5   How Free Is the Networked Press?

What kinds of separations and dependencies organize the contemporary networked press, and what sociotechnical dimensions of autonomy emerge from them? That is, if press freedom has always been a balance of positive and negative freedoms and the press’s power to convene publics has taken an infrastructural turn, what kind of autonomy does today’s networked press have—and what kinds of publics can it make? By examining the trade press on networked journalism—essentially, the stories that the press tells itself about itself—I find twelve sociotechnical dimensions of networked press freedom and offer a typology of sociotechnical relationships.

In previous work, I examined how the networked press emerges from sociotechnical spaces that are neither entirely journalistic nor exclusively technological but a hybrid of the two. Kate Crawford and I (Ananny & Crawford, 2015) traced how mobile news app designers see themselves as part of what we called a “liminal press” that is somewhere between algorithm design and editorial decision making. Leila Bighash and I (Ananny & Bighash, 2016) examined how online news organizations reveal their public service mandates when they periodically choose not to use paywall technologies to meter news. I showed how the sociotechnical ethics of witnessing and Roger Silverstone’s (2007) ideal of “proper distance” emerge from Google Glass infrastructure and design communities (Ananny, 2015), how the networked press’s rhythms, deadlines, and timelines emerge from a mix of social and technological forces (Ananny, 2016b), and how news organizations’ social media policies (Ananny, 2014) and application programming interfaces (Ananny, 2013) leave clues about how they see their work as both free from and reliant on audience participation. In each individual instance, the press was not found to be a set of journalistic actors pitted against software designers and their technologies. Rather, it was found to be a co-construction—a space of institutionalized communication that both depends on and distances itself from the people, norms, practices, and tools that live in an increasingly hybrid middle space between journalism and software.

There is a larger, field-level story to tell, though, about what kind of press autonomy exists in these hybrid spaces—what kind of networked press autonomy is emerging from these separations and dependencies. How does the press itself understand the field-level tradeoffs it finds itself making? Where does journalism see itself intersecting with technology, and what is at stake in these intersections? If the networked press’s autonomy depends on—and exists to protect—its power to arrange human and nonhuman actors into configurations that allow new material publics to emerge, then we need to know where these configurations are, how they work, and what they assume. One way to understand the networked press’s understanding of its own sociotechnical configuration is to look at how journalists describe their work among themselves and the conditions under which they say they do it.

Matt Carlson (2015b) calls this journalism’s “metajournalistic discourse.” Starting from the “premise that journalism should be understood as a cultural practice that is embedded in specific contexts, variable across time and space, and inclusive of internal and external actors” (p. 2), he shows how journalism’s significance exists, in part, in stories about journalism. Such narratives show how “various actors inside and outside of journalism compete to construct, reiterate, and even challenge the boundaries of acceptable journalistic practices and the limits of what can or cannot be done” (p. 1). Indeed, although scholars have not reflected a great deal on the use of such coverage in academic accounts of media industries (Wilkinson & Merle, 2013), communication researchers have a history of studying trade press and journalistic self-coverage to understand where a technology industry thinks it is going and how journalists understand their own role in institutional upheavals (Liao, 2014; Napoli, 1997, 2011; Turow, 1994). Although such accounts can give limited and ritualistic images of media industries—falling prey to the same patterns that constrain and standardize journalism’s coverage of other domains—and should not be mistaken for deep engagement with industry practices and actors (Wilkinson & Merle, 2013), they can act as “semi-embedded deep texts” that both “bring generalizing discussions of the nature and meaning of … production from one corporate media company or craft to another” and spur “discussion and eventual awareness in the public sphere of the consumer as well” (Caldwell, 2008, p. 203). These stories suggest how the press understands itself and how those outside of journalism might understand the field.

With these strengths and limitations in mind, I delve deeply into stories that the networked press tells about itself and to itself. I mine them to see how networked press freedom—as a system of sociotechnical separations and dependencies that gives rise to material publics—is being worked out in industry narratives of how and why journalism and technology intersect. (See the appendix for a discussion of the method that I used to build and analyze the corpus of trade press stories.) Within this chapter’s analysis, whenever one of the primary sources is referenced, I provide an endnote to the URL of the corpus primary source. These dimensions of press freedom emerge from evaluating the primary sources as metajournalistic discourse, not as technical descriptions. Journalists, technologists, and scholars may disagree with how primary sources understand these systems and their significance. I treat them as stories that journalists tell each other—what they think systems do and mean—that, taken together, paint a picture of contemporary networked press freedom.

Dimensions of Networked Press Freedom

So far in the book I have developed three interrelated claims—that the networked press is infrastructural relationships among human and nonhuman actors, that it derives its normative legitimacy partly from its ability to realize publics, and that the power to realize publics requires the freedom to create separations and dependencies among its human and nonhuman actors. I do not argue that the networked press needs to use its autonomy to create any particular type of public (as chapter 3 shows, publics are always normative tradeoffs). But its power to create material publics that emerge from relationships among human and nonhuman actors and not just from what people alone find interesting or relevant depends on how well it understands its own sociotechnical separations and dependencies—that is, how its own arrangements of human and nonhuman actors make it more or less likely that a public right to hear is realized.

But what are these arrangements, what types of separations and dependencies do they entail, and what type of public right to hear do they suggest? My aim here is to use stories that the networked press tells about itself as evidence of what Lisa Parks (2012) calls “infrastructural intelligibility”—the “process by which ordinary people use observations, images, information and technological experiences to infer or imagine the existence, shape or form of an extensive and dispersed infrastructure that cannot be physically observed by one person in its entirety” (p. 67).

Following the empirical approach described above, I identify twelve ways that the networked press reflects its institutional autonomy through stories about its sociotechnical infrastructure:

Observation: Watching social spaces for relevant people, events, and informational patterns.

Production: Creating the stories, representations, and interactions that circulate as news.

Alignments: Negotiating with platforms that deliver news and create the conditions under which people engage with news.

Labor: Defining operational roles and managing workforces.

Analytics: Measuring, describing, and creating abstracted accounts of networked news environments.

Timing: Setting and influencing the rhythms of networked news environments.

Security: Controlling conditions under which sources, journalists, and audiences are visible and protected.

Audiences: Defining news consumers and controlling their participation in networked news environments.

Revenue: Commodifying information and people to support networked news environments financially.

Facts: Defining the conditions under which information is seen as stable and legitimate.

Resemblances: Controlling the languages, media simulations, and international categories that make news seem familiar or foreign.

Affect: Setting stylistic and affective conditions of networked news environments.

I define each, offer examples from the journalistic trade press illustrating its sociotechnical separations and dependencies, and reflect on its stake in realizing a public right to hear. This purposefully exploratory and somewhat speculative approach does not claim to be a comprehensive image of the networked press but instead offers a typology of its dynamics to show how its autonomy is inseparable from its sociotechnical infrastructures.

Observation

Much of news work involves watching environments or beats where newsworthy people are expected to be and newsworthy events are expected to happen. In many ways, the networked press is no different, but its beats are sociotechnical observation systems of humans and nonhumans that scan environments, anticipate events, sense patterns, and watch locations. A mix of journalists and technologies perform acts of observing, and the sites they observe are hybrid spaces of people and digital infrastructure.

Observation takes one form in sociotechnical systems that news organizations have built to monitor social media.1 For example, in 2014, the New York Times created a section called “Watching” that both automatically and manually observes headlines, tweets, and multimedia streams from around the social Web, giving readers “a carefully filtered window into rest of the world of news, all with the NYT stamp of approval.”² Similarly, in 2013, ProPublica built a tool that lets its journalists monitor Instagram content for particular times and geographic coordinates (letting them see, for example, photos of the Boston Marathon finish line just before and after it was bombed³), and the New York Times, BuzzFeed, Gothamist, and the New York World were able to put reporters on the scene of a 2014 explosion in East Harlem because they were all using the CityBeat⁴ program to search social media metadata algorithmically and be alerted to events it deemed newsworthy.⁵

Journalists are also using data tools to anticipate stories they think might happen. The Anomaly Tracker tool developed by the Center for Responsive Politics watches databases of lawmaker votes and campaign contributions and triggers alerts if anything happens that is considered out of the ordinary. The Center suggests extraordinary events to journalists, highlighting events that it sees as suspicious and that its database can observe. Derek Willis, “when he was at the New York Times, used to have an alert any time there was a vote in Congress after midnight. ‘As I used to tell my kids, nothing good happens after midnight.’ He had that programmed.”⁶

ProPublica is a leader in data-centered reporting—making databases as part of reporting, reporting on databases, and creating databases that other organizations can use to find stories. For example, its Dollars for Docs database tracks payments that doctors receive from drug and medical device companies.⁷ Its School Restraints project not only produced a publicly accessible database of incidents in which children were physically restrained or secluded but also extensive guidance to other journalists on how to use the database and understand its limitations.⁸ Finally, its Data Store project serves as a clearinghouse of raw datasets from government sources (available for free), hand-made vetted databases that it creates or acquires during its reporting (available for a fee), and application programming interfaces that provide regularly updated access to fresh data that can be queried according to rules it sets.⁹ ProPublica is thus not only a data-driven news organization but a leading creator of what Michael Schudson (2010) calls “political observatories”—data collections that suggest, constrain, and empower journalistic work.

Finally, news organizations looking to observe geographic locations with drones must grapple with sociotechnical conditions.¹⁰ A number of news organizations—including the New York Times, the Washington Post, NBCUniversal, the Associated Press, and Getty Images—are experimenting with using drones in their reporting and are learning about the regulatory and technological limits of drone reporting.¹¹ The Federal Aviation Administration (FAA) has issued rules on commercial drone usage,¹² but there are currently no special allowances made for journalists or news organizations.¹³ Indeed, the company NoFlyZone maintains, for drone manufacturers, what it defines as a “comprehensive airspace database of critical infrastructure and sensitive sites” that include “airports, hospitals, schools, nuclear power plants, prisons” and other places it calls “sensitive locations.”¹⁴ What journalists are allowed to see with drones can be sociotechnically controlled. For example, databases can effectively create digital fences around sensitive locations (after one crashed at the White House, drones made by DJI now automatically land if they approach government locations),¹⁵ and law enforcement officials can pressure the FAA to create no-fly zones that prevent coverage of protest events (as the St. Louis County Police Department did when it wanted to prevent journalists from using drones to cover the Ferguson, Missouri, protests).¹⁶

The networked press uses a variety of sociotechnical infrastructures to scan social media environments, anticipate newsworthy events, sense data patterns, and watch geographic locations. Its freedom to observe is not simply about human journalists gaining unfettered access to data or to places. Rather, its freedom requires the shared sociotechnical capacity to query social media and understand what such queries can and cannot show, appreciate the limits of what algorithms are able to anticipate and what anomalies databases are primed to reveal, and see physical locations as long as technologies index them and governments allow such indexing. The public right to hear—to discover new publics—is inextricable from the networked press’s freedom to configure the sociotechnical separations and dependencies that enable observation.

Production

The networked press is also increasingly inseparable from infrastructures of humans and nonhumans that produce news together.17 Such infrastructures let journalists tell or publish stories faster, and they sometimes use “structured data”¹⁸—collections of highly structured and predictably organized information—to semiautonomously create content that then circulates in ways that are often indistinguishable from news created entirely by humans. The trade press accounts of such news production infrastructures center on issues of authorship, interaction, and tailoring.

Three of the most highly discussed authorship infrastructures¹⁹ are the Wordsmith platform that the Associated Press (AP) developed in partnership with Automated Insights to tell financial stories,²⁰ the Narrative Science Quill system that ProPublica used to report on educational data,²¹ and the QuakeBot system that the Los Angeles Times uses to report data from the United States Geological Survey.²² It is beyond the scope of this book to discuss these systems in detail, but they share a common process: journalists use their news judgments and domain-specific knowledge to create algorithmic processes that turn predictable, highly structured data—investment, educational, seismic—into human readable narratives. Authorship is a mix of human judgment, editorial tradition, data standardization, and algorithmic processing—with early research suggesting that people perceive “computer-written articles as more credible and higher in journalistic expertise but less readable” (Graefe, Haim, Haarmann, & Brosius, 2016, p. 1) and that algorithmically produced financial stories result in a greater number of trades.²³

Another form of infrastructural production exists in the form of news organizations’ semiautomated interactions with audiences—the creation of bots programmed to deliver news through interactive conversation-like experiences. For example, the New York Times used the Facebook messenger platform to create a bot that engaged readers conversationally about the 2016 presidential election, sending both prescripted messages and responses to users’ queries for electoral predictions and poll data.²⁴ The Washington Post is similarly using the Facebook platform to develop a conversational news bot,²⁵ and CNN²⁶ and NBC²⁷ have both built applications for the Kik bot platform designed to engage younger users with content through conversation-like prompts and story recommendations based on users’ responses. Quartz recently devoted its entire mobile news app to a chat experience that gives users “teases to stories in the form of chat bubbles, and then lets [them] select whether they want to get more information about a topic or move onto the next story.”²⁸ (In November 2016, the Knight Foundation awarded Quartz $240,000 to further develop its “bot studio” and to create apps that deliver news through conversations with Slack messenger, Amazon Echo, and Google Home).²⁹

The final area in which news production is intersecting with technological infrastructures is tailoring. News organizations recommend existing stories to readers and also create entirely new stories aligned with what their tracking systems record and different media require. For example, the Washington Post created personalized email newsletters with content related to stories that it observed users engaging with previously.³⁰ It also created a personalized reading experience called Re-Engage that tracks how readers interact with a mobile story, guesses when readers may be getting bored, and suggests at the optimal times alternative stories designed to keep that reader from closing the app.³¹ And the New York Times created healthcare policy stories personalized for where it sensed readers were located (“Consider Los Angeles, our best guess for where you might be reading this article”).³² The experience of having the middle of a news story personalized was so unnerving that it prompted the Times’s public editor to muse whether “personalization could deprive readers of a shared, and expertly curated, news experience, which is what many come to The Times for.”³³ Tailoring also appears as systems that automate translating a story from one medium to another. The AP created a system to solve the “manual labor nightmare” of translating a story written for print that needs to be rewritten for broadcast: “Stories are shorter, sentences are more concise, attribution comes at the beginning of a sentence, numbers are rounded.”³⁴ The system is designed to give reporters more autonomy. Because relying on it frees them from tasks seen to be mundane and routine, embedded in such separations and dependencies are visions of what reporters are supposed to do.

The networked press does not use technologies only to accelerate the production or dissemination of stories. Through authorship, interaction, and tailoring infrastructures, it also blends humans and nonhumans into assemblages that speak to, converse with, and target audiences. And such infrastructures are fallible. Wordsmith erroneously reported a drop in Netflix’s stock price (it had undergone a split, but the algorithm was not programmed to recognize this), and as Microsoft’s crowdsourced chat bot Tay showed when it gave racist, Holocaust-denying replies to users’ queries (reflecting what the crowd had taught its machine learning algorithm), designing artificially intelligent conversational agents with integrity is difficult.³⁵

Autonomy does not need to mean freedom from authorship, interaction, and personalization infrastructures, which can be valuable ways to let journalists focus on less mundane stories, create relationships with audiences at scale, and help people see different ways that issues are personally relevant to them. Rather, autonomy means engaging with infrastructures of production to ask questions about data and audiences: What types of data are algorithmic stories relying on, and does the data meet editorial standards? When and why should news organizations engage audiences in conversations (Schudson 1997 argues that conversations do not always serve democracy)³⁶ versus other modes of communication? And when do personalized stories let people retreat into private domains of concern versus help them see the inextricable conditions and consequences of shared public life?

Alignments

The networked press also increasingly needs to align with “sites and services that host public expression.” Such platforms do not “make [news] content, but they make important choices about that content: what they will distribute and to whom, how they will connect users and broker their interactions, and what they will refuse” (Gillespie, 2017, p. 255). Although companies like Facebook often eschew their roles as media companies (Napoli & Caplan, 2017)37 and people find news through a variety of different media, audiences are increasingly using such platforms to access and share news (Bright, 2016; Gotfried & Shearer, 2016; Trilling, Tolochko, & Burscher, 2017).

News organization alignments with platforms discussed in the trade press take the form of partnerships (platform-centered strategic publishing agreements), resemblances (news that looks like the content that platforms find valuable), and breakdowns (moments when infrastructural meetings show how platform priorities and editorial values can collide and cause discord).

Partnerships   Much of the aligning described in the trade press comes in the form of guidance and collaboration agreements between news organizations and social media platforms. Facebook,³⁸ Google,³⁹ and Twitter⁴⁰ offer instructions on how news organizations should create and format news in order to align with the strategies that platforms use to find, rank, and suggest content. With other collaborations, platforms commission original content from news organizations: Facebook essentially hires news organizations to provide exclusive content for its Facebook Live channels,⁴¹ and Twitter partnered with the BBC World Service’s Global News Podcast to embed exclusive, sponsored video content within promoted tweets as part of its Amplify program.⁴²

Other partnerships focus on creating what the trade press calls “distributed content”⁴³—designing special formats and subareas of platforms specifically for news content. For example, in exchange for housing content on Facebook’s servers, news organizations can join Facebook’s Instant Articles program (originally for only a select group of publishers⁴⁴ but now open to all⁴⁵) to load mobile content faster, reach advertising markets, and access exclusive interactive features.⁴⁶ (The New York Times demonstrated the value of distributing content around the Web before the Instant Articles program went live when it posted full stories to its Facebook page after its own website briefly went down.)⁴⁷ Google’s Accelerated Mobile Pages project similarly offers publishers access to faster loading times, analytics tools, and interactive and revenue-gathering features, without requiring news organizations to place their content on Google servers;⁴⁸ Twitter’s Project Lightning created Moments to let invited news organizations create digestlike summaries of tweets that they see as related to a particular news topic;⁴⁹ and several news organizations (from “BuzzFeed and Mashable to ESPN and National Geographic”⁵⁰) produce custom content exclusively for Snapchat’s Discover platform.

Finally, other partnerships focus not on producing content but on aligning customer bases and production work. For example, Amazon Prime subscribers get six months of free access to the Washington Post (owned by Amazon chief executive officer Jeff Bezos),⁵¹ and CNN collaborated with Twitter to produce a custom tool for letting it sort and publish tweets in near “real-time.”⁵²

Resemblances   Other forms of news-platform alignment are less programmatic. They are instead ongoing sociotechnical dances in which news organizations create content that they think platforms favor and try to adjust their production processes quickly when they see platforms changing. This is understandable: different platforms attract different types of readers, and once there, people read, share, and comment on news differently depending on the platform.⁵³

As with partnerships, the trade press describes different types of resemblance work. In the most extreme versions, news organizations decide to adopt platform formats entirely—as the Wall Street Journal did when it created a Facebook-only app in 2011,⁵⁴ as Vox did when it sequestered its Circuit Breaker within Facebook instead of as a separate Vox website,⁵⁵ and as many news organizations are currently doing as they create Flash Briefings customized for Amazon’s Alexa voice interface.⁵⁶ In other circumstances, news organizations tailor stories for technologies. The Verge did so when it showed different versions of a story to Apple versus Google versus Microsoft users,⁵⁷ the New York Times did so when it heavily peppered a story with “tweetable phrases” to make it easy for Twitter users to comment on and share the story,⁵⁸ and the Guardian does so when it hovers a Twitter icon above any story text that readers highlight, encouraging them to tweet the passage directly from its site.

News organizations can be rewarded greatly for such resemblances. For example, news stories that appear as Facebook Instant Articles are shared more frequently and widely than other types of news links.⁵⁹ And National Public Radio (NPR) drew attention to local stations’ stories when it used Facebook’s location-based system suggestions⁶⁰ to geo-index stories it thought had local relevance.⁶¹ Local traffic increased when Facebook’s way of encoding “local” aligned with NPR’s definition of a geographic news community.

When platforms change how they parse and rank content, news organizations pay attention and try to respond as quickly as possible. Facebook changed its News Feed algorithm in 2013 to limit what it called low-quality sites focused on spreading memes,⁶² in 2014 to increase what it saw as partisan political sources,⁶³ and in 2016 to surface more content about “friends and family”⁶⁴ and fight what it defined as clickbait.⁶⁵ Although some news sites saw traffic increases in 2013,⁶⁶ the trade press debated⁶⁷ whether the change harmed legitimate sites like Upworthy that tried to use memelike formats as a way of disseminating quality news. (Facebook does not release detailed statistics on such changes, so the trade press is left with anecdotal evidence of their effect on the industry as a whole. Mother Jones saw its traffic increase as a result of aligning with Facebook’s 2014 change,⁶⁸ but so did Mental Floss, a site that shares facts in memelike formats.)⁶⁹ Some news sites try to reduce their reliance on Facebook, as Bleacher Report did when it instead focusing on a partnership with Snapchat.⁷⁰

As often happens with infrastructure, some resemblances appear in sociotechnical forms that are largely invisible to end users or constrained to predefined categories. For example, Google News’s Spotlight architecture gives prominence to stories its algorithms see as “in-depth pieces of lasting value,”⁷¹ offers a “standout” metadata tag to let news organizations suggest pieces that they feel are high quality,⁷² and asks news organizations to use a finite set of Google-defined genre tags (press release, satire, blog, op-ed, opinion, user-generated) to categorize stories for the search engine.⁷³ Other platforms also mediate their interactions with news organizations through algorithmic architectures: Facebook’s Stories to Share feature alerts news organizations to content that Facebook thinks would circulate well but that it senses news organizations have not yet shared,⁷⁴ and Twitter has automatically promoted news articles that it determines have tweets embedded within them.⁷⁵

Breakdowns   Finally, news-platform alignments often are most visible during moments of breakdown—when cooperation agreements or sociotechnical relationships seem to fail.⁷⁶ Such moments are diagnostics of networked press freedom because they reveal limits on news-platform alignment. We hear calls for the networked press to create more separation from technology companies, and technology executives have tried to distance themselves from news organizations by focusing on the supposed neutrality of their algorithms, disavowing editorial responsibility, promising to do better, but ultimately claiming that the days of hand-curated news experiences are over.⁷⁷

There is no shortage of examples of such breakdowns, and they involve censoring or reformatting news or banning organizations and individuals. Some failures are accidental and quickly reversed, others are the product of platform policies and take considerable effort to overturn, others are unintended and inexplicable, and others are purposeful and defended. For example, in 2013, Twitter accidentally labeled an entire Philadelphia newspaper as spam and would not allow users to access the paper’s site through tweets.⁷⁸ Reporters protested and used Tumblr instead, and Twitter apologized and reinstated the account about a week later. Also in 2013, Facebook deleted several news organizations’ stories about same-sex marriage; the organizations received no explanation, and Facebook did not say why the stories had been removed.⁷⁹

In 2016, Facebook censored two different news sites after news stories containing images of naked bodies were found to violate the site’s community standards. In the first case, activist Celeste Liddle shared a story from the news site New Matilda that contained an image of Aboriginal women in ceremonial outfits that showed their breasts. The post was deleted, and Liddle’s account was suspended. Facebook claimed that it had to maintain “uniformity in global policies”—although New Matilda showed images of nude U.S. models published in Esquire magazine that were circulating uncensored on Facebook. In the second case, Facebook censored a Norwegian writer’s post containing Nick Ut’s Pulitzer Prize–winning “iconic photo of a girl, screaming and naked, fleeing napalm bombs during the Vietnam War.”⁸⁰ Facebook claimed that the post violated its community standards but eventually relented after the Norwegian newspaper Aftenposten reported on the suspension and included the story in its report; the paper had its own story removed and received a warning not to display nude images; the paper wrote an open letter claiming that Facebook was censoring the news; the leader of the Norwegian conservative party had her post removed for including the photo; and the Norwegian prime minister also posted the photo in protest over Facebook’s actions. The company eventually decided the photo had iconic, newsworthy status.⁸¹ Both the New Matilda and Aftenposten incidents highlighted the incompatibility of the platform’s community standards and news organizations’ editorial judgments and the idiosyncratic nature of how breakages are perceived and allowances for particular cases are made.⁸² (In a different moment fueling trade press perception that technology company standards are incompatible with news values, shortly after Amazon CEO Jeff Bezos bought the Washington Post, the paper briefly displayed Amazon Buy It Now links within the article texts—a move that was widely ridiculed by journalists claiming that Bezos was turning the Post into a marketing arm of Amazon.)⁸³

Although platforms usually claim that such incidents are unfortunate side effects of running global sites that serve massive scales of user-generated content in near real time and cannot be manually curated, sometimes a platform stands by its news censorship. In 2014, although some news organizations were showing the gruesome footage of the beheading of journalist James Foley, Twitter banned any accounts that displayed images of the murder, a move that many commentators said showed Twitter’s power to override editorial judgment selectively and purposefully.⁸⁴

Across different types of alignments—official partnerships, sociotechnical resemblances, and value-based breakdowns—news organizations and social media platforms find themselves in intricate and often inseparable relationships. Instead of seeing press freedom as the journalistic power to avoid aligning with platforms, we might trace what types of publics such alignments and breakdowns assume and create. Partnership agreements, sociotechnical resemblances, and failed collaborations show how the power is distributed unevenly between news organizations and technology companies. The networked press might demonstrate its need for autonomy—and what it could do with institutional freedom—by designing collaborations that ensure a public right to hear.

Labor

In his thoughtful study of digital news work, Matthew Powers (2012) observes that “Changes in the technologies of news production do not simply modify journalistic practices; they also introduce what might be considered technologically specific forms of work” (p. 24, emphasis in original). Such work, as we found in our study of mobile news app designers, is neither exclusively digital nor editorial but helps shape a “liminal press” (Ananny & Crawford, 2015) that both reifies and questions technological and journalistic traditions. If networked press freedom exists, in part, through envisioning and supporting different kinds of labor, what we see in the trade press are myriad initiatives, communities, and roles that attempt to balance separations and dependencies between journalists and technologists. What emerges is a complex story of how journalists both emulate and critique technological labor.

Some of these balances exist in entire news organizations centered on computational concepts and labor. Circa was founded on an “object-oriented” vision of news work that deconstructs the idea of the traditional news article, and it hired journalists who could think in “software development terms like forking and refactoring” in order to create a set of facts, stats, quotes, events or images that can be assembled and reassembled in the service of a news narrative.85 The company is one of several news organizations trying to create blended workforces that see news work as the creation and manipulation of structured content.⁸⁶ This approach to news work is both relatively new (connected to software engineering notions of reusable and easily extensible code) and quite old (rooted in longstanding journalistic forms like the inverted pyramid, the standalone lede, and editing stories from the bottom). The trade press is filled with stories of people in these hybrid workforces reflecting on their identities as both journalists and coders,⁸⁷ advocating for others to develop hybrid skills,⁸⁸ and debating whether journalists should think of themselves as coders with software engineering skills or as general builders with the ability to tinker whenever their journalistic instincts suggest doing so.⁸⁹ The New York Times even built an in-house computational tool to track its own staff’s Web browsing in an attempt to trace and summarize reporters’ intelligence work automatically and avoid duplication among workers.⁹⁰

These debates occurred alongside the rise of professional organizations designed to blend editorial and technological mindsets and offer networking opportunities for this new labor market. Akin to what Fred Turner (2005) calls “network forums”—entities that simultaneously convene and represent diverse people who may otherwise share little—the trade press contains many mentions of new groups designed to facilitate hybrid labor. Hacks/Hackers was founded to “help journalists (hacks) and developers (hackers) learn from each other and collaborate on projects,”⁹¹ the Knight-Mozilla OpenNews group was designed to “strengthen the bonds between the worlds of journalism and software development”⁹² and to execute open source journalism projects,⁹³ and the OpenNews initiative Source was designed to bring “wandering journonerds together under one roof” to share code and best practices.⁹⁴

Newsrooms and platform companies alike trade in hybrid labor. When Joanna Geary left the Guardian to lead Twitter’s UK news partnerships⁹⁵ and Liz Heron stepped down from head of news partnerships for Facebook to become HuffPost’s executive editor,⁹⁶ several trade press stories reflected on increasingly blurred differences between news and platform companies. The Wall Street Journal, Condé Nast, Vox Media, and the New York Times have hired platform specialists and platform relationship managers to be liaisons with technology companies and to advocate for social media forms of storytelling within newsrooms.⁹⁷ And although they eschew their roles as media companies, both Facebook⁹⁸ and Twitter⁹⁹ have created positions that are designed to lead partnerships with news organizations and that call for several years of experience within newsrooms. A controversial aspect of such shifts has been the question of whether journalists who have built social media brands at one news organization or platform should be able to take them to a new job¹⁰⁰—whether a journalist should be able to have a portable brand that travels easily between competitors and among editorial and technology contexts.

Visions of journalistic versus technological labor also play out in discussions about how and why to train the next generation of hybrid workers. There are calls for universities to change how they teach journalism students,¹⁰¹ a Facebook program to teach new journalists how to report using the platform,¹⁰² a focus in journalism schools on teaching students how to produce for social media,¹⁰³ scholarships to help software programmers learn about journalism¹⁰⁴ and journalism students learn to build technology projects,¹⁰⁵ an “in-house journalism school” at the New York Times designed to train future digital reporters,¹⁰⁶ and a traveling course where ProPublica and Times journalists teach technology skills to journalists in smaller media markets.¹⁰⁷ Each initiative is different, but they share an assumption that future journalists must have some literacy with technological concepts and systems. There is no sense that press freedom means sequestering journalism students from computational thinking.

Finally, there is evidence of some tension in this ideal of news organizations and platform companies educating the next generation of journalists together. When Gizmodo exposed the inner labor dynamics of the team behind Facebook’s Trending Topics product, the trade press questioned whether news-technology partnerships are truly equitable and what power individual journalists have in the face of massively funded technology companies. Gizmodo describes how Facebook hired approximately a dozen journalists—people who had previously worked at the New York Daily News, Bloomberg, MSNBC, and the Guardian—to be “news curators” who decided what should appear on the Trending Topics part of Facebook, wrote short summaries of topics, and chose relevant media. After the team was disbanded, Gizmodo reported that Facebook managers told the curators not to list on their resumes that they had worked on Trending Topics so that the appearance that the list was algorithmically curated could be maintained.¹⁰⁸ Writing for Forbes, a former curator said that the team was always given the impression that their job “was essentially to train an algorithm to write the news.”¹⁰⁹ The incident trigged a renewed discussion about how journalistic labor was valued by technology platforms and whether the much-heralded claims of partnership and cooperation were actually hiding technologists’ more insidious goal of studying journalistic labor only for the purpose of algorithmically modeling it.

Networked press freedom thus is emerging not as neat separations in which journalistic work or education is distinct from the goals and practices of technology companies. Nor do we see evidence of journalists having the power to depend selectively on the parts of computational labor that they see aligning with editorial goals. Rather, a messier and more complex interplay between traditional ideals of journalistic work and newer forms of technological work is developing in professional organizations, career moves, educational initiatives, and perceptions of exploitation.

Analytics

Several academic studies have analyzed how newsrooms using Web analytics—quantified information about how content travels online—to shape news work, imagine audiences, and drive editorial policy (Hanusch, 2016; MacGregor, 2007; Petre, 2015; Tandoc, 2014, 2015; Tandoc & Thomas, 2014; Usher, 2013; Vu, 2014; Welbers, van Atteveldt, Kleinnijenhuis, Ruigrok, & Schaper, 2016; Zamith, 2016). There is little doubt that such metrics are increasingly ubiquitous and powerful, but beyond calls for such tools not to overwhelm traditional editorial judgment, there is little understanding of how the press perceives the role such systems play in shaping its autonomy.

The trade press has long been tracking the appearance of Web analytics in digital news work, with companies like Google (Analytics), Chartbeat, Omniture, Newsbeat, and Parsely all promising news organizations the ability to provide rich and often real-time images of how audiences interact with their content110 or might interact with content.¹¹¹ Not long after they appeared, commentators and analysts began wondering whether knowledge about how stories were spreading would simply impact how stories were featured on news sites¹¹² or whether it would affect more fundamental thinking about which stories were told at all and how a story’s impact might be measured.¹¹³ The trade press conversation progressed from general worries about the influence of analytics on seemingly pure journalistic work that editors could direct to more specific questions: What is the right thing to measure? Where do analytics systems come from? And how should they be used within newsrooms? As the field struggled with each question, it left clues about what it thought its autonomy meant—how it could or should be free from knowledge about audiences and technology companies eager to provide rich metrics.

The first question—what is the right thing to measure?—is important to track because as the networked press evolved its vision of the ideal metric, it revealed what kind of knowledge it was willing to be influenced by. Counting page views for a story was an early and crude measure of how popular the story was and what reach editors imagined the story to have.¹¹⁴ This method was relatively quickly replaced with more sophisticated ways of measuring unique page views and sources of page views as the number of new readers and their origins became a more influential metric.¹¹⁵ Another metric became the time spent on an article¹¹⁶ as news sites tried to argue that time was a proxy for a reader’s interest and learning as well as a sign of an increased likelihood that a reader might click on a site’s advertisement.¹¹⁷ (Facebook agreed, publicly saying that its News Feed algorithm would use the time spent on a news story as a “signal” it would use to surface content.)¹¹⁸ These emphases on page views, unique visitors, and time on a page were gradually replaced with attention—an indication that readers were actually focusing on news stories. Focus was something that both news organizations and advertisers agreed could drive the news: the Guardian centered its analytics platform on attention,¹¹⁹ the analytics firm Chartbeat focused its dashboard on attention,¹²⁰ and the Media Ratings Council certified and endorsed Chartbeat’s model of attention as one that could be trusted to measure potential advertising impact.¹²¹ Other news organizations added the idea of engagement to the metrics mix, with the Daily Signal creating a platform to track how people commented on, shared, tagged, and otherwise interacted with its content across multiple social media platforms.¹²² The Christian Science Monitor interpreted the word engagement slightly differently, offering readers ways to take action after they read a story (by visiting a relevant organization, giving the Monitor feedback, or contacting an elected representative)—and then defining those actions as engagement.¹²³

The desire to understand engagement motivated a small-scale analytics platform called NewsLynx¹²⁴ (built by Columbia University researchers) to try to connect a story’s quantitative reach (such as number of tweets and Facebook posts) with its qualitative impact. It let news organizations add tags to a story’s analytical report indicating whether the story had—in the news organization’s judgment—increased awareness of an issue, spurred a government investigation, or prompted a policy change. This tool married an approach to analytics with a model of what it thinks news impact should mean (that is, more than a number of page views or unique readers, time spent on an article, attention paid, or social media engagements).

NewsLynx’s unique origin story and implicit normative goal suggest a second question to ask of the trade press’s discourse on news analytics: where do metrics systems come from, and why might such origins matter to networked press freedom? Some news organizations have enough resources to build their own analytics platforms and imbue them with their own editorial values. The Guardian’s Ophan system¹²⁵ was designed around its own understanding of engagement (the page “has to be in the foreground tab, and you have to be moving the mouse or scrolling, or clicking, or doing something like that”) and the ability for its editors to make changes to the system as theories occurred to them they wanted to test. The Financial Times wanted its metrics platform, Lantern, to be understandable to its journalists. It gives journalists access to information like “average time on page, retention rate, scroll rate, social performance, and what type of devices readers are coming from” as well as breakdowns “between subscribers and non-subscribers who are accessing a particular story.”¹²⁶ The New York Times similarly designed its platform, Stela, to be meaningful to journalists and “narrow the distance between reporters and analytics data,”¹²⁷ while the team designing NPR’s dashboards took a more participatory design approach, surveying competitors’ systems and interviewing NPR employees before building a platform designed to speak to all aspects of the organization.¹²⁸ NPR’s dashboards are not publicly available, but it has said that it will share them with the competitors they interviewed (the New York Times, BuzzFeed, the Atlantic, Huffington Post, USA Today, and the Guardian), suggesting that there are shared understandings of newsroom metrics among organizations with enough resources to develop in-house systems. Organizations without such resources still use third-party platforms like Chartbeat and Google Analytics, as well as platform-specific tools like those Facebook provides for news organizations to track content.¹²⁹ Several trade stories stress the importance of Facebook analytics.

The final question the trade press seems to be asking of analytics systems is, How should news organizations use them? Even though journalists often officially say that they pay little attention to traffic metrics, for years the trade press has been a place where journalists have acknowledged that analytics and content optimization impacts their work.¹³⁰ As Caitlin Petre (2015, n.p.) writes, “Metrics inspire a range of strong feelings in journalists, such as excitement, anxiety, self-doubt, triumph, competition, and demoralization.” Sometimes the power of such systems is openly acknowledged by news organizations themselves. The Washington Post acknowledges that its Bandito system lets editors “enter different article versions with varying headlines, images and teaser text” and then “detects which version readers are clicking or tapping on more, and automatically serves that version more frequently on the homepage and other areas of the Post’s site.”¹³¹ The New York Times’s Package Mapper tool took a similar approach but analyzed how audiences traversed the website and then dynamically offered personalized recommendations of paths that it thought particular users were most likely to follow.¹³² These systems show how analytics are balances between personalization and site-level patterns: news organizations are trying to create close relationships with individual readers while still giving the impression that they are offering a general news experience shared by all.

The trade press openly debates how metrics should influence news work. Although some news organizations that developed in-house analytics dashboards are trying to make reporters more aware of metrics on their stories, others keep those systems from writers. At the Verge and MIT’s Technology Review, only the “senior editorial leadership team”—not reporters—can see Chartbeat and Google Analytics dashboards. The San Francisco Chronicle’s managing editor disagreed with such an approach, saying “we pay [reporters] to sift through information and come up with logical stories and points of views,” so “why can they not do that with their own data?”¹³³ But famed New York Times media critic David Carr cautioned that “just because something is popular does not make it worthy.”¹³⁴

This debate plays out in the sociotechnical details of how news organizations use metrics. The Daily Caller created “a hybrid pay arrangement in which staffers receive base pay plus an incentive for the performance of their work on the Internet,” the details of which were not made public.¹³⁵ Gawker was seen as an industry leader in traffic incentives, creating a Recruits program that paid writers for each one thousand unique monthly visitors and rewarded them with longer-term contracts if they met their traffic targets.¹³⁶ Similarly, business news site TheStreet.com paid writers for page views ($20 for twenty thousand views in a seven-day period),¹³⁷ Advance gave bonuses to reporters who “post frequently and join comment chains,”¹³⁸ and the Apple news site 9to5Mac gave reporters a share of the Google AdSense revenue earned on their stories.¹³⁹ And in the process of trying to decide which writers it would lay off, Time Inc. ranked individual reporters from one to ten based on how “beneficial” they were to advertisers.¹⁴⁰ Some organizations—like Demand Media and AOL’s Seed, which described themselves as quasi-news organizations while others referred to them as Web content farms—went so far as to create on-demand workforces. They would analyze search engine query patterns and, in near real time, hire people to create content relevant to those patterns. Such companies were largely disavowed in the trade press, representing an extreme, unethical end of a spectrum of metrics uses.¹⁴¹

The use of analytics and traffic incentives highlighted divergent views of press freedom. For some journalists and news organizations, autonomy means mandating that reporters have freedom from metrics that are assumed to be corrupting, privileging the reporter’s image of the public interest. Others, though, say that such freedom harms journalism’s potential to impact the world, discover audiences’ interests, and generate revenue by writing popular stories. Most commentators discussed the need for proper metrics and good incentives, without articulating exactly how such balances should be struck.¹⁴²

News organizations define their relationships to reporters, technology companies, and audiences through analytics.¹⁴³ What they choose to measure indicates what they are willing to let influence them—that is, which forces they see as relevant and willing to engage with and which they see as incompatible and needing distance from. The trade press does not have a uniform appreciation of autonomy or what it needs to be free from. It is engaged in an open debate about which analytics are anathema and which are essential to what it sees as good journalism. Some see metrics dashboards as essential to bringing reporters closer to audiences, listening to them, and enacting audiences’ definition of the public interest. Others see such proximity as fundamentally problematic and emblematic of eroding press autonomy and journalists’ freedom to enact their vision of the public. Larger news organizations with strong resources and reputations can make their own in-house metrics systems, selectively use third-party systems, and take time to create the ideal distances between their workforces and metrics systems. Smaller, newer entrants to online journalism must more closely align with third-party systems created by Chartbeat, Google, and Facebook. A public right to hear comes not from journalists separating themselves from or integrating with dashboards. Such systems are actors in their own right in inseparable relationships with humans that together suggest what publics could be or might mean.

Timing

The press has always reflected and created structures of time,144 existing at and shaping the intersection of what I elsewhere (Ananny, 2016b) call the “inside-out” time of news organizations (which initiates, schedules, and controls coverage of events beyond the newsroom) and the “outside-in” time of factors outside the newsroom (which forces journalists to reorient their rhythms and deadlines). Although the internet is often spoken as a place of “timeless time”—where an “elimination of sequencing creates undifferentiated time, which is tantamount to eternity” (Castells, 1996, p. 494)—news time has not disappeared but has become embedded in the networked press’s sociotechnical structures. It continues to exist in the professional spaces and routines of traditional mainstream journalism (Carey, 1989b; Schudson, 1986), but it also is increasingly intertwined with how nonjournalists and technologies that are not overtly grounded in journalistic cultures organize journalists and audiences alike around new rhythms and deadlines.¹⁴⁵ The temporal patterns of the networked press—when audiences can participate, when sources are invoked, when accounts are produced, what kind of events register as news, and how often audiences encounter news—are all increasingly intertwined with a network of human and nonhuman actors. The power to speed up or slow down is not simply about journalists being free from audiences, advertisers, or technologies. As with other dimensions of networked press freedom, the networked press’s time—and thus, the timeliness of the publics it can create—emerges from sociotechnical negotiations.

The trade press is filled with commentaries and thought pieces on the novelty of contemporary news time,¹⁴⁶ the ways online reporters balance “immediacy versus importance,”¹⁴⁷ the possible need to change journalistic ethics during breaking news,¹⁴⁸ and the ways social media saves or consumes audiences’ time in ways that help or hinder journalism.¹⁴⁹ Beyond these general concerns, several patterns appear in the stories journalists tell themselves about networked news time—namely, how to align with or resist technology’s dominant focus on real time, manage journalistic memory and make room for historical points of view when relevant data often exists outside of news organizations and technology platforms favor newness, delay and protect news as a scarce resource that journalists primarily create, work with technology infrastructures that have their own often invisible temporal priorities, synchronize with the time habits of news readers who are also technology users, and anticipate scenarios beyond what current networks show.

Real-Time   First, the trade press is focused largely on creating real-time content that it thinks can compete with the real-time focus of social media. (In 2011, the Pulitzer Prize board changed a local “breaking news” category to reward “real-time” reporting.)¹⁵⁰ Relatively early on, HuffPost focused on hiring live bloggers who could practice what it saw as the lost journalistic genre of real-time reporting in ways that would attract the attention of search engines and social media platforms,¹⁵¹ and the New York Times recently created an Express Team tasked with covering “trends and breaking news that bubble up online and might fall between the cracks of our traditional desk structure.”¹⁵² In addition to initiatives focused on journalistic labor, news organizations also invested in creating technologies for real-time publishing: HuffPost and Boston.com both created live channels designed to attract and keep viewers;¹⁵³ ITV created a “rolling news stream” that aimed to mirror the real-time content of social media channels;¹⁵⁴ the Wall Street Journal created its Markets Pulse platform to provide continuously updated financial news; and BuzzFeed updated how it republished wire stores, creating sharable snippets of new visual information as opposed to following the traditional practice of updating wire stories from the top and editing from the bottom.¹⁵⁵ New York Magazine even boasted in 2011 that its new policy was to publish new content every six minutes between 8:30 a.m. and 7 p.m. (Eastern time),¹⁵⁶ and in 2016 NPR added the phrase “live from NPR News in Washington” to the beginning of its newscasts “to reinforce one of terrestrial radio’s greatest virtues, which is live-ness and a sense of immediacy” in an era where social media promises both.¹⁵⁷ And beyond journalists’ own real-time initiatives, they also debate the merits of having technology companies encourage and sometimes commission news for real-time formats such as Facebook Live, Twitter’s Periscope, Meerkat, and Kik.¹⁵⁸

This focus on real time manifests as custom news content and also as less visible internal infrastructures that are designed to represent (such as the New York Times’s social media–driven feed Watching),¹⁵⁹ quote (such as the Guardian’s use of NPR’s Quotable tool),¹⁶⁰ analyze (such as Quartz’s use of Spundge),¹⁶¹ or invite users into real-time social media streams (such as the Wikipedia-like startup Grasswire).¹⁶² The real-time speed of some domains also creates infrastructural feedback loops. For example, as high-frequency algorithms speed up stock trading, news organizations like the Associated Press (AP) (using Automated Insight’s Wordsmith platform) and Forbes (using Narrative Science’s Quill system) algorithmically write stories in near real-time to analyze the trades.¹⁶³ These algorithmically authored stories are then potential inputs to the Financial News system Thompson Reuters created to “scan and analyze stories on thousands of companies in real-time and feed the results” into traders’ stock-management software.¹⁶⁴ The news loop is closed by a real-time algorithmic arms race: high-frequency algorithms make trades in near real-time, news algorithms write stories about those trades, and other news algorithms analyze those stories in near real time and provide analysis to high-frequency traders. Sky News turns its journalists into real-time infrastructure, equipping them with mobile technologies and speed training that make them “broadcast-ready” in ninety seconds.¹⁶⁵

Readers often see the results of real-time infrastructures through push notifications. Such notifications are known to drive traffic to news apps¹⁶⁶ and are powerful because they can be updated automatically in light of new information,¹⁶⁷ personalized,¹⁶⁸ and targeted to readers in particular geographic locations.¹⁶⁹ But there is considerable concern that only news judgment and editorial principles trigger them,¹⁷⁰ not a desire to reflect social media trends or beat competitors.¹⁷¹ The AP worries that the special significance of its flash alert—historically and rarely used to tell newsrooms about a “transcendent development” such as the moon landing or the Twin Towers collapse—is diluted as readers fail to distinguish important from mundane interruptions.¹⁷²

Such worries are part of industrywide concerns over real-time news infrastructures, with many focusing on breaking news. News organizations are being told that they often spread news faster than Twitter,¹⁷³ that Twitter users follow breaking news,¹⁷⁴ and that readers look to mainstream news outlets¹⁷⁵ and elites¹⁷⁶ to confirm breaking news. In the face of such messaging, the trade press contains many warnings to journalists and readers alike about how to behave during breaking news: tweet “responsibly,”¹⁷⁷ offer live feeds without any commentaries or analysis,¹⁷⁸ avoid using personal accounts to prevent scooping the news organization you work for,¹⁷⁹ stay silent and resist any kind of online publishing,¹⁸⁰ or stay away from social media altogether.¹⁸¹ Most broadly, these concerns motivate a small but vocal group of commentators and practitioners calling for a “slow news” movement,¹⁸² and for greater attention to paid to the toll that 24/7 news cultures take on journalists’ physical and mental health.¹⁸³

Memory   The networked press struggles both to control its present and to define its past. The trade press describes two types of sociotechnical challenges inherent in news organizations’ memory work—being free from others to archive news as they wish and relying on the archives of others to ensure that they can report on the past.

Sociotechnical freedom from others takes a few different forms. The first is simply a freedom to delete news. U.S. News & World Report discovered how controversial this freedom was when it switched to a new content-management system and deleted much of its pre-2007 archive.¹⁸⁴ Commentators similarly challenged BuzzFeed’s freedom to reinvent itself after it removed over four thousand articles that it said “no longer met BuzzFeed’s updated editorial standards”¹⁸⁵ (that is, from before it considered itself a news site)—a move its editor in chief later said the organization “didn’t fully think through.”¹⁸⁶ In both cases, the sites learned that what they saw as internal reorganizations of their own archives triggered public concerns over whether news organizations had the right to alter their own pasts. Many in the trade press used such incidents to celebrate third-party projects—such as PastPages,¹⁸⁷ StoryTracker,¹⁸⁸ and NewsDiffs¹⁸⁹—designed to archive and monitor changes in news sites.

This fight to control past news also manifests as a conversation about how to monetize news organizations’ archives. Several news organizations routinely recycle news that has already proven valuable to advertisers and audiences alike—such as charging premium prices for access to past stories or packaging commercially valuable content for industry research firms¹⁹⁰ or republishing earlier stories under a new date stamp¹⁹¹ or in a slightly different format to appear new to search engines that favor novelty.¹⁹² And to protect the potential future value of their current work, some news organizations are trying to “future-proof”¹⁹³ their content-management systems to prevent their media from becoming inaccessible or corrupted as new technological formats are invented.¹⁹⁴

Finally, news organizations’ memories often are only as long as the data archives of the organizations they depend on. Journalists creating anniversary coverage of a protest where University of California at Davis police pepper-sprayed students and found themselves unable to access earlier media on the incident because the university had hired a Web reputation firm to remove online references to the protest.¹⁹⁵ And reporters wanting to cover when and why politicians delete tweets were hamstrung when Twitter blocked Politwoops (the Sunlight Foundation’s archive of public figures’ tweet streams) from accessing its data.¹⁹⁶ Twitter later restored Politwoops’s access,¹⁹⁷ but the competing site PostGhost was not so lucky because Twitter issued a cease-and-desist order and the archive shut down. Such sites and incidents may seem idiosyncratic, but reliable access to Web archives is key as journalists are being told to give contemporary stories historical context (the core motivation of the news startup Timeline),¹⁹⁸ use social media to find new stories,¹⁹⁹ and investigate how new histories can be told as organizations like the Library of Congress make their archival formats more relevant to journalists.²⁰⁰

Delays   Sometimes news organizations’ desires to postpone publishing may be at odds with the forces in the networked news environment. Several U.S. news organizations—and newslike information providers—have invoked the little-used “hot news doctrine” to argue that they should have the power to suspend temporarily other actors’ right to disseminate information.

This doctrine first appeared after a U.S. Supreme Court 1918 decision (International News Service v. Associated Press) gave the AP a “quasi-property” right, for a limited period of time, to the news it produced. It prevented competing news organizations—but not the general public—from having access to this news as a way to protect the investments that publishers made in the people, materials, and communication networks that created the news (Epstein, 1992). The problems with this doctrine are made even more salient in networked environments: How long, exactly, should a news organization be able to delay others from republishing news it reported? How can a networked news organization separate the resources that it invests in a story from the resources of other people and organizations? For example, if some of the reporting uses a social media platform, should the delay be shorter? Should only human actors (other journalists) be delayed from publishing, or are nonhuman actors like computational code and databases also under the embargo? And what does it mean to have a property right—as opposed to a copyright—on a story? How is it possible to own facts that can trace their origins and confirmations to multiple sources that extend beyond a particular newsroom? For example, if a Wikipedia page was consulted as part of creating a breaking-news account, can its editorial team end the embargo?

Despite these questions and the distributed nature of news work, the trade press contains several instances of news organizations claiming hot news doctrine protection. In 2009, in Associated Press v. All Headline News, the AP claimed that All Headline News repackaged stories it found online—including AP wire stories—and sold them to “newspapers, Internet web portals, websites, and other redistributors of news content.” The case was ultimately settled, but not before the U.S. District Court for the southern district of New York agreed to AP’s claim of “hot news misappropriation.”²⁰¹ In 2010, in Barclays Capital v. FlyOnTheWall.com (a financial news website), before the decision was overturned on appeal, the U.S. District Court for the southern district of New York required “Fly to wait until 10 a.m. EST before publishing the facts associated with analyst research released before the market opens, and to postpone publication for at least two hours for research issued after the opening bell.”²⁰² And in 2014, Dow Jones won a $5 million judgment against the UK-based financial news site Ransquawk after using the hot news doctrine to claim that the site was “pirating its content by broadcasting news within seconds of publication to traders and other subscribers.”²⁰³ It had earlier claimed hot news misappropriation and won an undisclosed settlement against Briefing.com for the site’s “systematic and often instantaneous misappropriation of Dow Jones headlines and articles.”²⁰⁴

Even without bringing suit or claiming hot news misappropriation, news can be delayed when private hosts of newsworthy events meter the speed of reporters’ social media usage. The University of Washington reprimanded a reporter with the Tacoma News Tribune for sending so many tweets during a basketball game (fifty-three) that they amounted to “live broadcasting,” a violation of the school’s policy on unlicensed real-time coverage.²⁰⁵

Temporal Infrastructures   Such temporal control can exist not only in the laws and policies regulating news technologies but also in the technical architectures of media infrastructures inseparable from digital news organizations. The trade press describes several types of control through temporal infrastructure. One entails content-neutral media platforms deciding—or being forced—to use time as an organizing strategy. For example, although the articles still exist on the news organizations’ own sites, the European Union’s “right to be forgotten” legislation forces intermediaries like Google to remove certain content from its search index, effectively making news from sites like the New York Times,²⁰⁶ the BBC,²⁰⁷ and others invisible to many Web users.

Other time-based decisions are made by the intermediaries themselves. When Google updated its search engine in 2011 to favor content that is about “recent events, hot topics, current reviews and breaking news items,”²⁰⁸ news organizations were forced to reformat stories quickly in ways that appealed to Google’s new algorithm.²⁰⁹ When news organizations choose to offer content through Snapchat, they know that their content will disappear from the platform “after 24 hours or much less.”²¹⁰ And when news organizations started sharing video through Facebook and Twitter, they found themselves beholden to the platforms’ decisions to make videos play automatically—decisions that the trade press quickly criticized as editorially unacceptable after, in 2015, social media users scrolling through Twitter and Facebook were forced to watch footage of two Virginia journalists murdered on air.²¹¹

In other instances, news organizations—especially those concerned with serving mobile content quickly²¹²—sign on to use media platforms in exchange for access to much faster and more robust infrastructures. In the case of Facebook’s Instant Articles, news content loads faster when news organizations agree to place content entirely within Facebook’s architecture, share usage data, and enter into advertising agreements.²¹³ In the case of Google’s Accelerated Mobile Pages (AMP) project (where content stays in the open Web and news organizations use an AMP version of HTML), increased mobile load times come with access to Google’s advertising network and paywall system.²¹⁴ Some publishers have expressed reservations about AMP, noting that, in exchange for speed, their stories appear with a google.com URL, and it is difficult for readers to reach the publisher’s own website (and advertisers).²¹⁵

Sometimes the sociotemporal patterns of social media platforms force news organizations to reorganize their own understandings of news time. For example, journalists are told that tweets from BBC News circulate on Twitter for the longest period of time.²¹⁶ Knowing that news spreads faster and further on social media depending on the day of the week and time of day,²¹⁷ the Miami Herald tries to tweet breaking news in the morning and conversational news in the afternoon.²¹⁸ In an attempt to maintain visibility to social media in time zones around the world, many publishers are told to devise hand-off procedures among their global newsrooms that ensure there are no temporal breaks in coverage.²¹⁹

Such persistence, though, brings journalists into tense relationships with social media infrastructure. They know they are in competition with platforms like Wikipedia²²⁰ and Reddit²²¹ that are developing their own in-house live and breaking news teams. They also are constantly having to rebut the timeliness of old or distant stories that, for unclear reasons, suddenly appear on social media platforms and seem new and relevant. For example, “news” of the death of Nigerian author Chinua Achebe widely circulated in March 2015, even though he died two years earlier in March 2013.²²² Journalists are told to expect such things when social media makes it look like “everything happens all the time.”²²³

Reader Rhythms   The sociotechnical dimensions of networked news time also lead news organizations to make guesses about how the rhythms of readers are changing in new technological contexts. Even if news organizations could create news times separate from platform infrastructures, they would reencounter them as soon as they tried to reach audiences whose paces and patterns are heavily influenced by technologies.²²⁴

For example, the Times of London and the Sunday Times changed when they updated their website, phone, and tablet apps after analyzing their traffic to find that four updates per day are optimal for reaching their subscribers.²²⁵ Vox decided that 8 p.m. is the best time to update its readers, and the Toronto tabloid Twelve Thirty Six is entirely organized around providing news that it says is most attractive to readers at 12:36 p.m.²²⁶

Beyond choosing the best time to update news for readers, news organizations also make content align with how long they think readers will read, using different technologies. The Guardian²²⁷ and the New York Times²²⁸ invented new one-sentence story forms designed to be read at a glance on the Apple Watch.²²⁹ The Financial Times created FastFT to give users “tweet-like” stories designed to keep readers on their site during breaking news events.²³⁰ The startup News on Demand was founded to give readers articles that people could read in the time they said they could spare.²³¹ And journalists are being told that readers will consumer longer stories on mobile phones, but only for about two minutes.²³²

Other consumer-oriented innovations in news time focus on managing time for readers. The Washington Post updated its mobile app with a display that shows Uber riders how long is left in their trip to reduce readers’ tendency to switch out of the news app.²³³ The New Yorker is similarly concerned with readers’ wandering attentions: it “tries to anticipate when a reader is about to close out of a story and pops up asking if the reader wants an emailed reminder to come back to the spot where they were left off.”²³⁴ The Economist refuses to link to other websites from within its stories in an effort to give its readers a “sense of completion” and calm any worries that “that they weren’t being completely informed,”²³⁵ while Newsbound ran a sitewide experiment to adjust how a story was segmented, finding an optimal size of information chunks that resulted in the greatest number of people finishing the story.²³⁶ KPCC radio’s website simply tells readers at the bottom of a story that it is no longer being updated and not to expect any new information,²³⁷ while Quartz, Time.com, and the Los Angeles Times use an “infinite scroll” technique to show readers new stories, keeping them in suspense and giving the impression that they will miss out if they leave the site.²³⁸

Anticipation   Finally, some news organizations often engage in what science and technology scholar Adele Clark (2016) calls “anticipation work”—“tacking back and forth” (p. 90) between empirical information and suspicions about what the future may hold; “deleting, sorting, rearranging, [and] (re)presenting” (p. 94) information with the goal of simplifying data and environments; and hoping for futures that align with suspicions and simplifications. There is little empirical evidence yet in journalism studies of the sociotechnical anticipation work Clark describes,²³⁹ but the trade press shows how some journalists focus on overcoming the surfeits of information they are expected to process by designing systems to predict and manage newsworthy events.

One such project is a relatively simple initiative to take information that reporters need to cover future, predictable events (event anniversaries, elections, holidays) that often exists in calendars, text files, spreadsheets, or human memory and move it to a standardized data structure that can easily be updated and used to organize reporting work.²⁴⁰

Other examples are more complex. For example, Breaking News created an “emerging story alerts” system that uses a mix of editorial judgment, eyewitness reports, and computational pattern matching to “notify users that a story could evolve into a major story.”²⁴¹ The aim is to scoop competitors, set readers’ expectations, and assemble an attentive audience that can be quickly informed if the event turns out to be newsworthy. Reuters has taken this notion in an even more computationally intensive direction, building its News Tracer platform to “monitor Twitter for major breaking news events” and algorithmically “assign verification scores to tweets based on 40 factors, including whether the report is from a verified account, how many people follow those who reported the news, whether the tweets contain links and images, and, in some cases, the structure of the tweets themselves.” Its creator says it works best for “‘witnessable events’ such as bombings and natural disasters” and helps shift some of the burden of witnessing” away from humans to algorithms that can act as pattern-searching watchdogs.²⁴²

This somewhat lengthy discussion—of how the networked press handles real-time news, creates its memory, delays publication, intersects with platform infrastructures, understands readers’ rhythms, and anticipates the future—is meant to demonstrate that a desire for autonomy always involves temporal separations and dependencies among humans and nonhumans. Journalists must fight for independence from what platforms define as the past or urgent or likely to happen, but they also must stay close enough to technological infrastructures to know how and when to assemble audiences and give historical context. As with the other dimensions of press freedom, these temporal twists and turns beg a fundamental question of the networked press: what kind of publics can and should it create with these separations and dependencies? If temporal infrastructures need to change in order to realize new types of publics, we need grounded theories of how sociotechnical infrastructures make time and why those infrastructures are most likely to give us the new publics we need.

Security

The press’s ability to guarantee safe and reliable collection and publication of information appears in the trade press as two concerns—the security of journalists and sources and the security of news audiences.

Protecting Journalists and Sources   Sociotechnical security for news organizations and sources manifests in several different ways. Journalists are encouraged to use a new set of technologies (such as encrypted phone and texting apps and traffic-anonymizing Tor networks that obscure packet origins and destinations) as well as new practices (such as creating strong passwords, locking devices, and directing colleagues and potential sources to similarly secure environments). The Freedom of the Press Foundation launched a campaign to fund journalistic security.243 Researchers at Columbia and the University of Washington created a security-minded journalist toolkit (Dispatch)²⁴⁴ for journalists who were being warned against using cloud-based services like Evernote, Dropbox, and Google Drive that may be mined for data or have connections to third-party data services.²⁴⁵ The story of Glen Greenwald, who almost missed out on connecting with Edward Snowden because Greenwald did not use PGP encryption, is now an object lesson in professional preparedness,²⁴⁶ and freelancer Quinn Norton tells in rich detail how her secure work practices helped ProPublica expose Russian state corruption.²⁴⁷ Snowden announced that he is planning to “develop a modified version of Apple’s iPhone for journalists who are concerned that they may be the target of government surveillance,”²⁴⁸ and reporters vulnerable to doxing—that is, having personal information or documents (dox) publicly revealed by hackers as a means of harassing or harming—are instructed in what to do in the event of an attack.²⁴⁹ Some journalists are even warned against engaging in investigative reporting practices that look like hackers’ methods to avoid being labeled as such.²⁵⁰ Journalists who expertly use information-security tools and habitually follow strong encryption practices may be seen by sources as more reliable reporters.²⁵¹ It is professionally beneficial for them to be close to sociotechnical security cultures, but such proximity brings personal risks. There is no one right distance that ensures both autonomy and security.

Much of the trade press discourse focuses on how to make news organizations become sociotechnically safe places for sources to approach.²⁵² This means hiring journalists who appreciate the value of secure tools and practices and creating initiatives that make it easy for sources to deliver sensitive information. SecureDrop—a tool originally developed by security and information advocate Aaron Swartz that lets whistleblowers leave files for news organizations anonymously using the Tor network—is now being used by the Guardian, the New Yorker, ProPublica, and the Intercept.²⁵³ The New York Times has a similar system called StrongBox,”²⁵⁴ and Forbes’s is called SafeSource.²⁵⁵ News organizations without such systems are seen as far less likely to receive sensitive information. Their freedom to do investigative reporting projects is intertwined with the stability and reputation of the Tor network and a host of other technologies that make it possible to mask internet addresses and protect source identities.

News organizations describe being concerned about their journalists’ security skills, their sources’ identities, and their own publishing systems. Google researchers warned news organizations that they are particularly vulnerable to attacks,²⁵⁶ and news organizations are regularly told to upgrade their sites’ security (for example, by using the HTTPS protocol)²⁵⁷ and respond quickly to security vulnerabilities (such as the OpenSSL authentication bug known as Heartbleed).²⁵⁸ Yet the trade press describes several incidents of news organizations being hacked. A public radio station nearly lost its entire archive of interviews, music, stories, donor information, and sales data when hackers demanded a bitcoin ransom in exchange for unlocking files it had encrypted;²⁵⁹ the New York Times website went down twice in one month after being attacked by the Syrian Electronic Army;²⁶⁰ the Guardian had eleven of its Twitter accounts hacked by the same group;²⁶¹ and a $136 billion U.S. stock market drop was attributed to the group after it hacked into the Associated Press’s Twitter account and falsely reported that the White House had been bombed and that President Obama had been injured.²⁶² In the case of the Twitter hacks, news organizations’ closeness to the platform—relying on its security infrastructure to archive and deliver news—meant that they shared in the platform’s security vulnerabilities. After the attacks, Twitter warned news organizations that more attacks were likely to come and asked news organizations to help thwart them.²⁶³ Securing the press’s freedom to report safely and reliably became a shared concern because news organizations and social media platforms had become tightly intertwined.

Protecting Identities of Audiences   The second main theme in the trade press’s discussion of security focused on protecting the identities of audiences—groups that are increasingly convened through technologies with uneven records of protecting peoples’ privacy. Although journalists were told that people do not mind trading personal information for free digital services,²⁶⁴ they also learned that people are most comfortable talking about sensitive topics like government surveillance anywhere but on social media platforms²⁶⁵—precisely the places where news organizations are increasingly trying to meet audiences. In response to these observations, many writers suggest that perhaps news organizations should be concerned about audiences’ privacy.²⁶⁶ They should alert readers to the fact that comments left under news stories may be subpoenaed by law enforcement because they are not covered under shield law protections,²⁶⁷ and they should create sites that limit the collection of personal information and insulate readers from systems that commodify them for advertisers.²⁶⁸ Such concern is rarely translated to action, though, because the trade press usually describes news organizations that are trying to learn about and commodify audiences as much as they can²⁶⁹ and are engaging in a technological arms race with ad-blocking software by devising new ways of blocking blockers.²⁷⁰ Although scholars have not yet investigated the issue in detail, these trade press conversations suggest a nascent concern that the right to hear news may require privacy, protection, and disconnection that news organizations and social media platforms are only beginning to appreciate—the opportunity to encounter news and participate in news environments without being tracked and named. When should news consumption be a solitary, private experience, and when should it be a collective, visible event?

News audiences also can find themselves unwittingly becoming sources in news stories as social media platforms mediate between them and journalists. If their social media accounts have open privacy settings, then posts, comments, and connections that make sense in one context can quickly become part of news stories in ways that the trade press struggles to describe: “Is it like quoting something overheard on the street or at a public event? Or is it more like eavesdropping on a conversation at a private party?”²⁷¹ Likewise, some writers warned journalists not to treat anonymous sources on secret-sharing sites such as Whisper and Secret—which traffic in both healthy self-disclosure and unsubstantiated gossip—as they would other social media platforms that require people to register.²⁷² Similarly, several writers posed an ethical dilemma to the press: during protests, how should journalists cover official police statements when they know that law enforcement surveils protesters’ social media posts without their knowledge or consent?²⁷³ Reporters could simply repeat official statements without comment because they believe it is not for journalists to decide if police surveillance is lawful or ethical, or they could let audiences know that such statements emerge from warrantless surveillance. Reporters can quickly learn much more about sources than they might ever have revealed in an interview, simply because both audiences and journalists are brought together by social media platforms that collapse contexts (Marwick & boyd, 2011).

Finally, a set of emerging technologies that is becoming part of journalism and publishing is beginning to raise privacy and security questions. Many news organizations are creating virtual reality experiences that immerse audiences in custom-built, three-dimensional worlds.²⁷⁴ These spaces can be powerful and evocative, but they also are highly instrumented environments that can track not just a user’s presence but a host of signals that news readers have never had monitored—exactly where someone looks, what she hears, how long she speaks, what facial expressions elicited responses, which virtual objects she manipulated and for how long, and more.²⁷⁵ The potential for surveillance in virtual news worlds where almost everything can be measured is unmatched. Similarly, as news organizations like Quartz create chat bots designed to integrate with Amazon Echo and Google Home, the voices and habits of news audiences will be recorded, modeled, and connected to third-party databases—actions that have already prompted law enforcement to try to gain access to such systems when they are thought to be relevant to investigations (Sauer, 2017).

Each new platform that news organizations design for or partner with is a potential surveillance environment. As journalists move closer to such technologies and create news experiences that are inseparable from media platforms, they bring audiences along with them. A public right to hear news in such spaces creates opportunities to listen in on audiences, often without their informed consent. If the press is to be seen as a trusted institution by sources and audiences who take risks by letting its infrastructures into their online environments and homes, it will need to show how it distances itself from surveillance architectures. This may mean that the networked press foregoes some connections, relationships, or sources of revenue, but such separations may give it greater legitimacy with publics who need the press’s discretion, in turn bringing sources and audiences closer to journalists.

Audiences

News organizations have long struggled with how to see readers simultaneously as consumers whose preferences they are trying to meet, citizens whose opinions they are trying to inform, and sources with information they need.276 Press freedom has not been about trying to achieve independence from audiences and their demands. It has been about negotiating an ideal distance that keeps journalists close enough to audiences to be relevant and distant enough to maintain a distinct journalistic perspective separate from what audiences might want. As with other dimensions of networked press freedom, what we see in the trade press are news organizations continuing to strike these balances but doing so in sociotechnical environments with new forces of proximity.

Most fundamentally, journalists find themselves trying to distinguish human versus computational audiences (Lokot & Diakopoulos, 2016; Woolley & Howard, 2016). They are warned that much of the internet’s traffic actually comes from automated bots. These rule-following computer programs can create expressions on social media platforms that are scheduled or instantaneous and are often indistinguishable from humans to all but skeptical observers aware of their ubiquity and sophistication. They generate large amounts of bogus traffic,²⁷⁷ corrupt advertising markets,²⁷⁸ are seen as responsible for significant amounts of misinformation during the 2016 U.S. election,²⁷⁹ and are forcing news organizations to be newly skeptical of their site analytics²⁸⁰—but they also are presented as potential tools for journalists and as ways to automate public-interest reporting²⁸¹ or counter misinformation instantaneously.²⁸² Underlying journalists’ concerns about how close they should be to online audiences is a more fundamental question of what exactly defines a news audience in sociotechnical environments populated by humans and nonhumans.

Beyond this existential question, two patterns emerge in trade press discourse showing how networked press freedom entails negotiating sociotechnical distances with audiences—in how news organizations interact with audiences through commenting and engagement systems and in how news organizations give audiences views inside newsrooms and invite collaboration.

Commenting and Engagement   Journalists are overwhelmingly focused on how they can communicate with audiences constructively when their systems and habits seem so ill-suited to the scale, speed, and diversity of online communication. Audiences are communicating with journalists in larger numbers, faster, and in more ways than ever before—and journalists are unsure how to respond.²⁸³

Although news organizations say they value reader comments,²⁸⁴ see commenters as potential sources,²⁸⁵ and are told that people leave comments for many different reasons,²⁸⁶ the vast majority of news organizations are shutting down or fundamentally rethinking their commenting systems.²⁸⁷ They offer different motivations for doing so. It is too much to ask readers to comment²⁸⁸ when good feedback is lost among vitriolic comments designed to cause conflict and confusion,²⁸⁹ and it is too much to ask journalists to reply to comments when news organizations lack the staff to moderate comments properly²⁹⁰ or the budget required to hire and manage outside moderators.²⁹¹ News organizations say that the stories with the most comments unintentionally end up being the most popular stories because the overwhelmingly negative comments draw traffic,²⁹² and although they are internally skeptical of the value of anonymous comments,²⁹³ they are confusingly told that requiring real names both improves comment quality²⁹⁴ or has no effect.²⁹⁵

They also find that comments can make it harder for readers to understand news stories. Popular Science discontinued comments on its stories because it saw hostile and ill-informed commenters spreading misinformation and unfounded doubt among its readers.²⁹⁶ (It did allow bloggers hosted on its site to decide for themselves whether to enable comments.)²⁹⁷ The Los Angeles Times similarly silenced voices when it announced that it would no longer publish letters denying that humans cause climate change.²⁹⁸ Both Popular Science and the Times essentially decided that to serve the parts of their audiences they wanted to engage with and keep close, they were willing to silence and distance themselves from others they thought were not qualified to speak. Many news organizations—Reuters,²⁹⁹ Motherboard,³⁰⁰ the Verge,³⁰¹ The Week,³⁰² HuffingtonPost,³⁰³ and NPR³⁰⁴—agreed and publicly stated that they were discontinuing comments because labor costs were high, better alternatives existed on social media sites, or as the Chicago Sun-Times said, it simply did not know how to “foster a productive discussion rather than an embarrassing mishmash of fringe ranting and ill-informed, shrill bomb-throwing.”³⁰⁵ Reddit founded its news spinoff, UpVoted, in part on the principle that comments harmed audiences.³⁰⁶

When news organizations decided to leave comments on their sites, the trade press offered advice on how to do it³⁰⁷ and news organizations partnered with technology companies³⁰⁸ and design communities³⁰⁹ to reinvent audience engagement. And news organizations individually experimented with different approaches. The Verge hosted general, open forums instead of story-specific comments.³¹⁰ The New York Times limited the number of stories it opened for comments each day³¹¹ and encouraged commenters to moderate themselves, creating badges and rewards systems that gave privileges to commenters with good track records.³¹² Reuters restricted comments to opinion pieces,³¹³ the National Journal let only subscribers comment,³¹⁴ and Tablet tried to charge readers for the privilege of commenting³¹⁵ before ultimately moving its audience engagement to Facebook.³¹⁶ Several news sites also edged closer to social media: BuzzFeed required Facebook accounts to access its chat bots,³¹⁷ and after finding itself deleting approximately 75 percent of its site’s comments,³¹⁸ the HuffingtonPost required commenters to log in with Facebook accounts³¹⁹ and also required that such accounts be verified by having users give Facebook their phone numbers.³²⁰ That is, to keep commenters on its site, HuffPo also forces them away: by outsourcing moderation and verification, it moves readers closer to Facebook, forcing them to accept that company in exchange for proximity to HuffPo.

Similarly, when news organizations abandoned custom commenting systems for social media platforms, the trade press traced their different approaches. Reuters conducted scheduled, topic-specific Twitter chats in which its journalists answered readers’ questions and listened to feedback. Mic chooses particular stories to post to its Facebook, Twitter, and Tumblr accounts, stressing that it “doesn’t just leave it alone from there” but moderates conversations on its stories wherever it finds them. Deciding that it was “fine” with its “communities organizing off-site, which is what they’re going to do anyway, whether we have comments or not,” USA Today’s FTW site added WhatsApp and SMS feedback channels; it also uses tools like SimpleReach, Chartbeat, and CrowdTangle to watch how its stories are being talked about on social media, choosing to “engage in those conversations where they’re happening, or not.”³²¹ The BBC more actively tries to engage audiences by hosting conversations on new platforms as they appear, most recently using WhatsApp, WeChat, Line, and Yik Yak to interact with readers.³²² Finally, although Popular Science uses social media platforms to engage with audiences, it also still exchanges hand-written letters with readers and hosts events where they can gather and converse.³²³

All of these negotiations involve tensions between positive and negative freedom. News organizations’ capacity to host verified commenters means that commenters must relinquish their freedom from social media platforms. News organizations’ ability to enact their editorial principles means accepting social media platforms’ community guidelines—tying their news judgments to the technology companies’ policies, norms, and algorithms and accepting the ways that platforms foster or silence debate.³²⁴ And getting closer to platforms’ users can mean getting further away from other audiences and demographics that use social media less or in different ways.³²⁵ Finally, meeting audiences off site does not necessarily alleviate the risks that news organizations will meet communities they were trying to avoid, as the Wall Street Journal learned when Anonymous held a “comment flashmob” on its Facebook page.³²⁶

Access and Collaboration   Although in much of the sociotechnical work of interacting with audiences, news organizations create strategic relationships to platforms that limit readers’ participation, at other times news organizations are trying to create closer direct relationships with audiences as they try to produce news with them. Early examples of these coproductions involved news organizations that invited readers to submit media and then tried to figure out how to distinguish this content from professionally produced media within existing notions of journalistic quality. CNN’s iReport was an early and particularly successful example,³²⁷ but initiatives like the Guardian’s GuardianWatch have continued to appear.³²⁸

Beyond systems for accepting user-generated content, several news organizations have created views into news work and infrastructures for collaborating with audiences. One tension these projects revealed was just how much access to give to the audience. In 2011, the Guardian created its Open Newslist project to let audiences see story topics and assignments as they evolved and provide feedback to editors and sources to reporters using the Twitter hashtag #opennews and an on-site liveblog.³²⁹ Although the initiative earned praise from some as an example of journalistic collaboration,³³⁰ others questioned whether it was just a performance of coproduction because ultimate power still rested with editorial staff and the added visibility guaranteed no additional media accountability.³³¹ Other organizations give audiences curated views or brief glimpses into their news work. The New York Times Insider offers premium subscribers behind-the-scenes accounts of how journalists work,³³² and BuzzFeed briefly made its internal analytics dashboard open for public viewing.³³³ In all cases, news organizations are not offering truly unfettered access to journalists and newsrooms but are carefully configuring systems and views of those systems in ways that give audiences limited opportunities to see and engage with news work.

Other collaborations are less performative and more complex and have journalistic outputs. For example, several news organizations offer application programming interfaces (APIs) that are essentially software toolkits to give outsiders access to some news organization data, under certain conditions, letting them create new sites and apps that repurpose journalistic work.³³⁴ The New York Times, the Guardian, NPR, and USA Today all offer such APIs, sustaining a small but active community of entrepreneurs, hobbyists, students, and researchers whose projects depend on how well news organization maintain their data infrastructures, what policies they set, and whether they answer programmers’ questions.³³⁵ In exchange for providing such access and services, news organizations gain visibility, fulfill organizational missions, discover technical problems with their infrastructures, and sometimes earn small amounts of licensing revenue from premium API subscribers.

An alternative form of collaboration entails crowdsourced work that requires no programming skills or interest. An early example was the Guardian’s Expense Scandal project in which it asked audiences to help find patterns and anomalies in large databases of government expenses records.³³⁶ NPR affiliate WNYC built a site for people to provide geotagged reports of cicada sightings during a once-in-seventeen-years surge in the insect’s population, and the NPR apps team built a mobile phone app to crowdsource reports of children’s playgrounds, trigging a series of stories about the lack of sites accessible to children with disabilities.³³⁷

ProPublica is a leader in this type of collaboration and has organized several similar initiatives, maintaining a Get Involved site to show readers how they can contribute to data-driven reporting.³³⁸ To track lawmakers’ positions on a gun reform bill, it asked readers to “find the contact information for their local senator’s office using ProPublica’s extensive Senate office database, find a statement or contact the office to get a comment on the bill, and fill out a quick survey on ProPublica’s website.”³³⁹ To analyze a large database of records on election spending quickly and uncover any potential abuses, it “got hundreds of readers to sift through television ad spending records” and report back with patterns.³⁴⁰ And it even created a “reporting recipe”³⁴¹ to help journalists at other news organizations use its free, publicly available, geotagged database of incidents in which local public schools restrained or secluded children.³⁴² (ProPublica once even asked Reddit users what data-driven investigative projects it should tackle next.)³⁴³ In all cases, in exchange for creating and managing such infrastructures of collaboration, news organizations are able to do large-scale data-driven projects that they could never accomplish with existing staff. Rather than trying to keep audience members at bay, these news organizations are trying to attract readers, make them collaborators, and benefit from their free labor. They use sociotechnical systems to bring themselves closer to audiences and enable their freedom to tell particular data-intensive stories.

Finally, some news organizations are working with audiences to develop entirely new products or improve existing ones. Finnish magazine Olivia developed custom collaboration software to let journalists and readers co-create an issue. They worked together to find “a story topic, angle, and interviewees” and shared “experiences and expertise,” and with “the crowd’s input,” journalists drafted stories, with audiences following along and chiming in as writing progressed.³⁴⁴ BuzzFeed prototyped a new version of its newsletter by showing drafts to audiences, asking for feedback, and implementing suggestions,³⁴⁵ and Gawker let readers create their own headlines and introductions for stories they shared from the site on social media.³⁴⁶ In other instances, audiences helped news organizations format existing content and, in turn, attract new audiences. Viewers of PBS’s NewsHour translated videos into non-English languages, helping the show attract new demographics,³⁴⁷ and New York Times readers helped the paper index its archives, creating new collections of content that the paper can market and sell.³⁴⁸

Across these points of access and collaboration, we see negotiations of what Jan-Hinrik Schmidt and Wiebke Loosen (2015) might call “inclusion distance”—expectations between readers and journalists alike about why audiences participate, how their roles differ from reporters, what types of participation are more or less important, and what impact audiences can have on news work when they are included in journalists’ networks. The added dimension here, though, is that many of these negotiations live among both people and nonhumans—social platforms with verification systems that require documentation, popularity algorithms whose calculations cannot distinguish well-meaning comments from uncivil trolling, analytics systems that track conversations across social media platforms, APIs that limit how often and under what conditions hobbyist programmers can query data, and large-scale databases with nonexistent or nonsensical metadata that cry out for human organization. It seems strange now to think that, to be free, news organizations need only distance from their audiences. In fact, they need to negotiate sociotechnical separations and dependencies—getting closer to or farther from humans and nonhumans—as they try to convene publics. I offer here no pronouncement of what kind of proximity these humans and nonhumans should construct but suggest that different versions of networked press freedom—and thus different types of publics—will emerge from different types of sociotechnical proximity.

Revenue

The networked press has been in a revenue crisis for years. Print circulation and its lucrative advertising revenue have been dwindling, and digital advertising fails to make up the difference; consumers accustomed to getting news for free are reluctant to pay for online journalism; and few news organizations know exactly what makes subscription plans successful (Arrese, 2015; R. Fletcher & Nielsen, 2016; Myllylahti, 2016; Pickard & Williams, 2014; Socolow, 2010). In this crisis, the press is searching for money, but it also is trying to find business models that align with its self-image as an industry that historically has carved out editorial independence and provided investors with lucrative returns (E. L. Cohen, 2002) while leveraging two main revenue sources—advertisers wanting to reach consumers and audiences willing to pay for news (J. T. Hamilton, 2006).

In the networked press, there is an additional actor beyond journalists, advertisers, and audiences—the sociotechnical systems whose configurations can commodify news and deliver revenues but sometimes at the expense of the industry’s self-image as an editorially independent field beholden to neither readers nor markets. As the press becomes increasingly intertwined with commodification technologies, the definitions and achievements of editorial independence are called into question anew. The trade press is preoccupied with this question of how to earn revenue through sociotechnical systems—a preoccupation that, when examined for its separations and dependencies, reveals a networked press struggling to reconcile economic security with institutional autonomy. The trade press reveals six dimensions to the sociotechnical construction of press revenue—configuring paywalls, commodifying readers, crowdfunding, commodifying journalistic expertise, creating sponsored content, and designing partnerships.

Paywalls   The trade press describes an ongoing search for revenue,349 primarily by figuring out how to create a virtual “barrier between an internet user and a news organization’s online content” (Pickard & Williams, 2014, p. 195) that can be crossed only by paying money. The short and ongoing history of news paywalls is the story of news organizations trying to band together, distinguish themselves from others, court some audiences while shunning others, and ask hard questions about how their value should be commodified through pricing technologies that could change at any time, for unclear reasons. Early stories revolved around the novelty that online news organizations would charge readers at all.³⁵⁰ Because audiences had grown accustomed to online news being free, there was a great deal of skepticism that anyone would ever pay for news.³⁵¹ Hoping for an industry-wide shift that would free them from having to negotiate their own individual strategies, news organizations wondered how many others would build paywalls,³⁵² how many readers paywalls would affect,³⁵³ how much they should charge and if charges should differ across devices,³⁵⁴ and whether their paper was sufficiently unique enough (that is, distant from competitors)³⁵⁵ to support paid subscriptions. They were consistently told that few people would pay for online news (only 10 percent in 2014)³⁵⁶ and that any increases in revenue would come with reduced traffic,³⁵⁷ which worried them that paywalls would drive away readers³⁵⁸ or enable journalists to connect only with affluent readers.³⁵⁹

News organizations’ concerns about how close to get to readers—and which readers—has manifested in a series of paywall experiments. How many articles should be offered for free before charging? Estimates range from one to twenty per month and have changed over time.³⁶⁰ Too many free articles makes readers think they never will be charged—and they are insulted when the time comes to pay—but not enough free articles makes it hard for readers to see the value of the product they eventually will be charged for.³⁶¹ Some news organizations have tried charging for access to certain content³⁶² (a move the Los Angeles Times quickly reversed after charging for its calendar of events caused a 97 percent drop in traffic),³⁶³ for each article,³⁶⁴ for access to particular journalists,³⁶⁵ or for premium sites with fewer ads and better interface designs.³⁶⁶ One proposal entailed letting readers themselves set the prices of articles, making the price of a story drop the more people clicked on it—essentially aligning news organizations with a logic of popularity that made some articles more likely to be purchased.³⁶⁷ In all these experiments, news organizations were trying to figure out which parts of themselves to commodify for which audiences, balancing unfettered connections with readers against the promise of revenue.

Another class of experiments focused on fundamentally rethinking the paywall. Some focused on language-specific content. The Dutch news startup Blendle banded Dutch newspapers and magazines together into a single repository of Dutch-language news in which readers would pay only for articles they read.³⁶⁸ Piano Media similarly tried to create a paywall covering all news organizations in Slovenia, but the experiment failed after the Slovenian market proved too small for a single paywall and individual news organizations left to try their own paywall configurations.³⁶⁹ The paywall itself became a diagnostic of how audiences and journalist alike saw news as like or unlike that of competitors.

Some news organizations—Slate,³⁷⁰ the Los Angeles Times,³⁷¹ and the Wall Street Journal³⁷²—tried to avoid the paywall stigma by calling subscriptions “memberships” that would give audiences closer, long-term relationships with news organizations. Others tried to capitalize on readers’ close associations with social media platforms, offering unmetered access to stories shared via Twitter or Facebook,³⁷³ or they leveraged readers’ own relationships, trying to convert nonsubscribers by giving subscribers “gift” articles that they could share and let others read for free.³⁷⁴

Despite these different experiments, the trade press still sees paywalls as dynamic and unstable. People routinely evade them by altering URLs, deleting browser histories, refusing cookies, and sharing passwords,³⁷⁵ and many news organizations have uneven policies on when to use paywalls, dropping them idiosyncratically for public emergencies, special events, and advertising partnerships.

Commodified Readers   News organizations also use their knowledge of audiences to turn readers and subscribers into revenue generators.³⁷⁶ Many news organizations sell to advertisers the data they gather when people sign up for subscriptions, the traffic patterns they observe on their websites, or the right to use tracking software to surveil visitors. Even readers who do not pay are still commodified as data sources: the closer news organizations can get to paying and nonpaying readers, they more they can earn from advertisers.³⁷⁷ Some news organizations also entice site visitors into answering surveys for technology companies and marketers, giving them news access in exchange for their opinions as consumers.³⁷⁸ The German news site Neue Zürcher Zeitung tracks all of its visitors and, without their knowledge, profiles them and targets personalized marketing and free content to those it thinks are most likely to become paid subscribers. The idea of being free from all readers and their preferences in order to pursue a pure editorial vision is not part of these tracking and commodification strategies. News organizations are using analytics and marketing systems to try to get very close to readers, especially those they see as being most valuable.

Crowdfunding   Another reason to get close to readers is to make them investors in addition to subscribers. Separate from news organizations’ paywalls, a flurry of crowdfunding platforms has arisen to generate funds for specific people and projects. These sites have fostered a new class of reader patrons who pledge small amounts of money to help journalists and news organizations realize projects. The closer they can get to potential donors, the greater chance they have of achieving financial freedom—although questions linger about what kind of professional autonomy emerges from such relationships.³⁷⁹

Many different people use news crowdfunding sites. Freelancers are able to raise money and reputations on such sites,³⁸⁰ for-profit news organizations have used them to fund beat reporting and get access to data,³⁸¹ nonprofit news organizations have used them to hire interns,³⁸² and a team of professional journalists crowdfunded money to teach a course on data journalism.³⁸³ Spot.us was an early example of a crowdfunding website focused on journalism, but several variants have arisen. The Dutch news site De Correspondent funds its own organization through crowdfunding,³⁸⁴ NewsFreed focuses on crowdfunding investigative reporting,³⁸⁵ Emphas.is focuses on photojournalism,³⁸⁶ Contributoria crowdfunds news projects that are then developed in the open while visible to the funding community,³⁸⁷ and Beacon places a fee on top of all crowdfunded stories to create a bonus pool shared among writers whose stories patrons recommend.³⁸⁸ The popular general crowdfunding site Kickstarter eventually created a special section for journalism projects.³⁸⁹ Across all variants, there are goals to create close relationships with particular funding communities to build with patrons a sense that projects are co-constructed and that donations are actually investments. The trade press is rife with stories about how to make journalism projects succeed on crowdfunding sites³⁹⁰ and how to create new crowdfunding communities.³⁹¹

As sociotechnical sites of networked press freedom, crowdfunding communities are one of the clearest examples of journalists using communication platforms to get close enough to people to make them patrons and audiences while trying to maintain enough distance to avoid being drawn into clientlike relationships in which journalists sacrifice their power to tell audiences things they do not want to hear. This concern appears in both the trade press and academic scholarship: some journalists see crowdfunding as a way to involve audiences and create what they see as publicly relevant news,³⁹² while other commentators caution that proximity to donors brings clientlike responsibility.³⁹³ This tension is confirmed by crowdfunding scholars who find that “journalists who crowdfund strongly believe in the journalistic norm of autonomy, but at the same time feel a great deal of responsibility towards their funders” (Hunter, 2015, p. 272).

Commodified Journalistic Expertise   Because news organizations are seen as sites of traditional journalistic practice as well as places where information is created, verified, and disseminated, news organizations are beginning to capitalize on their currencies as information professionals. For example, the nonprofit news site ProPublica earned approximately $30,000 in revenue by selling data that its journalists gathered, vetted, and offered access to through custom-built APIs.³⁹⁴ The organization’s data expertise—and its brand as a trusted source of information—leads purchasers to see it less as a traditional news organization and more as a trusted knowledge broker. Other news organizations have similarly positioned parts of themselves closer to data work than to journalism, creating APIs that give premium access—and the right to create new commercial ventures—to those who pay.³⁹⁵ The Guardian offers a “bespoke” API license to those wanting to place advertisements alongside API content, and USA Today similarly offers premium, paid API accounts to those building projects that require unlimited data access and permission to earn revenue.³⁹⁶ These two news organizations—through their API commodifications—move themselves closer to commercial domains. In contrast, public broadcaster NPR specifies in its API terms of use that API data is only for “personal, non-commercial use.”³⁹⁷

Sponsored Content   Academic research (Brandstetter & Schmalhofer, 2014; Carlson, 2015c; Wojdynski, 2016) and trade press discourse³⁹⁸ agree that sponsored content—sometimes called native advertising, paid content, or promoted stories³⁹⁹—confuses most readers by making content that advertisers commission resemble news that journalists produce. Despite (or because of) this confusion and because such content can earn far more revenue than other types of advertising,⁴⁰⁰ the trade press is filled with discussions of how to make sponsored content that both generates revenue and meets editorial standards.

Some news organizations, though, describe wanting little separation between commercial and journalistic content. BuzzFeed took media posted by Reddit users, repackaged it as sponsored content (which accounts for “nearly all the company's revenues”), and claimed that such advertisements were copyright-protected transformative uses.⁴⁰¹ The company’s strategy is acrobatic—staying close enough to Reddit to mimic its users, collapsing distinctions between its commercial and journalistic content, and then arguing that its sponsored content is sufficiently distinct from Reddit media, making its revenue generation acceptable under copyright law’s transformative-uses exemption. HuffingtonPost’s approach is simpler: it calls for sponsored content that “looks and feels similar to its editorial environment.”⁴⁰²

Other news organizations try to strike balances when displaying sponsored content. But such balances are hard to strike when online environments vary greatly in their visual styles and interaction designs. There are no clear standards to rely on when trying to distinguish commercial and journalistic content, so each news organization invents its own method. For example, in announcing the company’s new “native advertising platform,” the New York Times’s publisher promised clear differences between commercial and journalistic work: “a distinctive color bar, the words ‘Paid Post,’ the relevant company logo, a different typeface and other design cues to let readers know exactly what they are looking at.” (The Times also promised to use analytics to alert it if readers appeared to be confusing the two types of content but did not reveal how it would see such patterns of confusion.)⁴⁰³ The Washington Post allows paid posts within its comments sections but calls them “sponsored views.”⁴⁰⁴ Sports Illustrated experimented with forcing readers to watch a video advertisement before reading a story, creating a perceived connection between the story and the advertiser, even when none existed or was intended.⁴⁰⁵ Quartz created a new data-driven story format that uses statistics and numbers to explain issues, but it intersperses what it calls “sponsored data points” among news stories. Finally, the MinnPost created a “funded beats strategy” in which editorial leadership identifies newsworthy topics and issues that the board of directors then tries to match with individual funders, creating a close—but firewalled, the paper says—connection between wealthy donors interested in producing news on particular topics and journalists tasked with reporting on those beats.⁴⁰⁶ Another strategy is to create workforce separations—to hire former journalists to create branded online content that inevitably reads like news stories but that can honestly be described as originating from the business part of an organization and not from its news part.⁴⁰⁷

Sometimes, trying to separate sponsored and journalistic content implicates social media platforms, relying on their designs, habits, and expectations. For example, in 2013, the Associated Press created a minor concern in the trade press⁴⁰⁸ when it began putting out sponsored tweets under its own Twitter handle. Such tweets began with the words “SPONSORED CONTENT,” but some worried that such labels made little sense outside of news site contexts, that the AP could not adequately track whether the tweets confused users, that it was hard to understand what retweets of such advertisements meant, and that such tweets seemed similar to the “promoted tweets” format that Twitter introduced in 2010⁴⁰⁹ but that AP was not using. Although the Twitter corporation did not mind the potential confusion, Google has for several years warned publishers that it would not include in its search index pages that were solely or primarily paid links and has told news organizations that they must clearly separate sponsored content from their journalistic stories or risk having their entire sites removed from the search engine’s index.

In sponsored content, we see several kinds of separations and dependencies at play. Some news organizations seem apathetic about separating commercial and journalistic content or are designing for some amount of revenue-generating confusion. Others are resigned to the fact that some readers will be confused but try to create distinctions through design techniques and labor strategies. And others seem content to let distinctions reside on social media platforms that they use to serve both commercial and journalistic work. Through this messy mix, the press must find a way to signal that its journalism is independent from sponsored content. But such signals must not be so strong that they ward off the revenue-generating clicks, tweets, and likes of confused readers or readers who see the differences but are unbothered by news organizations that intersperse commercial and journalistic content.

Partnerships   Some forms of revenue come when news organizations partner with technology companies or create partnerships through new technology infrastructures, with trade press stories focusing on how news organizations can earn revenue by staying close to and distinct from competitors and social media platforms. The Associated Press showed how news organizations can band together to fight a technology that was thought to be harming revenues. In 2012, the AP and twenty-eight other news organizations launched NewsRight, a consortium designed to charge licensing fees to aggregators that were algorithmically repackaging and selling its news content.⁴¹⁰ The consortium was meant to demonstrate the power that news organizations can wield when they set aside competitive rivalries in the service of industrywide action, but the initiative eventually fell apart as the technology to track aggregated content failed and individual publishers decided to pursue their own individual monetization strategies.⁴¹¹ Other technology-focused partnerships among news organizations have proved more successful. After the Washington Post offered free digital access to its site for any subscriber to a local newspaper within the consortium it assembled,⁴¹² both Post traffic and local paper subscriptions increased.⁴¹³ The strategy showed how news organizations can create complementary online access schemes by agreeing to leverage their respective differences and segment audiences and advertising markets.

In other partnerships, news organizations have leveraged the publishing channels of third-party platforms, both using and critiquing their revenue models. In 2011, the trade press tried to make sense of Apple’s plan to let content providers sell subscriptions on apps that ran on Apple hardware. The company originally said that price of subscriptions sold outside the apps “can be no lower than the price offered inside the app” (effectively capping news organizations’ subscription revenue) and that “customer data for in-app purchases will remain with Apple” (cutting news organizations off from a vital source of information about their audiences and a potential source of revenue).⁴¹⁴ After industry uproar, exceptions to these policies eventually were made for news organizations, and not long afterward, Google offered a competing platform that publishers saw as more flexible and lucrative,⁴¹⁵ and news publishers focused on creating mobile websites that existed outside of apps and thus beyond the reach of app store makers like Apple and Google.⁴¹⁶ These news app incidents revealed a situation in which news organizations were sufficiently beholden to Apple to make such exceptions necessary, Apple was eager enough to host news apps that it conceded, and Google was interested in enough in news that it offered a competing product. The definitions and viabilities of news apps emerged only when journalists forced a technology company to distinguish the news business from other types of content production and when they began exploring technologies that would let them avoid app markets altogether. The networked press’s freedom to set subscription prices and reach mobile audiences required negotiations of policy and technology that both connected and distanced news organizations from platform makers.

Other news and tech partnerships are more focused on generating new content, sharing advertising revenue, and creating new audiences. Snapchat’s media partnership program shares advertising revenue with news organizations,⁴¹⁷ and Facebook’s Live program pays news organizations to create original and exclusive video experiences that are posted directly to Facebook, generating advertising income for both the platform and news organizations. For news organizations, the tradeoff with such proximity to Facebook is agreeing to work within its video format, house news content on its system, and know that only Facebook users see that news.⁴¹⁸ Other partnerships look more like brand relationships. The New York Times offered people free access to its website but only when readers accessed the sites through a Starbucks network,⁴¹⁹ guests staying at Trump hotels got free online access to the New York Times because it was “Donald Trump’s favorite newspaper,”⁴²⁰ the Associated Press experimented with technology that would automatically print its latest bulletins on restaurant receipts,⁴²¹ and a limited-term subscription to the Washington Post (owned by Amazon founder and CEO Jeff Bezos) offered a membership to Amazon’s Prime program.⁴²² As news organizations affiliate with brands through infrastructural partnerships, they become closer to certain consumers and demographics and potentially further from an image as a general public information provider.⁴²³

Finally, some types of partnerships are virtually invisible to readers but still powerfully shape news revenue generation. Many news organizations use software to block ad-blocking technologies, essentially engaged in an ongoing, largely hidden technological arms race to ensure that their advertisers’ messages reach readers actively trying to distance themselves from advertising.⁴²⁴ And separate from news organizations and their use of paywalls, several of the companies that make paywall technologies have merged, rapidly shrinking the competitive landscape of paywall infrastructures.⁴²⁵ Piano Media⁴²⁶ and Press+⁴²⁷ were founded to create and manage paywalls, direct news organizations on how best to configure their systems, and share research on what made paywalls successful (for example, sites with more daily stories sold more subscriptions),⁴²⁸ but as they merged with each other and then paywall maker Tinypass, they created a single entity that is in a far more powerful position to dictate the features of paywall technologies and the terms of agreements with news organizations. (This power may be challenged with Google’s addition of a paywall feature to its AMP publishing system.)⁴²⁹

Across these different partnerships, the networked press emerges as a distributed and somewhat unstable entity whose autonomy depends on why and how it affiliates with or breaks free from competitors, sponsors, and technology companies, each of which values news differently. Networked press freedom is not simply about being free from markets or commercial revenue. Indeed, it is often difficult for news organizations to trace exactly where funding originates, under what circumstances it flows, and whom it benefits. But revenue could become a diagnostic of press freedom—a way for news organizations to take stock of where their infrastructures intersect with those of revenue sources and ask what kind of publics such intersections underwrite. The key lesson here is not that news organizations are too dependent or not dependent enough on online revenue sources but that through their separations from and relationships with online funders, news organizations have an opportunity to indicate what kind of publics their funders support.

Facts

Journalists make and analyze what they see as facts. The history of press objectivity is, in part, the story of how social, technological, economic, and cultural forces shape how news is seen as separate from reporters’ own interests and values and instead simply reflects truths that are assumed to exist regardless of whether reporters provide accounts of them. Although there are stable, news-relevant facts that are beyond the reach of this constructivist view of truth, scholars have largely debunked the presumption that facts are easily separated from values. Most scholars see journalistic truth as an achievement: forces align to make value-laden accounts born of particular contexts into facts that are stable and legitimate enough to circulate largely without challenge.430 The press’s autonomy requires freedom from forces that undermine its ability to make and publish what it sees as facts, as well as the capacity to connect to and work with sources that enrich and protect its definitional work. Because social facts are largely made and not found, the freedom to create truths requires both separations and dependencies.

Today, many scholars see the power to create journalistic truths as intertwined with sociotechnical systems: structured and standardized content-management systems archive events and classify information, journalists acquire and make databases in the course of investigative reporting, fact-checking software and websites constantly rank and grade the truthfulness of sources, and algorithms claim to provide neutral results but often are inseparable from systematic bias (C. W. Anderson & Kreiss, 2013; Bucher, 2017b; Dörr & Hollnbuchner, 2016; Godler & Reich, 2012; Graves, 2016; S. C. Lewis & Westlund, 2015; Parasie, 2015; M. L. Young & Hermida, 2015). The likelihood that journalists think a particular piece of evidence counts or can be trusted emerges less from traditional processes of reporting and verification and more from what is known about the source of the assertion (C. W. Anderson, 2018). In some ways, this is an old story of journalists figuring out how to trust sources, but in other ways, it is a newer phenomenon in which trust cannot emerge from the newsroom and be created by journalists but lives instead in a distributed set of human and nonhuman actors working together to make truths legitimate, stable, and portable. The networked press’s autonomy lies in its freedom from some of these actors and its capacity to engage, configure, and challenge others. Three dynamics characterize much of the trade press discourse on the press’s sociotechnical construction of truth—collaborative facts, architectural facts, and probable facts.

Collaborative Facts   Much of the trade press describes journalists, technologists, and technologies as co-constructing facts. Andy Carvin (then a reporter with NPR) provided a relatively early example of collaborative fact making when, from Washington, DC, he used social media to report on the 2011 Arab Spring uprisings. He openly acknowledged that he shared information that he could not verify himself and asked his network of followers to determine whether claims about people and events were true.⁴³¹ Such outreach verification also can happen less urgently. After working with experts, New York Times reporter C. J. Chivers could not verify the “identify an unusual cluster bomb found in Libya,” so he asked for help from a blog he trusted to give accurate information and not scoop his story,⁴³² and several reporters use the question-and-answer platform Quora to learn about locations they cannot visit and gauge how much interest there is in a story before investing time and energy verifying information.⁴³³ This same ethic drives the news aggregator Storyful, which published a list of questions it asks of social media content as it tries to determine its veracity, including “Are streetscapes similar to geo-located photos on Panoramio or Google Street View?” and “Are there any landmarks that allow us to verify the location via Google Maps or Wikimapia?”⁴³⁴ Carvin and Storyful expect their audiences to be active, committed to the search for high-quality information, and skilled with the social media platforms that can provide the accuracy journalists cannot. Several dangers arise from such closeness to audiences and technologies. For example, 23 percent of U.S. adults have knowingly or unknowingly shared a false story;⁴³⁵ until it made changes to its algorithm, Google regularly ranked holocaust deniers high in its search rankings;⁴³⁶ and Twitter quickly spread a false report that South Carolina Governor Nikki Haley was about to be indicted,⁴³⁷ a claim that could not be practically countered or corrected because it was virtually impossible to reassemble the Twitter audience that had seen and shared the lie. (The AP includes in its social media guidelines instructions on how to correct erroneous tweets.)⁴³⁸ Relying on audiences and platforms to produce high-quality information and act as verification networks requires that journalists have a kind of second-order trust—a faith in people, policies, and technologies that are often invisible or unexaminable but that become journalists’ collaborators.

In 2016, there were several instances of failed collaborative fact production among humans and nonhumans. After firing the team of former journalists that it had hired to curate its Trending Topics section, Facebook saw a dramatic uptick in the amount of fake news being spread through the section.⁴³⁹ Facebook’s failure to find a good balance between editorial and algorithmic labor meant that its site became a source of misinformation and “fake news,” creating a major scandal several months later and after the U.S. presidential election as the dissemination of fake news became a major focus of the field. Journalists debated how much fake news had spread, who was responsible for its dissemination, why anyone would produce such information, what partnerships (among journalists, technology companies, audiences, educators, and government regulators) would fix the problem, and whether the problem was new and worthy of attention.⁴⁴⁰ In the wake of the fake news scandal, a number of sociotechnical initiatives emerged, including a crowdsourced list of design ideas for journalists and technology companies alike, started by Eli Pariser, who coined the term “filter bubble”;⁴⁴¹ a Harvard-MIT hackathon to invent new truth-making technologies;⁴⁴² an Indiana University project to visualize false news and show people how sharing spreads misinformation;⁴⁴³ and HoaxBot, a Twitter account that automatically responded to people who shared fake news.⁴⁴⁴

Recognizing their joint responsibility—Twitter’s former head of news said that “social media companies alone shouldn’t have to manage fake news”⁴⁴⁵—news organizations and technology companies created a formal consortium to address the problem. Called FirstDraft, it aims to define fake news and work to limit its production and dissemination⁴⁴⁶ and recently created a “trust kit” of best practices and technological tools designed to “rebuild trust with communities.”

These collaborative efforts at creating facts and policing those who spread misinformation reflect a new kind of press autonomy. Journalists cannot simply distance themselves from audiences for fear of encountering falsities because those same audiences are often the ones they need to work with to verify information. Similarly, journalists cannot separate themselves from technology companies that may be more tolerant of misinformation than they are because these same companies are also news distribution channels. The public right to hear and determine the veracity of information is intertwined with a network of journalistic and technological actors with both competing and complementary truth goals.

Architectural Facts   In other moments, the construction of journalistic truth lives in technological architectures—databases and interface designs that certify and signal the veracity of data. Not only is data-driven journalism an increasingly visible and prominent part of the field—there is now an entire news awards program dedicated to data journalism⁴⁴⁷—but data journalism has long been offered as a reason that government databases need to be opened and cleaned and that state officials need to be able to describe the data contained therein.⁴⁴⁸ The data that such journalism uses are often seen to possess an inherent truth separate from reporters’ analyses of the data or journalistic judgments about newsworthiness of the stories that use the data. (In contrast, although the Guardian maintains an extensive data journalism team under the motto “Facts Are Sacred,”⁴⁴⁹ it is self-described as an editorial staff whose work is integral, not ancillary, to story judgments.⁴⁵⁰ And although ProPublica provides and sells data through its online data store, it offers rich context on how data were generated, what they may mean, and how it thinks reporters should use them.)⁴⁵¹ Others, though, seem to suggest separations. The Dallas Morning News is partnering with government officials to collect and visualize data that combines state and user-submitted reports. The project removes data generation from the newsroom, with a Morning News editor saying “we’re the experts at communicating to the public, and they’re the experts at gathering police data.”⁴⁵² Similarly, Detective.io, a platform created by a “network of data journalists and developers” called Journalism++, “lets users input large data sets, predominantly of people or organizations, and then maps the connections between them into an easily searchable database.”⁴⁵³ Some writers suggest that data are facts that come first and are distinct from stories about data, while others caution that data and the facts they suggest must appear in social contexts⁴⁵⁴ (people may agree on a fact’s existence but disagree about its significance) and that “interpreting and presenting data requires making judgments and possibly mistakes” that journalists need to reflect on and audiences deserve to know about.⁴⁵⁵

Outside of news organizations, social media platforms are wrestling with the issue of whether they are or should be sources of facts versus sites of conversation. After the 2016 presidential election’s uproar about fake news, one Facebook executive—after several weeks of disclaiming responsibility for the site hosting fake news and learning that even its “verified” sites spread misinformation⁴⁵⁶—admitted that “we resisted having standards on fake news. That was wrong.”⁴⁵⁷ The company then created an initiative to prevent fake news providers from using its advertising network, made design changes that would caution users against sharing fake news, and gave fact-checking sites access to parts of its architecture that could label misinformation—but Facebook did not define how it defined “fake news,”⁴⁵⁸ nor did it provide funding to fact-checking sites it partnered with (ABC News, Snopes, PolitiFact, FactCheck.org, and AP),⁴⁵⁹ expecting the extra work to occur without compensation.⁴⁶⁰ It seemed to presume that an understanding of the phenomenon existed somewhere, elsewhere, and that its job was to implement that understanding. Similarly, but before the presidential election, Google added “fact-checked” tags to news articles, working with a community of news organizations to “follow the commonly accepted criteria for fact checks” and create an algorithm that “determines whether an article might contain fact checks.”⁴⁶¹ In both cases, platforms are struggling to locate fact checking within computational infrastructure, developing with journalists automated and algorithmic processes designed to approximate their expertise.

Probable Facts   The idea of approximation—Google’s fact-checking algorithm promises to determine “whether an article might contain fact checks”⁴⁶²—and related notions of probability and likelihood appear throughout trade press discussions of the sociotechnical systems that make journalistic facts. The networked press’s freedom to make and publish what it considers to be “true” information is only a probable freedom—an autonomy tied to information infrastructures designed to offer likelihoods, not categorical conclusions. The chance of the press trafficking in what it would consider truthful information lives in distributed architectures that, at any given moment and for any particular distribution, only approximate facts.

For example, several news organizations have created fact-checking algorithms that aim to operate at large scale and in real time and to address as many falsities as architectures allow. In 2012, the Washington Post proposed the idea of a real-time fact-checking app that would dynamically “parse audio from a speech or TV ad” and “use speech-to-text technology to gather the audio and compare it against a database of statements, stats and other data that had been fact-checked.” Ambitiously called Truth Teller, the project was scaled back to a simpler internal database of fact-checked statements⁴⁶³ and later expanded to a real-time video analyzer that evaluated transcripts for misinformation,⁴⁶⁴ but the vision of a real-time, real-world truth meter appears in several discussions of what journalists ideally want in order to manage the misinformation in media environments.⁴⁶⁵ Although skepticism of automated fact-checking persists among journalists experienced with the challenges of fact-checking political statements,⁴⁶⁶ the United Kingdom’s independent fact-checking organization Full Fact published a “roadmap” of automated fact-checking technologies,⁴⁶⁷ including the University of Texas at Arlington’s ClaimBuster platform, which uses machine learning techniques to identify potentially false statements that users can manually annotate;⁴⁶⁸ Le Monde’s “decoding machine,” which helps readers search through politicians’ previously fact-checked statements and reduce the likelihood that they will believe and share erroneous statements;⁴⁶⁹ the news photography service Scoopshot and the Guardian’s GuardianWitness project, which rely partly on analyzing an image’s embedded metadata and comparing it against an analysis of the image itself, with the aim being to alert readers and journalists alike to images that may have been tampered with;⁴⁷⁰ and Reuters’s News Tracer algorithm, which was designed to predict the likely emergence of breaking news on Twitter and uses a complex set of dynamic signals to evaluate a tweet’s authenticity (its designer remarked, for example, that “a tweet that is entirely in capital letters is less likely to be true”).⁴⁷¹

No sociotechnical system claims to guarantee the truth or accuracy of any of the information it processes, but each is designed to reduce the likelihood that people share misinformation and increase the chances that machines will flag and reject falsities.

Resemblances

A somewhat small and abstract—but growing and powerful—dimension of networked press freedom centers on the sociotechnical creation of familiarity or difference—technologies and practices designed to automate news translations, simulate social environments, and collapse international distances. The Washington Post, for example, is experimenting with processes to translate its online stories into languages other than English so that it can create dynamic multilingual versions of content that will let users switch between languages. The main hurdle to overcome in creating more multilingual content is finding staff time for the translations and thinking through the implications of hosting multiple versions of a single story, which are bound to have different nuances in different languages.472 The European network VoxEurop has solved this by creating a consortium of news organizations that share stories and translations, coordinating through a custom platform designed to create fast and vetted translations.⁴⁷³ Some writers are concerned, though, that projects such as these—which focus on human translations by news professionals—will be rapidly eclipsed by systems that automatically translate news stories. In the summer of 2016, for example, Facebook introduced its “multilingual composer” system, which lets authors create posts that Facebook can then translate to forty-five other languages.⁴⁷⁴ The company acknowledges the limitations of machine translation but gives authors the opportunity to edit translations and thus improve its algorithm. For news organizations, there is a tradeoff: they have an opportunity to translate content into a host of other languages automatically (and thus gain access to new audiences and advertisers), but they also risk publishing translations that contain errors and that lack the subtle nuances reporters use to create compelling narratives that people enjoy reading.

A related challenge is the opportunity to create computationally simulated news environments—augmented reality (AR) and virtual reality (VR) experiences designed to create empathy among viewers as well as contexts they could never experience without such technology.⁴⁷⁵ The New York Times,⁴⁷⁶ the Guardian,⁴⁷⁷ the Verge,⁴⁷⁸ Frontline,⁴⁷⁹ and several other Knight Foundation projects have developed journalism AR and VR projects.⁴⁸⁰ Some commentators claim that they foster empathy among audiences⁴⁸¹—a goal in Schudson’s (2008b) ideal model of the press—and that they offer a new form of storytelling that news organizations should invest in.⁴⁸² Others question whether journalism should be in the business of simulating environments that do not or could not exist without technology.⁴⁸³ The crux of the issue is one of separations and dependencies: new technologies may be able to collapse the distances audiences feel from people and places, but such proximity may not be what ethical journalistic witnessing needs, and such simulations may not bear enough resemblance to offline environments to create the kind of empathy that public change demands.⁴⁸⁴ Perhaps people should feel distance in order to realize the power they have to help sufferers, or it may be that too much closeness can create compassion fatigue that prevents long-term witnessing and sustained action (Ananny, 2015; Chouliaraki, 2006; Silverstone, 2007).

Finally, this question of “proper distance” (Silverstone, 2003) is addressed as news organizations both re-create and challenge what it means for online news to be “foreign.” WNYC’s On the Media weekly radio program asked why terrorist attacks in Paris received far more attention than those in Baga—a question it and its guest Ethan Zuckerman answered, in part, by arguing that online news bubbles leave many readers in highly localized contexts, despite the potential for news technology to help them care about other places.⁴⁸⁵ BuzzFeed addressed this question by jettisoning the structure of foreign news entirely, building its coverage around what it called “themes” instead of geography.⁴⁸⁶ And Facebook’s safety notification system is a kind of foreign news agenda setter, checking in with users and issuing alerts after disasters in certain locations only—notifications that trigger news coverage.⁴⁸⁷ Through a set of sociotechnical moves, the networked press signals which proximities are salient and powerful—which publics and issues it sees as within its purview and relevant to its autonomy.

Affect

Although the first image of the Western press is often that of a deliberative institution designed to equip rational citizens with good information (Schudson, 2003a), emotion has always played an important but usually overlooked role in journalism (Beckett & Deuze, 2016; J. M. Fishman, 2003). Journalists often create stories that are purposefully designed to move audiences and enroll them as collective witnesses, helping them feel connected to each other and able to appreciate the shared “structures of feeling” that make empathy and collective action possible (Maier, Slovic, & Mayorga, 2016; R. Williams, 1977; Zelizer, 2010). Emotional storytelling is often a way for journalists to sidestep the usual rituals of detached objectivity, revealing their own feelings and modeling for audiences how they might feel, as long as such feelings align with dominant cultural expectations of how people are supposed to feel in a particular context. Many of history’s Pulitzer Prize–winning stories are not exemplars of unemotional neutrality but are accounts that adhere to rituals of objectivity while simultaneously using “subjective language” to make “emotive appeals” to readers (Wahl-Jorgensen, 2013, p. 305). Although still understudied, emotion also plays a key role in the networked press, with audience emotions (Papacharissi, 2015) and algorithmically induced feelings (Bucher, 2017a) found to be key organizing structures of online culture.

Affect is a key dimension of press autonomy: it involves the power to author a story’s tone, imbue feelings with audiences, set the emotional tenor of media events, and select which audience emotions to sense and acknowledge. These dynamics are not only about journalists becoming free to pursue emotion as they see fit but also about the networked press’s collective capacity to realize affect—and publics through affect.

As with the theme of resemblance, affect is a small but increasingly prevalent element of trade press discourse on the press’s sociotechnical dynamics. For example, the New York Times highlighted its “Snow Fall” story—a “compelling destination” with a highly evocative aesthetic in which readers were immersed in a multimedia narrative—as a powerful and emotionally suggestive piece,488 and other news organizations immediately followed suit. A Quartz editor remarked, “We are focused on building tools to create Snow Falls every day, and getting them as close to reporters as possible.… I’d rather have a Snow Fall builder than a Snow Fall.”⁴⁸⁹ The story stood out not just as an example of evocative storytelling but as an inspiration to develop infrastructures—tools, people, practices—that could infuse affect across the news.

The affective dimension of networked press autonomy is largely the story of having the power to control emotion through infrastructure. Shortly after Facebook expanded its Like button to include a fixed set of emotions that people can use to standardize reactions to stories (and make such emotions easier to track and commodify), technology startup Antenna attempted to replicate the model for news. Its software—which requires a Facebook account to use—lets publishers create a customized set of emotional responses that readers can then apply to headlines, story text, or associated media. It is billed as “a nearly abuse-proof way to get feedback from an audience, and in a way that reinforces the brand language.” Predefined categories of emotions are seen as a way to get closer to audiences, be consistent with a brand, and build data models of readers’ affective states.

Sometimes, though, infrastructures force readers to be close to or far from highly charged media. For example, Google News’s editorial staff decided that it wanted to highlight upbeat stories while creating “factoids” of 2014 World Cup searches, but it directed its team of writers and data scientists to avoid stories it saw as negative (those containing words like destroy, defeat, humiliate, and shame). In choosing which search patterns to highlight as trends, it avoided language it thought was pessimistic or off-putting. As one staffer put it “a negative story … won’t necessarily get a lot of traction in social [media].”⁴⁹⁰ Journalists citing Google’s online trends as neutral statistics would leave audiences with the impression that people were far more upbeat and positive than their searches suggested.

Such close associations with technology companies can create incorrect assumptions for news audiences and also elicit unintended emotions. When “horrifying video of two journalists being shot and killed … in Virginia during a live broadcast on WDBJ-TV”⁴⁹¹ was copied from news sites’ online streams and shared on social media, some people became unintended audiences. At the time, Facebook and Twitter both played videos automatically, so people scrolling through their tweet streams or news feeds were immediately shown an emotionally charged scene that they may not have chosen to watch and that news organizations customarily issue warnings about before showing.⁴⁹² In the past, news organizations have mistakenly shown graphic scenes because of their desire to stay close to live events. For example, against the wishes of the anchor, Fox News mistakenly showed a suicide during live coverage of a car chase and immediately apologized for doing so.⁴⁹³ The difference in this case is that the decision to broadcast or watch emotionally charged material was not a conscious one made by news organizations, social media platforms, or audiences but instead was the product of a confluence of forces. The public right to hear—or avoid—disturbing content was an infrastructural outgrowth of editorial judgment, technological design, and audience practices.

Conclusion

Networked press freedom is the story of people and machines coming together and pulling away from often invisible and unacknowledged assumptions about what kind of press publics need. Paywalls bring revenue but shut out unpaying audiences; social media platforms deliver news faster to more audiences but at the expense of privacy, fact checking, and emotional ethics; data-driven newsrooms offer real-time images of audiences and the value of journalists’ labor but often uneasily integrate editorial and commercial standards; and crowdfunding sites bring new sources of revenue but can turn readers into clients and subtly suggest that journalists should work on what crowds see as popular instead of what they see as important. The separations and dependencies that structure each dimension of press autonomy beg larger questions: what exactly should the press be, and how can normative balances be struck through sociotechnical infrastructures?

I do not claim that this typology captures every dimension of the networked press from 2010 to 2016. I could have discussed different examples, and I am probably unaware of others. Additionally, the stories analyzed here do not necessarily contain the conclusions that scholars would draw about these people, technologies, organizations, or events. Researchers would likely dispute many of the accounts that journalists offer about what they do, why it matters and what role technologies play in networked journalism.

Instead, the typology is an account of the themes that dominate the stories the networked press tells itself about itself as well as an interpretation of how these themes reveal separations and dependencies among the humans and machines that, together, create the conditions under which journalists work, news is produced, audiences convene—and publics form. It is meant to be both a snapshot of a moment in time as well a framework for examining future examples and sociotechnical forces. New stories will arise, new technologies will be invented, and new claims will be made, but what I suspect will endure is a new way of thinking about press autonomy. Instead of seeing press freedom as journalists’ freedom from anyone but themselves to pursue their visions of the public, it might more accurately be described as the coming together of humans and nonhumans—often unintentionally, invisibly, and with different types of power—to make publics. If you want a particular type of public, you need a certain type of free press and thus a specific set of relationships among journalism’s humans and machines. Conversely, going forward, as journalists and technologists invent new types of human and machine relationships, they might do so with the knowledge that they are simultaneously creating publics. If such arrangements result in publics that are defensible—that ensure, for example, the public right to hear that individuals need to self-govern—then we might look to them as exemplars of how to create proximity among journalism’s humans and machines. But if we find ourselves with unacceptable publics or too narrow a set of possible publics, then we might purposefully intervene in the networked press’s separations and dependencies to create new normative visions of journalism and self-governance.

Notes

1. http://niemanreports.org/articles/a-brief-guide-to-robot-reporting-tools.

2. http://www.niemanlab.org/2014/09/watching-what-happens-the-new-york-times-is-making-a-front-page-bet-on-real-time-aggregation.

3. http://www.poynter.org/2013/propublica-releases-simple-tool-for-searching-instagram/212719.

4. http://www.niemanlab.org/2015/06/how-a-group-of-researchers-tried-to-use-social-media-data-and-algorithms-to-find-breaking-news.

5. Although it did not appear in the trade press corpus, see Neil Thurman, Steve Schifferes, Richard Fletcher, Nic Newman, Stephen Hunt, and Aljosha Karim Schapals (2016) for a thoughtful scholarly account of a similar tool, SocialSensor, which was designed by European researchers to help journalists identify trending stories.

6. http://www.theatlantic.com/business/archive/2016/12/information-bubbles/509039; https://www.opensecrets.org/resources/learn/anomalies.php.

7. https://projects.propublica.org/docdollars.

8. https://www.propublica.org/getinvolved/item/reporting-recipe-investigating-restraint-and-seclusion-in-us-schools.

9. https://www.propublica.org/datastore.

10. http://www.dronejournalismlab.org/manual.

11. http://www.nytimes.com/2015/01/16/business/media/10-companies-join-effort-to-test-drones-for-newsgathering.html.

12. http://www.poynter.org/2015/what-the-faas-newly-proposed-drone-rules-mean-to-journalists/320079.

13. http://www.cjr.org/q_and_a/new_drone_rules_journalists_faq.php.

14. https://www.noflyzone.org.

15. http://www.roboticstrends.com/article/dji_blocks_drone_flights_in_washington_dc_after_white_house_crash.

16. https://www.nytimes.com/2014/11/03/us/police-targeted-media-with-no-fly-zone-over-ferguson-tapes-show.html.

17. See Matt Carlson (2016) and Andreas Graefe (2016) for recent summaries of this domain; Michael A. Beam (2014); Graefe, Mario Haim, Bastian Haarmann, and Hans-Bernd Brosius (2016) for a study of people’s perception of algorithmically generated news stories; Neil Thurman et al. (2016) for an account of creating algorithmically produced news; and Tetyana Lokot and Nicholas Diakopoulos (2016) for a review of bot-based journalism.

18. See Christopher W. Anderson (forthcoming a) for a historical and contemporary account of structured data’s influence on journalistic production.

19. http://www.niemanlab.org/2014/03/when-robots-help-human-journalists-this-post-was-created-by-an-algorithm-written-by-the-author; http://www.nytimes.com/2015/03/08/opinion/sunday/if-an-algorithm-wrote-this-how-would-you-even-know.html.

20. http://niemanreports.org/articles/automation-in-the-newsroom.

21. https://www.propublica.org/nerds/item/how-to-edit-52000-stories-at-once.

22. http://www.theatlantic.com/technology/archive/2014/03/earthquake-bot-los-angeles-times/359261; http://www.theatlantic.com/technology/archive/2014/03/how-a-california-earthquake-becomes-the-news-an-extremely-precise-timeline/284506.

23. https://insights.ap.org/industry-trends/study-news-automation-by-ap-increases-trading-in-financial-markets.

24. http://www.niemanlab.org/2016/10/the-new-york-times-is-using-a-facebook-messenger-bot-to-send-out-election-updates.

25. http://digiday.com/platforms/washington-chatbots-news.

26. http://www.niemanlab.org/2016/07/developed-with-young-readers-in-mind-cnns-new-kik-bot-gives-you-the-basics-on-big-news-stories.

27. http://variety.com/2015/digital/news/nbc-news-is-first-tv-news-service-to-connect-with-kik-mobile-chat-service-1201462575.

28. http://www.niemanlab.org/2016/02/with-an-interface-that-looks-like-a-chat-platform-quartz-wants-to-text-you-the-news-in-its-new-app.

29. http://www.niemanlab.org/2016/11/quartz-launches-its-bot-studio-with-a-quarter-million-from-knight-and-plans-for-slack-and-amazon-echo.

30. www.niemanlab.org/2016/05/the-washington-post-tests-personalized-pop-up-newsletters-to-promote-its-big-stories.

31. http://www.niemanlab.org/2016/02/how-the-washington-post-built-its-tool-to-re-engage-the-attention-of-distracted-readers-on-mobile.

32. http://www.nytimes.com/interactive/2015/12/15/upshot/the-best-places-for-better-cheaper-health-care-arent-what-experts-thought.html.

33. http://publiceditor.blogs.nytimes.com/2015/12/17/when-news-gets-too-personal.

34. http://www.niemanlab.org/2016/10/the-ap-wants-to-use-machine-learning-to-automate-turning-print-stories-into-broadcast-ones.

35. For overviews of computational and communicative issues in designing embodied conversational agents, see Timothy Bickmore and Justine Cassell (2005) and Cassell, Joseph Sullivan, Scott Prevost, and Elizabeth F. Churchill (2000).

36. He defines the best dialogues as civil, governed by norms and theories of publicness, and integrated with other types of media.

37. In December 2016, Facebook CEO Mark Zuckerberg acknowledged that Facebook is not a “traditional” media company, saying “we do a lot more than just distribute news, and we’re an important part of the public discourse” (Gibbs, 2016, n.p.). And in 2011, Google’s Larry Page stated that “trying to improve media” was part of Google’s responsibility (Myers, 2011, n.p.).

38. https://www.facebook.com/notes/facebook-media/12-best-practices-for-media-companies-using-facebook-pages/518053828230111.

39. https://support.google.com/news/publisher/answer/40787?hl=en.

40. http://www.niemanlab.org/2011/06/twitter-for-newsrooms-as-a-relationship-building-guide.

41. http://www.wsj.com/articles/facebook-signs-deals-with-media-companies-celebrities-for-facebook-live-1466533472.

42. http://adage.com/article/media/twitter-partners-bbc-news-embargo-8/244897.

43. http://www.niemanlab.org/2015/03/a-wave-of-distributed-content-is-coming-will-publishers-sink-or-swim.

44. http://www.nytimes.com/2015/05/13/technology/facebook-media-venture-to-include-nbc-buzzfeed-and-new-york-times.html.

45. http://www.poynter.org/2016/facebook-is-opening-up-instant-articles-to-newsrooms-everywhere-will-a-flood-of-distributed-content-follow/397305.

46. https://media.fb.com/2015/05/12/instantarticles.

47. http://www.adweek.com/socialtimes/the-new-york-times-notes/426652.

48. http://www.niemanlab.org/2016/02/diving-all-in-or-dipping-a-toe-how-publishers-are-approaching-googles-accelerated-mobile-pages-initiative; http://www.niemanlab.org/2016/02/a-qa-with-googles-head-of-news-richard-gingras-on-its-vision-for-the-accelerated-mobile-pages-project.

49. http://www.niemanlab.org/2015/10/twitter-unveils-its-own-news-digests-with-mentions-and-some-news-orgs-can-participate.

50. http://www.niemanlab.org/2015/09/snapchat-wants-to-slip-a-little-news-into-teens-social-smartphone-time.

51. http://www.niemanlab.org/2015/09/from-two-day-shipping-to-news-a-cheaper-washington-post-is-now-an-amazon-prime-benefit.

52. http://www.businessinsider.com/facebook-news-feed-update-banishes-memes-2013-12.

53. http://www.journalism.org/2015/07/14/news-habits-on-facebook-and-twitter; http://www.poynter.org/2011/yahoo-users-searched-for-more-international-news-than-bing-users/154690.

54. http://www.niemanlab.org/2011/09/with-its-standout-tag-google-news-is-giving-publishers-new-incentive-to-credit-the-competition.

55. http://www.nytimes.com/2016/04/25/business/vox-media-tries-something-old-on-something-new.html.

56. http://www.niemanlab.org/2016/08/alexa-give-me-the-news-how-outlets-are-tailoring-their-coverage-for-amazons-new-platform.

57. http://www.niemanlab.org/2014/01/now-thats-adaptive-design-the-verge-show-different-versions-of-a-story-to-apple-google-or-microsoft-users.

58. http://www.poynter.org/2013/new-york-times-experiments-with-tweetable-highlights-in-snl-story/221911.

59. http://www.niemanlab.org/2015/10/instant-articles-get-shared-more-than-old-fashioned-links-plus-more-details-from-facebooks-news-push.

60. https://media.fb.com/2016/12/01/how-instant-articles-benefits-local-news.

61. http://www.niemanlab.org/2012/02/how-npr-drove-traffic-to-a-local-station-by-geotargeting-stories-on-facebook.

62. http://www.businessinsider.com/facebook-news-feed-update-banishes-memes-2013-12; http://www.adweek.com/news/technology/facebook-algorithm-tweaks-hurt-viral-sites-more-other-publishers-156533.

63. http://mashable.com/2014/07/13/facebook-politics-partisan-newsfeed.

64. http://www.nytimes.com/2016/06/30/technology/facebook-to-change-news-feed-to-focus-on-friends-and-family.html.

65. http://www.nytimes.com/2016/08/05/technology/facebook-moves-to-push-clickbait-lower-in-the-news-feed.html.

66. http://digiday.com/publishers/publishers-enjoy-facebook-traffic-boon; http://www.niemanlab.org/2014/08/sharing-on-facebook-surges-at-the-telegraph.

67. https://theawl.com/platform-patched-38e0a127576f#.kjzrboa4d; https://gigaom.com/2014/02/18/reality-check-no-upworthys-traffic-didnt-get-crushed-by-facebooks-algorithm-change.

68. http://www.niemanlab.org/2014/08/mastering-the-dark-arts-facebook-has-been-the-key-to-mother-jones-growing-popularity-online.

69. http://www.poynter.org/2014/mental-floss-a-big-winner-after-facebooks-mysterious-high-quality-algorithm-change/239917.

70. http://digiday.com/publishers/bleacher-report-facebook-proofing-traffic.

71. http://www.niemanlab.org/2010/01/what-qualifies-as-a-spotlight-story-on-google-news-heres-a-few-clues; http://www.niemanlab.org/2013/08/googles-new-search-feature-makes-it-easier-to-find-seminal-articles-on-big-topics.

72. http://www.niemanlab.org/2011/09/with-its-standout-tag-google-news-is-giving-publishers-new-incentive-to-credit-the-competition.

73. https://support.google.com/news/publisher/answer/93992?hl=en; https://support.google.com/news/publisher/answer/4582731.

74. http://www.poynter.org/2013/new-facebook-feature-suggests-which-content-news-orgs-should-share/226689.

75. http://www.adweek.com/socialtimes/twitter-embedding-articles/487203.

76. As Susan Leigh Star (1999, p. 382) argues, the “normally invisible quality of working infrastructure becomes visible when it breaks … [creating] the basis for a much more detailed understanding of the relational nature of infrastructure.”

77. http://www.cjr.org/analysis/facebook_and_media.php; http://www.cjr.org/tow_center/facebook_zuckerberg_trump_election.php.

78. http://www.poynter.org/2013/twitter-apologizes-for-blocking-newspapers-urls/228372.

79. http://www.pottsmerc.com/article/20130810/NEWS01/130819954#full_story.

80. http://www.npr.org/sections/thetwo-way/2016/09/09/493267919/after-facebook-censors-iconic-photo-norwegian-newspaper-pushes-back.

81. http://www.nytimes.com/2016/09/10/technology/facebook-vietnam-war-photo-nudity.html.

82. Although not a news organization, the American Civil Liberties Union had its account suspended after it “posted a link on Facebook to a blog post about the debate over a statue in a public part outside of Kansas City, which depicts a nude woman taking a photo of her exposed breasts” (D. Cohen, 2013, n.p.).

83. http://mashable.com/2014/08/18/washington-post-amazon-buy-button/#yE0OgGN2asqL.

84. https://www.theguardian.com/technology/2014/aug/21/twitter-free-speech-champion-selective-censor.

85. https://www.fastcompany.com/3008881/tracking/circas-object-oriented-approach-to-building-the-news.

86. http://www.niemanlab.org/2013/06/when-editors-design-controlling-presentation-in-structured-content.

87. https://www.propublica.org/nerds/item/why-develop-in-the-newsroom; https://hackerjournalist.net/2013/07/18/why-i-work-in-a-newsroom.

88. https://source.opennews.org/en-US/articles/boyer-interview; https://www.journalism.co.uk/news/recalculating-the-newsroom-the-rise-of-the-journo-coder-/s2/a555646.

89. http://www.niemanlab.org/2013/03/jake-levine-why-learning-to-code-isnt-as-important-as-learning-to-build-something.

90. http://www.niemanlab.org/2014/01/the-new-york-times-rd-lab-is-building-a-quantified-self-semantic-analysis-tool-to-track-web-browsing.

91. http://www.niemanlab.org/encyclo/hackshackers.

92. http://www.niemanlab.org/2013/03/opennews-learning-wants-to-provide-lessons-to-developers-in-and-out-of-newsrooms.

93. http://www.niemanlab.org/2011/08/transparency-iteration-standards-knight-mozillas-learning-lab-shares-lessons-of-open-source-for-journalism.

94. http://www.niemanlab.org/2012/07/context-code-and-community-source-is-one-stop-shopping-for-newsroom-developers.

95. http://www.niemanlab.org/2013/10/qa-guardian-social-and-community-editor-joanna-geary-heads-off-to-twitter-u-k.

96. http://money.cnn.com/2015/09/29/media/liz-heron-facebook-huffington-post/index.html.

97. http://www.wsj.com/articles/the-rise-of-the-publishing-platform-specialist-1458896683.

98. https://www.theguardian.com/technology/2016/dec/12/facebook-advertises-for-head-of-news-us-election-concerns-fake-news.

99. http://www.poynter.org/2015/eric-zuckerman-named-news-partnerships-manager-at-twitter/355615.

100. http://www.poynter.org/2011/bbc-reporter-takes-her-60000-twitter-followers-to-competitor-itv/140566.

101. http://mediashift.org/2014/07/remix-put-data-journalism-into-every-entry-level-j-school-class; http://www.cjr.org/analysis/data.php; http://www.niemanlab.org/2011/01/dave-winer-how-can-universities-educate-journo-programmers.

102. http://www.niemanlab.org/2016/10/facebook-launches-online-courses-for-journalists-about-what-else-using-facebook.

103. http://www.niemanlab.org/2015/12/a-new-class-at-usc-is-teaching-students-how-to-produce-stories-specifically-for-social-media.

104. http://knightlab.northwestern.edu/2016/09/15/software-developers-interested-in-journalism-northwestern-and-the-washington-post-want-you.

105. https://journalists.org/programs/ap-google-scholarship.

106. http://www.niemanlab.org/2011/11/how-time-inc-is-preparing-for-a-future-in-digital-news-with-a-j-school-of-its-own.

107. http://www.adweek.com/fishbowlny/code-with-me-journalist-training-workshop-headed-to-portland/259146.

108. http://gizmodo.com/they-treated-us-like-garbage-former-trending-news-writ-1789248355; http://gizmodo.com/want-to-know-what-facebook-really-thinks-of-journalists-1773916117.

109. http://www.forbes.com/sites/leifwalcutt/2016/08/26/facebook-lays-off-journalists-from-trending-topics-replaces-with-algorithm/#71e0d9fa341b.

110. http://www.cjr.org/innovations/tony_haile_chartbeat.php.

111. http://www.poynter.org/2011/new-generation-of-web-analytics-applies-big-data-to-newsroom-decisions-visual-revenue-jumptime/143389.

112. http://www.nytimes.com/2010/09/06/business/media/06track.html.

113. http://www.niemanlab.org/2012/08/metrics-metrics-everywhere-how-do-we-measure-the-impact-of-journalism; http://www.greglinch.com/2012/01/quantifying-impact-a-better-metric-for-measuring-journalism.html.

114. https://www.journalism.co.uk/news/measuring-the-news-what-are-the-alternatives-to-pageviews-/s2/a556640.

115. https://source.opennews.org/en-US/articles/promotion-pageviews.

116. http://www.niemanlab.org/2013/11/the-only-metric-that-matters-total-time-reading.

117. http://www.adweek.com/fishbowlny/upworthy-ditches-pageviews-and-uniques/266828.

118. http://newsroom.fb.com/news/2015/06/news-feed-fyi-taking-into-account-time-spent-on-stories.

119. http://www.adweek.com/fishbowlny/guardian-attention-analytics/266853.

120. http://streetfightmag.com/2014/06/12/attention-and-timing-are-the-new-clicks-chartbeat-says.

121. https://gigaom.com/2014/09/29/chartbeat-gets-certified-to-measure-attention-tries-to-move-advertising-away-from-clicks-and-pageviews.

122. www.niemanlab.org/2016/08/the-daily-signal-created-a-custom-engagement-index-to-help-it-prioritize-all-its-platforms.

123. http://digiday.com/publishers/christian-science-monitor-going-maximum-engagement.

124. http://www.niemanlab.org/2015/06/tracking-impact-newslynx-goes-beyond-google-analytics-to-measure-a-storys-path-in-the-real-world.

125. http://thenextweb.com/media/2015/04/13/how-the-guardians-ophan-analytics-engine-helps-editors-make-better-decisions.

126. http://www.niemanlab.org/2016/03/the-ft-is-launching-a-new-analytics-tool-to-make-metrics-more-understandable-for-its-newsroom.

127. http://www.niemanlab.org/2016/07/the-new-york-times-is-trying-to-narrow-the-distance-between-reporters-and-analytics-data.

128. http://www.niemanlab.org/2014/04/building-an-analytics-culture-in-a-newsroom-how-npr-is-trying-to-expand-its-digital-thinking.

129. http://www.niemanlab.org/2015/09/facebook-rolls-out-new-tools-to-help-reporters-share-their-work-and-choose-who-sees-it.

130. http://www.niemanlab.org/2014/08/when-it-comes-to-chasing-clicks-journalists-say-one-thing-but-feel-pressure-to-do-another; http://www.niemanlab.org/2010/09/what-impact-is-seo-having-on-journalists-reports-from-the-field; http://towcenter.org/research/traffic-factories.

131. http://www.wsj.com/articles/washington-posts-bandit-tool-optimizes-content-for-clicks-1454960088.

132. http://www.niemanlab.org/2014/05/watching-the-audience-move-a-new-york-times-tool-is-helping-direct-traffic-from-story-to-story.

133. http://ajr.org/2014/03/19/analytics-news-sites-grapple-can-see-data.

134. https://www.nytimes.com/2014/03/24/business/media/risks-abound-as-reporters-play-in-traffic.html.

135. https://www.washingtonpost.com/blogs/erik-wemple/wp/2014/02/26/daily-caller-implements-hybrid-salaryinternet-incentive-pay-plan.

136. http://ajr.org/2014/03/27/pay-per-visit-debate-chasing-viral-traffic-hurting-journalism.

137. http://talkingbiznews.com/1/thestreet-com-seeks-contributors.

138. http://www.poynter.org/2014/advance-defends-bonuses-for-reporters-who-post-frequently-and-join-comment-chains/244808.

139. http://www.niemanlab.org/2014/10/an-interesting-way-to-pay-reporters-give-them-the-revenue-from-their-own-google-adsense-ads.

140. http://gawker.com/time-inc-rates-writers-on-how-beneficial-they-are-to-1623253026.

141. http://pressthink.org/2010/10/the-100-percent-solution-for-innovation-in-news; http://mediashift.org/2010/07/your-guide-to-next-generation-content-farms200.

142. http://www.poynter.org/2014/as-reporters-get-measured-why-even-buzzfeed-upworthy-arent-beholden-to-numbers/244541; http://ajr.org/2014/04/15/give-journalists-right-metrics-pay-performance.

143. http://www.niemanlab.org/2016/02/the-next-step-moving-from-generic-analytics-to-editorial-analytics.

144. A complete review of journalism’s temporalities is beyond the scope of this book, but for excellent general discussions of how journalism shapes and reflects social senses of time, see Kevin G. Barnhurst (2011); Allan Bell (1995); Dan Berkowitz (1992); Daniel Dayan and Elihu Katz (1994); Jill A. Edy (1999); Richard B. Kielbowicz (2015); Harvey Molotch and Marilyn Lester (1974); Motti Neiger, Oren Meyers, and Eyal Zandberg (2011); Neiger and Keren Tenenboim-Weinblatt (2016); Michael Schudson (1986); Tenenboim-Weinblatt (2013, 2014); and Zelizer (2014).

145. For discussions of contemporary networked news temporality, see Mike Ananny (2016b); Henrik Bødker (2017); Frode Guribye and Lars Nyre (2016); Matthew Jones, Kamal Munir, Wanda Orlikowski, and Jochen Runde (2008); Christine Larson (2015); Zvi Reich and Yigal Godler (2014); James Stanyer and Sabrina Mihelj (2016); Judy Wajcman (2014); Esther Weltevrede, Anne Helmond, and Carolin Gerlitz (2014).

146. http://www.niemanlab.org/2013/09/journalism-has-become-unstuck-in-time.

147. http://www.niemanlab.org/2014/05/immediacy-vs-importance-the-tension-underlying-how-the-nytimes-com-homepage-gets-made.

148. http://www.bbc.com/news/blogs-echochambers-27553248.

149. http://www.theatlantic.com/business/archive/2015/03/facebook-google-and-the-economics-of-time/387877.

150. http://www.poynter.org/2011/pulitzers-shift-to-all-digital-entry-format/154546.

151. http://www.niemanlab.org/2011/10/is-the-huffington-post-reinventing-the-art-of-liveblogging; http://www.niemanlab.org/2012/03/practice-pray-and-have-redundancies-on-the-imperfect-science-of-liveblogging.

152. https://www.washingtonpost.com/blogs/erik-wemple/wp/2015/10/14/what-is-the-new-york-times-doing-with-breaking-news-unit.

153. http://www.niemanlab.org/2012/08/live-broadcast-why-the-huffington-post-and-boston-com-are-getting-into-streaming-media.

154. http://www.niemanlab.org/2012/03/embracing-the-stream-itvs-new-twitter-inspired-news-site-breaks-the-days-news-into-pieces.

155. http://www.niemanlab.org/2012/07/how-buzzfeed-wants-to-reinvent-wire-stories-for-social-media; http://www.theatlantic.com/technology/archive/2013/04/breaking-news-is-broken-could-buzzfeed-be-the-one-to-fix-it/275310.

156. http://www.niemanlab.org/2011/10/adam-moss-ny-mag-publishes-new-content-every-six-minutes.

157. http://www.npr.org/sections/ombudsman/2016/02/18/467123974/live-from-d-c-its-the-npr-newscast.

158. https://www.poynter.org/2015/how-should-journalists-navigate-meerkat-periscope-kik-and-what-the-heck-is-tarsii/331251; http://www.ibtimes.com/twitters-periscope-infiltrates-tv-newsrooms-reporters-cnn-cbs-nbc-others-try-live-1866848; http://www.recode.net/2016/4/6/11585888/facebook-paying-media-partners-like-buzzfeed-to-livestream.

159. http://www.niemanlab.org/2014/09/watching-what-happens-the-new-york-times-is-making-a-front-page-bet-on-real-time-aggregation; https://www.journalism.co.uk/news/why-the-new-york-times-has-launched-new-homepage-stream-watching-/s2/a562568.

160. http://sarahmarshall.io/post/107708899448/using-the-quotable-tool-for-real-time-social.

161. http://www.niemanlab.org/2012/09/first-look-spundge-is-software-to-help-journalists-to-manage-real-time-data-streams.

162. https://gigaom.com/2014/05/16/grasswire-founder-austen-allred-is-trying-to-build-a-wikipedia-style-platform-for-real-time-news.

163. http://www.nytimes.com/2015/03/08/opinion/sunday/if-an-algorithm-wrote-this-how-would-you-even-know.html.

164. http://thomsonreuters.com/content/dam/openweb/documents/pdf/tr-com-financial/news-analytics-product-brochure--oct-2010.pdf.

165. https://www.journalism.co.uk/news/mobile-journalism-sky-news-reporters-broadcast-ready-in-90-seconds-/s2/a555282.

166. http://www.niemanlab.org/2016/06/the-guardian-is-experimenting-with-interactive-auto-updating-push-alerts-to-cover-big-stories.

167. http://www.niemanlab.org/2016/06/the-guardian-is-experimenting-with-interactive-auto-updating-push-alerts-to-cover-big-stories.

168. http://www.niemanlab.org/2016/04/you-can-now-get-personalized-breaking-news-alerts-on-slack.

169. http://www.niemanlab.org/2014/06/the-notification-knows-where-you-are-breaking-news-debuts-news-alerts-tied-to-your-location.

170. http://www.cjr.org/tow_center/mobile_notifications_changing_new_york_times.php; http://www.nytimes.com/times-insider/2015/04/01/this-just-in-behind-the-news-alert.

171. https://www.washingtonpost.com/pr/wp/2016/03/23/how-the-washington-post-improved-the-speed-of-breaking-news-email-alerts/?utm_term=.710e92e75e1f; http://bleacherreport.com/articles/1124741-inside-media-on-breaking-news-sourcing-credit-scoops-in-sports.

172. http://www.poynter.org/2013/ap-sends-rare-flash-for-mandelas-death/232611.

173. http://www.adweek.com/socialtimes/study-twitter-rarely-beats-news-outlets-to-breaking-news/132842.

174. http://www.journalism.org/2015/07/14/the-evolving-role-of-news-on-twitter-and-facebook.

175. http://www.poynter.org/2012/when-news-breaks-most-americans-seek-a-second-trusted-source-for-more-info/190640.

176. http://www.niemanlab.org/2014/05/how-is-twitter-different-during-a-big-news-event-personal-chatter-goes-down-attention-shifts-to-elites.

177. http://www.theatlantic.com/technology/archive/2012/11/how-to-tweet-responsibly-during-a-breaking-news-event/264386; http://www.adweek.com/fishbowlny/5-things-tweeting-a-tragedy/257761; https://www.washingtonpost.com/lifestyle/style/news-outlets-often-stumble-in-quest-for-speed-and-accuracy/2013/09/16/e5444820-1f19-11e3-8459-657e0c72fec8_story.html.

178. http://www.adweek.com/lostremote/ap-offering-slow-tv-for-breaking-news/50267.

179. http://www.poynter.org/2012/bbc-news-when-to-break-news-on-twitter/162588.

180. http://www.niemanlab.org/2013/04/breaking-news-pragmatically-some-reflections-on-silence-and-timing-in-networked-journalism.

181. https://gigaom.com/2013/07/08/if-you-dont-like-the-chaos-of-breaking-news-you-should-probably-stay-off-twitter.

182. http://www.niemanlab.org/2012/12/from-real-time-to-slow-social; http://www.cjr.org/the_kicker/what_we_need_is_a_slow_news_mo.php; https://www.theguardian.com/commentisfree/2013/apr/23/ap-twitter-hack-hoax-digital-media.

183. http://www.nytimes.com/2012/05/13/magazine/joe-weisenthal-vs-the-24-hour-news-cycle.html; http://www.poynter.org/2012/kim-bui-growing-demands-on-reporters-time-is-worrisome/161267.

184. http://jimromenesko.com/2014/02/18/u-s-news-deletes-content-published-before-2007.

185. http://gawker.com/over-4-000-buzzfeed-posts-have-completely-disappeared-1619473070.

186. http://www.poynter.org/2014/buzzfeeds-ben-smith-we-didnt-fully-think-through-the-removal-of-old-posts/264007.

187. http://www.niemanlab.org/2012/06/great-idea-pastpages-offers-automated-citations-for-researchers.

188. http://www.niemanlab.org/2014/11/storytracker-is-a-new-tool-to-track-how-news-homepages-change.

189. http://www.cjr.org/watchdog/newsdiffs_new_york_times.php.

190. http://www.niemanlab.org/2013/04/the-newsonomics-of-recycling-journalism.

191. http://digiday.com/publishers/publishers-evergreen-content-strategy-make-the-old-new-again.

192. http://open.blogs.nytimes.com/2016/07/26/the-future-of-the-past-modernizing-the-new-york-times-archive.

193. http://mediashift.org/2014/04/future-proofing-news-apps.

194. http://www.cjr.org/behind_the_news/minus_proper_archives_many_new.php.

195. http://www.sacbee.com/news/local/education/article71659992.html.

196. http://www.huffingtonpost.com/2015/06/04/twitter-revokes-politwoops_n_7512892.html.

197. http://mobile.nytimes.com/2016/01/01/technology/twitter-politwoops.html.

198. https://www.fastcompany.com/3040983/meet-timeline-a-mobile-news-app-with-a-long-long-memory.

199. http://www.poynter.org/2013/5-ways-journalists-can-use-social-media-to-resurface-old-content/197991.

200. https://medium.com/@adriennelaf/what-will-yesterdays-news-look-like-tomorrow-7f82290ab8d0#.t58uvn3ig.

201. http://www.niemanlab.org/2010/09/whats-the-law-around-aggregating-news-online-a-harvard-law-report-on-the-risks-and-the-best-practices.

202. http://www.niemanlab.org/2010/03/the-barclays-case-will-hot-news-limit-the-right-to-aggregate-news.

203. http://www.niemanlab.org/2014/10/the-hot-news-doctrine-lives-and-earns-dow-jones-a-5-million-ruling.

204. http://globenewswire.com/news-release/2010/11/15/434451/207063/en/Dow-Jones-Successfully-Resolves-Lawsuit-Aimed-at-Protecting-Its-Content.html.

205. http://www.poynter.org/2012/university-of-washington-limits-reporters-live-tweets-during-games/195653.

206. https://www.nytimes.com/2014/10/04/business/media/times-articles-removed-from-google-results-in-europe.html.

207. https://www.americanpressinstitute.org/need-to-know/offshore/bbc-listing-its-pages-removed-from-google-as-part-of-europes-right-to-be-forgotten-law.

208. https://techcrunch.com/2011/11/03/googles-new-algorithm-update-impacts-35-of-searches.

209. http://mobile.nytimes.com/blogs/bits/2011/11/03/google-changes-search-algorithm-trying-to-make-results-more-timely.

210. http://www.nytimes.com/2016/04/25/business/media/snapchat-election-campaign-news.html.

211. http://digiday.com/platforms/news-crew-shooting-shows-perils-auto-play-videos-twitter-facebook.

212. http://www.niemanlab.org/2014/09/how-nytimes-com-cut-load-times-and-got-faster-for-users.

213. http://www.poynter.org/2016/facebook-is-opening-up-instant-articles-to-newsrooms-everywhere-will-a-flood-of-distributed-content-follow/397305.

214. http://www.niemanlab.org/2016/02/a-qa-with-googles-head-of-news-richard-gingras-on-its-vision-for-the-accelerated-mobile-pages-project; http://www.niemanlab.org/2016/02/diving-all-in-or-dipping-a-toe-how-publishers-are-approaching-googles-accelerated-mobile-pages-initiative; http://www.nytco.com/the-new-york-times-participates-in-google-amp-project.

215. http://www.nytimes.com/2017/01/01/technology/google-amp-mobile-publishing.html.

216. http://www.poynter.org/2012/study-bbc-articles-live-longest-on-twitter/191793.

217. http://www.adweek.com/socialtimes/best-time-to-post-social-media/504222; https://www.americanpressinstitute.org/publications/reports/survey-research/how-americans-get-news/, https://blog.bufferapp.com/best-time-to-tweet-research.

218. http://www.niemanlab.org/2013/05/at-the-miami-herald-tweetings-about-breaking-news-in-the-a-m-and-conversation-in-the-p-m.

219. http://www.niemanlab.org/2014/07/report-around-the-clock-how-some-news-orgs-use-time-zones-to-their-advantage-to-operate-247.

220. http://www.niemanlab.org/2014/02/who-edits-breaking-news-articles-on-wikipedia.

221. https://gigaom.com/2014/02/25/reddit-embraces-its-role-as-a-journalistic-entity-with-new-live-reporting-feature.

222. http://www.niemanlab.org/2015/03/r-i-p-chinua-achebe-again-the-unstuck-in-time-life-of-social-media.

223. http://www.theatlantic.com/technology/archive/2015/03/on-social-media-everything-happens-all-the-time/388469.

224. See Wajcman (2014) for an excellent survey of the technological dimensions of contemporary time cultures.

225. http://www.niemanlab.org/2016/03/the-uks-times-and-sunday-times-are-structuring-their-new-apps-and-website-around-peak-traffic-times.

226. http://www.niemanlab.org/2015/10/is-there-an-ideal-time-of-day-to-read-news-a-site-called-twelve-thirty-six-has-one-idea-youll-never-guess.

227. https://www.theguardian.com/help/insideguardian/2015/mar/09/introducing-moments-from-the-guardian-app-built-for-apple-watch.

228. http://www.nytco.com/the-times-rolls-out-one-sentence-stories-on-apple-watch.

229. http://www.niemanlab.org/2015/03/the-next-stage-in-the-battle-for-our-attention-our-wrists.

230. https://gigaom.com/2013/05/29/ft-launches-breaking-news-tool-when-140-characters-doesnt-cut-it.

231. https://www.journalism.co.uk/news/news-on-demand-to-offer-articles-based-on-readers-attention/s2/a556587.

232. http://www.niemanlab.org/2016/05/good-news-publishers-people-will-read-your-long-stories-on-their-phones-for-two-minutes-anyway.

233. http://www.niemanlab.org/2016/11/you-can-now-connect-your-washington-post-app-to-uber-for-more-efficient-in-car-reading.

234. http://www.niemanlab.org/2015/10/cant-finish-a-new-yorker-story-online-the-magazine-will-now-send-you-an-email-reminder-to-come-back.

235. http://www.mathewingram.com/work/2015/04/01/the-economist-wants-to-give-you-a-sense-of-completeness-even-if-thats-an-illusion.

236. https://medium.com/@jkalven/watch-out-a-cliff-d2312cbb1fc4#.xy3dejhe8.

237. For example, http://www.scpr.org/news/2016/08/30/64142/fast-moving-riverside-county-fire-burns-120-acres.

238. http://digiday.com/publishers/four-tactics-publishers-use-get-readers-stick-around; http://digiday.com/brands/definitive-digiday-guide-whats-whats-2017

239. For more on anticipation as sociotechnical work, see Steven J. Jackson (2016); Jackson, David Ribes, Ayse Buyuktur, and Geoffrey C. Bowker (2011); and Stephanie B. Steinhardt and Jackson (2014, 2015).

240. http://www.niemanlab.org/2014/06/can-you-turn-future-news-events-into-structured-data; https://wicknews.wordpress.com/2014/06/12/is-future-news-the-news-of-the-future.

241. http://www.niemanlab.org/2015/08/push-it-how-breaking-news-notifies-users-of-news-stories-before-they-become-big; http://blog.breakingnews.com/post/124267649514/know-faster-with-emerging-story-alerts.

242. http://www.niemanlab.org/2016/11/reuters-built-its-own-algorithmic-prediction-tool-to-help-it-spot-and-verify-breaking-news-on-twitter.

243. http://mediashift.org/2013/12/freedom-of-the-press-foundation-steps-up-encryption-efforts-for-journalists; http://www.niemanlab.org/2013/12/91274.

244. http://www.cjr.org/analysis/security_tools.php.

245. http://www.niemanlab.org/2014/12/algorithm-fatigue-what-evernotes-news-recommending-product-can-tell-us-about-privacy; http://www.newsweek.com/wary-privacy-issues-ditch-dropbox-and-avoid-google-says-edward-snowden-276956.

246. http://www.huffingtonpost.com/2013/06/10/edward-snowden-glenn-greenwald_n_3416978.html.

247. http://www.cjr.org/feature/the_back_door.php.

248. http://www.nytimes.com/2016/07/22/technology/edward-snowden-to-help-develop-a-safer-phone-for-journalists.html.

249. http://niemanreports.org/articles/how-to-deter-doxxing.

250. http://www.niemanlab.org/2014/03/hacking-in-the-newsroom-what-journalists-should-know-about-the-computer-fraud-and-abuse-act.

251. In the opening of his book No Place to Hide, Glenn Greenwald (2014) describes how Edward Snowden refused to communicate with him until Greenwald had created an encryption key and demonstrated his proficiency with secure communication techniques.

252. http://niemanreports.org/articles/how-to-keep-sources-secure-from-surveillance; http://towcenter.org/digital-security-and-source-protection-for-journalists-research-by-susan-mcgregor.

253. http://www.niemanlab.org/2014/06/the-guardian-introduces-securedrop-for-document-leaks.

254. http://www.niemanlab.org/2013/05/the-new-yorker-launches-strongbox-what-are-the-experts-saying.

255. http://www.cjr.org/behind_the_news/securedrop_etc.php.

256. http://www.reuters.com/article/us-media-cybercrime-idUSBREA2R0EU20140328.

257. http://www.cjr.org/first_person/_by_joel_simon_the.php; http://www.niemanlab.org/2016/04/wireds-making-the-long-and-slow-switch-to-https-and-it-wants-to-help-other-news-sites-do-the-same; http://www.niemanlab.org/2015/07/s-is-for-secure-why-news-organizations-are-ditching-or-should-ditch-http-for-https.

258. http://www.niemanlab.org/2014/04/what-does-heartbleed-mean-for-journalists.

259. http://www.niemanlab.org/2014/11/how-a-virus-demanding-a-bitcoin-ransom-almost-destroyed-a-public-radio-stations-archives.

260. http://www.theatlantic.com/national/archive/2013/08/new-york-times-website-crashes-again/311706.

261. http://www.theverge.com/2013/4/29/4282202/the-guardian-falls-victim-syrian-electronic-army-11-twitter-accounts-hacked.

262. https://www.washingtonpost.com/news/worldviews/wp/2013/04/23/syrian-hackers-claim-ap-hack-that-tipped-stock-market-by-136-billion-is-it-terrorism/?utm_term=.950ff350da43.

263. https://www.buzzfeed.com/jwherrman/twitter-warns-journalists-we-believe-that-these-attacks-will?utm_term=.lxV29535G#.if4GYkBk0; http://www.marketingpilgrim.com/2013/04/twitter-doesnt-exactly-instill-confidence-in-memo-to-journalists.html.

264. http://www.poynter.org/2014/new-pew-study-finds-most-people-ok-trading-privacy-for-valued-digital-services/309055.

265. http://www.pewinternet.org/2014/08/26/social-media-and-the-spiral-of-silence.

266. http://www.niemanlab.org/2013/05/should-media-companies-be-doing-more-to-protect-their-readers-and-commenters-right-to-privacy.

267. http://www.niemanlab.org/2013/05/should-media-companies-be-doing-more-to-protect-their-readers-and-commenters-right-to-privacy; http://www.poynter.org/2012/judge-orders-spokesman-review-to-id-anonymous-commenter/183698; http://www.poynter.org/2014/judge-oks-subpoena-for-nola-com-commenters-identities/236722.

268. http://www.niemanlab.org/2015/12/love-thy-reader-securely.

269. https://theconversation.com/think-youre-reading-the-news-for-free-new-research-shows-youre-likely-paying-with-your-privacy-49694.

270. https://www.buzzfeed.com/matthewzeitlin/the-washington-post-begins-blocking-ad-blockers?utm_term=.woRVeX0Xr#.fcJp3odo4.

271. http://www.cjr.org/realtalk/facebook_as_a_reporting_tool.php.

272. http://www.niemanlab.org/2014/04/how-some-journalists-are-using-anonymous-secret-sharing-apps.

273. http://www.reuters.com/article/facebook-privacy-warrants-idUSL1N10116220150721?irpc=932; https://www.engadget.com/2016/10/11/aclu-police-use-twitter-facebook-data-to-track-protesters; http://m.computerworld.com/article/2494830/data-privacy/twitter-gives-up-user-data-to-feds-69--of-the-time.html.

274. http://www.cjr.org/the_feature/virtual_reality_facebook_second_life.php; http://www.niemanlab.org/2015/11/the-new-york-times-hopes-its-first-virtual-reality-film-the-displaced-kicks-off-mass-adoption-of-vr; http://www.niemanlab.org/2015/10/what-was-it-like-watching-a-vr-stream-of-last-nights-democratic-debate-hot-and-heavy; https://medium.com/viewing-the-future-virtual-reality-in-journalism.

275. https://theintercept.com/2016/12/23/virtual-reality-allows-the-most-detailed-intimate-digital-surveillance-yet.

276. For an excellent review of the gap between what journalists and audiences see as valuable news in online environments, see Pablo J. Boczkowski and Eugenia Mitchelstein (2013).

277. http://www.niemanlab.org/2013/10/when-readers-are-robots-tracking-bogus-botnet-traffic.

278. https://www.bloomberg.com/features/2015-click-fraud.

279. http://www.nytimes.com/2016/11/18/technology/automated-pro-trump-bots-overwhelmed-pro-clinton-messages-researchers-say.html.

280. http://www.niemanlab.org/2015/09/beware-the-bots-some-best-practices-to-combat-the-fraudulent-traffic-industry.

281. http://www.niemanlab.org/reading/can-news-bots-do-public-interest-journalism.

282. https://medium.com/thoughts-on-journalism/we-built-a-twitter-bot-that-replies-to-people-who-share-fake-news-d23127c1ed15#.aw8h9uiww.

283. The academic scholarship on comments is large and tends to focus on what kind of commenting features news sites offer, what motivates readers to comment, what the quality of comments is, whether anonymity helps or hurts comments, how journalists perceive such comments, and whether they influence news production. For good recent studies, see Annika Bergström and Ingela Wadbring (2015); Kevin Coe, Kate Kenski, and Stephen A. Rains (2014); Stephanie Craft, Tim P. Vos, and J. David Wolfgang (2016); Sanne Hille and Piet Bakker (2014); Laura Hlavach and William H. Freivogel (2011); Thomas B. Ksiazek, Peer, and Lessard (2015); Thomas B. Ksiazek, Limor Peer, and Kevin Zivic (2016); Fabian Prochazka, Patrick Weber, and Wolfgang Schweiger (2016); Bill Reader (2012); Zvi Reich (2011); Nina Springer, Ines Engelmann, and Christian Pfaffinger (2015); Patrick Weber (2014); Rodrigo Zamith and Seth C. Lewis (2014); and Marc Ziegele, Timo Breiner, and Oliver Quiring (2014).

284. http://www.niemanlab.org/2014/07/newspaper-editors-strongly-support-online-comments-about-their-daily-content.

285. http://publiceditor.blogs.nytimes.com/2014/03/17/for-some-reader-contributions-become-a-new-reporting-tool.

286. http://www.niemanlab.org/2016/03/who-is-posting-comments-on-news-stories-and-why-do-they-do-it.

287. https://www.wired.com/2015/10/brief-history-of-the-demise-of-the-comments-timeline; http://www.cjr.org/behind_the_news/demise_of_comment_sections_or.php; http://www.nytimes.com/2013/09/22/magazine/no-comments.html; https://www.washingtonpost.com/lifestyle/style/some-news-sites-cracking-down-on-over-the-top-comments/2014/05/07/4bc90958-d619-11e3-95d3-3bcd77cd4e11_story.html?utm_term=.de633af18333; http://www.poynter.org/2012/why-well-never-stop-struggling-over-comment-sections/167214.

288. http://www.niemanlab.org/2014/04/some-news-orgs-are-killing-comments-but-not-just-because-their-commenters-are-terrible-at-being-humans.

289. http://www.poynter.org/2013/nyt-community-manager-good-comments-shouldnt-sit-among-those-designed-to-cause-conflict/224101.

290. http://www.niemanlab.org/2015/09/what-happened-after-7-news-sites-got-rid-of-reader-comments; http://www.popsci.com/science/article/2013-09/why-were-shutting-our-comments.

291. http://www.niemanlab.org/2011/04/the-writing-on-the-wall-why-news-organizations-are-turning-to-outside-moderators-for-help-with-comments.

292. http://www.niemanlab.org/2015/09/what-happened-after-7-news-sites-got-rid-of-reader-comments.

293. http://www.nytimes.com/2010/04/12/technology/12comments.html.

294. http://www.poynter.org/2012/new-study-real-names-improve-quality-of-website-comments/183275.

295. https://techcrunch.com/2012/07/29/surprisingly-good-evidence-that-real-name-policies-fail-to-improve-comments. For a thoughtful recent review of literature on requiring real names in online environments, see J. Nathan Matias (2017).

296. http://www.popsci.com/science/article/2013-09/why-were-shutting-our-comments.

297. http://www.adweek.com/fishbowlny/popular-science-blog-network-comments-jacob-ward/98632.

298. https://www.americanpressinstitute.org/need-to-know/up-for-debate/la-times-wont-print-letters-denying-humans-cause-global-warming.

299. http://blogs.reuters.com/great-debate/2014/11/07/editors-note-reader-comments-in-the-age-of-social-media.

300. http://motherboard.vice.com/read/im-on-twitter-too.

301. http://www.theverge.com/2015/7/6/8901753/so-comments-have-been-turned-off.

302. http://theweek.com/articles/441774/theweekcom-closing-comments-section.

303. http://www.niemanlab.org/2013/08/the-huffington-post-is-killing-anonymous-comments.

304. http://www.npr.org/sections/ombudsman/2016/08/17/489516952/npr-website-to-get-rid-of-comments.

305. http://www.poynter.org/2014/sun-times-kills-comments-until-it-can-fix-morass-of-negativity-racism-and-hate-speech/247525.

306. https://www.wired.com/2015/10/no-comments-allowed-reddits-new-news-site-upvoted.

307. https://www.americanpressinstitute.org/publications/reports/strategy-studies/choose-commenting-platform/single-page.

308. http://www.niemanlab.org/2014/06/why-the-new-york-times-and-the-washington-post-and-mozilla-are-building-an-audience-engagement-platform-together; http://www.niemanlab.org/2015/10/the-coral-project-is-building-its-first-product-a-listening-tool.

309. http://www.niemanlab.org/2016/06/here-are-some-of-the-best-and-most-interesting-ideas-from-the-guardians-comments-focused-hack-day.

310. http://www.niemanlab.org/2015/09/what-happened-after-7-news-sites-got-rid-of-reader-comments.

311. http://publiceditor.blogs.nytimes.com/2012/10/15/questions-and-answers-on-how-the-times-handles-online-comments-from-readers.

312. http://publiceditor.blogs.nytimes.com/2015/11/20/change-needed-for-commenting-that-favors-the-verified; http://www.nytimes.com/interactive/2015/11/23/nytnow/23commenters.html.

313. http://www.niemanlab.org/2015/09/what-happened-after-7-news-sites-got-rid-of-reader-comments.

314. http://www.poynter.org/2014/national-journal-eliminates-comments-from-non-members/252404.

315. http://www.niemanlab.org/2015/02/troll-toll-tablet-is-now-charging-its-readers-for-the-right-to-comment.

316. http://www.niemanlab.org/2015/06/after-deciding-to-charge-for-comments-tablets-conversation-moves-to-facebook.

317. https://www.buzzfeed.com/amandahickman/in-cleveland-meet-buzzbot.

318. http://www.poynter.org/2013/huffington-post-deletes-75-percent-of-incoming-comments/222059.

319. http://www.niemanlab.org/2014/06/want-to-comment-on-a-huffington-post-article-youll-need-to-use-facebook-now.

320. http://www.poynter.org/2013/want-to-comment-on-huffpost-just-give-facebook-your-phone-number-first/233193.

321. http://www.niemanlab.org/2015/09/what-happened-after-7-news-sites-got-rid-of-reader-comments.

322. http://digiday.com/publishers/how-bbc-is-using-yik-yak-to-talk-to-millennials-and-get-them-to-talk-back.

323. http://www.niemanlab.org/2015/09/what-happened-after-7-news-sites-got-rid-of-reader-comments.

324. https://www.nytimes.com/2014/08/27/upshot/how-social-media-silences-debate.html.

325. https://www.americanpressinstitute.org/publications/reports/survey-research/millennials-news; http://www.journalism.org/2016/05/26/news-use-across-social-media-platforms-2016.

326. http://www.poynter.org/2012/wsj-facebook-pages-hit-by-comment-flashmob/163822.

327. http://www.niemanlab.org/2011/08/ireport-at-5-nearly-900000-contributors-worldwide.

328. http://www.adweek.com/fishbowlny/guardian-launches-user-generated-content-platform-guardianwatch/259500?red=kw.

329. https://www.theguardian.com/news/series/open-newslist; https://www.theguardian.com/help/insideguardian/2011/oct/17/guardian-newslist; https://www.journalism.co.uk/news/open-journalism-guardian-newslist/s2/a549794.

330. https://gigaom.com/2011/10/10/memo-to-newspapers-let-your-readers-inside-the-wall.

331. http://www.adweek.com/fishbowlny/new-experiment-lets-readers-influence-editorial-decision-making-process-at-the-guardian/248469?red=kw.

332. http://www.nytimes.com/subscriptions/switch/lp8964W.html.

333. https://gigaom.com/2010/09/02/buzzfeed-opens-up-access-to-its-viral-dashboard; https://storify.com/patrickruffini/what-buzzfeed-s-analytics-dashboard-tells-us-about.

334. For a scholarly discussion focused on how news organizations use APIs to negotiate distances with audiences, see Mike Ananny (2013). For other studies of journalistic uses of APIs, see Tanja Aitamurto and Seth C. Lewis (2011, 2012), Michael Lahey (2016), and Yuanbo Qiu (2016).

335. http://www.niemanlab.org/2012/05/the-guardian-creates-an-api-for-n0tice-its-open-news-platform; http://www.poynter.org/2011/beginners-guide-for-journalists-who-want-to-understand-api-documentation/138211; https://www.poynter.org/2012/8-apis-your-news-organization-should-start-using-today/165347.

336. http://www.niemanlab.org/2009/06/four-crowdsourcing-lessons-from-the-guardians-spectacular-expenses-scandal-experiment.

337. http://towcenter.org/npr-news-app-team-experiments-with-making-data-driven-public-media-with-the-public.

338. https://www.journalism.co.uk/news/propublica-puts-spotlight-on-collaborative-opportunities-for-readers/s2/a552286.

339. http://www.adweek.com/fishbowlny/propublica-crowdsources-gun-control-bill-with-trackthevote/259280?red=kw.

340. http://www.niemanlab.org/2012/12/crowdsourcing-campaign-spending-what-propublica-learned-from-free-the-files.

341. http://www.niemanlab.org/2014/08/how-propublica-uses-a-reporting-recipe-to-cook-up-collaboration.

342. https://www.propublica.org/getinvolved/item/reporting-recipe-investigating-restraint-and-seclusion-in-us-schools.

343. http://www.adweek.com/fishbowlny/propublica-asks-reddit-what-should-we-cover/258733?red=kw.

344. http://mediashift.org/2013/06/challenges-cocreating-magazine-journalism-reader-input.

345. https://www.buzzfeed.com/millietran/what-we-learned-from-a-week-of-prototyping-a-newsletter-in-p?utm_term=.tfwZKWYWv#.hlZx8kRk3.

346. http://www.niemanlab.org/2013/07/gawker-is-letting-readers-rewrite-headlines-and-reframe-articles.

347. http://www.niemanlab.org/2012/08/pbs-newshours-viewers-are-translating-its-videos-into-52-languages-and-counting.

348. http://www.niemanlab.org/2014/10/why-the-new-york-times-built-a-tool-for-crowdsourced-time-travel.

349. https://theconversation.com/newspapers-ongoing-search-for-subscription-revenue-from-paywalls-to-micropayments-40726.

350. http://www.wsj.com/articles/SB10001424052748704213404576100033883758352.

351. http://www.poynter.org/2011/why-would-anyone-pay-to-read-the-new-york-times-online/142936; http://www.cjr.org/the_audit/shirky_and_paywalls.php; http://www.npr.org/2012/01/30/146093302/how-online-paywalls-are-changing-journalism.

352. http://www.niemanlab.org/2012/04/wait-so-how-many-newspapers-have-paywalls.

353. http://www.poynter.org/2012/paywalls-now-affect-one-third-of-daily-newspaper-readers/180323.

354. http://www.niemanlab.org/2012/04/the-newsonomics-of-99-cent-media.

355. http://www.adweek.com/fishbowlny/usa-today-publisher-paper-not-unique-enough-for-paywall/257265?red=kw.

356. http://www.poynter.org/2014/only-about-10-percent-of-online-readers-pay-for-news/255507.

357. https://www.minnpost.com/david-brauer-blog/2011/12/strib-metered-pay-wall-web-traffic-down-10-15-percent-revenue#.Tt5Um96dWws.twitter.

358. http://www.poynter.org/2012/new-york-times-traffic-flat-since-paywall/160780.

359. http://www.niemanlab.org/2011/03/the-price-you-pay-for-asking-people-to-pay-the-price-gerry-marzorati-on-class-and-the-nyt-paywall.

360. http://www.niemanlab.org/2012/04/the-newsonomics-of-99-cent-media; http://www.poynter.org/2012/naa-list-shows-newspaper-paywalls-typically-allow-11-free-articles/183295; http://www.poynter.org/2012/raju-narisetti-says-free-samples-can-lead-to-paid-subscribers/180709; http://www.niemanlab.org/2013/02/the-boston-globe-tightens-up-as-executives-seek-the-optimal-balance-between-free-and-paid.

361. http://www.poynter.org/2012/changes-to-new-york-times-paywall/167147.

362. https://www.cnet.com/news/sf-chronicle-erects-paywall-for-premium-content; http://www.medialifemagazine.com/slate-tries-new-sort-paywall.

363. http://www.niemanlab.org/2009/02/will-paid-content-work-two-cautionary-tales-from-2004.

364. http://www.niemanlab.org/2015/04/the-winnipeg-free-press-is-launching-a-paywall-that-lets-readers-pay-by-the-article; https://medium.com/on-blendle/blendle-a-radical-experiment-with-micropayments-in-journalism-365-days-later-f3b799022edc#.ufmqg34fw.

365. https://gigaom.com/2013/02/08/five-ways-media-companies-can-build-paywalls-around-people-instead-of-content; http://www.niemanlab.org/2013/04/getting-personal-a-dutch-online-news-platform-wants-you-to-subscribe-to-individual-journalists.

366. http://ajr.org/2013/12/16/beyond-paywalls-new-ways-charge-news.

367. https://gigaom.com/2013/07/11/what-if-the-price-you-had-to-pay-to-read-a-story-dropped-as-more-people-clicked-on-it.

368. https://medium.com/on-blendle/the-dutch-revolution-in-journalism-all-newspapers-behind-one-paydike-a2031594e430#.an6omo2xz.

369. http://digiday.com/publishers/countrywide-paywall-faltered.

370. https://gigaom.com/2014/04/21/slate-tries-to-buck-the-paywall-trend-by-focusing-on-membership.

371. http://www.niemanlab.org/2012/03/what-kind-of-challenges-does-the-l-a-times-face-in-creating-a-membership-program.

372. http://www.niemanlab.org/2014/09/wall-street-journal-turns-subscribers-into-members-with-wsj.

373. http://www.recode.net/2016/3/28/11587326/the-new-york-times-may-make-it-harder-to-use-facebook-and-twitter-to.

374. http://www.niemanlab.org/2013/11/minnpost-funds-reporting-through-new-donor-backed-beats.

375. http://nymag.com/daily/intelligencer/2013/02/new-york-times-closes-url-paywall-loophole.html; http://publiceditor.blogs.nytimes.com/2013/04/10/times-writer-tangles-with-the-ethics-of-password-sharing.

376. For excellent general discussions of how readers, users, and their attention streams are commodified for advertisers (even when they are consuming seemingly free content) and how hard it is for people to extract themselves from such commodified surveillance, see Mark Andrejevic (2003), Julia Angwin (2014), Joseph Turow (2011), and Tim Wu (2016).

377. https://theconversation.com/think-youre-reading-the-news-for-free-new-research-shows-youre-likely-paying-with-your-privacy-49694.

378. http://www.poynter.org/2012/google-customer-survey-questions-are-an-offbeat-alternative-to-a-paywall/168333; http://www.niemanlab.org/2011/10/how-google-is-quietly-experimenting-in-new-ways-for-readers-to-access-publishers-content; http://www.columbiamissourian.com/opinion/local_columnists/dear-reader-missourian-site-is-ending-its-time-paywall-and/article_38bb43f7-3e37-5a14-8132-2ce04a95d9e3.html.

379. For theorizations and case studies of crowdfunding public-facing industries like journalism and the arts, see Tanja Aitamurto (2011); Daren C. Brabham (2017); Miguel Carvajal, José A. Garcia-Aviles, and José L. Gonzalez (2012); Roei Davidson and Nathaniel Poor (2015); Rodrigo Davies (2014); Andrea Hunter (2015); and Lian Jian and Jieun Shin (2015).

380. http://mediashift.org/2013/05/crowdfunding-journalism-a-new-financing-model-for-freelancers.

381. http://www.adweek.com/fishbowlny/kickstarter-campaign-the-magazine/265648?red=kw; http://www.niemanlab.org/2014/08/the-huffington-post-wants-you-to-help-pay-for-its-future-coverage-in-ferguson; http://www.niemanlab.org/2011/10/bringing-a-crowd-to-a-records-fight-the-columbia-missourian-uses-spot-us-to-raise-money-for-data-request.

382. http://www.adweek.com/fishbowlny/how-propublica-used-kickstarter-to-fund-a-reporting-internship/261742?red=kw.

383. http://www.adweek.com/fishbowlny/for-journalism-news-developers-launch-kickstarter-to-raise-money-to-teach-data-journalism-for-all/257959?red=kw.

384. http://www.niemanlab.org/2013/04/a-dutch-crowdfunded-news-site-has-raised-1-3-million-and-hopes-for-a-digital-native-journalism.

385. http://techcitynews.com/2013/05/29/newsfreed-crowdfunding-investigative-journalism.

386. http://thenextweb.com/media/2012/02/26/like-a-kickstarter-for-photojournalism-emphas-is-helps-finance-visual-storytelling.

387. https://gigaom.com/2014/01/08/contributorias-founders-talk-about-why-they-are-building-an-open-community-for-crowdfunding-journalism.

388. https://techcrunch.com/2014/02/12/beacon-y-combinator-launch.

389. http://www.poynter.org/2014/kickstarter-adds-official-category-for-journalism/255526.

390. http://www.niemanlab.org/2012/10/how-to-make-your-journalism-project-succeed-on-kickstarter; http://www.adweek.com/fishbowlny/how-to-crowdfund-your-journalism-project/257893?red=kw.

391. http://mashable.com/2013/06/26/kickstarter-propublica-journalism/#G0EQ.ysTXPq8; http://www.niemanlab.org/2015/02/heres-a-recipe-for-successfully-crowdfunding-journalism-in-2015.

392. http://www.journalism.org/2016/01/20/crowdfunded-journalism.

393. https://www.journalism.co.uk/news/-value-for-money-how-crowdfunding-journalism-affects-reader-relationships/s2/a562942.

394. http://www.niemanlab.org/2014/07/propublica-sees-30000-in-new-revenue-from-data-store.

395. For a more complete discussion of APIs as infrastructures of networked press autonomy, see Ananny (2013).

396. http://www.niemanlab.org/2011/10/usa-today-toys-with-a-side-business-selling-commercial-access-to-its-data.

397. http://www.niemanlab.org/2011/04/new-name-new-mission-npr-digital-services-expands-hoping-to-help-streamline-local-journalism; http://www.npr.org/about-npr/179876898/terms-of-use.

398. http://www.niemanlab.org/2016/12/native-ads-are-still-very-confusing-to-many-readers-a-new-survey-suggests; https://www.theguardian.com/commentisfree/2014/feb/25/yahoo-opens-gemini-native-advertising.

399. http://www.niemanlab.org/2016/09/recommended-content-widgets-still-have-major-disclosure-and-clickbait-problems-says-a-new-report.

400. http://www.politico.com/media/story/2013/12/sulzbergers-native-advertising-manifesto-001772.

401. http://www.poynter.org/2013/the-problem-with-buzzfeeds-sponsored-posts/200214.

402. https://www.journalism.co.uk/news/native-advertising-how-news-sites-separate-church-and-state/s2/a554648.

403. http://www.politico.com/media/story/2013/12/sulzbergers-native-advertising-manifesto-001772.

404. http://www.adweek.com/fishbowlny/post-announces-sponsored-views/261134?red=kw.

405. http://www.adweek.com/news/press/sports-illustrated-testing-new-type-paywall-152935.

406. http://www.niemanlab.org/2013/11/minnpost-funds-reporting-through-new-donor-backed-beats.

407. https://www.journalism.co.uk/news/native-advertising-how-news-sites-separate-church-and-state/s2/a554648.

408. http://www.niemanlab.org/2013/01/the-ap-is-selling-ads-in-its-tweets-but-twitter-doesnt-mind; http://www.adweek.com/socialtimes/ap-sharing-sponsored-tweets-and-panicking-reporters/475311.

409. https://blog.twitter.com/2010/promoted-tweets-testing-in-the-timeline.

410. http://www.poynter.org/2012/ap-28-news-orgs-launch-newsright-to-collect-licensing-fees-from-aggregators/157817.

411. http://www.niemanlab.org/2013/05/the-newsonomics-of-where-newsright-went-wrong.

412. http://www.niemanlab.org/2014/03/the-washington-post-goes-national-by-offering-free-digital-access-to-readers-of-local-newspapers.

413. http://www.niemanlab.org/2014/06/that-washington-post-deal-with-local-newspapers-is-generating-decent-numbers.

414. http://www.niemanlab.org/2011/02/what-apples-new-subscription-policy-means-for-news-new-rules-new-incentives-new-complaints.

415. http://www.niemanlab.org/2011/02/take-that-cupertino-google-undercuts-apples-subscription-plan-with-a-cheaper-one-of-its-own.

416. http://www.poynter.org/2012/with-new-tablet-web-app-new-york-times-may-avoid-apples-fees/190297.

417. http://www.recode.net/2016/10/18/13326196/snapchat-discover-ad-sales-plan-change.

418. http://www.poynter.org/2016/heres-why-facebook-is-paying-news-organizations-millions-of-dollars/418132; http://www.wsj.com/articles/facebook-signs-deals-with-media-companies-celebrities-for-facebook-live-1466533472.

419. https://gigaom.com/2013/02/27/new-york-times-gives-starbucks-visitors-15-free-stories-a-day.

420. See http://www.poynter.org/2014/trump-hotel-guests-get-free-online-access-to-the-new-york-times-the-donalds-favorite-paper/255532.TheNewYorkTimesmaynolongerbeDonaldTrump’sfavoritenewspaper.

421. http://www.poynter.org/2013/sponsored-treats-ap-prints-news-on-restaurant-receipts/200081.

422. http://www.niemanlab.org/2015/09/from-two-day-shipping-to-news-a-cheaper-washington-post-is-now-an-amazon-prime-benefit.

423. News organizations and commercial brands have a long history of intermingling. In the early days of radio and television news, entire bulletins and programs were visibly sponsored by advertisers like Goodyear, Glo-Coat, and Camel cigarettes, with on-air personalities reading endorsements (Ponce de Leon, 2015).

424. http://www.niemanlab.org/2016/07/the-financial-times-is-testing-blocking-the-adblockers-by-blocking-actual-words-from-its-stories; http://fortune.com/2016/06/08/the-new-york-times-is-preparing-to-step-up-its-war-on-ad-blockers.

425. http://www.niemanlab.org/2014/09/the-newsonomics-of-the-pianopress-merger-creating-the-worlds-largest-paywall-tech-company; http://www.niemanlab.org/2015/08/tinypass-and-piano-media-merge-and-embrace-paywall-as-a-fluid-concept.

426. http://www.niemanlab.org/2012/02/piano-media-bumps-up-slovak-paywall-price-25-percent.

427. http://www.poynter.org/2012/press-passes-300-clients-now-using-metered-model-to-charge-for-online-news/168712.

428. http://www.poynter.org/2012/more-stories-result-in-more-subscription-revenue-press-says/180554.

429. https://www.bloomberg.com/news/articles/2016-02-24/google-gives-publishers-what-facebook-apple-haven-t-a-paywall.

430. For excellent discussions of epistemological tensions underlying journalistic professionalism, see Mark W. Brewin (2013), Theodore L. Glasser and James S. Ettema (1993), Raymond W. K. Lau (2012), Dan Schiller (1979), Michael Schudson (1978), Schudson and Christopher W. Anderson (2008), Steven C. Ward (1996), Stephen J. A. Ward (2005), and Kate Wright (2011).

431. http://www.poynter.org/2012/andy-carvin-explains-how-twitter-is-his-open-source-newsroom/157874.

432. http://www.poynter.org/2012/stumped-new-york-times-reporter-crowdsources-id-of-a-rare-bomb/161533.

433. http://www.poynter.org/2011/6-ways-journalists-can-use-quora-as-a-tool-for-reporting-sharing-ideas/114314.

434. http://niemanreports.org/articles/finding-the-wisdom-in-the-crowd; http://niemanreports.org/articles/the-process-of-verification.

435. http://www.journalism.org/2016/12/15/many-americans-believe-fake-news-is-sowing-confusion.

436. https://gizmodo.com/google-changes-algorithm-to-remove-holocaust-denying-re-1790500861.

437. http://www.nytimes.com/2012/04/10/us/politics/false-nikki-haley-twitter-report-spreads-fast.html.

438. http://www.poynter.org/2012/ap-tells-staff-how-to-correct-erroneous-tweets-in-new-social-media-guidelines/159622.

439. https://www.washingtonpost.com/news/the-intersect/wp/2016/08/29/a-fake-headline-about-megyn-kelly-was-trending-on-facebook/?utm_term=.f78bd9b2b01e.

440. https://www.buzzfeed.com/craigsilverman/partisan-fb-pages-analysis?utm_term=.sfpL5R6Rm#.unABJjbjv; https://medium.com/@BiellaColeman/on-truth-and-lies-in-a-pragmatic-performative-sense-with-my-respects-to-nietzsche-or-why-reality-5c8400bd9ac2#.z0fi6lac4; https://firstdraftnews.com/almost-half-of-americans-believe-government-responsible-for-tackling-fake-news; http://www.npr.org/sections/ed/2016/12/22/505432340/the-classroom-where-fake-news-fails; http://www.politico.com/magazine/story/2016/12/fake-news-history-long-violent-214535; https://medium.com/thoughts-on-journalism/we-built-a-twitter-bot-that-replies-to-people-who-share-fake-news-d23127c1ed15#.fwrcykuev.

441. https://www.theguardian.com/technology/2016/nov/29/facebook-fake-news-problem-experts-pitch-ideas-algorithms.

442. http://www.niemanlab.org/2012/03/hacking-for-truth-whatever-that-is-ideas-to-fight-misinformation.

443. http://hoaxy.iuni.iu.edu.

444. https://medium.com/thoughts-on-journalism/we-built-a-twitter-bot-that-replies-to-people-who-share-fake-news-d23127c1ed15#.fwrcykuev.

445. http://www.npr.org/2016/11/20/502770866/ex-head-of-twitter-news-social-media-companies-alone-shouldn-t-regulate-fake-new.

446. http://www.niemanlab.org/2016/09/facebook-twitter-and-30-other-orgs-join-first-drafts-partner-network-to-help-stop-the-spread-of-fake-news.

447. https://www.theguardian.com/technology/blog/2013/jul/01/data-journalism-awards-improving.

448. http://radar.oreilly.com/2010/12/data-journalism.html.

449. https://www.theguardian.com/news/series/facts-are-sacred.

450. https://www.theguardian.com/news/datablog/2013/apr/04/data-journalism-guardian-rusbridger.

451. https://www.propublica.org/datastore/about.

452. http://www.niemanlab.org/2013/03/dallas-morning-news-partners-with-local-government-group-to-gather-clean-universal-crime-data.

453. https://www.journalism.co.uk/news/detective-io-launches-to-search-connections-in-datasets/s2/a555266.

454. http://towcenter.org/what-the-tesla-affair-tells-us-about-data-journalism.

455. https://qz.com/189703/the-problem-with-data-journalism.

456. http://www.huffingtonpost.co.uk/entry/facebook-fake-news_uk_582c20b7e4b08b8e51a43bc2.

457. http://www.vox.com/policy-and-politics/2016/12/2/13812818/facebook-fake-news-think-before-you-share.

458. http://newsroom.fb.com/news/2016/12/news-feed-fyi-addressing-hoaxes-and-fake-news.

459. https://www.ap.org/ap-in-the-news/2016/facebook-finally-gets-serious-about-fighting-fake-news.

460. https://www.technologyreview.com/s/603188/the-limits-of-fact-checking-facebook.

461. http://www.pri.org/stories/2016-10-14/fact-check-tags-added-google-news. https://blog.google/topics/journalism-news/labeling-fact-check-articles-google-news.

462. http://www.pri.org/stories/2016-10-14/fact-check-tags-added-google-news.

463. http://www.poynter.org/2012/washington-posts-truthteller-project-hopes-to-birth-real-time-fact-checking/183774.

464. https://www.washingtonpost.com/news/ask-the-post/wp/2013/09/25/announcing-truth-teller-beta-a-better-way-to-watch-political-speech/?utm_term=.62d0bbe2c7c5; http://www.knightfoundation.org/articles/debuting-truth-teller-washington-post-real-time-lie-detection-service-your-service-not-quite-yet.

465. http://thefjp.org/2013/03/can-robots-tell-the-truth.

466. http://www.poynter.org/2016/who-decides-whats-true-in-politics-a-history-of-the-rise-of-political-fact-checking/429326.

467. https://fullfact.org/blog/2016/aug/automated-factchecking.

468. http://www.poynter.org/2016/report-automated-fact-checking-is-coming-and-soon/426370.

469. http://www.poynter.org/2016/le-monde-wants-to-build-a-b-s-detector-2/440078.

470. http://www.poynter.org/2013/new-guardian-scoopshot-efforts-bring-elements-of-automation-to-photo-verification/212585; http://mashable.com/2013/04/16/the-guardianwitness/#Z75KU_idQPqQ.

471. http://www.niemanlab.org/2016/11/reuters-built-its-own-algorithmic-prediction-tool-to-help-it-spot-and-verify-breaking-news-on-twitter.

472. http://www.niemanlab.org/2016/06/the-washington-post-is-dabbling-in-translations-to-reach-a-growing-non-english-speaking-audience.

473. http://www.niemanlab.org/2016/08/european-publishers-are-teaming-together-to-translate-the-news-to-reach-broader-audiences.

474. https://www.wired.com/2016/07/fb-2.

475. https://www.gitbook.com/book/towcenter/virtual-reality-journalism/details; http://www.theverge.com/a/virtual-reality.

476. www.poynter.org/2016/how-the-new-york-times-used-its-archives-to-make-the-past-a-virtual-reality; http://www.niemanlab.org/2016/05/the-new-york-times-is-trying-to-make-vr-films-that-arent-one-offs-and-that-keep-readers-coming-back.

477. http://www.theguardian.com/world/ng-interactive/2016/apr/27/6x9-a-virtual-experience-of-solitary-confinement.

478. http://www.theverge.com/a/virtual-reality.

479. http://www.niemanlab.org/2015/11/what-frontlines-team-learned-producing-its-new-vr-documentary.

480. http://www.knightfoundation.org/media/uploads/publication_pdfs/VR_report_web.pdf.

481. http://www.niemanlab.org/2016/08/designing-news-products-with-empathy-how-to-plan-for-individual-users-needs-and-stresses.

482. http://www.cjr.org/innovations/virtual_reality_journalism.php.

483. http://www.cjr.org/the_feature/virtual_reality_facebook_second_life.php; http://www.wired.com/2016/02/vr-moral-imperative-or-opiate-of-masses.

484. For a more in-depth discussion of how sociotechnical separations and dependencies influence journalistic ethics of witnessing, see Mike Ananny (2015).

485. http://www.wnyc.org/story/why-paris-not-baga.

486. http://www.niemanlab.org/2013/12/where-in-the-world-is-buzzfeed-building-foreign-news-around-themes-rather-than-geography.

487. http://www.cjr.org/analysis/facebook_as_state.php.

488. http://publiceditor.blogs.nytimes.com/2012/12/27/snow-fall-tells-a-story-about-an-avalanche-and-a-newspapers-reinvention.

489. http://www.niemanlab.org/2014/05/the-leaked-new-york-times-innovation-report-is-one-of-the-key-documents-of-this-media-age.

490. http://www.npr.org/sections/alltechconsidered/2014/07/09/330003058/in-google-newsroom-brazil-defeat-is-not-a-headline.

491. https://digiday.com/media/news-crew-shooting-shows-perils-auto-play-videos-twitter-facebook.

492. http://digiday.com/platforms/news-crew-shooting-shows-perils-auto-play-videos-twitter-facebook.

493. http://mediadecoder.blogs.nytimes.com/2012/09/28/as-it-followed-a-car-chase-fox-news-showed-a-man-kill-himself.