideological exposure. For example, Mutz and Martin (2001) found that individuals are exposed to far more dissimilar political views via mass media (broadcast) than through interpersonal communication ...
Context 14
... Figure 1A shows, a message posted by seed user S could spread in four different paths: S→1; S→4; S→2→5; S→3→6; and ...
Context 16
... collection of diffusion paths is called a diffusion cascade, of which Figures 1A and 1B are two examples. A broadcast model means that all people share the message directly from the seed user ( Figure 1B). On the other hand, diffusions following the viral model have many intermediaries. The diffusion trees are composed of many person-to-person diffusion paths ( Figure 1A). A central difference between the figures is the cascade depth, which is the number of generations or steps in a diffusion tree. A large depth value suggests a long chain of SELECTIVE SHARING AND DIFFUSION CASCADES information diffusion and thus implies viral spreading. The length of the chain indicates how far the original message has spread. In Figure 1A, the cascade depth is three. Individuals 1-4 are the sharers at step 1, individuals 5-7 are the sharers at step 2, and individual 8 is the sharer at step ...
Context 17
... collection of diffusion paths is called a diffusion cascade, of which Figures 1A and 1B are two examples. A broadcast model means that all people share the message directly from the seed user ( Figure 1B). On the other hand, diffusions following the viral model have many intermediaries. The diffusion trees are composed of many person-to-person diffusion paths ( Figure 1A). A central difference between the figures is the cascade depth, which is the number of generations or steps in a diffusion tree. A large depth value suggests a long chain of SELECTIVE SHARING AND DIFFUSION CASCADES information diffusion and thus implies viral spreading. The length of the chain indicates how far the original message has spread. In Figure 1A, the cascade depth is three. Individuals 1-4 are the sharers at step 1, individuals 5-7 are the sharers at step 2, and individual 8 is the sharer at step ...
Context 18
... collection of diffusion paths is called a diffusion cascade, of which Figures 1A and 1B are two examples. A broadcast model means that all people share the message directly from the seed user ( Figure 1B). On the other hand, diffusions following the viral model have many intermediaries. The diffusion trees are composed of many person-to-person diffusion paths ( Figure 1A). A central difference between the figures is the cascade depth, which is the number of generations or steps in a diffusion tree. A large depth value suggests a long chain of SELECTIVE SHARING AND DIFFUSION CASCADES information diffusion and thus implies viral spreading. The length of the chain indicates how far the original message has spread. In Figure 1A, the cascade depth is three. Individuals 1-4 are the sharers at step 1, individuals 5-7 are the sharers at step 2, and individual 8 is the sharer at step ...
Context 19
... collection of diffusion paths is called a diffusion cascade, of which Figures 1A and 1B are two examples. A broadcast model means that all people share the message directly from the seed user ( Figure 1B). On the other hand, diffusions following the viral model have many intermediaries. The diffusion trees are composed of many person-to-person diffusion paths ( Figure 1A). A central difference between the figures is the cascade depth, which is the number of generations or steps in a diffusion tree. A large depth value suggests a long chain of SELECTIVE SHARING AND DIFFUSION CASCADES information diffusion and thus implies viral spreading. The length of the chain indicates how far the original message has spread. In Figure 1A, the cascade depth is three. Individuals 1-4 are the sharers at step 1, individuals 5-7 are the sharers at step 2, and individual 8 is the sharer at step ...
Context 20
... we converted a list of retweeters (Figure 2A Cross-ideological sharing was measured by comparing the ideological difference between adjacent participants in a diffusion tree. For example, if A (who leans left) retweets a message from B (who leans right), that qualifies as a case of cross-ideological sharing (see Figure 1). If retweeters are from the same ideology group, it is classified as a within-ideology case of sharing. The present study coded cross-ideological sharing as 1 and within-ideological sharing as 0. However, before that step, we had to identify the ideological preferences of all 297,566 retweeters in the diffusion ...
Similar publications
Objective: The study aims to analyze the week's day's effect on Twitter users' engagement. Method: This is quantitative, conclusive and, descriptive research, with data collection by observation and analysis via a T-test. We analyzed 51,967 posts made by 45 professional profiles with the highest number of followers from five countries (Argentina, B...
Citations
... The main approach used in the literature to distinguish broadcast from viral spreading is by computing the structural virality index [11]. Structural virality has been widely used in different studies [11][12][13][14][15]. However, when there are multiple users involved in the information-spreading process, this approach may cause causality confusion that can result in misattribution. ...
Classifying information diffusion patterns is critical to many information analysis areas, e.g., misleading information detection. However, diffusion pattern classification remains challenging when multiple users are involved. To address this challenge, this study aims to classify how information diffuses, distinguishing between broadcast and viral spreading, solely through the analysis of observational data from retweet networks on X (formerly known as Twitter). In broadcasting, most users directly receive information. However, viral spreading allows users the opportunity to receive information from a variety of sources. Therefore, viral spreading increases the likelihood of identifying misleading information. Existing methods classify diffusion types mainly through structural virality, which relies on the average distance between the users. However, when dealing with diffusion networks involving two or more information sources, these approaches can potentially lead to confusion regarding causality. To tackle this problem, we develop a deterministic causal inference method for categorizing information diffusion types. To the best of our knowledge, this is the first study investigating information diffusion types based on causality. This approach can be used to assess source credibility and assist in detecting misleading information. It can also be extended to other social media platforms.
Graphical Abstract
... Radical or extremist ideas originating from fringe communities, such as 4chan and Reddit, manage to spill over onto large social media platforms (e.g., Twitter; Zannettou et al., 2017). Once they have entered large mainstream platforms, these ideas can quickly be reproduced and spread to additional users and platforms via information cascades (e.g., Liang, 2018). ...
Over the past decade, extremists have increasingly aimed to integrate their ideologies into the center of society by changing the presentation of their narratives to appeal to a larger audience. This process is termed (strategic) mainstreaming. Although this phenomenon is not new, the factors that contribute to the mainstreaming of radical and extremist ideas have not been systematically summarized. To identify elements fostering mainstreaming dynamics, we conducted a systematic literature review of N = 143 studies. The results demonstrate that mainstreaming’s gradual and long-term nature makes it particularly difficult to operationalize, which is why it often remains a buzzword. In this article, we propose a novel conceptualization of mainstreaming, understanding it as two communicative steps (content positioning and susceptibility), and present 12 contributing factors. These factors can serve as starting points for future studies, helping to operationalize mainstreaming, empirically monitor it, and, subsequently, tackle its (long-term) effects.
... According to Mel and Vishwakarma (2020), individuals may either accept or reject facts and information based on their interests and knowledge. Research also indicates that individuals share attitudinally congruent content on social media for personal motives such as relevance, identity management and personal influence (Bobkowski, 2015;Liang, 2018). Another research on informational utility indicates individuals share news content that aligns with their personal experiences and political and social beliefs (Knobloch-Westerwick, 2015). ...
This study investigates the gatekeeping of fake news dissemination on X (formerly known as Twitter) in India and Bangladesh, aiming to identify influential actors in the online fake news ecosystem. Grounded in cascade theory, the research applies a network analysis approach to examine how fake news spreads in these countries, where political and religious misinformation is prevalent. Using data from 100,001 X posts in April 2024, the study analyzes key accounts with high betweenness centrality, eigenvector centrality, and modularity class. The primary data sources were publicly available social media posts, analyzed using Gephi software to visualize network structures and relationships. The findings reveal that individual personal accounts, rather than large institutional or media accounts, play a dominant role in shaping the conversation around fake news. The analysis demonstrates that while larger accounts have a broader reach, they are not as influential in controlling information flow. The primary topics driving fake news discussions are political and religious, with a clear focus on the upcoming elections in both countries. This study contributes to both theory and practice by offering insights into the misinformation landscape in non-Western contexts and suggesting future avenues for research, including surveys and platform-specific studies. Additionally, it provides practical implications for policymakers seeking to combat misinformation through enhanced digital literacy and regulatory frameworks.
... It has been found that highly clustered networks tend to contain redundant social ties (i.e., friends of friends are friends themselves; Granovetter, 1973;Watts & Strogatz, 1998) that increase the frequency of contacts, and hence the volume of messages communicated in general (Bond et al., 2012;Centola, 2010). Some studies have indeed found that local community structures, such as network redundancy or clustering affect the probability of message diffusion (Liang, 2018;Liang & Fu, 2016;Harrigan et al., 2012;Centola, 2010). This suggests that being part of highly clustered networks, as redundant ties increase the chances of message exposure, may induce individuals to post more messages than those in less clustered networks. ...
Most messages on social media platforms are reportedly posted by a small number of active communicators, while the great majority of users remain silent as lurkers who read but seldom write. Despite extensive research to date, it remains unclear why such a disparity in individuals’ participation in social media exists. Drawing on the behavioral data of 15,633 Facebook users nested in 73 local networks, this study attempted to examine how the structural properties of networks give rise to the highly skewed distribution of message contributions between individual users. Multilevel statistical analyses of the data revealed that the participation disparity among individuals might be in part a function of the structural characteristics of networks in which they are embedded, suggesting that being active or silent in the social media environment is largely conditional on the surrounding network structures.
... "What matters is going viral" is an ideology that hinders the eradication of hoaxes and fake news. (Bernatta & Kartika, 2020); (Liang, 2018) Meanwhile, capitalism in the mass media manifests in the size of clickbait, page view standards, and the production of news quantity (Hadiyat, 2019); (Munger et al., 2018); (Zhou, 2022); (Kaushal & Vemuri, 2021); (Lischka & Garz, 2021). Spreading information and news links is one way to viralize political support before the 2024 elections. ...
Political discourse preceeding the 2024 General Election has led to the contamination of the media landscape with hoaxes and fakenews. The public, candidates, and state actors not only compete with factual information and positive news, but also contends with the spread of misinformation and fake news. The prevalence of hoaxes on social media and fake news in mass media circulation prior to the 2024 General Election poses a tangible threat to social unity and national integration. This paper aims to analyze the model of public communication used to combat hoaxes and fakenews before the 2024 General Election done by the Ministry of Communication and Information and other stakeholders. The data for this paper is based on qualitative research carried out with data collection technique through literature searches, observation, and interview. The study discovered that the public communication model developed to combat hoaxes and fake news in advance of the 2024 General election has been ineffective. The fundamental weakness of the public communication model in eradicating hoaxes and fake news lies in the uneven application of the law inforcement and penalties. In addition, there are some elements of the public communication model that do not systematically show significant and consistent progress. At a macro level, this can be understood as the lack of clear roadmap and grand design to eliminate hoaxes and fake news in Indonesia. A clear plan or atrategy is needed for the long term establishment of a culture focused on the eradication of hoaxes and fake news.
... It is, however, reassuring that the patterns of political information-sharing do not always produce homogeneity. A study by Liang (2018) found that political messages are likely to travel across the ideological spectrum. That is, political messages can simply become viral and are widely shared regardless of ideology instead of being broadcasted in a top-down fashion. ...
The volume Divided: Open-Mindedness and Dogmatism in a Polarized World provides a current scientific understanding of open-mindedness and dogmatism, illuminates the nature and causes of polarization, and provides clues regarding how one might attempt to reduce pernicious forms of polarization. To do so, this volume brings together a diverse group of leading psychologists, political scientists, and communication scholars who investigate dogmatism and open-mindedness within social and political contexts and who provide novel insight into these pressing topics. Chapter authors cover a wide range of topics including key definitions of dogmatism and open-mindedness, the emergence of affective polarization, how open-mindedness relates to attitude formation and change, the correspondence between intellectual humility and open-mindedness, how social norms and situations shape open-minded cognition, authoritarianism among liberals and conservatives, the relation between political conservatism and dogmatic thought, and the role of affect in guiding cognitive processes. Authors consider both the beneficial and more problematic features of open-mindedness, dogmatism, and polarization. Collectively, this volume provides a format that enables readers to learn about creative approaches to understanding dogmatism and open-mindedness and, potentially, to generate innovative solutions that reduce polarization and increase constructive social compromise in the future.
... In terms of behavior, one study found a positive influence of homophilic peer influence on ILEs (Brugnoli et al. 2019), while another study found no impact of peer-induced mainstreaming or polarization on ILEs (Shore et al. 2018). Additionally, two studies focused on Twitter reported a negative association between cascading information dissemination among non-connected users (Liang 2018) and the frequency of inter-group exchange (Yang et al. 2021), and ILE emergence. However, direct interaction with seed users demonstrated a positive impact on ILEs, both on Facebook and Twitter (Bozdag et al. 2014, Shore et al. 2018). ...
Social media platforms offer a convenient way for people to interact and exchange information. However, there are sustained concerns that filter bubbles and echo chambers create information-limiting environments (ILEs) for their users. Despite a well-developed conceptual understanding, the empirical evidence regarding the causes and supporting conditions of these ILEs remains inconclusive. This paper addresses this gap by applying the triple-filter-bubble model developed by Geschke et al. (2019) to analyze empirical literature on the individual, social, and technological causes of ILEs. While we identify some factors that increase the probability of ILEs under certain conditions, our findings do not suffice to thoroughly validate conceptual models that explain why ILEs emerge. Therefore, we call for future research to investigate the causes of ILEs with higher external validity to develop a more comprehensive understanding of this phenomenon.
... Indeed, the social environment, at least in conventional offline settings, often gives females less access to information and resources in new venture creation [39], [40] so they have less access to information related to entrepreneurship. In contrast, information and knowledge via social media can be spread in a broadcast way and is generally not gender-specific [55], such that females can access information and knowledge as equally as their male counterparts. Hence, social media can, to some extent, help level the playing field for females who may struggle to get entrepreneurship-related information and knowledge from conventional offline channels. ...
As an advanced information technology, social media has increasingly become a significant and prevalent source of information for both firms and people, and has reshaped entrepreneurial activities. Our article draws on the information processing perspective to build a model to explore the relationship between social media use and entrepreneurial entry and whether the relationship differs for disadvantaged populations in terms of gender, ethnicity, geographic divide (rural versus urban), and regional marketization. Using a nationally representative survey of 22 614 adults in the China Family Panel Studies, we test our model quantitatively with regression analysis. The results reveal that social media use can have a positive effect on individuals’ entrepreneurial entry and that the relationship between social media use and entrepreneurial entry is stronger for disadvantaged populations in terms of gender, rural versus urban divide, and regional marketization. Our identification of females, rural populations, and people in low marketization regions is among the first to distinguish the effect of social media on entrepreneurship for distinct populations. By highlighting the role of social media use in shaping entrepreneurial entry among disadvantaged populations, this study opens the black box of social media vis-à-vis entrepreneurial entry, and highlights that the role of social media use on entrepreneurial entry is stronger for disadvantaged populations, who have greater barriers to accessing entrepreneurship-related information in their offline environment.
... We hypothesize public discourse on social media as a dynamic system that is driven by an oscillation between viral versus broadcast diffusion processes (Liang, 2018). These processes appear driven by endogenous and exogenous events, respectively (Crane & Sornette, 2008). ...
... Information diffusion on social media can be categorized into two types of processes: broadcast and viral spreading (Liang, 2018). Each has distinct good, bad, and ugly impacts on public discourse in aspects such as political variety (Liang, 2018), innovation adoption (Zhai et al., 2021), challenging "dominant knowledge" (Jackson & Foucault Welles, 2015), and the spreading of false information (Vosoughi et al., 2018). ...
... Information diffusion on social media can be categorized into two types of processes: broadcast and viral spreading (Liang, 2018). Each has distinct good, bad, and ugly impacts on public discourse in aspects such as political variety (Liang, 2018), innovation adoption (Zhai et al., 2021), challenging "dominant knowledge" (Jackson & Foucault Welles, 2015), and the spreading of false information (Vosoughi et al., 2018). ...
Understanding information diffusion is vital to explaining the good, bad, and ugly impacts of social media. Two types of processes govern information diffusion: broadcasting and viral spread. Viral spreading is when a message is diffused by peer-to-peer social connections, whereas broadcasting is characterized by influences that can come from outside of the peer-to-peer social network. How these processes shape public discourse is not well understood. Using a simulation study and real-world Twitter data (10,155 users, 18,000,929 tweets) gathered during 2020, we show that broadcast spreading is associated with more integrated discourse networks compared to viral spreading. Moreover, discourse oscillates between extended periods of segregation and punctuated periods of integration. These results defy simple interpretations of good or bad, and instead suggest that information diffusion dynamics on social media have the capacity to disrupt or amplify both prosocial and antisocial content.
... Moreover, issue moralization was found to amplify myside sharing above and beyond attitude extremity in the majority of the studies. Expanding prior research on selective communication, our work provides a clear demonstration that citizens' myside communicational preference is powerfully amplified by their moral and political ideology (18,19,(39)(40)(41)(42)(43). By examining this phenomenon across multiple experiments varying numerous parameters, we demonstrated the robustness of myside sharing and of its amplification by participants' issue moralization and attitude extremity. ...
We explored whether moralization and attitude extremity may amplify a preference to share politically congruent (“myside”) partisan news and what types of targeted interventions may reduce this tendency. Across 12 online experiments (N = 6,989), we examined decisions to share news touching on the divisive issues of gun control, abortion, gender and racial equality, and immigration. Myside sharing was systematically observed and was consistently amplified when participants (i) moralized and (ii) were attitudinally extreme on the issue. The amplification of myside sharing by moralization also frequently occurred above and beyond that of attitude extremity. These effects generalized to both true and fake partisan news. We then examined a number of interventions meant to curb myside sharing by manipulating (i) the audience to which people imagined sharing partisan news (political friends vs. foes), (ii) the anonymity of the account used (anonymous vs. personal), (iii) a message warning against the myside bias, and (iv) a message warning against the reputational costs of sharing “mysided” fake news coupled with an interactive rating task. While some of those manipulations slightly decreased sharing in general and/or the size of myside sharing, the amplification of myside sharing by moral attitudes was consistently robust to these interventions. Our findings regarding the robust exaggeration of selective communication by morality and extremism offer important insights into belief polarization and the spread of partisan and false information online.
