Original Research Article

Big Data & Society

January–June 2018: 1–12
Isomorphism through algorithms: ! The Author(s) 2018
DOI: 10.1177/2053951718757253

Institutional dependencies in the case of journals.sagepub.com/home/bds

Facebook

Robyn Caplan1 and danah boyd2

Abstract
Algorithms and data-driven technologies are increasingly being embraced by a variety of different sectors and institutions.
This paper examines how algorithms and data-driven technologies, enacted by an organization like Facebook, can induce
similarity across an industry. Using theories from organizational sociology and neoinstitutionalism, this paper traces the
bureaucratic roots of Big Data and algorithms to examine the institutional dependencies that emerge and are mediated
through data-driven and algorithmic logics. This type of analysis sheds light on how organizational contexts are
embedded into algorithms, which can then become embedded within other organizational and individual practices. By
investigating technical practices as organizational and bureaucratic, discussions about accountability and decision-making
can be reframed.

Keywords
Algorithms, accountability, Facebook, institutional theory, isomorphism, bureaucracy

This article is a part of special theme on Algorithms in Culture. To see a full list of all articles in this special theme,
please click here: http://journals.sagepub.com/page/bds/collections/algorithms-in-culture.

Concerns about the impact of data-driven interme- ecosystems. From the rise of ‘‘penny press’’ to the
diaries on the news media industry have been growing dynamics of government-driven propaganda, journal-
steadily over the last several years (Saurwein et al., ism has had to change depending on the broader land-
2015). Major social media and information companies scape (Schudson, 1987: 14). More recently, the internet,
like Facebook, Google, and Twitter play a central role social media, and algorithmic and data-driven systems
in what news and information people consume have altered many aspects of the news and information
(Gottfried and Shearer, 2016). The popularity of these landscape. Through the use of algorithms that rely on
systems—and the scale with which they impact both signals from both the content and interactions of con-
viewership and finances—has forced many news sumers, these technologies help curate what news and
media producers to alter how they produce and dissem- information is presented to whom. Furthermore, by
inate content for their audiences. In short, long-stand- providing services that allow everyday people to
ing news outlets must construct their content with actively serve as content distributors, their systems
algorithmic and data-centric intermediaries in mind.
Furthermore, a whole host of new digital-first outlets 1
Data & Society Research Institute, USA; School of Communication and
such as BuzzFeed and Breitbart have emerged to capit- Information, Rutgers University, USA
2
alize on the way in which this ecosystem is architected. Microsoft Research, USA; Data & Society Research Institute, USA
The news industry has long been interwoven with
Corresponding author:
other industries and institutions—most notably, adver- Robyn Caplan, Data & Society Research Institute, 36 West 20th Street,
tising and government. At various points in history, Floor 11, New York, NY 10011, USA.
news media has been reconfigured by shifts in those Email: [email protected]

Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (http://
www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further
permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-
at-sage).
2 Big Data & Society

help distribute some content more widely than others. led to confusion about their role and responsibilities
Traditional news enterprises, dependent on attention within the news media ecosystem. Ambiguity has
and clicks over digital advertising (often delivered by stemmed from the emergence of algorithmic and data-
programmatic advertising networks owned by driven platforms, built by people who typically lack
Facebook and Google), are forced to respond to these domain-specific expertise, entering into wide spectrum
shifts. As a result, these platforms have upended the of sectors and industries, such as transportation, public
organizational practices of news-producing platforms, health, criminal justice, or media. Often, technology
altering how both the newsroom and individual jour- companies position themselves as ‘‘platforms,’’ which
nalists operate (Christin, 2014; Petre, 2015). both serves to highlight their intermediary role and
This paper underscores how efforts to increase allow them to position themselves as ‘‘neutral’’ in
accountability within algorithmically-mediated fields ways that would make them more immune from more
need to consider the organizational values and institu- top-down regulation or from complaints by users
tionalized mechanisms embedded within algorithms within the United States (Gillespie, 2010). Julie Cohen
that have been driving organizational change across (2016) argues that this has created problems for a regu-
the news media industry. Part of the challenge for algo- latory environment developed during industrialism that
rithmic accountability work is to understand how algo- is dependent on ‘‘well-defined industries’’ with specifi-
rithms and data-driven technologies are both situated cations for what would ‘‘trigger regulatory oversight’’
within larger macro-social trends, such as the increased (p. 4). Information technology companies have come to
privatization of public services in the current era of mediate more and more of everyday life, without a clear
capitalism as well as changes in ownership structures understanding of how the incentives or goals of the
of industries, and also influence a wide-range of actors organizations developing technologies can affect
and organizations that have become dependent on diverse sectors or industries. Despite the ubiquity of
algorithmic and data-driven intermediaries. We draw digital and information technologies now, the language
on concepts from institutional theory, such as iso- of technology—in this case data and algorithms—is
morphism, to understand how algorithms structure dis- often used to make a company’s activities distinct
parate businesses and aims into an organizational field, from previously regulated institutions. At the same
leading them to change their goals and adopt new prac- time, the presence of these technologies also serves to
tices (DiMaggio and Powell, 1983: 148). This paper homogenize the sector’s practices and incentives.
provides an analysis of how the media industry has The process of homogenization is not unique to
shifted through its dependence on powerful algorithmic algorithms and data. Twentieth century organizational
intermediaries, such as Facebook. In doing so, we and neo-institutional scholarship focused on a different
examine both how technology has shaped media indus- set of mechanisms—namely, bureaucracy—that also
tries as well as how these systems have conferred value induced similar changes across sectors and industries.
and legitimacy to specific individuals and organiza- Scholars have previously highlighted this automation
tions. Using DiMaggio and Powell’s (1983) theory of of bureaucratic processes through software. For exam-
isomorphic change, we highlight how Facebook and its ple, James Beninger’s (1986) The Control Revolution
algorithmic and data-driven practices have become an focuses on these dynamics long before the totalizing
institutionalized organization within this domain, effects of Big Data and algorithmic intermediation
structuring the media system as an organizational had even begun to take shape. More recently, popular
field. In this sense, algorithms and data-centric technol- articles in publications like Slate and Real Life Mag
ogies, like bureaucracy, act as a mechanism of legitim- make connections between Big Data and algorithms
ation in the process of institutionalization, reflecting and their bureaucratic administrative predecessors
broader macro-structural social processes, inducing a (Clair, 2017; Elkus, 2015). Given long-standing efforts
process of isomorphism, or homogenization, among to introduce accountability into bureaucratic systems,
dependent organizations (DiMaggio and Powell, 1983; there is good reason to examine sociotechnical organ-
Meyer and Rowan, 1977). We argue that algorithmic ization practices through the lens of bureaucracy to
and data-driven technologies can be ‘‘de-mythified’’ open up ways of rethinking accountability in this
and viewed more akin to bureaucratic or administrative environment.
mechanisms than intelligent systems. Some who still consider social media platforms as
Viewing algorithmic systems as akin to bureaucratic focused on personal experiences rather than news may
or centralized administrative instruments can help re- think an analysis of Facebook’s effect on the news
orient technology companies that have eluded classifi- media an odd choice for an analogy of algorithmic sys-
cations, back into their regulatory domains. The fact tems as centralized bureaucratic institutions. Yet, con-
that social media and information technologies are not versations around algorithmic accountability often
automatically labeled as part of the news business has center on Facebook and Google in ways that reveal
Caplan and boyd 3

the entanglement of social media, news media, and (1986: 16). In this sense, algorithms that serve to pre-
algorithms (Napoli, 2015). Rooted in sociology and process, categorize, and classify individuals and organ-
political science, DiMaggio and Powell’s (1983) izations should be viewed as extensions of bureaucratic
theory of isomorphism provides a novel perspective tools such as forms, that have been associated with the
for debates about algorithms and culture, providing a state in the past. This comparison, however, has fallen
different vantage point for understanding the relation- by the wayside in contemporary studies of algorithms.
ships between organizations embedded within algorith- It is difficult to understand why, as much of the con-
mic logic. temporary information technology industry has
focused on the digitization of records and practices
that were previously done offline. For instance, early
Data-driven algorithms homogenize software development focused on business applications
Although algorithmic processes are shaping sociotech- like Microsoft Office and Lotus 1-2-3, which were
nical systems more than ever before, the notion of an designed to enable bureaucratic aims like the collection
‘‘algorithm’’ still lacks analytic stability and coherence and storage of records about actors (and relationships
(Seaver, 2017). Many scholars and computer scientists between actors), events, and processes. Algorithmic and
take as their starting point the definition provided by information systems have in many ways served to
Donald Knuth (1968), who argued that the word ‘‘algo- re-mediate the record-keeping function and standard-
rithm’’ refers to a ‘‘finite set of rules which gives a ization of bureaucratic mechanisms. In the process,
sequence of operations for solving a specific type of however, privately-owned software companies that
problem,’’ (p. 27). Meanwhile, there has been a growing have undertaken this work have fundamentally
understanding within the burgeoning field of ‘‘algo- transformed business and government, as well as
rithm studies’’ that algorithms are a powerful ‘‘rationa- consumer technologies, reconfiguring most sectors
lizing force’’ within the network society (Pasquale, into data-driven bureaucracies where algorithms promise
2015: 15). Thus, a number of researchers have been efficiency and optimization.
looking to other past methods of ‘‘rationalization’’ of Despite the obvious link between bureaucratic
societies to understand how to both frame an analysis modes of information production and online modes
of algorithms, as well as decrease its importance as the of data production using algorithmic models, there
object of analysis in algorithm studies. Algorithms that are some key differences in the current information
serve to rationalize industries, also work to homogenize environment which may have obscured this relation-
or make an industry more uniform and similar. ship. Firstly, bureaucratic modes of information pro-
Institutional and neoinstitutional definitions of algo- duction and management have often been associated
rithms examine at the role algorithms have begun to with the state and more centralized and hierarchical
play as mediators of macropolitical processes. Philip information organizational contexts. Bureaucracy, for
Napoli (2014) and Mike Ananny (2016) have both Weber, is a mechanism used by the state to induce
noted the usefulness of neoinstitutional theory for re- rationality within complex political and economic
locating algorithms within these complex social, polit- environments. The critiques we make of the character-
ical, and economic relationships. Both Napoli’s argu- istics of bureaucracy now in the twenty-first cen-
ment of ‘‘algorithms-as-institutions’’ and Ananny’s tury—that it is too hierarchical, too rigid, and
concept of ‘‘algorithmic assemblages’’ seek to under- requires workers to be specialized—was seen by
stand how algorithms come to mediate supra-organiza- Weber as precisely the instruments that could be used
tional processes, and automate them to directly by the state to create fairer, more just, and more equit-
‘‘structure user behaviors.’’ What is needed to expand able treatment of citizens within societies (Du Gay,
on both Napoli and Ananny’s theories is a way to 2000: 2; Green, 2008: 201). The ethos of bureaucracy,
understand how algorithms, as mechanisms of institu- according to Du Gay’s reading of Weber, was to be
tionalization, lead to broader system-wide changes ‘‘impersonal, expert, and procedural’’ through a com-
among organizations and individuals structured mitment and subordination to the bureaucratic hier-
through algorithms. archy (Du Gay, 2000: 4; Green, 2008: 201; Weber,
In the 1980s, James Beninger warned that the mech- 1978: 958ff). Like bureaucracies, algorithms are also
anisms of algorithms, which define individuals and often deployed with an expressed interest in limiting
actions into discrete categories (inputted as variable the subjectivity of decision-making systems (Beninger,
types) are underpinned by a belief in the value of 1986: 15; Du Gay, 2000: 2; Green, 2008: 201). Often
such processes of rationalization to organize societies, this is done as a way to make algorithmic systems
that ‘‘control can be increased not only by increasing appear more objective than their human counterparts,
the capability to process information, but also by even when humans directly play a role in the creation,
decreasing the amount of information to be processed’’ training, and deployment of algorithmic systems.
4 Big Data & Society

Viewing algorithmic systems as an extension of demonstrates that these processes are subject to similar
bureaucratic mechanisms can both serve to temper biases and concerns as previous bureaucratically driven
anxieties about the role algorithms are playing in institutions. Though this process of institutional de-
re-structuring industries, and highlight potential legitimization can find its roots in many histories, the
avenues for critique. Algorithms, like bureaucracies in narrative of technology as that which could disrupt
the past, have been positioned as the necessary antidote existing institutional structures can be traced to the
to subjective decision-making processes within large- ideologies embraced by many of early proponents of
scale and complex systems that are coordinating the internet.
between many individuals, industries, and organiza- This is where the theories of neoinstitutional scho-
tions, simultaneously. Companies that have emerged lars, such as DiMaggio and Powell (1983) and Meyer
to digitize records and automate the delivery of infor- and Rowan (1977), are especially useful for analyzing
mation and services to users, using proprietary algo- how these industry-wide changes can occur. For Weber,
rithms, have been able to enter into new industries bureaucracy rationalized society by trapping people
and spaces under the guise of internet exceptionalism into a structured order, or an ‘‘iron cage.’’ DiMaggio
(Wu, 2010). In the process, the bureaucratic mechan- and Powell (1983) revisit the concept of his ‘‘iron cage’’
isms that came to frustrate so many individuals have of bureaucracy, seeking to make sense of the existence
been closed down and hidden behind not only the of a system that persists and continues to structure
‘‘black box’’ of algorithms, but a mythology that sug- social life, despite its removal from the context in
gests that the work done by algorithms is fundamen- which that iron cage emerged (and despite it no
tally different from that done by offline administrative longer being an efficient way to structure society). The
mechanisms in the past (Pasquale, 2015). Viewing algo- concept of an ‘‘iron cage’’ is fruitful for considering the
rithms in this way also highlights how they can work to impact of algorithms, whose formalized logics often
organize, homogenize, and synthesize industries, such contradict the rhetoric of personalization, choice, and
as the news media industry, through processes of ‘‘iso- freedom.
morphism’’ studied by DiMaggio and Powell (1983), Algorithms and Big Data function in a similar
within the context of bureaucracy. way—in a world where surveillance is the norm,
merely existing in the world means you are structured
Algorithms as administrative into the technologies and systems of data collection,
production, and analysis that structure most of social
mechanisms
life today. For DiMaggio and Powell, the iron cage of
As technology companies have come to support state- bureaucracy persisted 80 years after Weber was writing,
based processes like predictive policing and algorithmic not because bureaucracy increased competition or
sentencing (Brayne et al., 2015; Christin et al., 2015), made the state more efficient or just or equal, but
administrative power has transferred from the state to because it served as a mechanism of rationalization
private enterprise, particularly as the technology indus- and structuration. In effect, bureaucracy demanded
try has advanced (Owen, 2015). Over the last two dec- legibility of every actor or organization that interacts
ades, the buzzwords disrupt and disruption have, with the state within the terms the state defined.
according to Taylor Owen, come to stand in ‘‘for a DiMaggio and Powell (1983) look to these mechan-
form of libertarianism deeply rooted in the technology isms of structuration to explain how organizations and
sector, a sweeping ideology that goes beyond the pre- individuals become more similar or ‘‘homogenous’’
cept that technology can engage social problems to the (p. 147). They use a concept, ‘‘isomorphism,’’ to
belief that free market technology—entrepreneurial- provide a way to understand system-wide changes in
ism—should be left unhindered by the state’’ (Owen, an industry, sector, or ‘‘organizational field’’ that
2015). During waves of hype surrounding automation forces ‘‘one unit in a population to resemble other
and ‘‘Big Data,’’ the mythology surrounding these tech- units that face the same set of environmental condi-
nologies implies it is more legitimate than existing insti- tions’’ (DiMaggio and Powell, 1983: 149). This concept
tutions, with more accurate claims to objectivity (boyd is helpful for understanding how both the mechanisms
and Crawford, 2012). Yet, the adoption of particular and the rhetoric of algorithms have become embedded
technological practices and vocabulary often serves within the structure of social life, despite rampant
more as a signal of ‘‘legitimacy’’ than an attempt to critiques about their capacity to accurately represent
improve productivity and performance of an industry reality, increase efficiency, or remain free from bias.
(DiMaggio and Powell, 1983; Meyer and Rowan, These theories provide an orientation to examine
1977). This has been seen in the adoption of algorithmic how organizations become more similar through
and data-driven processes across a wide spectrum of dependence of one organization on another organiza-
sectors and institutions, despite evidence that tion, using predictors such as ‘‘the greater the
Caplan and boyd 5

dependence of an organization on another organiza- who is exerting a dominant administrative function

tion, the more similar it will become to that organiza- (in this case, Facebook’s News Feed algorithm), the
tion in structure, climate, and behavioral focus’’ more that organization will be able to exert change
(DiMaggio and Powell, 1983: 154). DiMaggio and on other organizations that rely upon them
Powell offer three frames or forces with which one (DiMaggio and Powell, 1983: 154). By defining and
can analyze these changes—coercive, mimetic, and nor- re-defining the concept of relevance or ‘‘value’’ of infor-
mative—though these forces can hardly be considered mation and news media, Facebook increasingly writes
as separate. In analyzing news media industry in the rules, or code, that defines which content succeeds
response to algorithms and Big Data, what becomes or fails in no small part because Facebook is now play-
relevant is how these frames of analysis can help iden- ing an outsized role in how people access news content
tify new forms of legitimation within organizations, (Gottfried and Shearer, 2016). Furthermore, by altering
professionals, and audiences that can work to tease the economics of journalism through the reconfigur-
out what or whose values are being prioritized through ation of attention and advertising, Facebook drives
the current media ecosystem, that are then structured news media organizations to incorporate metrics such
through algorithms. The increasing dominance of the as click-rate, likes, and shares.
algorithmic form is thus analogous to the role played
by bureaucracy in neo-institutionalist work in soci-
ology, where bureaucracy is as much a carrier of legit-
News media and the news feed
imation, as it is the outcome of more macro-structural In their analysis of the bureaucratic state, DiMaggio
social processes. and Powell describe coercive forces that stem from pol-
itical influence and what they refer to as ‘‘the problem
Algorithmic isomorphism: Re-orienting of legitimacy’’ (DiMaggio and Powell, 1983: 150). In
the context of information intermediaries, Lawrence
the organizational field of news media Lessig (1999) takes a similar approach through his
The news media industry provides an interesting case adage ‘‘code is law,’’ which emphasizes the degree to
study to study the increasing dominance of the algo- which code and hardware serve to structure and regu-
rithmic form at competing levels of organizational con- late certain activities and outcomes within computa-
texts (the technology company versus the news media tional systems.
company) and individual practices (engineering versus Facebook itself has used its News Feed algorithm,
journalism). Technology companies that produce and changes being made to it, to exert powerful coer-
search engines, social media, aggregators, and recom- cive pressures on organizations operating within its
mendation engines are currently operating as interme- walls. Evidence that news media organizations are sub-
diaries in spaces where news media content is both ject to the informal and formal pressures Facebook’s
produced and distributed. Though platforms have platform places upon them can be seen in their relative
repeatedly tried to distance themselves from traditional success following changes to Facebook’s News Feed
classifications of sectors (Napoli and Caplan, 2017), algorithm. Publishers that had early success in News
theories of isomorphism can work to re-situate these Feed effectively subsumed their own organizational
technologies back into their domains. DiMaggio and practices to the logic of Facebook’s algorithms.
Powell’s theory of institutional isomorphism provides Outlets like BuzzFeed, EliteDaily, and UpWorthy were
one mechanism to re-orient these companies into the early winners in the Facebook ecosystem, using tactics
organizational field of media. This lens reveals the dis- like engineering headlines to include emotional direct-
connect between (1) the values and assumptions being ives for readers (to click or like and share) and creating
embedded into the technology that shapes the media digestible and relatable content that could be easily
industry and (2) the (often problematic) values that shared among users (Meyer, 2013; Oremus, 2016). In
have dominated journalism historically. This has been August 2013, Facebook engineer Lars Backstrom
made more complicated as platform companies authored a ‘‘News Feed FYI’’ for the corporate blog,
engaged in the prioritization and filtering of news explaining why sites engaging in ‘‘clickbait’’ practices
media content, like Facebook and Google, have repeat- were prioritized by Facebook’s algorithms. At that
edly sought to differentiate themselves from traditional time, Facebook’s algorithms prioritized stories with
media companies (Napoli and Caplan, 2017). many likes and comments, and re-prioritized content
The power and centrality of Facebook’s News Feed to the top of a user’s feed that had significant engage-
algorithm, in particular, is important. According to ment (Backstrom, 2013). The impact on the organiza-
DiMaggio and Powell’s predictors of isomorphic tions willing to play by these rules was significant. In
change (or more homogeneous and similar), the more October 2013, Facebook announced that referral traffic
dependent an industry becomes on one organization to media sites from Facebook had grown by over 170%
6 Big Data & Society

from the previous year. Publishers that engaged with (Efrati, 2016). This included a new emphasis on
this system saw even more significant growth, with ‘‘native videos’’ embedded directly in the News Feed,
BuzzFeed reporting referral traffic increases of 855% which was communicated out to news publishers dir-
and Bleacher Report at 1081% (Wong, 2015). It was ectly by Facebook (Oremus, 2016). It also included the
during this period that many of the dependencies launch of Instant Articles in May 2015, a platform
between Facebook and news publishers were strength- developed exclusively for the hosting of content from
ened. SimpleReach, a content measurement and distri- recognized news media publishers to reduce load time
bution company, announced in 2013 that Facebook for users clicking on news media stories (Reckhow,
was driving ‘‘more traffic than any other social 2015). This new platform was also the first step in a
network,’’ surpassing other social media sites popular revenue-sharing model between Facebook and news
at the time, such as Twitter and StumbleUpon publishers, albeit limited—they offered that publishers
(Scottberg, 2013). could sell ads in their articles and ‘‘keep the revenue’’ or
Facebook’s central position within this emerging use Facebook’s Audience Network, the site’s own tar-
organizational field led to repeated changes within the geted advertising product, already used by many
media industry as organizations adapted to Facebook’s brands and publishers (e.g. The Huffington Post) for
algorithms, and as Facebook changed its algorithms to audience measurement and targeted ad delivery
adapt to these organizations. In their effort to combat (D’Onfro, 2016). One report, by Digiday, said pub-
the dominance that news media organizations engaging lishers using Instant Articles saw a drop in referral traf-
in ‘click-bait’ were having over their network, fic over the last quarter of 2015, though the author
Facebook released another change to the News Feed notes that few publishers were willing to speak about
in late 2013 to identify ‘‘high-quality’’ news content declines in traffic on the record (Moses, 2016). During
(Meyer, 2013). ‘‘High-quality’’ was defined by another tweaking of its algorithm in June 2016,
Facebook as whether users continued to interact with Facebook’s algorithm re-prioritized friends and family
an article after-the-fact, which meant that some pub- over publishers, and news media organizations again
lishers saw older articles begin to re-emerge on the net- saw significant declines in referral traffic (Mosseri,
work, with traffic driven to this older content. In 2016). This corresponded (though there is no docu-
August 2014, Facebook released another change to mented causal link) to an uptick in a spread of misin-
the News Feed to address ‘‘Click-Baiting Headlines,’’ formation (commonly referred to as ‘‘fake news’’) over
further defining their concept of quality news sources. the Facebook network over the same period
In this version, Facebook used variables like ‘‘how long (Silverman, 2016).
people spend reading an article away from Facebook’’ Though Facebook has in many cases claimed that its
as a way to calculate how users determine content that algorithms merely neutrally reflect the aggregate activ-
is valuable to them (El-Arini and Tang, 2014). ities of users (Zuckerberg, 2016, 2017), the framing and
Facebook warned publishers relying on click-baiting re-framing of the News Feed’s prioritization of content
headlines, that their referral traffic may decrease. challenges this claim. This pattern of changing the algo-
Outlets like Eli Pariser’s UpWorthy were particularly rithm to meet their own organizational incentives also
affected by this change, with a decrease of 46% of refer- highlights how accountability proposals that focus pri-
ral traffic over two months (McArdle, 2014). Pariser marily on gaining access to algorithms or data will fall
responded to the change by re-evaluating the metrics short, given that changes can be made to the News Feed
by which UpWorthy calculated its own success, to be algorithm quickly and with widespread effects on indus-
more in line with Facebook’s own goals. Pariser is try practices. Of course, not all news organizations were
quoted as saying he was shifting his organization as responsive to the changes made to the News Feed
towards ‘‘Facebook’s focus on engaged time’’ (Kafka, algorithm. Understanding the broader contexts
2014). This shifting of goals in response to Facebook is through which some companies adapted to changes
perhaps indicative of the less explicit forms of coercive made to the algorithm, when others did not, would be
isomorphism described by DiMaggio and Powell a worthwhile area of future investigation.
(1983), in which organizations are driven to conform
to gain support from the organizations upon which
they are now dependent (p. 151).
‘Innovation’ through imitation
Over the course of 2015 and 2016, several other Changes stemming from coercive forces, especially
changes Facebook made to its News Feed had effects when frequent, lead to an environment of uncertainty
on news media organizations. As status updates and that prompts dependent organizations to learn from
personal sharing among users began to decline over other dependent organizations that have successfully
2015, Facebook began to invest more of their resources conformed to the structuring mechanisms. This process
in products geared towards news media distribution of ‘‘mimesis,’’ or imitating models for success, is
Caplan and boyd 7

another process DiMaggio and Powell (1983: 151) incorporating the organizational incentives of these
argue will induce similarity across an organizational technologies (how Facebook was structuring value of
field. In this sense, the dominant organization’s incen- an article) into the structure of their organizations,
tives or goals become embedded across an industry as well as into the system of incentives that were
through the borrowing of practices that lead to success being used to drive coverage among journalists.
over the network. In the case of Facebook, this was The New York Times addressed this issue of ‘‘disrup-
seen in the adoption of data-driven metrics and ana- tion’’ of the existing media industry by new technology
lytics into newsrooms, as well as the growth of a new set players directly within their report, in a section titled
of intermediaries that were fed directly by the ‘‘What is Disruption?’’ (The New York Times, 2014:
Facebook API, whose role it was to analyze and com- 16). Featured in the report were media businesses that
municate Facebook metrics back to publishers. had adopted metrics and analytics into their coverage
During the early era of the News Feed, relationships (such as BuzzFeed, ESPN, and Quartz) and who were
between Facebook and media organizations were far expanding their digital offerings rapidly by using
from static or one-directional. Rather, an ecosystem ‘‘social search and community-building tools and stra-
of social media analytics businesses, using the tegies’’ (p. 24). Other media organizations sought to
Facebook API, acted as intermediaries between similarly adapt their business models, highlighting the
Facebook and the news media industry which was role that data-driven and algorithmic processes can
growing dependent on the social media platform to take in compelling news media organizations to take
reach audiences. Throughout 2013, a number of tools on the characteristics of social media platforms—an
and products were rolled out for media organizations in example being Tribune Publishing rebranding itself as
order to bring in data or content from Facebook users the tech company ‘‘tronc’’ which purports to use
directly into their newsroom. In September 2013, machine learning to better serve audience interests
Facebook rolled out tools for publishers and ‘‘media (Napoli and Caplan, 2017). As DiMaggio and Powell
partners’’ including BuzzFeed, CNN, NBC’s Today argue, modeling of one organization’s practices by
Show, BSkyB, and Slate to integrate ‘‘public posts of another is ‘‘a response to uncertainty’’ through the
real-time activity about any given topic,’’ in the form of borrowing of practices that may enhance legitimacy
Keyword Insights API, and the Public Feed API or possibility for success, or to demonstrate to others
(Osofsky, 2016). The development of these products, that they are working to change their practices to be
as well as other partnerships, quickly lead to an add- in-line with those of the dominant organization
itional ecosystem of businesses who used Facebook’s (DiMaggio and Powell, 1983: 151). To that end,
public API and served as an intermediary between many of the practices that we now associate with
Facebook and media organizations. Facebook lists ‘‘innovation,’’ such as the adoption of Big Data meth-
these ‘‘media solutions’’ partnerships on their site, ods by the Tribune, is actually due to an uncertain
which includes CrowdTangle, a social media analytics environment which induces one organization to copy
company that was bought by Facebook in November the practices of another.
of 2016 (Newton, 2017).
As Facebook and other online intermediaries began
Influence (breadth) versus reputation (depth)
to take on a larger role in the distribution of journalism
and other news media content, the media industry con- The third source of isomorphic change described by
tinued to shift in respond to this algorithmic and data- DiMaggio and Powell occurs at the individual level
driven environment. While some content providers, during processes of professionalization of a workforce,
such as BuzzFeed and The Huffington Post, emerged which they refer to as ‘‘normative pressures’’ (p. 152).
out of these new algorithmic markets producing con- There are many reports that the work of journalism has
tent directly for Facebook and other social media net- changed significantly in response to digital media, how-
works (Herrman, 2016), other news agencies, including ever, to what extent this is due to Facebook in particu-
The New York Times, had to quickly grapple with how lar is not known. At the same time, broader trends—the
to incorporate metrics and analytics into their news- rise of citizen journalism, journalists adapting data-
room cultures (Sobel Fitts, 2015). These pressures driven metrics into communicating the value of their
were greater for some organizations who saw falling work, and the incorporation of computational skills
readership on both their website, and on mobile, as into journalistic work—need to be considered alongside
they competed with other digital content producers the emergence of Facebook and other data-driven
more able to quickly adapt to the algorithmic and intermediaries to assess how algorithms have changed
data-driven ecosystem. the journalist-audience relationship (Anderson, 2011).
The news industry was responding to the impact that Additional factors for consideration include algorith-
new digital technologies were having on their industry, mic methods of surfacing news content (through the
8 Big Data & Society

Trending Topics module on Facebook) and the produc- that newsmaking is an ongoing process and a collab-
tion of news content through automated methods orative venture between journalists and audiences.
(Podolny, 2015). Meyer (2013) found that, because of social media, audi-
As Anderson (2011) and other scholars (Tandoc, ences expected journalists to write stories they could
2014; Tandoc and Ferrucci, 2017; Vu, 2013) have ‘‘relate’’ to, mirroring expectations from the personal-
noted, as online metrics and data-driven processes ization of content that occurs over algorithmic and
gained an increased status in the news media ecosystem, data-driven networks, such as Facebook.
news media organizations began to adapt, using these Coercive, mimetic, and normative pressures, influ-
metrics to learn more about what their audiences search enced through algorithmic prioritization of content
for online and what topics can drive revenue. As a determined through metrics, can thus have important
result, they began to choose their ‘‘subjects solely on implications for the production of content. As social
these computer-generated metrics’’ (Anderson, 2011: media and search engines have centralized the produc-
536). Anderson stressed that media companies began tion and distribution of content, news media content in
to embrace an ‘‘algorithmic understanding of demo- particular, they have also shifted how publishers and
cratic processes,’’ viewing the process of news delivery journalists determine not only what is important to
as one in which data and algorithms are used to assess cover, but how the value of this coverage is communi-
individual user wants and needs, relying on a model of cated to a wide range of actors communicating value
‘‘individual consumer choice’’ that becomes equated through metrics. These actors include publishers com-
with democratic values (Anderson, 2011: 541). municating the reach of their stories and site, journal-
Journalists themselves were reporting pressures ists communicating their value to potential employers,
within newsrooms to adapt to the new digital ecosys- and the communication of value (and cost) to adver-
tem, as systems of incentives that valued views and tisers, using these same services to place and distribute
clicks began to dominate. Though media companies advertisements to consumers. Due to their centrality
have always used audience measurement to guide within this network of interrelated actors, however,
reporting, Petre (2015) argues that the ‘‘tracking cap- power to drastically alter these relationships has
abilities of the internet, as well as the ability to store become centralized within opaque proprietary compa-
and parse massive amounts of data, mean that audience nies like Facebook, which can work define, and re-
metrics have grown far more sophisticated in recent define the rules structuring these relationships with no
years.’’ Analytics companies contribute to the ecosys- accountability nor oversight.
tem of data-driven journalism, impacting the profes-
sional practices of journalists within newsrooms. Algorithms as
Reports by Nieman Lab (2014) have detailed the administration—Implications
manner by which journalistic and editorial practices
began to shift in response to the emergence of web met-
for oversight
rics, leading to the rise of a ‘‘culture of the click,’’ with Arguments both in favor of and against incorporating
web metrics often used as a management tool, particu- algorithmic decision-making tend to over-emphasize
larly when websites rely on ‘‘traffic-based financial the role algorithms specifically play in the construction
incentives’’ (Christin, 2014). Metrics influence journal- of reality. Instead, algorithms should be viewed more as
istic practices even when only editors—and not journal- administrative mechanisms that organize relationships
ists—have access to the data (Sobel, 2015). Even when between organizations and individuals. As part of this
boundaries between journalists and metrics exist, jour- organizational structure, not only are algorithmic sys-
nalists still seek out ways to numerically compare their tems working to automate the administrative mechan-
work to others (Petre, 2015). isms of a dominant organization, but they are also
A focus on numeracy and digital data shapes every providing a common language or structure that serves
aspect of how journalists are expected to do their work. a legitimizing function that affects other organizations
For example, journalists are increasingly encouraged to and individuals within that field. This is both positive
develop technical skills, become skilled at using social and negative for increasing oversight into this process
media, and learn to code (Broussard, 2015). The pres- of structuration. Using neoinstitutional theory, it is
sure to learn digital skills appears to have affected jour- possible to trace a network of organizational dependen-
nalism training as early as the 1990s, with roots cies and relationships that we cannot see through the
stemming even earlier to the digitization of some news- code alone.
papers in the 1970s (Fahmy, 2008: 24). What is unique Through the administrative function of algorithms,
to the current ecosystem in which intermediaries play a organizational incentives become deeply embedded
more central role are calls for journalists to engage in within many layers of an organizational field, making
‘‘journalism-as-a-process’’ which stresses to journalists the encoded algorithmic models almost invisible and
Caplan and boyd 9

less amenable to change. Within the area of media algorithmically-driven systems, humans are part and
policy, tracing this map of interdependencies, in terms parcel of how individuals, behaviors, and content
of both sociotechnical and economic dependencies, become classified and embedded within algorithmic-
begs us to question how one changes system-wide systems (e.g. the use of journalists to train the
incentives and organizational structures that become Facebook Trending Topics algorithm (Nunez, 2016)).
embedded at multiple levels. It also begs the question However, automation and algorithms may take over
of how, or according to what principles, we begin to many processes, leading to fewer points of access for
assess the constraints, limitations, and goals of the individuals to question or critique how they have been
dominant organization as their commitments become classified into the system (Eubanks, 2018).
embedded into algorithmic or computational mechan- Additionally, the human element of bureaucratic sys-
isms. As part of this re-definition of values and criteria, tems may in some way reduce the complexity of the
neoinstitutional scholars Meyer and Rowan (1977) bureaucratic systems. If humans are responsible for car-
stress the need to include mechanisms that would not rying out mechanisms of bureaucracy, to some degree,
necessarily be subsumed within this dominant formal they must understand how the bureaucracy works.
structure (p. 356). In other words, one approach Within algorithmic bureaucracies, code often serves to
Facebook could take would be to develop a mechanism mediate these relationships, as well as the relationships
by which news content is excluded from the same between data collected through the platform, reducing
metric-ization of clicks and likes as other content on the number of people capable of understanding the
Facebook. In this way, Facebook could ‘‘decouple’’ complexities of the whole system, as well as opportu-
its valuation of news from its organizational logics. nities for critiquing process or whistleblowing.
What makes algorithms so seemingly necessary and The news media industry has been irrevocably
powerful is the sheer amount of data they parse and sort shifted in the era of data and algorithms. In many
through—far beyond what individuals are capable of ways, these mechanisms brought about positive
processing on their own. And yet, evidence increasingly changes to an industry that has long been hegemonic,
suggest that algorithms are no less immune to the biases top-down, and controlled by powerful corporate inter-
and inefficiencies of the humans that created them, ests that were allowed to consolidate in the late 1990s
within the domains or sectors within which they are after changes to the Telecommunications Act (Croteau
situated (Kroll et al., 2017). Consistently, technology and Hoynes, 2006). While algorithms are only a small
remains embedded within the specific historical social, part of what is inducing change across the news indus-
political, and economic contexts—and existing systemic try, the Facebook example shows that power rests not
social injustices—in the domains and sectors in which in an algorithm’s capacity to induce a new logic into an
they are deployed, despite their owner’s hopes for their industry, but in its function as an administrative mech-
disruptive power (Owen, 2015). anism that is embedded with the values and cultural
Understanding algorithms as an extension of the and economic environment of their creators. In this
concept of bureaucracy, in terms of both their organiz- process, algorithms shape all other organizations and
ing and legitimizing functions, is one step towards individuals operating within a given landscape.
understanding how organizations have become more In studying the role data and algorithms have in
homogenous in the era of algorithms and data-driven reproducing the structure of social life, there has been
processes. In their analysis about the similar function a significant emphasis placed on gaining access to the
bureaucracy played centralizing the power of the state, specific technologies themselves. Efforts to increase the
DiMaggio and Powell provide one set of mechan- transparency or auditability of algorithms and data rest
isms—predictors of isomorphic change—that is useful on the assumption that gaining access to code, or even
for conceptualizing how organizational contexts the data used to train machine learning models, can pro-
become embedded within other dependent organiza- vide us insight into what goes wrong when there are
tions, through the administrative processes of data biases or inefficiencies within systems (Diakopoulos,
and algorithms. This work is useful for re-orienting dis- 2014; Sandvig et al., 2014). However, institutional
cussions about how to assess the values driving the theory shows that, to the extent that one organization
shaping of an organizational field, such as the news is able to control the behaviors, actions, or incentives of
media industry. other organizations through changing the structure of
Where the analogy between algorithms and bureau- their system unilaterally, the ripple effects go far
cracy falls short can also illuminate broader concerns beyond the code itself. More research is needed to under-
about introducing accountability mechanisms within stand how the interplay between organizations in the
algorithmic systems. Bureaucracies were implemented news ecosystem are influencing each other and being
by human beings and tend to have humans at most implemented through code. Increasing oversight or
ends of the bureaucratic process. Within these more accountability into how an industry has been shaped
10 Big Data & Society

by a dominant organization is incredibly complex, and Christin A, Rosenblat A and boyd d (2015) Courts and pre-
entails a system-wide analysis of how organizational dictive algorithms. Data & Civil Rights: A New Era of
incentives, built into algorithms, both operate as a con- Policing and Justice. New York, NY: Data & Society.
straint across a field, and are themselves constrained by Clair A (2017) Rule by nobody: Algorithms update bureau-
cracy’s long-standing strategy for evasion. In: Real Life
larger macro-sociological trends.
Mag. Available at: http://reallifemag.com/rule-by-
nobody/.
Acknowledgments Cohen J (2016) The regulatory state in the information age.
We thank Philip M Napoli, CW Anderson, Emmanuel Moss, Theoretical Inquiries in Law 2: 370–414.
and our anonymous reviewers for their generous feedback on Crain M (2009) The rise of private equity media ownership in
this paper. Participants at three workshops also helped us the United States: A public interest perspective.
hone early versions of this document: the Digital International Journal of Communication 3: 208–239.
Technologies and Democratic Theory workshop at Stanford Croteau D and Hoynes W (2006) The Business of Media:
University in May 2017, the Algorithms in Culture workshop Corporate Media and the Public Interest. Thousand
held at UC Berkeley in November 2016, and the Eclectic Oaks, CA: Pine Forge Press.
Workshop held at Data & Society in December 2016. We Diakopoulos N (2014) Algorithmic Accountability Reporting:
are deeply grateful for all of their feedback, critique, and On the Investigation of Black Boxes. New York, NY:
recommendations. Columbia Journalism School, Tow Center for Digital
Journalism.
DiMaggio J and Powell W (1983) The iron cage revisited:
Declaration of conflicting interests
Institutional isomorphism and collective rationality in
The author(s) declared no potential conflicts of interest with organizational fields. American Sociological Review 48:
respect to the research, authorship, and/or publication of this 147–160.
article. D’Onfro J (2016) How an ‘oddball’ team created one of
Facebook’s biggest threats to Google. Business Insider.
Funding Available at: http://www.businessinsider.com/what-is-
The author(s) disclosed receipt of the following financial sup- facebook-audience-network-and-why-does-it-matter-2016-
port for the research, authorship, and/or publication of this 2 (accessed 15 October 2016).
article: in part by a grant from the Open Society Foundations. Du Gay P (2000) In Praise of Bureaucracy. London: SAGE.
Efrati A (2016) Facebook struggles to stop decline in ‘ori-
ginal’ sharing. In: The Information. Available at: https://
References www.theinformation.com/facebook-struggles-to-stop-
Ananny M (2016) Toward an ethics of algorithms: decline-in-original-sharing (accessed 14 November 2016).
Convening, observation, probability, and timelines. El-Arini K and Tang J (2014) News Feed FYI: Click-baiting.
Science, Technology, & Human Values 4(1): 93–117. Facebook Newsroom. Available at: https://newsroom.fb.
Anderson CW (2011) Deliberative, Agonistic, and com/news/2014/08/news-feed-fyi-click-baiting/ (accessed
Algorithmic Audiences: Journalism’s Vision of its Public on 16 October 2016).
in an Age of Audience Transparency. International Elkus A (2015) You can’t handle the (algorithmic) truth. In:
Journal of Communication 5: 529–547. Slate / Future Tense. Available at: http://www.slate.com/
Backstrom L (2013) News Feed FYI: A window into news articles/technology/future_tense/2015/05/algorithms_
feed. Available at: https://www.facebook.com/business/ aren_t_responsible_for_the_cruelties_of_bureaucracy.
news/News-Feed-FYI-A-Window-Into-News-Feed. html (accessed 25 January 2017).
Beninger J (1986) The Control Revolution: Technological and Eubanks V (2018) Automating Inequality: How High-Tech
Economic Origins of the Information Society. Cambridge, Tools Profile, Police and Punish the Poor. New York,
MA: Harvard University Press. NY: St. Martin’s Press.
boyd D and Crawford K (2012) Critical questions for Big Fahmy S (2008) How Online Journalists Rank Importance of
Data: Provocations for a cultural, technological, and News Skills. Newspaper Research Journal 29(2): 23–39.
scholarly phenomenon. Information, Communication, & Gillespie T (2010) The politics of ‘platforms’. New Media &
Society 15(5): 662–679. Society 12(3): 347–364.
Brayne S, Rosenblat S and boyd d (2015) Predictive Policing. Gottfried J and Shearer E (2016) News Use Across Social
New York, NY: Data & Society. Available at: http://www. Media Platforms 2016. Washington, DC: Pew Research
datacivilrights.org/pubs/2015-1027/Predictive_Policing. Center. Available at: http://www.journalism.org/2016/05/
pdf (accessed 10 October 2016). 26/news-use-across-social-media-platforms-2016/
Broussard M (2015) Teaching Coding in Journalism Schools: (accessed 10 October 2016).
Considerations for a Secure Technological Infrastructure. Green JE (2008) Max Weber and the Reinvention of Popular
Computation þ Journalism, October 2–3, 2015, New York, Power. Max Weber Studies 8(2): 187–224.
NY. Herrman J (2016) Inside Facebook’s (Totally insane,
Christin A (2014) When it Comes to Chasing Clicks, Unintentionally gigantic, Hyperpartisan) Political-Media
Journalists Say One Thing but Feel Pressure to do Machine. The New York Times Magazine. Available at:
Another. Cambridge, MA: Nieman Lab. https://www.nytimes.com/2016/08/28/magazine/inside-
Caplan and boyd 11

facebooks-totally-insane-unintentionally-gigantic-hyper- gizmodo.com/want-to-know-what-facebook-really-thinks-
partisan-political-media-machine.html (accessed 17 of-journalists-1773916117 (accessed 15 October 2016).
October 2016). Oremus W (2016) Who controls your Facebook feed. In:
Kafka P (2014) Upworthy’s traffic is still headed down. Slate. Available at: http://www.slate.com/articles/technol-
Blame us, not Facebook, says Upworthy. In: Recode. ogy/cover_story/2016_01_how_facebook_s_news_feed_
Available at: http://www.recode.net/2014/5/14/11626848/ algorithm_works.html (accessed 15 October 2016).
upworthys-traffic-is-still-headed-down-blame-us-not-face- Osofsky J (2016) Information about trending topics. In:
book-says. Facebook Newsroom. Available at: http://www.news-
Knuth D (1968) The Art of Computer Programming. Boston, room.fb.com/news/2016/05/information-about-trending-
MA: Addison-Wesley. topics/ (accessed 17 October 2016).
Kroll JA, Huey J, Barocas S, et al. (2017) Accountable algo- Owen T (2015) Disruptive Power: The Crisis of the State in the
rithms. University of Pennsylvania Law Review 165: 633– Digital Age. Oxford: Oxford University Press.
704. Pasquale F (2015) The Black Box Society: The Secret
Lessig L (1999) Code and Other Laws of Cyberspace. Basic Algorithms that Control Money and Information.
Books. Cambridge: Harvard University Press.
McArdle M (2014) Facebook puts a downer on upworthy. In: Petre C (2015) The traffic factories: Metrics at Chartbeat,
Bloomberg. Available at: https://www.bloomberg.com/ Gawker Media, and The New York Times. In: Tow
view/articles/2014-02-11/facebook-puts-a-downer-on- Center for Digital Journalism. Available at: http://www.
upworthy (accessed 10 October 2016). towcenter.org/research/traffic-factories (accessed 17
McQuail D (1992) Media Performance: Mass Communication October 2016).
and the Public Interest. Thousand Oaks, CA: Sage. Podolny S (2015) If an algorithm wrote this, how would you
Meyer J and Rowan B (1977) Institutionalized organizations: even know?. The New York Times. Available at: https://
Formal structure as myth and ceremony. American Journal www.nytimes.com/2015/03/08/opinion/sunday/if-an-algo-
of Sociology 83(2): 340–363. rithm-wrote-this-how-would-you-even-know.html
Meyer R (2013) Why are upworthy headlines suddenly every- (accessed 15 November 2016).
where? The Atlantic. Available at: http://www.theatlantic. Reckhow M (2015) Introducing Instant Articles. Facebook
com/technology/archive/2013/12/why-are-upworthy-head- Media. Available at: https://media.fb.com/2015/05/12/
lines-suddenly-everywhere/282048/ (accessed 12 October instantarticles/ (accessed 15 October 2016).
2016). Tandoc EC (2014) Journalism is twerking? How web ana-
Meyer R (2016) Facebook purges journalists, immediately lytics is changing the process of gatekeeping. New Media
promotes a fake story for 8 hours. The Atlantic. & Society 16(4): 559–575.
Available at: http://www.theatlantic.com/technology/ Tandoc EC Jr and Ferrucci PR (2017) Giving in or giving up:
archive/2016/08/facebook-steps-in-it/497915/ (accessed 15 What makes journalists use audience feedback in their
October 2016). news work? Computers in Human Behavior 68: 149–156.
Moses L (2016) Uh-oh, some publishers see a drop in Sandvig C, Hamilton K, Karahalios K, et al. (2014) Auditing
Facebook traffic. In: Digiday. Available at: http://www. algorithms: Research methods for detecting discrimination
digiday.com/publishers/publishers-just-saw-decline-face- on internet platforms. In: Data and discrimination: convert-
book-traffic/. ing critical concerns into productive inquiry (preconference
Mosseri A (2016) News feed FYI: Addressing hoaxes and at international communications association ‘14), 22 May
fake news. In: Facebook Newsroom. Available at: 2014, Seattle.
https://newsroom.fb.com/news/2016/12/news-feed-fyi- Saurwein F, Just N and Latzer M (2015) Governance of algo-
addressing-hoaxes-and-fake-news/ (accessed 15 January rithms: Options and limitations. info 17(6): 35–49.
2017). Schudson M (1987) Discovering the News: A Social History of
Napoli PM (2014) An institutional theory perspective on American Newspapers. New York, NY: Basic Books.
algorithmic media production and consumption. Scottberg E (2013, October 21) Facebook continues to be the
Communication Theory 24(3): 340–360. biggest driver of social traffic. In: SimpleReach Blog.
Napoli PM (2015) Social media and the public interest: Available at: http://blog.simplereach.com/facebook-con-
Governance of news platforms in the realm of individual tinues-to-be-the-biggest-driver-of-social-traffic/ (accessed
and algorithmic gatekeeprs. Telecommunications Policy 39: 15 October 2016).
751–760. Seaver N (2017) Some tactics for the ethnography of
Napoli P and Caplan R (2017) When Media Companies Insist algorithmic systems. Big Data & Society. Epub ahead of
They’re not Media Companies, Why They’re Wrong and print 2017. Available at: https://doi.org/10.1177/
Why That Matters. First Monday 22(5). 2053951717738104.
Newton C (2017) Instant recall. In: The Verge. Available at: Silverman C (2016) This Analysis shows How Viral Fake
https://www.theverge.com/2017/4/16/15314210/instant- Election News Stories Outperformed Real News.
articles-facebook-future-ads-video (accessed 15 August BuzzFeed News. Available at: https://www.buzzfeed.
2017). com/craigsilverman/viral-fake-election-news-outper-
Nunez M (2016) Want to Know What Facebook Really formed-real-news-on-facebook?utm_term¼.
Thinks of Journalists? Here’s What Happened When It drE8WWBDW1#.aa17xxwKxa (accessed 15 January
Hired Some. Gizmodo.com. Available at: https:// 2017).
12 Big Data & Society

Sobel Fitts A (2015) When metrics drive newsroom culture. Wu T (2010) Is internet exceptionalism dead? In: Szoka B and
Columbia Journalism Review. New York. Available at: Marcus A (eds) The Next Digital Decade – Essays on the
https://www.cjr.org/analysis/how_should_metrics_drive_ Future of the Internet. Washington, DC: TechFreedom.
newsroom_culture.php (accessed 17 October 2016). Zuckerberg M (2016) I want to share some thoughts about
The New York Times (2014) 2015 Annual report. The New trending topics. Available at: Facebook/Mark Zuckerberg
York Times. Public Posts: https://www.facebook.com/zuck/posts/
Vu HT (2013) The online audience as gatekeeper: The influ- 10102830259184701?pnref¼story (accessed 10 October
ence of reader metrics on news editorial selection. 2016).
Journalism 15(8): 1094–1110. Zuckerberg M (2017) Building global community. In:
Weber M (1978) Economy and Society. Berkeley: University Facebook.com. Available at: https://www.facebook.com/
of California Press. notes/mark-zuckerberg/building-global-community/
Wong D (2015) In Q4, social media drove 31.24% of overall 10154544292806634 (accessed 14 July 2017).
traffic to sites. Report. In: Shareholic. Available at:
https://blog.shareaholic.com/social-media-traffic-trends-
01-2015/ (accessed 15 October 2016).