Click, Connect, Conspire: Unraveling Facebook’s Global Misinformation Web

Authors: (1) Esteban Villa-Turek, Corresponding author; (2) Rod Abhari, Collaborator; (3) Erik C. Nisbet, Collaborator; (4) Yu Xu, Collaborator; (5) Ayse Deniz Lokmanoglu, Collaborator. Table of Links Abstract Transnational Network Dynamics of Problematic Information Diffusion Theoretical Framework Methodology Statistical Analysis and Results Discussion and Limitations Conclusions and Policy Implications, and References Theoretical Framework The notion of transnationalism in social media content diffusion is certainly not new. For instance, it has been employed to study the networks of far-right actors on Twitter to show how like-minded people across national borders form communities based on a mutually understood, shared worldviews and interpretative frames (Caiani & Della Porta, 2011; Froio & Ganesh, 2019). In other words, transnational online communication represents an additional channel for groups to consolidate narratives and audiences, which may in turn translate into offline activism and mobilization efforts, like those arranged and promoted by MPV in European and Latin American countries (Gardel, 2020; Maldita.es, 2021). Increasingly, social networking sites (SNS) enable and mediate instances of human communication forming networks that are virtually unknown and in which flow is much harder to track, let alone to predict. This has brought about what has been termed the “dark social” (Madrigal, 2012; Swart et al., 2018) and we argue that we find ourselves in the age of “dark networks”. Instances of this phenomenon can be found in the emergence of online communities of different sizes, on different platforms and for different purposes. In general, the theoretical emphasis around defining online communities has revolved around their two main components: what communities are and their online nature. Regarding the first element, organizational theorists have argued that organizations precede communities, but the case has been made for postindustrial communities to precede and mediate the emergence of organizations (O’Mahony & Lakhani, 2011). Most importantly, however, is the idea that these contemporary forms of community can be better understood from a procedural standpoint by characterizing them as “a voluntary collection of actors whose interests overlap and whose actions are partially influenced by this perception” (O’Mahony & Lakhani, 2011). These communities therefore enable the convergence of a multitude of diverse backgrounds, interests, locations, etc., under the banner of some narrowly conceived perception of overlapping interests (O’Mahony & Lakhani, 2011), which mainly occurs due to the pervasiveness of networked digital communication technologies. With regards to the second element, their digital nature, online communities can be thought of as those who exhibit the properties of pervasive awareness and persistent contact (Hampton, 2016). Both properties relate to the person-to-network (Hampton, 2016) nature of online communities, in which actors are perennially connected across time and space, and where actors can broadcast information without the ability to confirm its reception (Hampton, 2016). Online communities therefore allow us to understand how problematic information was able to flow from the Global North to the Global South during the coronavirus pandemic. In other words, by allowing otherwise very dissimilar and distant actors to establish affinity ties by means of having overlapped and narrow interest (as signaled by being affiliated to public Facebook groups), online communities enabled a type of collective action - connective action - which was characterized by the existence of key groups that acted as transnational brokers. The theory of connective action (Bennett & Segerberg, 2012) was introduced in an effort to make sense of the multitude of digitally enabled organizational phenomena that were being sparked across the globe, from the Arab Spring to Occupy Wall Street to los indignados. More traditional theories on the emergence and organizational logic of offline mobilizations, such as the theories of collective action, were not able to fully explain the new geographically distant yet paradoxically unified instances of social mobilizations happening around very diverse issues across the world. In other words, the theory of connective action helped bridge the gaps that were starting to become evident with cases in which digital communication platforms afforded new kinds of networked communities, characterized by pervasive awareness and persistent contact (Hampton, 2016). Thus, the theory of connective action enabled the characterization and theorization of particular communities that have two distinctive characteristic (Bennett & Segerberg, 2012). First, they are unified, not around a central organizing entity or idea, but around personalized and actionable ideas that are intended to relate to individuals at personal scale through the perception of narrow, overlapping interests (O’Mahony & Lakhani, 2011), even if there are other overarching differences among those individuals. This, in turn, eliminates some of the barriers that might otherwise inhibit the formation of communities (O’Mahony & Lakhani, 2011) in general and transnational online movements (Hampton, 2016) in particular. Second, the organizations that emerge following the logic of connective action should have a technological component such that it enables communication among those individuals that connect around shared personalized ideas (Bennett & Segerberg, 2012). Although previous theories of collective action were not able to fully explain the existence of these technologically enabled cases of online communities, as (Bennett & Segerberg, 2012) propose, the logics of collective and connective action are not mutually exclusive but rather allow for a spectrum of hybrid categories of action logics. This hybridity is useful in cases where, although some elements of connective action logic are present, others are distinctively characteristic of the collective action logic, like the existence of concomitant propinquity or spatial proximity that could help connective action communities grow and thrive (O’Mahony & Lakhani, 2011) through the use of collective action elements like traditional resource mobilization (such as on-the-ground rallies). Another instance of hybridity can be observed in the fluidity of connective action roles (Bennett & Segerberg, 2012). As we will see, the seminal advocacy roles were fulfilled by more prominent and central actors, specifically those who ignited the spark by means of their perceived role of authority and expertise. These initial, somewhat static, roles did resemble those of traditional collective action theory, whereas the dynamic emergence of key transnational brokers resembled the connective action logic more. The two case studies analyzed here involved problematic information, in the form of COVID-19 problematic information, that spread transnationally across countries and continents. This occurred specifically because they were enabled to do so by an underlying digital communication platform that allowed users to find and join groups of like-minded users around personalized and narrow interests (Bennett & Segerberg, 2012; O’Mahony & Lakhani, 2011), despite being located in distant geographies or having nothing else in common. The possibility of networked yet diverse and possibly distant actors was possible therefore thanks to the connective affordances that digital communication platforms represent (Vaast et al., 2017). In such cases, the platform affords otherwise dissimilar individuals to connect and form networks based on shared interests or around distinctive issues. In these networks, roles are not fixed like in traditional theories of collective action, but rather allow for a fluid and dynamic emergence and cessation of roles in the network, like in the case of roles of support, advocacy and amplification (Vaast et al., 2017) We argue that Facebook public groups are the quintessence of connective affordance: they are sets of very diverse actors that nevertheless coalesce around a very specific issue of interest, that in turn informed the creation of the group in the first place. In fact, studies have shown that, in general, Facebook users rely on the platform to connect with otherwise unknown people while at the comfort of home, in other words to expand their networks’ weak ties in a form of bridging social capital (Papacharissi & Mendelson, 2010). In fact, recent studies have found that Facebook groups afford several different kinds of connective action, for better or worse. For instance, a study of an extremist Swedish Facebook group found that the platform’s architecture enabled the digitization of old anti-immigration discourse by allowing users to connect and assume different roles in an effort to increase the group’s persuasive power (Merrill & Åkerlund, 2018). On the other hand, several other studies have documented how Facebook groups afford connective action for information seeking purposes. Specifically, another study of a Facebook group of international (non-Swedish) mothers living in Sweden found that, through the group’s brokerage, otherwise dissimilar women were able to connect based on their shared motherhood in a foreign land to search and provide information relevant to their wellbeing (Mansour, 2021). Similarly, another study found that a Facebook group of Iranian women denouncing their experiences with hijab norms in the country afforded the creation of a crucial collective identity in the protest build-up process (Khazraee & Novak, 2018). Therefore, it follows that Facebook groups can be understood as critical elements for connective action in that they broker information diffusion among otherwise geographically distant people who nevertheless overlap around narrow and shared particular interests represented by the thematic nature of the Facebook groups. By employing a network perspective, we are able to identify dynamic brokerage roles similar to those of connective action (advocacy, support or amplification), but with the crucial function of operating at a transnational level. We argue that distinctively active groups served as transnational brokers in the process of problematic information diffusion between the Global North and the Global South during the coronavirus pandemic. This is the case, because the fluid nature of emergent roles in the logic of connective action allows for actors (public Facebook groups in this study) to broker between otherwise distant countries and continents. In fact, brokerage roles have been found to play a critical role in social movements by fulfilling the need to bridge between otherwise disconnected actors, therefore ensuring cohesion among the collective’s members (Diani & McAdam, 2003). This potential brokerage role, however, has been found to depend greatly on certain actor attributes (Diani & McAdam, 2003). In fact, we expect influential groups to operate enabled by the fact that there is a shared linguistic element (a “postcolonial affordance” of sorts), which supported the emergence of the transnational misinfodemic network in the first place. Therefore, we set out to investigate which group characteristics explain why they became or not influential actors in the diffusion process and whether the same group-level characteristics are able to explain the emergence of ties in the network as a whole. As has been explained above, online communities operate by linking dissimilar actors under a narrow and personalized overlapping area of interest (O’Mahony & Lakhani, 2011). Thus, we hypothesize first that thematic similarity between groups increases the likelihood of tie formation in the network, since the initial narrow yet shared issue of interest informed the creation and growth of the groups in the first place. Second, it has been noted that the logic of connective action allows for hybridity in connective affordances, meaning that online communities can also be strengthened by physical proximity (Bennett & Segerberg, 2012; O’Mahony & Lakhani, 2011). To capture possible geographical proximity effects in the emergence of the network, we employ a measure of geographic location of the groups at the country level, hypothesizing that geographic proximity among groups increases the likelihood of tie formation in the network. Third, we argue that there are latent cultural similarity effects among groups, and that they increase the likelihood of tie formation in the network. For this purpose we employ a measure of that captures cultural similarities among countries well, Facebook’s Social Connectedness Index (SCI) (Bailey et al., 2018). In fact, the SCI has proven useful in correlating transnational connectedness with historic intranational and international migratory patterns (Bailey et al., 2018), as well as international trade patterns (Bailey et al., 2021), capturing therefore intangible relationships that serve as a proxy for cultural proximity in this study. Finally, as noted above, the appearance of influential roles in the network, such as brokerage, is conditional on the existence of certain characteristics in the actors (Diani & McAdam, 2003). Therefore, given that shared linguistic attributes among postcolonial countries is a sine qua non characteristic for the emergence of the networks under study, it is necessary to include them as group-level attributes to estimate their role in its emergence, arguing that a shared language increases the likelihood of tie formation and, subsequently, enables the emergence of dynamic roles within the network. This study uses a network approach that is sensitive to transnational communication dimensions, specifically the geographic proximity between Facebook groups, the cultural proximity of audiences in diverse parts of the world, all while upholding the importance of shared linguistic attributes in the process. The latter is a crucial element for understanding the rapid diffusion of COVID-19 problematic information in Spanish and French, which emerged in Europe and spread rapidly across Latin American and francophone countries, respectively. Isolating the network structure of conspiracy theory and misinformation diffusion exposes their distinctive structural characteristics, which enables researchers, healthcare workers and policymakers to develop targeted responses and contributes to a better evidence-based approach to global communication governance efforts (GCG) (Padovani & Pavan, 2011). This paper is available on arxiv under CC BY 4.0 DEED license. Authors: (1) Esteban Villa-Turek, Corresponding author; (2) Rod Abhari, Collaborator; (3) Erik C. Nisbet, Collaborator; (4) Yu Xu, Collaborator; (5) Ayse Deniz Lokmanoglu, Collaborator. Authors: Authors: (1) Esteban Villa-Turek, Corresponding author; (2) Rod Abhari, Collaborator; (3) Erik C. Nisbet, Collaborator; (4) Yu Xu, Collaborator; (5) Ayse Deniz Lokmanoglu, Collaborator. Table of Links Abstract Abstract Transnational Network Dynamics of Problematic Information Diffusion Transnational Network Dynamics of Problematic Information Diffusion Theoretical Framework Theoretical Framework Methodology Methodology Statistical Analysis and Results Statistical Analysis and Results Discussion and Limitations Discussion and Limitations Conclusions and Policy Implications, and References Conclusions and Policy Implications, and References Theoretical Framework The notion of transnationalism in social media content diffusion is certainly not new. For instance, it has been employed to study the networks of far-right actors on Twitter to show how like-minded people across national borders form communities based on a mutually understood, shared worldviews and interpretative frames (Caiani & Della Porta, 2011; Froio & Ganesh, 2019). In other words, transnational online communication represents an additional channel for groups to consolidate narratives and audiences, which may in turn translate into offline activism and mobilization efforts, like those arranged and promoted by MPV in European and Latin American countries (Gardel, 2020; Maldita.es, 2021). Increasingly, social networking sites (SNS) enable and mediate instances of human communication forming networks that are virtually unknown and in which flow is much harder to track, let alone to predict. This has brought about what has been termed the “dark social” (Madrigal, 2012; Swart et al., 2018) and we argue that we find ourselves in the age of “dark networks”. Instances of this phenomenon can be found in the emergence of online communities of different sizes, on different platforms and for different purposes. In general, the theoretical emphasis around defining online communities has revolved around their two main components: what communities are and their online nature. Regarding the first element, organizational theorists have argued that organizations precede communities, but the case has been made for postindustrial communities to precede and mediate the emergence of organizations (O’Mahony & Lakhani, 2011). Most importantly, however, is the idea that these contemporary forms of community can be better understood from a procedural standpoint by characterizing them as “a voluntary collection of actors whose interests overlap and whose actions are partially influenced by this perception” (O’Mahony & Lakhani, 2011). These communities therefore enable the convergence of a multitude of diverse backgrounds, interests, locations, etc., under the banner of some narrowly conceived perception of overlapping interests (O’Mahony & Lakhani, 2011), which mainly occurs due to the pervasiveness of networked digital communication technologies. With regards to the second element, their digital nature, online communities can be thought of as those who exhibit the properties of pervasive awareness and persistent contact (Hampton, 2016). Both properties relate to the person-to-network (Hampton, 2016) nature of online communities, in which actors are perennially connected across time and space, and where actors can broadcast information without the ability to confirm its reception (Hampton, 2016). Online communities therefore allow us to understand how problematic information was able to flow from the Global North to the Global South during the coronavirus pandemic. In other words, by allowing otherwise very dissimilar and distant actors to establish affinity ties by means of having overlapped and narrow interest (as signaled by being affiliated to public Facebook groups), online communities enabled a type of collective action - connective action - which was characterized by the existence of key groups that acted as transnational brokers. The theory of connective action (Bennett & Segerberg, 2012) was introduced in an effort to make sense of the multitude of digitally enabled organizational phenomena that were being sparked across the globe, from the Arab Spring to Occupy Wall Street to los indignados. More traditional theories on the emergence and organizational logic of offline mobilizations, such as the theories of collective action, were not able to fully explain the new geographically distant yet paradoxically unified instances of social mobilizations happening around very diverse issues across the world. In other words, the theory of connective action helped bridge the gaps that were starting to become evident with cases in which digital communication platforms afforded new kinds of networked communities, characterized by pervasive awareness and persistent contact (Hampton, 2016). Thus, the theory of connective action enabled the characterization and theorization of particular communities that have two distinctive characteristic (Bennett & Segerberg, 2012). First, they are unified, not around a central organizing entity or idea, but around personalized and actionable ideas that are intended to relate to individuals at personal scale through the perception of narrow, overlapping interests (O’Mahony & Lakhani, 2011), even if there are other overarching differences among those individuals. This, in turn, eliminates some of the barriers that might otherwise inhibit the formation of communities (O’Mahony & Lakhani, 2011) in general and transnational online movements (Hampton, 2016) in particular. Second, the organizations that emerge following the logic of connective action should have a technological component such that it enables communication among those individuals that connect around shared personalized ideas (Bennett & Segerberg, 2012). Although previous theories of collective action were not able to fully explain the existence of these technologically enabled cases of online communities, as (Bennett & Segerberg, 2012) propose, the logics of collective and connective action are not mutually exclusive but rather allow for a spectrum of hybrid categories of action logics. This hybridity is useful in cases where, although some elements of connective action logic are present, others are distinctively characteristic of the collective action logic, like the existence of concomitant propinquity or spatial proximity that could help connective action communities grow and thrive (O’Mahony & Lakhani, 2011) through the use of collective action elements like traditional resource mobilization (such as on-the-ground rallies). Another instance of hybridity can be observed in the fluidity of connective action roles (Bennett & Segerberg, 2012). As we will see, the seminal advocacy roles were fulfilled by more prominent and central actors, specifically those who ignited the spark by means of their perceived role of authority and expertise. These initial, somewhat static, roles did resemble those of traditional collective action theory, whereas the dynamic emergence of key transnational brokers resembled the connective action logic more. The two case studies analyzed here involved problematic information, in the form of COVID-19 problematic information, that spread transnationally across countries and continents. This occurred specifically because they were enabled to do so by an underlying digital communication platform that allowed users to find and join groups of like-minded users around personalized and narrow interests (Bennett & Segerberg, 2012; O’Mahony & Lakhani, 2011), despite being located in distant geographies or having nothing else in common. The possibility of networked yet diverse and possibly distant actors was possible therefore thanks to the connective affordances that digital communication platforms represent (Vaast et al., 2017). In such cases, the platform affords otherwise dissimilar individuals to connect and form networks based on shared interests or around distinctive issues. In these networks, roles are not fixed like in traditional theories of collective action, but rather allow for a fluid and dynamic emergence and cessation of roles in the network, like in the case of roles of support, advocacy and amplification (Vaast et al., 2017) We argue that Facebook public groups are the quintessence of connective affordance: they are sets of very diverse actors that nevertheless coalesce around a very specific issue of interest, that in turn informed the creation of the group in the first place. In fact, studies have shown that, in general, Facebook users rely on the platform to connect with otherwise unknown people while at the comfort of home, in other words to expand their networks’ weak ties in a form of bridging social capital (Papacharissi & Mendelson, 2010). In fact, recent studies have found that Facebook groups afford several different kinds of connective action, for better or worse. For instance, a study of an extremist Swedish Facebook group found that the platform’s architecture enabled the digitization of old anti-immigration discourse by allowing users to connect and assume different roles in an effort to increase the group’s persuasive power (Merrill & Åkerlund, 2018). On the other hand, several other studies have documented how Facebook groups afford connective action for information seeking purposes. Specifically, another study of a Facebook group of international (non-Swedish) mothers living in Sweden found that, through the group’s brokerage, otherwise dissimilar women were able to connect based on their shared motherhood in a foreign land to search and provide information relevant to their wellbeing (Mansour, 2021). Similarly, another study found that a Facebook group of Iranian women denouncing their experiences with hijab norms in the country afforded the creation of a crucial collective identity in the protest build-up process (Khazraee & Novak, 2018). Therefore, it follows that Facebook groups can be understood as critical elements for connective action in that they broker information diffusion among otherwise geographically distant people who nevertheless overlap around narrow and shared particular interests represented by the thematic nature of the Facebook groups. By employing a network perspective, we are able to identify dynamic brokerage roles similar to those of connective action (advocacy, support or amplification), but with the crucial function of operating at a transnational level. We argue that distinctively active groups served as transnational brokers in the process of problematic information diffusion between the Global North and the Global South during the coronavirus pandemic. This is the case, because the fluid nature of emergent roles in the logic of connective action allows for actors (public Facebook groups in this study) to broker between otherwise distant countries and continents. In fact, brokerage roles have been found to play a critical role in social movements by fulfilling the need to bridge between otherwise disconnected actors, therefore ensuring cohesion among the collective’s members (Diani & McAdam, 2003). This potential brokerage role, however, has been found to depend greatly on certain actor attributes (Diani & McAdam, 2003). In fact, we expect influential groups to operate enabled by the fact that there is a shared linguistic element (a “postcolonial affordance” of sorts), which supported the emergence of the transnational misinfodemic network in the first place. Therefore, we set out to investigate which group characteristics explain why they became or not influential actors in the diffusion process and whether the same group-level characteristics are able to explain the emergence of ties in the network as a whole. As has been explained above, online communities operate by linking dissimilar actors under a narrow and personalized overlapping area of interest (O’Mahony & Lakhani, 2011). Thus, we hypothesize first that thematic similarity between groups increases the likelihood of tie formation in the network, since the initial narrow yet shared issue of interest informed the creation and growth of the groups in the first place. Second, it has been noted that the logic of connective action allows for hybridity in connective affordances, meaning that online communities can also be strengthened by physical proximity (Bennett & Segerberg, 2012; O’Mahony & Lakhani, 2011). To capture possible geographical proximity effects in the emergence of the network, we employ a measure of geographic location of the groups at the country level, hypothesizing that geographic proximity among groups increases the likelihood of tie formation in the network. Third, we argue that there are latent cultural similarity effects among groups, and that they increase the likelihood of tie formation in the network. For this purpose we employ a measure of that captures cultural similarities among countries well, Facebook’s Social Connectedness Index (SCI) (Bailey et al., 2018). In fact, the SCI has proven useful in correlating transnational connectedness with historic intranational and international migratory patterns (Bailey et al., 2018), as well as international trade patterns (Bailey et al., 2021), capturing therefore intangible relationships that serve as a proxy for cultural proximity in this study. Finally, as noted above, the appearance of influential roles in the network, such as brokerage, is conditional on the existence of certain characteristics in the actors (Diani & McAdam, 2003). Therefore, given that shared linguistic attributes among postcolonial countries is a sine qua non characteristic for the emergence of the networks under study, it is necessary to include them as group-level attributes to estimate their role in its emergence, arguing that a shared language increases the likelihood of tie formation and, subsequently, enables the emergence of dynamic roles within the network. This study uses a network approach that is sensitive to transnational communication dimensions, specifically the geographic proximity between Facebook groups, the cultural proximity of audiences in diverse parts of the world, all while upholding the importance of shared linguistic attributes in the process. The latter is a crucial element for understanding the rapid diffusion of COVID-19 problematic information in Spanish and French, which emerged in Europe and spread rapidly across Latin American and francophone countries, respectively. Isolating the network structure of conspiracy theory and misinformation diffusion exposes their distinctive structural characteristics, which enables researchers, healthcare workers and policymakers to develop targeted responses and contributes to a better evidence-based approach to global communication governance efforts (GCG) (Padovani & Pavan, 2011). This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv