Theory Research Paper #1
Journal of Communication ISSN 0021-9916
ORIGINAL ARTICLE
The Strength of Peripheral Networks: Negotiating Attention and Meaning in Complex Media Ecologies W. Lance Bennett1, Alexandra Segerberg2, & Yunkang Yang1
1 Department of Communication, University of Washington, Seattle, WA 98195, USA 2 Department of Political Science, Stockholm University, Stockholm, SE-106 91, Sweden
Networked content flows that focus or fragment public attention are key communica- tion processes in multimedia ecologies. Understandings of events may differ widely, as networked attention and framing processes move from core participants to more distant spectator publics. In the case of the Occupy Wall Street protests, peripheral social media networks of public figures and media organizations focused public attention on eco- nomic inequality. Although inequality was among many issues discussed by the acti- vists, it was far less central to the protest core than problems with banks or democracy. Results showed how public attention to inequality was constructed through pulling and pushing interpretive frames between the core and periphery of dense communication networks. Various indicators of public attention—such as search trends, Wikipedia arti- cle edits, and legacy media coverage—all credited the protests with raising public aware- ness of inequality, even as attention to problems with banks grew at the protest core.
Keywords: Attention Economy, Social Movement Communication, Media Ecology, Networked Framing, Hybrid Media.
doi:10.1093/joc/jqy032
The subprime mortgage market in the United States began to collapse in 2007, and by 2008, the Lehman Brothers investment bank became insolvent, producing a cas- cading banking crisis. Bank failures, home mortgage defaults, and loss of liquidity in dubious mortgage-backed investment products soon spread to Europe, where more bank failures and impending national defaults led to a sovereign debt crisis that threatened the Euro currency. The resulting global economic downturn was com- monly called the Great Recession, reflecting its status as the worst financial crisis since the Great Depression of the 1930s. Citizen anger also swept the world, taking different forms, from the Tea Party in the United States to ethnic nationalist move- ments and parties on the right in Europe, and vast mobilizations, such as the
Corresponding author: Lance Bennett; e-mail: lbennett@uw.edu
659Journal of Communication 68 (2018) 659–684 © The Author(s) 2018. Published by Oxford University Press on behalf of International Communication Association. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
Spanish M-15 (“indignados”) and Occupy Wall Street (OWS). While some of those citizen initiatives became enduring movements and parties, the legacy of OWS remains more enigmatic.
This article examines the legacy of the Occupy protests as recorded in social media traces and other indicators of public attention and participation in the fram- ing process in the United States, where OWS began and attained the largest scale. The empirical questions are variations on classic queries about social and political communication surrounding disruptive events such as protests: what were the pre- dominant societal understandings of these social mobilizations and how closely did those public understandings correspond to the primary messages projected by the protesters themselves? These questions take on new theoretical importance in con- temporary media ecologies, which are less dominated than in the past by official framing of events that cued mass audiences through traditional journalistic gate- keeping in highly-institutionalized media systems. Multimedia ecologies are charac- terized by more porous information flows across social and legacy media, offering publics opportunities for selective attention and active content production.
The article makes three general theoretical contributions, based on modelling the interactions among different layers of Occupy protest communications and peripheral social attention. First, the study shows that establishing the popular sig- nificance of events such as OWS in complex media ecologies involves the iterated negotiation of attention and meaning across many types of media and content sources. The push and pull of content through differently-located networks is at the core of networked gatekeeping and framing processes. Far from consisting of minimally-involved “clicktivists,” such peripheral networks can play a significant role in defining and responding to social events. A second, and related, contribution is showing how these networked attention and framing processes can have a variety of outcomes affecting the public reach of a movement’s messages and concerns, including amplifying, burying, or distorting them. This sets up our third general point: as public attention flows become shaped by participatory media embedded in hybrid systems (Chadwick, 2013), standard communication constructs such as media effects, agenda setting, and framing become challenging to define and mea- sure (Bennett & Pfetsch, 2018; Neuman, Guggenheim, Mo Jang, & Bae, 2014). We show how introducing various types of attention measures based on digital trace data can help sort out some of these challenges.
The puzzling legacy of Occupy Wall Street
The Occupy protests began in New York on September 17, 2011, and soon swept through cities across the country and around the world. Crowd-sourced estimates of events recorded demonstrations in 951 cities, located across 82 countries, includ- ing some 600 locations in the United States (Wikipedia, 2017a). Tens of thousands camped in public spaces and took part in protest actions, and millions more fol- lowed, interacted with, and cheered them online. Many added to the evolving
660 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
protest discourse online, for example by posting their own stories to the “We are the 99%” Tumblr blog, which launched the popular slogan that was coined for the movement. Like many movements in the digital age, OWS used an impressive array of social media and digital platforms to reach and include large audiences. As with other such protests around the world, the growth of large online support networks on the periphery made OWS highly visible and difficult for authorities to manage or ignore (Barberá et al., 2015). In addition, mediated interaction processes between direct and online participants accomplished various kinds of organizational work, such as publicizing legal services for arrested protesters, offering food or shelter, coordinating protest actions, and reporting on events. High volumes of media con- tent, from activist videos to legacy news reports, flowed in and out of these multilay- ered networks (Bennett, Segerberg, & Walker, 2014).
Like many observers, we were initially persuaded by the idea that the legacy of OWS involved sparking attention to problems of inequality in society. Indeed, as our data show, that was the main legacy of the protests with respect to public under- standing. A consensus emerged fairly early in online media, legacy media, and other pulse points of societal attention that the protests deserved credit for establishing inequality on the public agenda. However, our data also show that inequality was far from the main concern at the core of the protests. Inequality was certainly one concern raised by the protesters, but among the activist core, it was vastly out- weighed by other issues, such as fixing banks, the economy, and democracy. It is not surprising that the dominant issues in social media exchanges involving the most mentioned and followed group protest accounts (such as @occupyla or @occupywallst) were mainly focused on banks and the banking crisis. After all, the movement named itself Occupy Wall Street, and the first camp was in the heart of the Manhattan financial district. Even our examination of social media communica- tion surrounding the broad “We are the 99%” meme (see below) showed that the theme of inequality was similarly outweighed by other issues there, as well.
This suggests a puzzle about Occupy Wall Street and its public legacy. The pro- tests succeeded in gaining positive societal attention, and influenced the public agenda. Yet various indicators of public attention, along with framing in the legacy media, focused on an issue that was not the most important at the protest core. It appears that, as growing public attention focused on interpreting the meaning of the protests, the inequality frame captured the greatest social attention, and this shaped longer-term public understanding. Our point is not that inequality was of no concern to many activists, or that the inequality frame was simply imposed on the protests by legacy media or other actors. Rather, it appears that something more intriguing took place: complex networked processes of negotiating and focusing societal attention shaped the broadly-enduring public meaning of the movement.
The puzzle of the Occupy legacy, therefore, revitalizes a classic set of questions to do with communication between movements and broader society. In the mass media age, impressions of protest movements formed by spectator publics were often cued by officials in mainstream news reports that typically framed protests in
661Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
negative terms (Benford & Snow, 2000; Gitlin, 1980). In the multimedia age, as Tufekci (2013) has argued, the capacity of movements to shape public attention has changed with access to social media by citizens. Activists have, in some ways, gained more control over their public messages, yet the processes shaping societal under- standing of events are complex and require better understanding.
We suggest that the processes of negotiating attention and meaning are inter- twined, and that social media networks interact with legacy media, and other plat- forms, in ways that depart from earlier mass media research. The interpretive puzzle of “what were the OWS protests about?” points to a key underlying dynamic: how things appear to some observers in multimedia communication ecologies is the result of densely-networked framing processes that may look quite different from other network positions. The construction of attention and meaning between the core and periphery in technology-enabled crowds involves an iterative mix of selec- tive attention, network interactions, narrative construction, and selective uptake by legacy media. The resulting narratives are networked productions that blur the lines between conventional journalism and citizen reporting (Papacharissi & de Fatima Oliveira, 2012; Russell, 2016), and between social and legacy media. In all, this dynamic interrelation of attention and meaning can combine to amplify, bury, or distort the key protest message. Even in relatively supportive public environments, networked information flows can translate and transform key ideas as they move from the most committed direct participants to the most distant spectator publics, who pay attention and contribute through a combination of social and legacy media.
So how did densely-networked information flows affect public attention and understanding of the Occupy movement? Answering this question involves tracking information moving through complex media networks, as variously-positioned actors and their networks pushed, pulled, and shared content through digital intermediaries. In this case, we argue, selected movement messages were both amplified and transformed in the networked flows. Among the many networks shaping attention surrounding OWS and other events were prominent public figures with large followings (celebrities, writers, film makers, politicians). These peripheral actors fed their support into—and pulled meanings out of—the protests, while sharing their interpretations with large social followings. Journalists from both alternative and mainstream news organizations drove the inequality message to even larger publics in a second wave of framing that followed the attention curve shaped by public figures. This process of selectively translating the concerns of the protesters into terms that drew the attention of more distant and successively larger media networks, produced the disparities that we see between the priorities of the core protesters and the narratives that captured the attention of much of society. The next section presents the theoretical model for understanding and operationa- lizing the attention and framing processes underpinning this analysis. Later sections test the evidence for these propositions, along with offering empirical demonstra- tions of how these networked attention and framing processes worked.
662 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
Networked framing, attention, and the role of peripheral networks
Processes that operated in simpler ways in the mass media age have become far more complicated, and in some cases displaced, in the multimedia age. In line with this, changes to legacy communication concepts have been recommended for agenda-setting (Neuman et al., 2014), gatekeeping (Barzilai-Nahon, 2008), and framing (Meraz & Papacharissi, 2013; Tufekci, 2013), along with rethinking theoret- ical frameworks pertaining to changing media spheres (Bennett & Pfetsch, 2018). Since information flows around contemporary protests often span several overlap- ping layers of publics, complex media ecologies engage multiple and differently- interacting audiences. To examine the relationship between movements and societal understandings under these conditions, a crucial question is how information flows across differently-positioned social networks shape broader public attention and related understandings of events.
It is clear that networked gatekeeping and framing processes (Meraz & Papacharissi, 2013) feature centrally in this context. The push and pull of informa- tion across socially-networked crowds selectively pushes to prominence both con- tent and specific contributors. In the iterated process, it often becomes difficult to separate the roles of different types of media (e.g., legacy vs. social) and content for- mats (e.g., journalism vs. activist reports; Barzilai-Nahon, 2008; Papacharissi, 2015; Tufekci, 2013). It is also difficult to neatly identify system boundaries or hierarchies of functions, as ordinary citizens, journalists, elites, and other actors all become entwined in the construction and distribution of both attention and meaning (Meraz & Papacharissi, 2013; Russell, 2016).
The role of core and periphery in the intersecting networks that become involved in networked framing deserves particular attention. Earlier work has shown that actors and networks on the periphery of technology-supported crowds (e.g., engaged online supporters, critical commentators, bots, trolls) may greatly affect the scale and impact of protests and protest communication visibility (Barberá et al., 2015; Steinert-Threlkeld, 2017; Zhang, Wells, Wang, & Rohe, 2017; see also Diani, 2003). These are actors that may be peripheral to events, in terms of not being close to the protest (e.g., not directly involved in organization of, partici- pation in, or concern with the direct action) and not having close network ties to the centralized and well-connected protest core. The strength of such peripheral actors in expanding the reach and impact of protest communication has been traced to the volume of their aggregated network followers and connections, but also in the diversity or distinctness of their connections as compared to the core. Following this, peripheral actors might be expected to be significant not only for spreading information and inspiring participation, but also shaping meaning in the iterated processes of pushing and pulling content.
Actors who are peripheral to core protest events but powerful or central in their own circles may play a particularly important role in networked framing processes. In particular, people or organizations with large social media followings of their
663Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
own have been shown to be critical to focusing broader public attention at the blurred interface beween social and legacy media. Alternative and mainstream jour- nalists can be active in this dynamic. But so can other actors, such as public figures, as illustrated in the case of the death of Trayvon Martin, a young Black man shot by a neighbourhood vigilante in a Florida suburb (Freelon, McIlwain, & Clark, 2018; Graef, Stempeck, & Zuckerman, 2014). The news stories about the shooting were fading when activists directed a change.org petition to celebrities with large social media followings. The story spiked across the web, and was pushed back into the mainstream media, which focused continuing public attention on the shootings of young Black men in America. In this case, the peripheral networking process helped to amplify concerns at the core of the protests, but it is conceivable that it could work differently in other cases, burying or distorting the protest message.
Drawing from these insights about networked information flows, we focused on how differently-positioned networks, from the center to the periphery of OWS, pushed and pulled content in a networked framing process. Such framing processes involved activist groups managing social media accounts, public figures with large followings, and journalists who interacted with various media feeds and shaped leg- acy media framing that pushed particular meanings to even broader publics. These iterative processes conditioned the communication between movement and society, and shaped public attention and meaning.
The article therefore addresses three research questions. The first establishes the empirical grounds for the networked framing puzzle outlined above, and the second two examine the underlying processes.
RQ1: What were the predominant societal understandings of OWS, and did they differ from the concerns expressed in the public communication at the core of the protests? RQ2: How did networks at the core and periphery interact to shape the attention of different audiences to OWS? RQ3: What is the relationship between social and legacy media in these network framing processes?
We began our analysis by establishing that societal attention became focused on inequality as the legacy of OWS, and that this differed from the issue attention closer to the core of the protests. We then moved forward to examine the role of peripheral networks in the processes of networked gatekeeping and framing, and in mediating the complex relationship between social and legacy media.
Conceptualizing and measuring social attention to protest themes
There are many indicators of public attention in multimedia environments. In establishing that OWS became widely associated with popular attention to inequal- ity, we looked at three of these indicators: the succession of crowd-sourced edits in Wikipedia articles about OWS; Google search trends before, during, and after the
664 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
protests; and the association of OWS with inequality in legacy press accounts. These indicators offered different ways of observing public attention: news frames pro- vided indirect measures of public attention, cued by legacy media content, while Wikipedia edits and Google searches offered more direct measures of particular types of content that publics actively created, engaged with, and sought out. We began with the more direct measures.
Within a few weeks of the beginning of the protests, and continuing to the time of this writing some 6 years later, the Wikipedia article on Occupy Wall Street opened with minor variations on this statement: “Occupy Wall Street (OWS) was a protest movement that began on September 17, 2011 in Zucotti Park, located in New York City’s financial district, receiving global attention and spawning the movement against economic inequality worldwide” (Wikipedia, 2017b). An inspec- tion of more than 8,000 edits on this article showed that the framing followed the arc of our analysis of Twitter trends reported below: attention in the first two weeks focused mainly on the financial crisis, banks, and corruption, but inequality soon took over as the primary frame. For example, the version of the Wikipedia article published on September 30, 2011, led with the observation that the protesters opposed “negative corporate influence over U.S. politics and a lack of legal repercus- sions over the global financial crisis” (Wikipedia, 2011a). Within a few days, by October 3, inequality entered the opening list of protest causes, and by the end of that day, a flurry of edits moved inequality to the top of the list, saying that partici- pants “are mainly protesting against social and economic inequality and corporate greed, among other concerns” (Wikipedia, 2011b). Other issues changed or were added in the coming months, but inequality remained the first in the list. By the time that edits tailed off in June 2012, inequality was given a section of its own, and other issues were included in a subsequent paragraph titled “goals.” This framing persisted to the time of this writing, and inequality remained in a primary place in the opening paragraph.1 As the protests unfolded, links were added to other articles on wealth inequality and income inequality in the United States. By early 2013, the Occupy Wall Street article cited various sources, including academics and media monitoring organizations, affirming that OWS had brought America’s attention to problems of inequality (Wikipedia, 2013).
Another interesting indicator that OWS triggered public attention to inequality entails comparative Google search trends. Figure 1 shows relative search trends for several topics relevant to the financial crisis, banks, and inequality in the United States between 2004 and 2017. The measures indicate search interest in a topic, nor- malized on a 100-point scale, relative to total searches for all other topics at each point in time for a given region (in this case, the United States). While attention to inequality was clearly evident before the financial crisis and the Occupy protests, it surged at the time of the protests and has continued to climb since then, even as rel- ative search interest in the financial crisis and bank bailouts declined over the same period. We attributed the post-OWS attention trends to a combination of increased legacy media coverage and public policy debates (some led by President Obama
665Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
during 2012) about inequality-related issues, such as the minimum wage. Searches on the financial crisis peaked after the crisis dominated the headlines in 2008, and tailed off steadily thereafter, with little impact from OWS. Similarly, search interest in bank bailouts (the most common search involving variations on the term “bank”) also peaked at the start of the financial crisis years earlier, and tailed off during and after the OWS period. These search trends did not reflect the the prominence of protest themes about bailing out banks at public expense, or a failure to punish representatives of those institutions for causing the crisis.
It may seem obvious why OWS did not become credited with drawing public attention to the banking and financial crisis. After all, that crisis had already attracted substantial public attention, as indicated in Figure 1. However, given pro- test anger over lack of punishment for bankers, and other criticisms of how the gov- ernment handled the crisis, it is conceivable that OWS could have triggered a resurgence in public attention. It is also conceivable that the protests could have reframed the crisis around protest themes of harsh penalties, stronger bank regula- tion, or corporate money corrupting democracy. Indeed, those were top issues named by many at the core of the protests, who occupied public spaces and marched against financial corruption (Bennett & Segerberg, 2013).
As an indirect measure of public attention, legacy media also cued attention to inequality and linked it to OWS. Our searches of Lexis-Nexis news archives in the United States before, during, and after the protest period showed that stories about inequality spiked during the protests, and that far greater percentages of those stor- ies mentioned OWS than did stories on banks and the financial crisis.2 News items with headlines or lead paragraphs mentioning the topic of inequality increased
Figure 1 Google search trends in the United States for inequality, financial crisis, and bank bailout, 2004 to 2017. Protest period bolded. Source credit: Google Trends.
666 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
significantly in the last quarter of 2011 as compared with the same period in 2010 (+376%), and remained substantially higher in 2012 than in 2010 (+183%). This shift in attention to inequality was broadly associated with the Occupy protests. References to OWS appeared in more than half (52%) of the headlines and lead news and editorial items on inequality in the last quarter of 2011, compared with OWS references appearing in just 6% of bank stories. Since the banking and finan- cial crises had been huge news stories for many years, and the economic disruption continued at the time of the protests, the total volume of news about banks and the financial crisis remained larger than reports on inequality during the protest period. However, press attention to bank issues was steadily declining, while attention to inequality was spiking. For example, our news searches yielded 3,007 stories in 2010 on the banking crisis, with volumes declining to 2,497 in 2011 and 2,124 in 2012. Stories on inequality rose from 200 in 2010 to 522 in 2011 and 566 in 2012. A com- parison between 2010 and 2012 revealed that headline/lead mentions of bank cor- ruption or reform decreased by 29%, while headline/lead mentions of inequality jumped by 183%. In addition, the press prominently linked OWS to the inequality narrative nine times more often than to bank stories.
Whereas a standard mass media framing analysis might simply count the num- ber of stories during the protest period, we focused on attention dynamics by look- ing at story volume in the context of trends over time and interpretive cues embedded in that coverage. This attention-based approach showed that the press focus on banks was fading, while the uptick in inequality stories was sharp, sus- tained, and associated explicitly with OWS. All of this suggests that the media atten- tion economy may have been driven less by the sheer volume of stories, than with the sudden upshift in volume and with changes in the cues that engaged attention.
A look at the qualitative news data illustrated these points even more clearly. Beyond shifting attention to inequality, the dominant narratives in mainstream news coverage explicitly credited OWS for putting the issue on the public agenda. Following roughly the same time sequence of the Wikipedia edits discussed above, the legacy press narrative associating OWS with inequality emerged after the second week of the protests, and became fully established as the main news frame by late October/early November, as shown in Figure 3D below. For example, Nicholas Kristoff’s (2011) New York Times column of October 15 headlined Occupy as “America’s Primal Scream,” a line taken from former Vice President Al Gore’s char- acterization of the protests as a “primal scream of democracy.” However, in his ver- sion, Kristoff claimed that interviews with protesters in Zuccotti Park revealed the primal scream to be mainly about economic inequality (Kristoff, 2011). Kristoff shared the article with his Twitter following of over 1.3 million, from where it circu- lated back into protest networks and was pulled out again by different social networks.
As the iterative push and pull of interpretive frames among peripheral networks progressed, many other mainstream press outlets, such as CNN, MSNBC, and the Wall Street Journal, among others, issued similar proclamations. Online magazines,
667Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
political blogs, and commentators added to an echo chamber on how OWS was shaping the public debate about inequality in the media and politics. Politico pub- lished a much-cited quantitative analysis on the surge of media attention to inequal- ity (Byers, 2011). Soon, leading business publications, such as Forbes and Business Week, credited Occupy with the national concern about inequality, with headlines like “Occupy Protesters Inject Income Inequality into Political Debate” (Deprez & Dodge, 2011) and “Occupy Wall Street: Income Inequality And The Burden Of Action” (Fontevecchia, 2011).
Following the early framing by leading news organizations, and the subsequent echo chamber of blogs, alternative media, and business news, the OWS inequality story developed a life of its own that dominated the media agenda (see also Gaby & Caren, 2016). End-of-the-year news retrospectives credited OWS with focusing attention on the issue of inequality. Wikipedia articles and legacy media reflections on the lasting effects of the movement on pushing inequality onto the public agenda continue to list high in Google search returns at the time of this writing in 2018. These various indicators show that public attention became focused on the link between OWS and inequality. However, the data we present in the following sec- tions show that this was not the main concern in communication emanating from the core of the protests.
Conceptualizing and measuring issue attention at the core and periphery
OWS was credited with drawing social and political attention to inequality, yet communication closer to the core of the protests was more focused on other issues. Protest records issued from city camps and recorded in online wikis emphasized that the movement’s consensus was to endorse no single issue or common demand. Indeed, assembly meetings and crowd-sourced wikis compiled and published many long lists of social, political, and economic problems (Bennett & Segerberg, 2013, p. 187). However, some of those issues were given more attention by the protesters than others.
Our analysis indicated that, among those that were closely attentive to the core of protest action in terms of organization, participation in, or concern with direct action, far greater attention was directed beyond inequality to issues such as banks and bankers, the corruption of democracy, and a host of problems with the econ- omy. Of particular interest was how core and peripheral networks interacted to shape the attention of different audiences to OWS. We also showed how these net- work dynamics shaped the relationship between social and legacy media in the net- work framing process. The nature of these questions made Twitter data a prime focus for analysis.
Why Twitter?
Focusing on Twitter was analytically useful in this case, because it offered a public window on communication by and among activists, as well as a spectrum of
668 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
communication about the protests that occurred in identifiable spectator networks, including media organizations. Our methods of gathering and structuring the data made it possible to study the sequences of interactions among different networks as they intersected through mentions, hashtags, and retweets on Twitter and beyond.
Far from being just an available data source among many others, Twitter was the most central communication network linking core activists, peripheral suppor- ters, media organizations, and spectator publics. While a variety of platforms may have been more central in other mobilizations, Twitter played an important role in OWS. Twitter officially claimed that there were more than 100,000 Occupy-related hashtags (many of them playful, such as #occupybritney or #occupychipotle). Protesters, themselves, verified nearly 900 protest accounts, with over 11 million fol- lowers (Bennett & Segerberg, 2013, pp. 181–183). Our previous work on OWS showed that among the dozens of social media platforms involved in the communi- cative organization of the protests, Twitter operated as the most commonly used integrative mechanism, enabling activists to communicate across many scattered protest sites, reach broader publics, and target information and attention to specific, localized events and activist sites (Bennett & Segerberg, 2013; Bennett, et al., 2014). Repeated network crawls of digital media sites, starting variously from local Facebook sites, city and national websites, and coordinating sites for particular pro- test events, all showed that Twitter received the most in-links from other activist sites and social media platforms. This suggests that activists and supporters used Twitter to focus attention on events, interconnect other platforms, share resources via links, and reach broader publics.3
Particularly important for the present study was that Twitter offered an interface between legacy and social media. Many legacy media organizations followed the protests on Twitter, and used Twitter to report on articles about the protests. Moreover, Twitter had, by that time, become a relatively important source of news in the U.S. social media ecology. Although less widely adopted than Facebook as a news feed among the general public, Twitter users considered it a broader and more diversified news source than Facebook family and friends’ feeds (Mitchell, Rosenstiel, & Christian, 2012).
Identifying key issues in Occupy Twitter
We were able to access Twitter for the historical period surrounding the protests through a license from Crimson Hexagon. This media analytics platform provides “firehose” historical Twitter data.4 We created various collections of tweets in analy- sis interfaces, termed “monitors” by Crimson Hexagon. These monitors enabled the time sampling, filtering, and sequencing we used to study protest discourses, and their networked distribution via followers, mentions, and retweets. We also tracked issue associations over time with different types of accounts (e.g., protest core, pub- lic figures on the periphery of the protests, and media organizations).
We set up a general monitor to collect broad communication about OWS using Boolean search strings that encompassed a wide sample of OWS-related group
669Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
accounts and OWS-related keywords, developed in previous studies of OWS (Bennett et al., 2014). The OWS monitor was designed to capture the broadest pos- sible spectrum of messages about OWS at the national and local levels.5
Figure 2 illustrates what the overall OWS monitor looked like for the period starting before the protests, from July 1, 2011, up to the end of 2011, when the occu- pation sites and camps were largely removed by police. Since the inequality narra- tive had been firmly established by early November, much of our analysis focused on the first 7 weeks of the protests. The long tail in Figure 2 shows that we let the monitor continue to gather data for years after the protests, but we have only reported 2011 data related to our research questions here. The sample of 15,547,005 Tweets was by no means all of the OWS Twitter universe, but is representative of communication at different levels of the crowd (local to national) and across differ- ent city sites and protest events.
Next, we created four topical sub-monitors by adding each issue term (banks, democracy, economy, and inequality) to the same search string that generated the large collection from Figure 2.6 It is important to emphasise that we checked to make sure that this sampling method was not biased by using the four core issue terms, as opposed to adding dozens of variations for each term. Before settling on this approach, we tested a number of combinations of other issue terms associated with each of the four general issue areas, such as adding “financial crisis,” “bailouts,” and other terms to banks. We also worked with expanded vocabularies around each of the issue areas in order to make sure that we captured the respective issue to a satisfactory extent. In the end, the four core terms remained dominant and their rankings in proportion to the big data set did not change when we added more terms to our search strings. Having a manageable set of terms greatly simplified an
Figure 2 Collection of Occupy Wall Street–related Tweets, gathered in full corpus search on key Occupy protest terms and city protest sites. N = 15,547,005 (July-December 2011). Credit: Crimson Hexagon.
670 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
already complicated analysis, and enabled easier comparisons with other data sets, such as the news data reported above.
The four issue monitors were not mutually exclusive, meaning that the same issue theme could appear in multiple monitors. However, examination of overlap- ping content from each issue group confirmed that the terms were generally exclu- sive to each issue category more than 95% of the time, suggesting that the character limits in tweets kept the users focused on specific topics rather than lists. For exam- ple, fewer than 4% of tweets from the large collection that mentioned other eco- nomic problems (N = 94,390) contained mentions of inequality (N = 3,451). A qualitative examination of random sub-samples also supported the conclusion that the messages in the monitors covered meaningfully distinct areas. Table 1 gives examples of typical tweets in each area, drawn from the top three retweets in each category between September 17 and December 31, 2011. The important point is that the collection of inequality tweets from the large data set contained all refer- ences to inequality, as the collections of bank and democracy tweets contained all references to those terms. Since we were interested in tracking these general themes as they flowed in and out of the protest crowd, we chose to keep the key issue search terms simple, and feel confident that this is a better solution than trying to track and report dozens of related terms for each cluster.
In addition to the general Occupy Wall Street dataset and the four issue sub- monitors, we also created a dataset that focused on messages incorporating the “We are the 99%” meme.7 The dedicated dataset allowed us to explore the possibility that inequality messages were being communicated using this meme. Within this monitor, we then systematically explored the scope of different keyword filters (inequality, banks, democracy, economy) following a procedure similar to that used with the sub-monitor sampling. As reported below, the 99% meme did not contain any greater emphasis on inequality than on the other terms. Table 1 also shows typ- ical examples of messages in this dataset.
Identifying peripheral networks in Occupy Twitter
The choice to focus on Twitter and the sampling of the Twitter firehose served the purpose of encompassing a wide spectrum of public communication around OWS. Most importantly, this captured messages from those closer to the core of the pro- test events, as well as those further away in the broader periphery of observers, com- mentators, and critics. In the analyses below, the most-mentioned OWS group accounts provided indicators of concerns associated with sites, actors, and actions near the protest core, while the most-mentioned accounts from celebrities, public figures, and media organizations provided rough indicators of concerns capturing attention at the periphery.
Since we wanted to associate different content flows with different types of net- works over time, we analyzed the weekly top 10 most-mentioned accounts associ- ated with each of the four issues. To classify types of accounts, we used historical (fall 2011) account profiles, along with our previous research on Twitter networks
671Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
in the protests. The basic account classification scheme identified the following cate- gories of account holders that occupied different locations in (and interactions with) the protest twittersphere: core protest accounts included Occupy Wall Street group accounts (e.g., @occupyla, @occupywallst, @occupywallstnyc), Anonymous group accounts (e.g., @anonymous_sa, @anonyops), and other core activist suppor- ters, such as @adbusters and @theyesmen; mainstream media outlets included the New York Times, the Economist, the Globe and Mail, the Guardian, and accounts of journalists affiliated with those organizations; alternative media included Think Progress, Democracy Now, Wikileaks, Mother Jones, Truthout, the Rolling Stone, and people affiliated with these organizations; and public figures included authors
Table 1 Typical Messages in Each Issue Category
Inequality My column on Occupy Wall St as a primal scream against inequality: http:// nyti.ms/oFeVDY US is less equal than Egypt!
It’s the inequality, stupid: 11 charts that explain what’s wrong with America http://motherjones.com/politics/2011/02/income-inequality-in-america- chart-graph … #OccupyWallStreet #ows #OccupyWallSt
The #OccupyWallStreet protesters deserve a lot of credit for focusing attention on the income inequality in our country. #OWS.
Banks #OWS Fact: More people have now been arrested for protesting financial crimes than the # of bankers arrested for committing those crimes
Dear Bankers, “Surrender Offender, The 5th of November.” ~SN #Quote #OWS #OccupyWallStreet
My advice to the OWS protesters: hit bankers where it hurts http://www. rollingstone.com/politics/news/my-advice-to-the-occupy-wall-street- protesters-20111012 … via @rollingstone
Economy The world’s economy has been wrecked by these rapacious traders. Yet it is the protesters who are jailed. #occupywallstreet
Wall Street Bankers Jailed for Destroying the World Economy = 0 | #OWS Demonstrators Jailed for Protesting >1,000.
To Occupy is to simply take charge: “This is OUR country, OUR Congress, OUR economy, OUR jobs, OUR future-& we refuse to let the 1% own it!”
Democracy Slavoj Zizek on the #ArabSpring, #Occupy and the rise of #China:“the marriage between capitalism and democracy is over“http://aje.me/rHbLis
Tens of thousands #occupy the #BrooklynBridge chanting, “This is what democracy looks like! This is what America look like.” #N17 #OWS”
“This is what democracy looks like” #OccupyWallStreet. March begins Foley Sq to Wall Street
99% The battle is not left vs. right. The battle is corporatism vs. humans. #occupywallstreet #99%
I am the #99%. RT if you are. http://t.co/NG36jZRl #ows It’s the 1% vs. the 99%—which side are you on, members of Congress? The
99% Party has awakened. #OccupyTogether
672 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
and intellectuals (e.g., Salman Rushdie, Nouriel Roubini), politicians (e.g., Bernie Sanders), actors (e.g., Mark Ruffalo, Susan Sarandon), and filmmaker Michael Moore, along with other celebrities. We also tracked top-mentioned individual accounts, ranging from protesters to critics. The few accounts that did not fit the above categories were added to an “other” category.
Top-mentioned accounts served as a proxy for measuring attention in different networks engaged by the protests. Mentions of an account indicated content that was recognized and shared by others, as well as identified which types of actors were most important in shaping attention flows in networked framing processes at different points in time. Tracking the types of content contained in the most- mentioned accounts from different categories of actors offered a rough indicator of which issues were deemed most relevant in different types of networks. More gener- ally, we were able to determine what account types and content were “crowdsourced to prominence” (Meraz & Papacharissi, 2013 p. 158). We limited our week-by-week analysis to the top 10 most-mentioned accounts to keep the analyses manageable, as numbers of followers for just the top 10 accounts associated with our issue themes numbered in the millions. We also analyzed retweet patterns for the different issues.
In order to develop a measure of issue attention in the different top-mentioned network types, it was important to know how many followers each account had at the time of the protests. Determining this piece of metadata enabled us to estimate the scale of the networks promoting different issues. We searched our Crimson Hexagon OWS database and established historical follower numbers for all but one of the weekly top 10 most-mentioned accounts over the entire span of the protest.8 We also excluded four instances of @youtube from the analysis of 171 different top-mentioned accounts (72 different accounts in the bank monitor and 112 in the inequality monitor with 13 accounts appearing in both monitors) over the September to December time period, because YouTube did not produce its own content and it had huge follower numbers unrelated to protest content that would wildly distort the results.
Analysis of attention flows and networked framing
Although the predominant societal understanding of Occupy revolved around inequality, the theme of inequality was far less common in Occupy Twitter messages in our big data sample than references to banks, general economic problems, or democracy. Table 2 demonstrates the relatively low volume of messages about inequality in the general data set of OWS tweets. The gap between low references to inequality and much higher references to the other protest issues was even larger during the critical period in which OWS became tied to inequality in public under- standing (September 17–October 31).9
A possible counter-argument to the present analysis would be that the overrid- ing identity frame of the movement—“We are the 99%”—was really about inequal- ity, and that this is where the inequality messages were to be found. While retrospective interpretations may make this seem self-evident, the evidence does not support this assumption. A similarly low volume of inequality references is shown
673Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
in Table 2 for the 99% meme data. Indeed, fewer issues of any kind were mentioned in association with the 99% meme than in the general sample in the period of September 17 to December 31, 2011. Moreover, as Table 2 shows, when particular issues were attached to 99% tweets, they appeared roughly in the same rank order as issues in the general population of tweets, with mentions of banks and demo- cratic corruption far surpassing mentions of inequality. This pattern held for both the entire protest period and for the crucial period up to October 31 when the inequality frame was generally adopted by the larger public. Many people with whom we have discussed these findings continue to believe that the 99% meme was really a code for inequality. However, we propose that the causality actually ran in the opposite direction: people who were cued by the processes described here to focus attention on inequality tended to make the assumption that the protests in general, and the 99% meme in particular, were mainly about this theme.
While many members of the general public were forming the impression that OWS was about inequality, the protest culture favored showing how the concentra- tion of political and economic power in the hands of the few—the 1%—had created dozens, and by some counts, hundreds of problems for ordinary members of soci- ety, with inequality being just one of them. Rather than a coded statement about inequality, the 99% meme appears to have operated as what Papacharissi (2015) describes as an affective identity process. It was a rallying cry, a shared identity, and a general meme that traveled into society, and still remains popular.
When interpreting these patterns, it is important to note that information carried by Twitter and other social media formats addressed many problems and concerns beyond the issues at the heart of the protests. The vast majority of mediated commu- nication focused on practical and logistical issues. For example, Twitter served multi- ple roles, including allowing for real-time action coordination, resource distribution, and network weaving (e.g., linking multiple media platforms), and offering a news feed and issue forum. Moreover, many of the protest activities (occupying public spaces, shutting down ports, organizing general strikes) resulted in harsh police reac- tions, with accompanying bursts of social media reactions. There were more than 1.4 million references to police in the OWS tweet collections reported in Table 2 and
Table 2 The Numbers and Relative Volumes of Tweets About Democracy, Economy, and Banks, Compared With References to Inequality During the Peak Protest Period
September 17–December 31 OWS tweets (N = 15,547,005)
September 17–December 31 99% tweets (N = 367,237)
Inequality Baseline (N = 28,979) Baseline (N = 1,403)
Democracy 314% (91,080) 215% (3,019) Economy 324% (94,390) 242% (3,390) Banks 640% (185,544) 252% (3,530)
Notes: Data collection credit: Crimson Hexagon. OWS = Occupy Wall Street.
674 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
Figure 2. Most of the Occupy twittersphere focused attention on the movement and protest actions. In a random sample of 10,000 tweets from the dataset of fifteen mil- lion shown in Figure 2, the most frequently-mentioned terms were: protesters, police, people, movement, new, protest, #p2 (a hashtag used by progressives), live, protests, #tcot (a hashtag used by conservatives), and arrested.
It is not surprising that Twitter content focused more on protest dynamics than issue discourse (which was far more developed in face-to-face assemblies). Beyond the rallying, news feed, and networking functions of Twitter, the issue rankings reported in Table 2 reflected patterns similar to those found in earlier qualitative analyses of websites, documents, and interviews with activists (Bennett & Segerberg, 2013; Bennett et al., 2014). The broad cross-section of actors using Twitter in this case enabled us to construct indicators of the relative importance assigned to vari- ous issues by differently-located actor networks.
If references to banks were by far the largest of the issues circulating in the Occupy protest culture and twittersphere, and inequality the least mentioned, what explains the focus of public attention on inequality? Moreover, how did inequality emerge so quickly to capture public attention? The next section addresses the sec- ond research question, regarding the workings of the core and peripheral networks in the iterative networked framing processes. We also answer the third research question by showing how social and legacy media can operate together to focus the attention of large audiences on particular ways of understanding complex events.
The push and pull of attention in core and peripheral networks
How did networks at the core and the periphery interact to shape the attention of different audiences to different issues? What was the relationship between social and legacy media in these network framing processes? Our analysis shows that the inequality theme was initially promoted by prominent public figures who operated at the periphery of the protest events as commentators, witnesses, cheerleaders, and messengers to their (often large) followings. These public figures included movie stars, writers, and politicians. Other actors on the protest periphery were the many commentators and journalists for various alternative and mainstream media organi- zations. By tracking information flows over time and through these different kinds of peripheral networks, we were able to show how the inequality frame was selec- tively pulled from the protest crowds by these networks and spread so pervasively through society.
In order to analyze the issue attention and interaction of differently-positioned core and peripheral networks, we documented patterns of mentions of different types of accounts over time. To keep the analysis within space limits, we compared the networked attention to inequality and banks, since they were the issues that best captured the attention divide between general publics and communication at the protest core. The analysis reported on the types of accounts that were highly men- tioned in relation to the two issues, along with the numbers of their followers, through November 4, by which time the inequality story had been firmly linked to
675Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
Occupy, as indicated by the various popular attention and framing measures pre- sented earlier.
Figure 3 shows the selective attention to banks and inequality found in different types of accounts over time. Note that the scales on the Y axis (total followers of top 10 most-mentioned accounts) differ due to the need to compress the figures to fit the same visual format. Also note that each Y value corresponds to the center of each pair of ticks on the X axis. These results showed that attention to banks per- sisted and even grew over time in the core protest networks, as measured by total followers of the top-mentioned accounts managed by Occupy protesters. Over the same time periods, networks at the periphery (first public figures, then media orga- nizations) increasingly attended to the inequality frame.
Although some core protest accounts had large followings (two of the Occupy accounts had more then 170,000 followers), many of the peripheral accounts had far more followers (Salman Rushdie, 467,000; Mark Ruffalo, 330,000; Nicholas Kristoff, over 1.3 million). As noted earlier, this pattern of expanded total followings moving from the core to the periphery of dense, distributed networks has been
Figure 3 Attention to inequality and banks in top 10 most-mentioned Twitter accounts over time, compared to top-mentioned accounts of protesters, public figures, and media. Weekly analysis of different account types and their followers. Data credit: Crimson Hexagon. Total weekly followers are shown for (a) all top mentioned accounts, (b) top pro- test accounts, (c) top public figure accounts, and (d) top media accounts.
676 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
associated with the scale and visibility of protests (Barberá et al., 2015). Our analysis shows that large peripheral networks can also focus broader public attention on dif- ferent aspects of the event than those that attract the attention of the core.
Figure 3C shows that public figure accounts initially featured strongly in refer- ences to banks. This was largely due to the activity of filmmaker Michael Moore and politician Bernie Sanders, who pulled the banking themes from the protest core, but ended up mainly having networked conversations with the core protest networks and alternative media sites. Meanwhile, momentum for inequality quickly grew, as other public figures promoting inequality were pushed to the top of men- tioned accounts. By the third week of October, attention volumes for public figures promoting bank themes had faded out of the top 10 public figure mentions, indicating that the pace of these shifting attention flows was remarkably fast. Another indicator of the speed and sequential nature of the attention flows was the shift in top mentions of inequality from public figure to media accounts. Note that public figure accounts dropped entirely from the top 10 mentioned accounts in the final week of the analysis in Figure 3. The public figures were displaced by a mix of media and activist accounts, with the former distributing the inequality frame to far larger networks than the bank frame pushed by the activists.
To provide further contextual information about these attention flows, we also added up total impressions about banks and inequality produced by all types of top- mentioned accounts between September 17 and November 4. Tweets mentioning inequality that made the weekly top 10 most-mentioned accounts in the Occupy data set reached a combined following of 23,161,136, compared with 17,595,619 total followers receiving tweets about banks. As this networked cascade shifted toward inequality, impressions on over 20 million of those 23 million followers were made in the final 4 weeks, compared to just 10 million for the top-mentioned accounts associated with banks. This means that during the month in which the inequality story became dominant, the top-mentioned Twitter accounts, alone, made 10 million (twice as many) more impressions than did the top accounts asso- ciated with references to banks.
To further address the question about the relationship between social and legacy media in these networked framing processes, it was interesting to dissect the overall pattern of media attention to banks and inequality by breaking down the roles of alternative and mainstream media. Figure 3D shows combined alternative and main- stream media account mentions. A deeper analysis showed that the attention to banks in this figure was produced almost entirely by alternative media, with the exceptions of one mention of banks by Kristoff in the New York Times in early October and one MSNBC mention in early November. While alternative media were generally more focused on core protest messages about banks, several alternative media outlets had been reporting on the growth of inequality even before the protests began, and pushed those reports into the crowd in the early days of the uprising. One early introduction of inequality in the OWS twittersphere involved multiple waves of messages circulat- ing links to a story on inequality that Mother Jones magazine published in spring
677Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
2011. The Mother Jones account itself entered the stream in early October, and remained among the most-mentioned media accounts associated with inequality throughout the period shown in Figure 3D. Within weeks, however, this alternative media buzz about the importance of inequality was overshadowed (in terms of men- tions and size of followings) by mainstream media organizations such as the Economist, the Guardian, National Public Radio, and Nicholas Kristoff of the New York Times. Here, we saw how complex interactions between alternative and main- stream media, as well as social, online, and legacy media, interconnected in the net- worked framing processes surrounding OWS.
These dynamics are better displayed in Figure 4, which shows the disproportion- ate role of core protest and alternative media accounts in promoting attention to banks, and the tandem roles of public figures and mainstream media sequentially focusing (and together amplifying) attention to inequality. Figure 4 shows how atten- tion to core protest accounts with reference to banks actually grew over this period, while attention to inequality around the core fell to zero in the top 10 account men- tions. At the same time, the attention of both alternative and mainstream media to inequality grew, and those top accounts had far more followers than accounts at the protest core, as shown in Figure 3. These measures are, of course, rough indicators of information and attention flows, but they fit the overall pattern of the multiple indica- tors used to assess the various elements of our analysis.
Drilling even deeper into these figures revealed other measures of the different issue attention patterns between core and peripheral networks. A striking difference is that there were 13 unique core protest accounts in the top 10 mentions in connec- tion with banks over the 7-week analysis shown in Figures 3 and 4, compared with just 4 different core protest accounts associated with inequality. The pattern for public figure accounts was almost the exact opposite, with 13 unique accounts asso- ciated with inequality compared with just 3 in the top mentions for banks. Interesting differences also showed up in relation to the media accounts. While
Banks Inequality
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
9/17–9/23 9/24–9/30 10/1–10/7 10/8–10/14 10/15–10/21 10/22–10/28 10/29–11/4 0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
9/17–9/23 9/24–9/30 10/1–10/7 10/8–10/14 10/15–10/21 10/22–10/28 10/29–11/4
AlternativeMedia MainstreamMedia CoreGroupProtest Public Figure Individual Other
Figure 4 Top 10 Occupy Twitter Accounts Associated With Banks and Inequality, by Week and Type of Account, from September 17 to November 4, 2011. Data credit: Crimson Hexagon.
678 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
there were nearly equal numbers of unique media accounts engaged with the two issues (14 for banks, 16 for inequality), all but 2 of those tweeting about banks were classified in the alternative media category, while 9 of the top accounts associated with inequality were in the mainstream category. This means that almost half of the alternative media accounts communicated about both inequality and banks.
Not only were different types of networks involved with boosting attention to inequality, but important characteristics of the actual tweets in those networks dif- fered for banks and inequality in ways that contributed to the shaping of attention and meaning. Although the percentages of retweets mentioning the two topics were roughly the same (51.2% of inequality tweets were retweeted vs. 49.3% for banks), fully 72% of the inequality retweets contained links (mostly to media content about the topic), compared to just 43% of bank retweets. This suggests that the inequality information flows were focused more on deep content, while many bank references echoed protest identity and rallying calls about events that were more salient at the protest core where most of the bank tweets originated. In some weeks, the top hash- tags in tweets mentioning banks included #NYPD, but police hashtags did not occur in the top hashtags in tweets about inequality, providing another indicator that banks were closer to the heart of the protest action. In these many ways, networks on the periphery amplified and focused attention on messages that were not as cen- tral to the protest core.
Conclusion
It is, by now, commonplace to say that we live in an attention economy, but it is less evident how the complex information exchanges that shape attention actually work and how they play into the negotiation of meaning. What is clear is that high- impact social events such as OWS involve participants, spectator publics, and multi- media networks in the dynamic production of attention and meaning. Not only do various types of actors share in networked gatekeeping and framing processes, but peripheral networks can play a significant role.
In the case examined here, OWS protesters were most centrally concerned about democratic corruption, punishing and regulating wayward banks, and a host of eco- nomic problems, but broader public attention to the financial crisis and the role of banks had peaked and begun to decline as the movement emerged. Actors on the periphery activated large social media networks and linked Occupy to the inequality theme, even though it was far from the most important theme at the protest core. Of all the issues circulating in and out of the protests, the networked gatekeeping and framing processes focused the greatest public attention on problems associated with inequality in society. Various background factors helped explain why this issue was pulled so readily from the crowd and pushed into society. For example, alterna- tive media such as Mother Jones had reported on growing problems of inequality before the protests began. Many people also seemed to have experienced the exis- tential realities of inequality long before these communication processes made it the
679Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
focus of so much attention. Polls taken around the time the mainstream media linked inequality to OWS revealed high levels of public concern about the issue. These favorable polling results drew Democratic politicians into the inequality dis- cussion, led by then-President Obama. The arrival of official political voices in the conversation engaged traditional journalistic indexing processes that further expanded coverage in the legacy media. Policy debates in areas such as raising the minimum wage followed, and attention to inequality continued to grow for years afterwards. During the same period, political and media attention to banking poli- cies faded, despite continuing public dissatisfaction with both financial institutions and government regulation efforts.
We drew four general conclusions from this analysis that may be applied to other cases. First, the persistent conceptual divide between social and legacy media is not helpful to understanding how various actors and platforms now interact in complex media ecologies to spread information that captures the attention of large audiences (Entman & Usher, 2018). Contrary to the assumptions underpinning the divide, there are many interactive paths along which content may travel across social and legacy media in the networked framing process.
Second, gatekeeping and framing in multimedia ecologies are decreasingly uni- form or top-down processes controlled by traditional communication (e.g., journal- ism) norms. The dynamics of attention and meaning reflect the push and pull of information over different communication networks with different gatekeeping pro- cesses, content distribution methods, and combinations of human and algorithmic decisions. Consequently, the framing process becomes a matter of complex and iter- ated negotiation.
Third, it is not always the core actors that dominate the outcomes. Peripheral networks may play a significant role in the networked framing process, particularly at the blurring interface of social and legacy media. This means that, in networked communication processes, distant audiences may be cued by differently-located peripheral actors, with important effects on public attention and framing.
Finally, in order to understand these complex processes of gatekeeping, atten- tion, and framing, communication research must continue to move beyond conven- tional approaches to news framing and content analysis. While our analysis suggests that legacy media continue to matter, they may no longer be the prime movers in framing and related processes. The diffusion of ideas in multimedia ecol- ogies clearly operates differently than in the earlier mass media communication regime (Williams & Delli Carpini, 2011).
These generalizations may serve as heuristic starting points for approaching research design for the purposes of analyzing the relationship between movements (and other events) and societal understandings in the multimedia age.
Nevertheless, it is important to note that our case represents just one networked attention and framing pattern among many that can occur. Indeed, a fifth conclu- sion is that these complex networked framing processes may be associated with diverse outcomes. Actual cases will display various information flow patterns, which
680 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
may serve to amplify, bury, or distort the public import of movement messages. For example, when perspectives are well aligned between core and peripheral actors, peripheral networks may amplify ideas and support actions circulating at the activ- ist core. Cases such as the Arab Spring uprisings, #Black Lives Matter, or the #MeToo movement suggest this general pattern. In other cases, political opponents or hostile governments may disrupt the coherent framing of protests and other political events, such as elections, with surveillance, trolls, and fake media accounts. In yet another variation, when messages circulating at the center of the action are numerous and/or peripheral actors with large network followings select some mes- sages over others, even broadly supportive public understandings may drift away from the dominant messages projected at the center of action. As an example of the latter, our analysis illustrates how a protest movement more concerned about one thing (a banking crisis) was selectively credited with alerting society to another problem (inequality). While some protesters were concerned about inequality, more were concerned about other problems. As one November 2011 tweet reported: “One protester reminded me: ‘The #OccupyBoston & #OWS movements aren’t just about inequality…we think democracy is broken’.”
Our research design and methods reflect the reality that contemporary media ecologies contain multiple pulse points of public cueing, attention, and framing. These pulse points include core and peripheral social media content flows, opinion polls, editing sequences of Wikipedia articles, societal search trends, and news media narratives. Patterns of attention revealed by these multiple indicators offer insights about the origins of social consensus, conflict, or confusion surrounding social and political events. We hope this analysis stirs thinking about the processes that shape attention and understanding of events in rapidly-changing media sys- tems. As media ecologies grow ever more complex and boundaries among different types of media and their audiences become blurred, it seems time to rethink basic communication processes and research paradigms.
Acknowlegments
This work received funding support from the Alexander von Humboldt Foundation and the Swedish Research Council (Grant 2015-01835).
We thank Richard Rogers for his helpful comments and suggestions, as well as the anonymous reviewers.
Notes
1 The article was not labeled as conflicted, and of the more than 8,000 edits, 64% were done during the protest period between September–December, 2011, and more than 90% were done by mid-2012.
2 Search in Lexis Nexis Academic Database. Sources were U.S. newspapers and broadcast transcripts (selection: ABC News, CBS News, CNBC News, CNN, Fox News Network, MSNBC, National Public Radio, and NBC News). The search terms for articles with
681Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
headlines and lead stories mentioning inequality were: HLEAD (inequality). The results were as follows, by date: N = 200 in 2010, 522 in 2011, and 566 in 2012, with 75 from September–December, 2010 and 357 from September–December 2011. To search for references to Occupy Wall Street within this sample, we bracketed the inequality search string and added “AND (“Occupy Wall Street” OR “Occupy protest” OR “Occupy movement”)”. The results for September–December 2011 were 185 articles. The search terms for articles with headlines and lead stories mentioning the bank crisis were: HLEAD (bank AND crisis) OR HLEAD (bank AND reform) OR HLEAD (bank AND corruption). The results were as follows: 3,007 in 2010, 2,497 in 2011, and 2,124 in 2012, with 802 from September-December 2010 and 1,127 from September–December 2011. To search for references to Occupy Wall Street within this sample, we bracketed the bank crisis search string and added “AND (“Occupy Wall Street” OR “Occupy protest” OR “Occupy movement”)”. The results for September–December 2011 were 71 articles.
3 These properties of Twitter have also been observed in other protest studies (e.g., Barberá et al., 2015).
4 Crimson Hexagon preserves the metadata for all tweets captured by our search terms, but the content of any subsequently-deleted tweets was removed from the database. We have no reason to think that the removal of tweets in our issue categories was biased for or against any particular issue. For more information, see https://www.crimsonhexagon.com/.
5 The Boolean search terms for our OWS monitor were: occupy OR ows OR occupywallstreet OR occupywallst OR occupyoakland OR occupyla OR occupyatlanta OR occupyboston OR occupydenver OR occupyportland OR occupyphilly OR occupydallas OR occupydc OR occupychicago OR occupytogether.
6 For the OWS bank monitor, we bracketed the same search string described in footnote 5, and added the filter “AND” followed by “banks OR bankers.” For the OWS inequality, OWS democracy, and OWS economy monitors, we also began with the search string from footnote 5 and added, respectively, “AND” “inequality,” “democracy,” and “economy.”
7 The Boolean search terms for the OWS 99% meme monitor were: “(99 OR 99% OR wearethe99% OR wearethe99prct OR #99% OR #99%) AND (occupy OR ows OR occupywallstreet OR occupywallst OR occupyoakland OR occupyla OR occupyatlanta OR occupyboston OR occupydenver OR occupyportland OR occupyphilly OR occupydallas OR occupydc OR occupychicago OR occupytogether).
8 The historical number of followers were identified by examining samples of historical tweets exported from Crimson Hexagon. Crimson Hexagon allowed for a random sample of 10,000 tweets for time periods of any length. By adjusting the time frames for our samples of bank and inequality tweets, we were able to collect all of those tweets. For each post in the sample, the historical number of followers for the Twitter account at the time when the tweet was posted was recorded. These collections of bank and inequality tweets were then searched for top-mentioned accounts, and we were able to extract all historical follower numbers for all top-mentioned accounts in the categories that were critical to our analysis. Only 7 accounts (6 of them in the “other” category) were missing, because they were removed. The single case of missing followers concerned a core protest account that was among the top-mentioned during the week of September 24, 2011, but the exclusion of this account did not change the overall trends for attention to banks in this category.
9 During this time, messages about democracy, the economy, and banks appeared at rates of 374%, 377%, and 682%, respectively, compared to those focused on inequality.
682 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
References
Barberá, P., Wang, N., Bonneau, R., Jost, J. T., Nagler, J., Tucker, J., & González-Bailón, S. (2015). The critical periphery in the growth of social protests. PLoS One, 10(11), e0143611. doi:10.1371/journal.pone.0143611
Barzilai-Nahon, K. (2008). Toward a theory of network gatekeeping: A framework for exploring information control. Journal of the Association for Information Science and Technology, 59(9), 1493–1512. doi:10.1002/asi.20857
Benford, R. D., & Snow, D. A. (2000). Framing processes and social movements: An overview and assessment. Annual Review of Sociology, 26, 611–639. doi:10.1146/annurev. soc.26.1.611
Bennett, W. L., & Pfetsch, B. (2018). Rethinking political communication in a time of disrupted public spheres. Journal of Communication, 68(2), 243–253. doi:10.1093/joc/jqx017
Bennett, W. L., & Segerberg, A. (2013). The logic of connective action: Digital media and the personalization of contentious politics. New York, NY: Cambridge University Press.
Bennett, W. L., Segerberg, A., & Walker, S. (2014). Organization in the crowd: Peer production in large-scale networked protests. Information, Communication & Society, 17 (2), 232–260. doi:10.1080/1369118X.2013.870379
Byers, D. (2011, November 11). Occupy Wall Street is winning. Politico. Retrieved from https://www.politico.com/blogs/media/2011/11/occupy-wall-street-is-winning-040753
Chadwick, A. (2013). The hybrid media system. Politics and power. Oxford UK: Oxford University Press.
Deprez, E., & Dodge, C. (2011, November 9). Occupy protesters inject income inequality into political debate. Bloomberg/Business Week. Retrieved from https://www.bloomberg. com/news/articles/2011-11-09/occupy-wall-street-protests-inject-income-inequality- into-political-debate
Diani, M. (2003). Networks and social movements: A research programme. In M. Diani & D. McAdam (Eds.) Social movements and networks: Relational approaches to collective action (pp. 299–317). Oxford, England: Oxford University Press.
Entman, R.M., & Usher, N. (2018). Framing in a fractured democracy: Impacts of digital technology on ideology, power, and cascading network activation. Journal of Communication, 68(2), 298–308. doi:10.1093/joc/jqx019
Fontevecchia, A. (2011, November 17). Occupy Wall Street: Income inequality and the burden of action. Forbes Magazine. Retrieved from https://www.forbes.com/sites/ afontevecchia/2011/11/17/occupy-wall-street-income-inequality-and-the-burden-of- action/#5015ff6e2904
Freelon D., McIlwain, C., & Clark, M. (2018) Quantifying the power and consequences of social media protest. New Media & Society, 20(3), 990–1011, doi:10.1177/ 1461444816676646
Gaby, S., & Caren, N. (2016). The rise of inequality: How social movements shape discursive fields. Mobilization: An International Quarterly, 21(4), 413–429, doi:10.17813/1086- 671X-21-4-413
Gitlin, T. (1980). The whole world is watching: Mass media in the making & unmaking of the New left. Berkeley: University of California Press.
Graef, E., Stempeck, M., & Zuckerman, E. (2014). The battle for “Trayvon Martin”: Mapping a media controversy online and off-line. First Monday, 19(2). doi:10.5210/fm.v19i2.4947
683Journal of Communication 68 (2018) 659–684
W. L. Bennett et al. The Strength of Peripheral Networks
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
Kristoff, N. (2011, October 15). America’s primal scream. New York Times. Retrieved from http://www.nytimes.com/2011/10/16/opinion/sunday/kristof-americas-primal-scream.html
Mitchell, A., Rosenstiel, T., & Christian, L. (2012). What Facebook and Twitter mean for news. Pew Research Center. Retrieved from http://assets.pewresearch.org/wp-content/ uploads/sites/13/2012/03/28180923/PJ_2012.03.18_STONM_SpecialReport_ FacebookandTwitterMeanforNews_FINAL.pdf
Meraz, S., & Papacharissi, Z. (2013). Networked gatekeeping and networked framing on #Egypt. The International Journal of Press/Politics, 18(2), 138–166, doi:10.1177/ 1940161212474472
Neuman, W. R., Guggenheim, L., Mo Jang, S., & Bae, S. Y. (2014). The dynamics of public attention: Agenda‐setting theory meets big data. Journal of Communication, 64(2), 193–214. doi:10.1093/joc/jqx019
Papacharissi, Z. (2015). Affective publics: Sentiment, technology and politics. Oxford, UK: Oxford University Press.
Papacharissi, Z., & de Fatima Oliveira, M. (2012). Affective news and networked publics: The rhythms of news storytelling on #Egypt. Journal of Communication, 62(2), 266–282. doi:10.1111/j.1460-2466.2012.01630.x
Russell, A. (2016). Journalism as activism: Recoding media power. Cambridge, England: Polity.
Steinert-Threlkeld, Z. (2017). Spontaneous collective action: Peripheral mobilization during the Arab Spring. American Political Science Review, 111(2), 379–403. doi:10.1017/ S0003055416000769
Tufekci, Z. (2013). “Not this one”: Social movements, the attention economy, and microcelebrity networked activism. American Behavioral Scientist, 57(7), 848–870. doi:10.1177/0002764213479369
Wikipedia. (2011a). Occupy Wall Street [September 30, 2011, version of webpage]. Retrieved from https://en.wikipedia.org/w/index.php?title=Occupy_Wall_Street&oldid= 453289685
Wikipedia. (2011b). Occupy Wall Street [October 3, 2011, version of webpage]. Retrieved from https://en.wikipedia.org/w/index.php?title=Occupy_Wall_Street&oldid= 453790434
Wikipedia. (2013). Occupy Wall Street [February 9, 2013, version of webpage]. Retrieved from https://en.wikipedia.org/w/index.php?title=Occupy_Wall_Street&oldid= 537314159#Income_inequality
Wikipedia. (2017a). List of Occupy Movement protest locations. Retrieved from https://en. wikipedia.org/wiki/List_of_Occupy_movement_protest_locations
Wikipedia. (2017b). Occupy Wall Street. Retrieved from https://en.wikipedia.org/wiki/ Occupy_Wall_Street
Williams, B. A., & delli Carpini, M. X. (2011). After broadcast news: Media regimes, democracy, and the new information environment. New York, NY: Cambridge University Press.
Zhang, Y., Wells, C., Wang, S., & Rohe, K. (2017). Attention and amplification in the hybrid media system: The composition and activity of Donald Trump’s Twitter following during the 2016 presidential election. New Media & Society, doi:10.1177/ 1461444817744390
684 Journal of Communication 68 (2018) 659–684
The Strength of Peripheral Networks W. L. Bennett et al.
Downloaded from https://academic.oup.com/joc/article-abstract/68/4/659/5052234 by Adam Ellsworth, Adam Ellsworth on 13 August 2018
Copyright of Journal of Communication is the property of Oxford University Press / USA and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.