Media and power in times of hegemonic crisis: exploring contentious climate politics on Twitter

ABSTRACT During moments of hegemonic crisis, it is crucial to understand the dynamics of media power. This article identifies key structural forces and disruptive tendencies that are factors in a current crisis of the neoliberal hegemonic order. This intellectual background underpins an empirical analysis of Twitter data collected during the UN climate summit in Glasgow, Scotland (COP26). We identify large attention clusters around established political actors and their counter-hegemonic challengers; concluding that during these intense global political events a hegemonic crisis produces spaces where systemically marginalized actors can harness relational attention power and mobilize it toward distinct political purposes.

pp. [83][84].The turmoil of modernization in emerging western democracies was uprooting populations from older habits, rituals, networks, and common sense (Tönnies, 1887(Tönnies, /2001)).Consequently, the loyalty of the 'mass audience' -imagined as a crowd of atomized individuals -seemed vulnerable to the persuasive power of the then emerging new media (i.e.radio, film, popular press).In the aftermath of the First World War, questions concerning 'mass communication' and its effects would greatly influence the consolidation of media research into an institutionalized discipline (Pietilä 2008).
We seem to be living through another one of these moments.The world-shaping power of western neoliberal globalization (Harvey 2005) has been challenged by its own internal contradictions, geopolitical rivals, and environmental limits.Economic restructuring has produced growing levels of social inequality within nations (e.g.Savage 2021, Piketty 2022).The rapid rise of digital, networked, and datafied communication infrastructures has fueled the disruption of conventional forms of political and cultural authority (e.g.Cmiel andPeters 2020, Calhoun et al. 2022).The fossil fuel-driven energy supply of modernity has ushered in an existential environmental crisis and transformative demands at all scales, ranging from globalized social systems to localized everyday life (e.g.Mitchell 2011, Malm 2016, Hulme 2021).Taking stock of media power whilst living inside this conjuncture is as crucial as it is complex.
This article combines attention to wider historical conditions and forces with a detailed empirical analysis of contemporary media content.We begin with a snapshot of the past that anchors our analysis of the present systemic crisis.Drawing parallels between the early 20 th century and today reminds us that understanding media power is crucially important and particularly interesting in moments where hegemonic powers and societal orders are under question.This motivates a retrospective consideration of historical communication research and its ways of making sense of media power.Paying attention to parallels and discontinuities between this earlier period of instability we pinpoint new aspects of the current media system and its place in social and political power.We then operationalize these reflections through an empirical analysis of communication pertaining to the root cause of the current crisis conjuncture: widely shared consciousness of the projected instability of Earth's climactic system.
More specifically, we examine social media interactions that occurred during the UNFCCC global climate summit, held in Glasgow, Scotland (COP26) from 28 October to 12 November 2021.By analyzing a large sample of Twitter interactions (11,667,143 million tweets), we exemplify how the current global crisis conjuncture concretely and materially played out on one influential social media platform.Furthermore, we discuss how differentiated resources of political power translate into attention capital in a social media network.Our analysis of patterns in the COP26 retweet network reveals large clusters of activity both around hegemonic, systemic actors and their challengers.Closer investigation of similarities and differences between two counter-hegemonic 'crisis communities' helps to illustrate how relational attention power is constructed in the current social media landscape.We conclude that exceptionally intense global political moments, like UN climate summits, reveal how fissures in the hegemonic order produce spaces where systemically marginalized actors can harness attention power and mobilize it toward distinct political purposes.

The context: an anatomy of hegemonic crisis
In the 2020s, our argument is that any consideration of media power must be situated against a deeper contextual understanding of a systemic crisis of the hegemonic order and the shifting horizons of common sense upon which this order is constituted.In this article, we highlight three main aspects of this crisis conjuncture.First, the climate crisis which points to material limits and a growing self-awareness of the unsustainability of existing social structures as well as to the temporal urgency of adapting to these environmental limitations.Second, lessons from historical conceptualizations of media effects which highlight connections between shifting imaginaries about audiences and media power, as well as the ways they differ in the context of crisis and stability.Third, we consider the changing role of 'social networks' in a contemporary infrastructure of mediated, public communication.These remarks provide a selective anatomy of the hegemonic order at stake in our empirical analysis of socially mediated climate politics.We contend that understanding of structural power (a)symmetries, contradictions and fissures in this hegemonic arrangement are essential ingredients for analyzing media-power relations today.

The context: climate crisis as a systemic problem
Climate change is quite literally a constitutional problem for modern societies (Jasanoff 2012, pp. 247-248), exposing material limits to the constitution of society.Fossil fuel dependency persists as a necessary element of our entire way of life and the social order of contemporary societies.This dependency stretches from our energy systems to our everyday habits, from our institutional structures and beliefs about legitimate governance to our perceived identities and systemic power relations both within and between nations (Mitchell 2011, Dryzek et al. 2013).This spotlights the material connection between 'power' as the source of concrete energy and the structures of social and political power (Malm 2016, pp. 16-17).Simultaneously, a widely shared scientific consensus points to the material unsustainability of this order and the imperative for rapid economic, societal and political transition.Yet the 'super wicked' (Levin et al. 2012) challenge of decarbonization entangles climate politics with other contemporary social issues: inequality (e.g.Piketty 2022, pp. -225-237), the promise of artificial intelligence (Crawford 2021, Brevini 2022), human mobility and migration (Ghosh 2021, Vince 2022), among many others.
The hard external, environmental limits of carbon-based modern societies and lifestyles are creating frictions inside societies that are increasingly felt in everyday lifeworlds (Friedland and Kunelius 2023).The economic, social and cultural feedback effects of these externalities are shaking up existing social structures and life expectations.This is carving out new spaces of political division (Norris and Inglehart 2018) while offering fertile ground for backward looking, reactionary (Lilla 2017) or defensive (Jackson and Kreiss 2023) identity politics.The global rise of populist movements (Moffitt 2016) and openly authoritarian leaders (Keane 2020) across political contexts represents one obvious symptom of this increasingly volatile systemic order.

Lessons from media research: from mass society to social networks
Historical sociology has thoroughly examined the structural causes and contingent events that spiraled much of the world into the catastrophic period of the two World Wars (e.g.Polanyi 1944, Arrighi 1994, Mann 2013).These volatile decades of the 20 th century also triggered increasing scholarly interest in the role of media and communication (Peters 1999).The First World War produced influential analyses of propaganda (Bloch 1920/2013, Lippmann 1922, Lasswell 1927, Bernays 1928) as scholars came to view wartime communication as an opportunity to better understand the diffusion of news and dynamics of public opinion.Some scholars, such as Lasswell (1927), saw the realities of mass societies almost forcing political leaders to resort toward the steering of public opinion and the manufacturing of consent (Lippmann 1922).Horkheimer andAdorno (1944/1997) viewed mass communication itself as having an 'isolating effect', enforcing the lonely and susceptible figure of an atomized modern individual.This atomized social fabric, they argued, functioned as a counterpart for authoritarian leaders to mount successful propaganda campaigns (see also Kracauer in Kang et al. 2022).
The legacy of these decades was to enforce a view of media audiences as 'masses', and scholarly interest in understanding the effects of messages on individuals.Subsequently, the overlapping imperatives of propaganda with the commercial interests of advertising would fuel a wave of empirical research, particularly after World War II.This work produced more nuanced readings and insights into the dynamics of media effects, ultimately leading to more modest views of the media's persuasive power over individual attitudes and behaviours (Klapper 1960).For our purposes here, one study stands out as particularly important in this regard: Personal Influence by Katz et al. (1955Katz et al. ( /2006)).
At the time, Katz and Lazarsfeld's findings fit neatly into their contemporary narratives about the moderate opinion power of media effects, which was later duly criticized (e.g.Gitlin 1978).However, a more lasting contribution to the study of media power was their move away from the imaginary of an 'atomized' mass audience of individuals to focus on social networks of media reception. 1 Katz andLazarsfeld (1955/2006) saw these networked relationships as the channels which mediate (and moderate) media effects; suggesting that the possession and transformation of opinions is fundamentally a social activity.In everyday interaction, people play differentiated roles in collective opinion formation, based on their specific areas of interest and expertise (public affairs, fashion, film, etc.).As such opinion leaders in different fields of interest, people -not just media -were crucial factors in information seeking, in directing attention to issues and trends, and thus in the construction of public opinions.
This idea of chains of influence mediated by personal networks obviously resonates with contemporary reflections on the power of social media; as do the critical arguments.Gitlin (1978), for example, draws attention to a point particularly pertinent to our analysis in arguing that Personal Influence focused empirically on an exceptionally stable and homogenous social milieu -a mid-western, suburban city in the United States.Thus, while Katz and Lazarsfeld may have found evidence of stable social networks acting as buffers against direct media power, these findings cannot be extended to the role of social networks in more volatile, ideologically polarized contexts.

The nature of digital social networks
Our current focus on 'networks' then is not new.Indeed, even Katz and Lazarsfeld argued for the 'rediscovery' of social networks in referring to the work of Tarde (1898/1989) as one early example of such analysis (see also Katz 2006).Yet the questions we raise about media power in contemporary social networks are distinctive, at least in four main ways.
First, social networks and digital platforms are now the infrastructure of media and political power (Peters 2015, Nielsen andGanter 2022).Legacy media, political institutions and their actors are now deeply embedded in these corporate-controlled social media networks (Chadwick 2017, Hallin et al. 2021).This same network infrastructure constitutes a key component of everyday lifeworld communication.If social networks in the mass media era were imagined as an 'intervening' factor in the process of mass communication, social networks today function materially as the 'channels' where audiences are exposed to mediated political communication and information (see e.g.Andi 2020, Newman 2023) Second, the dynamics of digital networks are different.Barabási and Bonabeau's (2003) theory of 'scale-free' networks captures this by arguing that their structure and emergence escapes material and contextual constraints (geography, physical infrastructure, institutional boundaries).This allows for connections between nodes to be made more widely, rapidly, and easily than before.Scale-free networks may present new opportunities for making connections and different forms of resistance (Benkler 2006, Castells 2009), since any node in a network is -theoretically -free to connect with any other node.Somewhat counter-intuitively, however, this leads to a 'natural' tendency to produce highly uneven attention hierarchies.When new nodes (actors) join a communication network, they tend to attach to 'hub nodes' with many preestablished connections in an existing network structure (Barabási and Bonabeau 2003, p. 63).This 'power-law' (Barabási and Bonabeau 2003, pp. 62-63) underscores the importance of existing power resources and structures which can tilt the field of attention and network centrality to their advantage.
Thirdly, digital networks are susceptible to political polarization.Partly, the strong clustering of social media networks into 'communities' relates to the power-law of scalefree networks.The 'logic of social media' (van Dijck and Poell 2015), with its grammar of shares and likes, can work to mobilize and accentuate 'real' world social homophily and strengthen communication between like-minded people.In a lively recent debate about 'filter-bubbles' or 'echo-chambers' (Pariser 2012, Sunstein 2018, Benkler et al. 2018, Bail 2021, Bruns 2021), some scholars focused on the polarizing effects of digital media infrastructure while others pointed to the political interests that strategically drive polarization.While complex casual factors evidently demand closer scrutiny (Brüggemann and Meyer 2023), what connects these explanations thus far is mutual recognition of the fragmentation and polarization of public attention in digital social networks.
Finally, contemporary digital social networks are materially mappable and traceable, heightening their susceptibility to surveillance and manipulation.For some critics, these developments illuminate the historically diminishing role played by human agency in the digital infrastructure where socially networked communication occurs today (Couldry and Meijas 2019, Wu 2017, Zuboff 2019).This centralized capacity of social media companies to harvest, analyze and sell user data is combined with the power to target and algorithmically manipulate the attention networks people are exposed to.These developments constitute a historically new power resource -data -which can redefine the very notion of 'public opinion', while eroding our capacity to distinguish between public and private life (Habermas 2022, Splichal 2022).Notably, this data-driven power to track and target specific audiences emerges simultaneously with new potential for socially and politically motivated relational power, which we investigate further below.

Studying climate change communication in digital social networks
As public discourses increasingly unfold online, a growing body of research has studied climate change communication on social media (Schäfer 2012, Tegelberg et al. 2014, Williams et al. 2015, Yagodin et al. 2017, Schäfer and Hase 2022).In a critical literature review, Pearce et al. (2018, p. 10) observe strong bias toward Twitter in 26 of 35 (74%) research articles published prior to 2018.They attribute this bias to the openness of Twitter's API -now demolished -relative to data sharing policies at other dominant social media platforms (Pearce et al. 2018, p. 10).We acknowledge that Twitter -today labelled X -has recently undergone radical and opaque changes.However, at the time of our data gathering, it provided a historically unique international platform for climate politics and thus a critical medium for studying the resources of attention power in socially mediated climate politics (e.g.Falkenberg et al. 2022, Gilardi et al. 2022).While all findings on specific social media platform data are partly shaped by the algorithms they run on, their networking affordances and user practices point to a shared 'social media logic'.Retweeting, that is paying attention to and sharing content in personalized networks, serves as one example of such practices.This logic and its relationship to power are what motives this research.
Others have critically assessed the potential and limitations of employing computational methods to study climate change communication on social media platforms (Schäfer and Hase 2022, pp. 3-4).Critics posit that bridging quantitative, data-driven methods with qualitatively rich interpretive analyses can yield greater insights into the diverse public discourses and climate imaginaries that emerge within and across different social media 'platform cultures' (Pearce et al. 2018, Schäfer andHase 2022).Similarly, after finding evidence that network analyses of climate activism largely concentrate on network structures among actors and organizations, Baran and Stoltenberg (2023) conclude that ' . . .discursive as well as social connection between different sites of environmental/climate content are needed' (Baran and Stoltenberg 2023, p. 466).
Recent large-scale social network analyses of Twitter activity provide convincing evidence of a striking rise in levels of political polarization around climate change (Falkenberg et al. 2022, Chavalarias et al. 2023).After years of stable climate discourse on Twitter between COP20 (2014) and COP25 (2019), Falkenberg et al. (2022Falkenberg et al. ( , p. 1117) identify a considerable spike in polarized political discourses during COP26 (2021).This is attributed to a growing presence of contrarian and right-wing political ideologues activated by the political momentum youth climate strikers gained beginning in 2019 (Falkenberg et al. 2022(Falkenberg et al. , p. 1116)).Similarly, in the wake of the COVID-19 pandemic, which drew public attention away from climate change, Chavalarias et al. (2023, p. 11) describe the global climate debate on Twitter as one that has become 'highly bipolarized' due to intensification in the activity of climate skeptics and denial groups.Over a threeyear study period, Chen et al. (2022) identified similar peaks in Twitter activity occurring around key political events, such as UN climate summits, when dominant framings of US climate movement actors congealed.Because attention is rarely sustained outside these staged political events (Wozniak et al. 2021), Chen et al. (2022, p. 405) conclude that interactions during these major events represent critical moments in the longer 'life cycles' of climate change discourses on Twitter.
These findings align with broader calls for researchers to develop more nuanced understandings of relationships between social media and political polarization (Barberá 2020, Brüggemann et al. 2020, Brüggemann and Meyer 2023) by extending analysis beyond 'echo-chambers' (Sunstein 2018) as the dominant causal factor in online polarization.Barberá, for one, offers evidence that just because social media users ' . . .self-select themselves into homogeneous communities does not necessarily imply that they are never exposed to dissonant political messages' (Barberá 2020, p. 38).Tyagi et al. (2020) demonstrate how this occurs in a network analysis of climate change discourses on Twitter where 'affective polarization' is what drives hostile but active interactions between opposing ideological groups (i.e.climate 'believers' and 'disbelievers').Jackson and Kreiss' ( 2023) longitudinal network analysis of climate change attention structures on Twitter, in turn, reveals how such antagonism between opposing ideological camps gets amplified during key triggering events in climate politics.In a study of Twitter activity during the 2019 Finnish election, Chen et al. (2021, p. 2) found strong evidence of a 'partisan sorting' phenomenon where 'positions on climate issues are strongly aligned with positions on political parties . . .[and] climate attitudes [that] have become strongly aligned with attitudes toward immigration'.This occurred in the Finnish context even though climate change had not previously been highly politicized.In a study of hyperlinking in German online climate networks, Kaiser and Puschmann (2017, p. 381) identified similar 'alliances of antagonism' evidenced by ' . . .structural connections between climate skeptics, conspiracy theorists, and other factions on the right political fringes'.The authors refer to these alliances as counterpublics united through shared antagonism toward mainstream political elites; and suggest that such alliances can transcend nationally specific political contexts (Kaiser and Puschmann 2017, p. 382).These research findings mirror wider polarization across Europe, where political parties with universal and communitarian values increasingly stand in opposition to right-wing populist parties.Evidently, antagonism between opposing ideological camps (Tyagi et al. 2020, Meyer et al. 2023), the network influence of 'highly visible partisan individuals' (Barberá 2020, p. 37, King et al. 2022, Brüggemann and Meyer 2023), and growing intersections between climate politics and other politically polarizing topics (Kaiser andPuschmann 2017, Chen et al. 2021), such as immigration and security, demand greater scrutiny from climate researchers.
We have now highlighted the current sense of systemic crisis and volatility of hegemonic structures, offered selective reflections of the dynamics between such crisis and theories of media power, and discussed some of the radically new potentials of digitalized social networks, particularly as these wider trends pertain to increasingly polarized online climate politics.With these resources at hand, we now turn attention to a concrete moment in global climate politics where we can study these relationships as they played out on Twitter.

Research design and questions
Building on this earlier research, our analysis contributes to the study of climate politics on Twitter in two main ways.First, we combine computational methods with critical qualitative analysis to empirically map and interpretively highlight distinct resources of political power mobilized by different actors in the social media sphere around COP26.Second, we concentrate on community structures and the retweeted content of social actors representing ideologically opposing camps within this social media climate sphere: climate contrarians and youth climate activists.Analysis of two diametrically opposed, counter-hegemonic groups reveals how both take advantage of the intersection between a sense of hegemonic crisis and available networking affordances on Twitter.
There are many ways to represent social media interactions analytically, which depend upon the functionalities of the platform under study.While Twitter data offers multiple ways to capture and represent relationships between actors (accounts) -including by mentions, replies, and retweets -it has become commonplace to construct maps of retweet networks.On Twitter, a retweet (RT) refers to any instance where a user selectively chooses to publicly recognize a specific tweet by sharing it among their followers.Past research demonstrates that such retweeting is generally considered to represent an act of supporting shared content (Barberá et al. 2015, Meyer et al. 2023).Through retweeting individual user accounts, the user actualizes the primary power resource of any social media actor: the ability to increase the potential attention that other actors gain within the network.Sharing, in this respect, is an act of symbolic recognition.As previously discussed, because such acts essentially shape the 'channels' of communication in the contemporary media landscape, retweets can be seen as the primary unit of attention power in these relationships.
As a methodological entry point, we mapped the broad landscape and contours of political attention power in the retweeting networks that surfaced during COP26.Several computational methods and social network analysis tools were used to visualize the COP26 retweeting network (see Figure 1). 2 During the climate summit, daily Twitter API requests were made between October 28 and November 12 for all tweets using the keywords 'climate crisis', 'climate change', 'COP26', and 'IPCC'. 3This produced a preliminary sample of 11,667,143 tweets.To facilitate analysis of relatively active retweeting networks, a sample of 8,921,103 million retweets was filtered to remove all user accounts with less than four daily retweets on average.This produced a final retweet network with 48,470 nodes and 77,824 edges.These nodes and edges were then spatially positioned and colourized, using community detection (de Meo et al. 2011) and layout algorithms (Jacomy et al. 2014).The network data was then exported to Gephi, an opensource tool for network visualisation.Degree centrality, which refers to the total number of direct links into and out of a specific node, helped us identify nodes with the most retweets in the COP26 Twitter network.The Louvain modularity community detection algorithm 4 assigned community memberships to each node based on the similarity of their action with other nodes in the network.Each Twitter user (i.e.node) in the data set is assigned to a community, with the nodes and edges colourized to visually represent community membership and structure (see Table 1).To make this visualization more legible to human readers, a customised ForceAtlas2 layout algorithm 5 positioned these nodes spatially to differentiate the relevant Twitter users for analysis.With relevant Twitter users and communities identified and visualised, the full dataset was subset based on the community memberships of the nodes to facilitate qualitative analysis.
By combining these tools, we have produced a mapping of attention power in climate politics that can be read simultaneously as an algorithmically derived network of 'clusters', representing the mathematical frequency of connections between nodes, and as a qualitatively produced visualization of human-perceived connections in this network that take the form of community structures.In discussing our results, we refer to 'clusters' as the calculated connections between nodes detected by the social network analysis software.When naming and describing 'communities', we refer to our interpretive analysis of relationships perceived between different political actors based on their positioning within the mapped clusters.That is, we bridge quantitative (clusters) and qualitative (communities) analysis of the COP26 data to better understand the power relations inherent within this material.
Our empirical analysis proceeds in four steps, each of which is informed by a specific research question.We interpret and discuss these in the following order: RQ1: What were the dominant structures of attention power on Twitter during COP26?RQ2: What resources of social and political power do the most dominant clusters rely on?
The first two steps lead us to identify two clusters of attention that are especially interesting for understanding the dynamics of power in the context we have sketched above.The following research questions, then, zoom in on the communities of 'relational' power which emerge from challenging the hegemonic order of climate politics.
RQ3: How are 'relational' attention communities structured and who are their key actors?
RQ4: What kind of messaging and rhetorical strategies dominate communication in the 'relational' communities?

Dominant structures of attention (RQ1)
Figure 1 provides an overview of the retweet attention economy of COP26 in Glasgow, Scotland (2021).We immediately notice that attention accumulates and concentrates heavily around several large nodes.These node sizes are relative to the amount a user is retweeted by other users within the same cluster (indegree centrality).The eight largest clusters take on one of two distinct community structures.The first are mono-nodal clusters where the community is structured around one massively retweeted node (actor), such as Greta Thunberg (@GretaThunberg), the UK COP26 Presidency (@COP26), US President Joe Biden (@POTUS), and Indian Prime Minister Narendra Modi (@narendramodi).Notably, the mono-nodal clusters differ in their shape and location in the overall map.The star-shape and relative isolation of the Modi cluster suggests loyal followers to a leader but a weak presence beyond that.Dense links from the official COP26 account to the periphery (relatively disconnected retweeters), as well as to the other central clusters points to wider influence, both in informing a general 'audience' and engaging with other powerful clusters.
Multi-nodal clusters have a structure where dominant influence is spread out across multiple user accounts.That is, the communication flow between users in multi-nodal clusters is less centralized than in mono-nodal clusters, with multiple users holding together the network rather than a singularly influential user.Again, there are different cases of these diffuse clusters.Near the center of the map, the blurred attention boundaries between science and NGO-actors point to their intensive interaction and investment in a field where no single actor has a dominant role (in the shadow of the massive attention to Thunberg).At the bottom of the map, a diffuse but large cluster hosts the climate contrarians, who have weak links to other clusters and seem to be led by multiple relatively large nodes.

Power resources (RQ2)
From the perspective of power, the attention economy mapped out in Figure 1 raises interesting questions regarding the distinctive bases of social and political power that different types of dominant actors rely upon.Of course, a retweet mapping cannot reveal the full complexity of this, but even a relatively descriptive analysis serves to complement the analysis of cluster structures by illuminating the differentiated power resources from which they draw.
Table 1 identifies the eight largest, coloured clusters in our COP26 retweet network map by name, nodal structure, power resource and retweet total.What we mean by power resource for the purpose of this network analysis is twofold.First, we found that some actors in the network translate conventional, institutional resources into attention power.But we also find less established and institutionalized actors, whose 'relational' power is based on both their ability to build attention communities within network itself and on their counter hegemonic posture vis-à-vis the established power players of the COP event.
Our findings reveal that institutionalised and established resources of power are highly significant, and that some actors can effectively translate accumulated influence into attention power in the COP26 retweet network.The respective power resources of these actors are not the same.Rather, they clearly reflect some inherited, institutional differentiations of political power.We see evidence, for example, of hegemonic (geo)political resources (Biden, Modi, etc.) and signs of bureaucratic or administrative 'event' power (e.g.COP26 official user account, UNFCCC, and the presence of local UK-Scottish hosts).Further, we see indications of the weight of scientific expertise, signs of an earlier, reputational symbolic authority (or capital), exemplified by ENGOs, such as Greenpeace and Avaaz, and traditional media power observed through the presence of large mainstream media nodes (e.g.Reuters, Sky News, NY Times).These media nodes logically serve as a bridging community between others.In the COP26 retweet network, we classify these as conventional, systemic power resources which are evidently historically constituted.The strong positions of these actors within the attention economy testify to the enduring power of such 'primary definers' (Hall et al. 1978, p. 58).This suggests a thread of continuity between the attention economy of mass media systems and the more hybrid, platformed attention dynamics of social media platforms.
A related observation reminds us of the two-sided nature of the relationship between attention and political power.Albeit a counterfactual point, one notable absence from  (Dahlberg 2023).Despite the intensive lobbying presence of corporate actors at COP summits (Chandrasekhar et al. 2022), the retweeting map shows limited representation of corporations, and fossil fuel lobbyists in particular.For these corporate actors, maintaining an inconspicuous presence within this online social climate sphere may represent, in itself, a capacity to wield conventional political power by avoiding critical attention.This relative invisibility in the social media event is of course much a part of managing attention as successful visibility.
On a more theoretical note, these reflections point to the specific value and challenge of social media analysis for theorizing contemporary political power.They suggest, on the one hand, that earlier analyses of differentiated power resources from historical sociology (e.g.Mann 2013) or social theory (e.g.Thomson 1995) still remain valid: resources accumulate and they matter.On the other hand, in the context of a political, globally mediated event that simultaneously takes 'place' in a networked communication infrastructure, such institutionalized and structurally accumulated resources are intertwined with other capacities of power.A case in point is that two of the largest attention clusters (Greta Thunberg and Contrarian) are not evidently linked to inherited power resources.These two clusters offer provisional examples of 'hub nodes' (Barabási and Bonabeau 2003, p. 67) within the socially mediated climate sphere since they capitalize on the accumulation of sharing activity within their respective communities.Rather than translating institutionalized resources into attention they construct attention capital through their oppositional relation to the constellation of established, legacy actors.We tentatively name these relational (context, situation specific) clusters, and examine them more carefully.In the following sections, we explore how key actors in these communities -who, in different ways articulate the 'crisis of hegemony' we sketched earlier -mobilize a relational network power resource that generates attention capital from the structuration of the social media platform itself. 6

Relational attention power: structures and dominant actors (RQ3)
Figure 2 visually juxtaposes our two relational attention power clusters.We refer to these as the Greta Thunberg and Contrarian communities.Analysis of their structure and most dominant actors reveals two crisis communities that display radically different attention structures, shedding greater light on the dynamics of media power in digital social networks.

Greta Thunberg
Retweeting activity around Greta Thunberg's central node forms a one-to-many community structure, where Greta stands alone as the primary influencer supported by a dedicated core of retweeting followers.Users in this community retweeted Greta Thunberg (@GretaThunberg) 120,596 times during COP26.This represented 20% of the total retweets (588,608 retweets) produced by all users in this cluster (and approximately 1.3% of total retweets in our full dataset).Thunberg's attention power is remarkable in itself, given that she only joined Twitter in 2018.In less than 4 years, she rose to yield more attention power in the COP26 context than any establishment political actor, including the US President (@POTUS) and the UNFCCC/COP-organization (@COP26), among many others.The exceptional strength of her attention power is well illustrated by the fact that a video clip of Thunberg's speech at a Glasgow demonstration was the most widely shared tweet by Reuters -a world leading news agency with global, institutionalized media attention power.This suggests that Thunberg is, practically speaking, a personalised global media channel.Further, the retweet-pattern around her extends not only to high-level political actors (see center of Figure 1) but to a widely dispersed following of individual user accounts (see periphery of Figure 1).This ability to gather such a wide following of 'ordinary' users can be viewed as a key factor in how Thunberg commands recognition from elite actors as well.

Contrarian
Despite its seemingly peripheral location (indicating fewer connections to the core of the retweet network in Figure 1), the Contrarian cluster represents a significant one, accounting for approximately 6% (526,863) of total retweets (8,921,103).Here, a more diffuse retweeting community structure includes several nodes that generate considerable attention and influence over communication and (dis)information flows (see Figure 2).For example, the largest node (@PeterSweden7) was retweeted 35,183 times (6% of total retweets).Similarly, the second (@JamesMelville) and third-largest node (@PrisonPlanet) were retweeted 30,082 times (5%) and 29,439 times (5%).Several other influential user accounts shared between 2% and 4% of the total retweets.We consider this a several-to-many community structure where multiple political actors are influencers.The communication flow between users in such communities is evidently less centralized than in the case of Greta Thunberg, with multiple dominant users holding together the community rather than one user with a large following.Table 3 shows that the largest node represents the Twitter account for Peter Imanuelsen (@PeterSweden7), a contrarian ideologue known for spreading far-right extremist content on social media platforms (King et al. 2022).Imanuelsen's past tweets have denied the Holocaust and directed hatred toward Muslim communities, and other minority groups (King et al. 2022, p. 64).A report on online climate disinformation and denial campaigns lists Imanuelsen among a small but highly influential group of 'climate misinformation super-spreaders' (King et al. 2022, p. 32; see also Chavalarias et al. 2023).With 35,183 retweets, Imanuelsen's node is the fifth largest in the whole COP26 retweet network.This attests to his capacity to generate relational attention power by functioning as a critical 'hub' node between climate politics and a larger right-wing ideological agenda.Some of this attention power is drawn from Imanuelsen's capacity to become an establishing 'channel' for right wing ideology on Twitter and other social media platforms.He uses the event of the UN climate summit to bring his views on the climate politics to the attention of his broader following.@PrisonPlanet represents another influential node within the Contrarian community.This account is run by Paul Joseph Watson, another high-profile online contrarian and conspiracy theorist with over 1 million followers on Twitter and a YouTube channel with over 500 million views.Watson is closely tied to other leading propagators of conspiracy and far-right extremism, including InfoWars owner and founder Alex Jones, British alt-right political commentator Milo Yiannopoulos (UKIP), and US-based white supremacist Paul Nehlen (DeSmog 2020).Watson was retweeted 29,439 times during COP26, representing 5% of the total retweets in this community.Table 3 shows several other top retweeted users that represent political actors on the UK far-right, including right-wing media personality James Melville (5%), and right-wing political leaders Martin Daubney (2%) and Nigel Farange (1.5%).Similarly, the retweeting data underscores the influence of specific media outlets known to traffic in junk news and misinformation, such as @disclosetv (5%) and @GBNEWS (4%).The only edge-lines that travel outside of this largely self-contained community originate from the node for GB News (@GBNEWS).Notably, only four of the top retweeted users in the Contrarian community traffic exclusively in climate denial and draw influence from prior connection to the climate denial movement.They are Michael Shellenberger (@ShellenbergerMD), Patrick Moore (@EcoSenseNow), whose bio describes him as former Greenpeace founder and now 'sensible environmentalist', Peter D. Clack (@PeterDClack) and Steve Milloy (@JunkScience), a Fox news contributor who claims to be 'Perhaps the most influential climate science contrarian'.By contrast, other influential users in the Contrarian community can be viewed as actors who take advantage of the attention structures that develop around the UN climate talks to spread a broader right-wing political agenda.Climate contrarianism is just one variant of the right-wing ideology advanced by the retweeting activity around these users.For example, we find users, such as @PrisonPlanet and @PeterSweden7, who leverage the attention around COP26 to spread hateful and extremist content relating to other social contexts, including migration, border security, and COVID-19 vaccination policies.
By not relying exclusively on the climate crisis as a topic of mobilization, in network terms, the Contrarian community reduces its dependency on any single account in the top retweeted user list.This suggests a certain durability of the network and calls for more research on how the same actors are connected in other event contexts (e.g.migration).Further, we observe how several dominant actors serve as bridging nodes to other platforms and blog accounts where similar and often more extreme ideological content is promoted.While from the perspective of climate politics, this community appears to be relatively 'closed' and isolated from powerful, institutional actors (and thus weak), this finding may be deceiving.Closer scrutiny of the most retweeted content produced by key actors in the Contrarian community reveals a strategy of communication that bridges different political issues and contexts which, in turn, may establish relational linkages with institutional power (parties, politicians, corporations, states) in other issue contexts.

Messaging and rhetoric in relational attention power (RQ4)
To develop a more nuanced understanding of the specific nature of these 'relational' attention communities (Greta Thunberg and Contrarian), we examined the most widely distributed content in these clusters.The highly concentrated nature of social media sharing informs our shift away from computational analysis towards interpretative inquiry.In both clusters, we coded and closely read the content in the top 50 retweets, which were all in the English-language.Although the total number of actual retweets was modest, their attention power in these communities was considerable.In Greta Thunberg's community, the 50 most shared tweets represented 37% (217,719) of total retweeting activity while in the Contrarian community the same figure was 31% (161,309).Bearing in mind that by focusing on viral content we may have overlooked interesting threads, analyzing this content produces insights on the symbolic resources that such 'relational' communities successfully draw upon in event contexts like COP26.
Table 4 provides an overview of our analysis.We identify two dominant categories of content: 1) criticism or shaming of elites and decision-makers and 2) calls for some type of action.In addition, we coded for two dominant rhetorical styles: 1) uses of irony and 2) exposure of hypocrisy.Each tweet was coded freely into two of these four categories.Then we compared salient themes and rhetorical styles in the top 50 tweets for both communities.We readily see how these relational communities are tied together by heavy emphasis on elite decision-makers.The attention they gather during COP26 constructs the political world through a division or cleavage between the 'people' (i.e. a community of followers) and the decision-makers of global climate politics.In this rudimentary sense, they are both 'populist', counter-hegemonic communities with strong moral messages about the way elites have abandoned their followers.In both communities, this binary world view and its 'critique' are often formulated by explicitly accusing elites of hypocrisy (saying one thing, acting differently) or through the use of irony.
Several popular retweets in each community direct attention to the massive carbon footprint generated by the luxury travel preferences of world leaders and corporate elites.In the Contrarian community, @PrisonPlanet (RT: 11515) tweets that 'wealthy technocrats arriving on private jets staying at luxury 5 star hotels and being chauffeured around in large entourages [will] . . .spend the next 2 weeks lecturing us about how we need to reduce our living standards'.In the Greta Thunberg community, one popular tweet (5440 retweets) makes an almost identical critique: '@NadiaWhittomeMP: A private jet produces 10 times the emissions per person as an economy class flight.World leaders used 400 private jets for #COP26, emitting as much CO2 as 1,600 Scottish people do a year.Lifestyle changes should start with the rich and powerful'.Nadia Whittome is a member of the UK Labour Party and the Member of Parliament for Nottingham East.She refers to herself as a Democratic Socialist with politics on the opposite end of the ideological spectrum to @PrisonPlanet.Irony and sarcastic exposure of double-standards are rhetorical modes that build on the relational demarcation lines between community members and political power brokers.That is, they focus their communicative effort on creating and sustaining community engagement (as a source of relational power) by distancing themselves from elite actors in global climate politics.These similarities are important for theorizing the nature and dynamics of political power in the current era where new social media attention economies intersect with a sense of systemic or hegemonic crisis.But they should not blind us to key differences between the popular content that circulates in these communities.Our analysis reveals differences that are indeed quite radical: while both communities are anchored in moral critiques and blaming, this elite shaming leads to very different calls for (in)action.Among Contrarian influencers, the dominant message is to ignore the global elite hypocrites in power.For Greta and her followers, by contrast, the shameful inaction of the same elites is often explicitly connected to demands for collective action (by followers) to exert pressure on them to change.In the Contrarian community, none of the 50 most widely shared tweets call upon followers to act.In the Greta Thunberg community, three out of the four most popular tweets make explicit calls for climate action.Popular Greta tweets suggest alliances and connections between this community and other influential nodes in the COP26 retweet network.As do written words, slogans and multimedia content, including several tweets that hyperlink to reporting on Greta's activities by legacy news media, such as Reuters and The Guardian.Other popular tweets contain embedded video links to prominent messaging from the UNFCCC or to raw footage of climate justice activists marching through the streets of Glasgow.Finally, several popular retweets provide Thunberg's followers with prescriptions on how to act.One popular retweet (6963) invites followers to 'join the climate strike' on Saturday, November 2. Another links to an Avaaz.orgpetition, crafted by Thunberg and three other prominent youth climate activists, that makes an emergency appeal for climate action.By the end of the summit, the petition had collected over 1.85 million signatures.
In both relational communities, the most popular tweets present followers with different visions for the outcomes of COP26.Greta Thunberg's tweets call for collective, civil society action to pressure world leaders into action.These tweets often pit science and advocacy communities against national political actors who have failed to heed their demands for rapid climate action.In the Contrarian community, by contrast, the most popular tweets combine irony with apathy to direct attention away from the climate crisis toward a far more sinister, radical extremist agenda.Peter Imanuelsen's (@PeterSweden7) most popular tweet downplays the threat of climate change by positioning an ongoing erosion of civil liberties as a greater threat.As he puts it (to over 280,000 followers): 'I'm not worried about climate change or global warming.I'm worried about the global tyranny taking over the world'.

Discussion and conclusions
We began this article by sketching a systemic, global crisis of hegemony driven by the interplay of hard material conditions of sustainability, rapid changes to social and communicative infrastructures, and their local political consequences.Our understanding of this crisis is twofold.On one hand, while this systemic crisis is ultimately humanmade, it is underpinned by an external, hard fact of anthropogenic climate change that demands radical transformations to societies and ways of life.On the other hand, this is a hegemonic crisis that is constructed by particular forms of communication that unfold in historically new and rapidly changing media platforms.Such complexity demands a multidimensional reconsideration of the relationship between media and political power.
After developing a historical and theoretical argument about how a crisis of systemic, hegemonic power amplifies concerns about media power, we sketched the contours of the contemporary era of crisis.We showed how our inherited ways of conceptualizing media audiences still resonate in our current conjuncture but also demonstrated how they must be redefined.These reflections were operationalized through an analysis of social media interactions during a global event in climate politics.Two clusters of social media interaction that we classified as 'crisis communities' were situated within the larger field of power surrounding digitally mediated climate politics during COP26 in Glasgow (2021).We conclude this article by working our way back from this empirical evidence toward larger questions of power, crisis and (social) media.
Our analysis located, analyzed and compared two ideologically antagonistic crisis communities in the COP26 Twittersphere.Despite diametrically opposed political positions on climate policy, they share some common features (see Table 5).Their attention power is communicational in the sense that they both do not lean on any obvious, previously accumulated power resource.They both attack global elites, the key hegemonic, systemic actors of this event.They also partly favor similar kinds of rhetorical strategies (irony, exposure of hypocritical posing and PR).These similarities, we suggest, show how the crisis of established actors provides openings for relational attention power.Key actors in these communities take advantage of a major climate event -and the public exposure of hegemonic actors -to build their collective attention power through dispersed individual acts of retweeting.Their significant presence within the dataset (about 12% of total retweets) points to the radical potential of networked media infrastructures to gather together communities that need not rely on any preestablished institutional or systemic accumulation of political power (national political power, scientific authority, etc.).This generation of power, however, is not merely a consequence of the affordances of the new media landscape.Rather we argue that this power is fundamentally relational, as it depends on the apparent inability of hegemonic actors to perform with authority in their attempts to solve the crisis.The power of these social media communities to engage their followers and to generate attention and recognition ultimately flows from the disrupted lifeworld horizons of their followers.An essential part of creating such relational power is to produce a heightened sense of violated common sense.With elites as shared targets of blame, and a shared sense of being located outside the domain of political (decision-making) power, it is not surprising that both communities show fondness to ironic, sarcastic messaging (see also, Falkenberg et al. 2022Falkenberg et al. , pp. 1115Falkenberg et al. -1116)).These similarities do not negate key differences between these two large clusters of retweeting activity (see Table 5).In terms of community structure, Greta Thunberg's mono-nodal community enjoyed remarkable network power within COP26.Despite its stinging critique of elites, however, Thunberg's presence connects different types of elite actors (science, politics, mainstream media, ENGOs) at the political core while, at the same time, drawing a together a large, diffuse group of movement followers (that bring a representational, political flavor to her power).In contrast, the Contrarian community relies on a more widely spread group of leading actors that each draw attention from smaller groups of retweeting actors.This community is far more excluded from established political actors, with only a few weak links to popular right-wing news outlets.While Thunberg's engagement with differentiated forms of power potentially increases its political significance, her dominance also means a centralised reliance upon one node, which can be viewed as a structural weakness.The Contrarian community illustrates a more evenly distributed network with a network structure that is arguably more sustainable and resilient, albeit one that comes with the cost of not assuming the same commanding presence in the context of the event.These differences are crucial for thinking through the substantial political nature of these relational, seemingly counterhegemonic publics.Indeed, following Jackson and Kreiss (2023), we could label the Contrarian community as a more 'defensive' public that -despite its superficial antagonism toward elites -is ultimately conservative in relation to existing power hierarchies and ways of life: these actors do not want the climate crisis to be taken politically seriously.This suggests retaining the concept of 'counter' publics for the Greta Thunberg community because of its radical calls for systemic change.
These observations concerning the construction of different kinds of relational, communicative power illustrate how contemporary relationships between media and power partly escape earlier binary conceptualizations of isolated masses in contrast to socially embedded citizens.While digital social networks still function as channels of influence, the very same networks can serve as sites where power and influence can be created, constructed and wielded when political events allow for it.Indeed, nodes like Thunberg are perhaps best understood as momentary, pop-up personalized mass media outlets rather than everyday 'opinion leaders' à la Katz and Lazarsfeld.It is also important to be mindful that our empirical analysis focuses on the nature and amount of attention power in a given, specific moment.We thus cannot make claims concerning how such relational attention power might later be translated into, for instance, political power (e.g. through electoral success) or to expert or moral authority in other moments and contexts.
As events that embody an era of crisis, global climate summits call for elite political actors to perform to and redeem their hegemonic positions.In the context of global, scale-free attention networks, this is a tall order (Alexander 2011).We have sought to demonstrate and theorize on how relational aspects of power materialized in the two counter-hegemonic crisis communities we investigated.While our analysis only scratches the surface of these new dynamics, we hope to have shown that simultaneous analysis of networked 'communication flows' and 'shared stories and collective concerns' can indeed help us to make sense of the power dynamics that underlie relational publics (see Starr 2021, p. 19).Our analysis also lends credence to arguments concerning the strong role of power as an asymmetrically accumulated resource: the 'social media sphere' around COP26 was powerfully saturated by the presence of 'usual suspects'.Namely, the politicians, experts, legacy news media outlets, and recognized celebrity activists who successfully mobilized and translated their existing power resources to gain attention in the social media sphere.In sum, the sense of an emerging and deepening crisis not only informs counter-hegemonic critiques of the inherited interaction patterns of powerful systemic actors but also mobilizes these same powerful stakeholders to take advantage of the dynamics of social networks.By returning to parallels and differences between historical and present moments of hegemonic crisis, we hope that our analysis of relational attention power in climate politics illuminates some of the drawbacks and potential that a rapidly changing socially networked media environment poses to the broader task of guiding the systemic transformation that the climate crisis so urgently demands.

Figure 2 .
Figure 2. Close-up of Greta Thunberg and Contrarian communities.

Table 4 .
Salient themes and styles in top retweeted content.

Table 5 .
Elements of relational political power in two crisis communities.