Chapter 18: The Media — From Walter Cronkite to TikTok, How Information Shapes Democracy

18.1 Why a Chapter on Media Belongs in a Chapter on Power

Public opinion, the topic of the previous chapter, does not form in a vacuum. It forms out of the information environment people encounter. What citizens believe about the economy, about crime, about the conduct of officials, about what other Americans think — almost all of it is mediated. Almost none of it comes from direct personal experience. A voter in suburban Pennsylvania does not personally observe what is happening at the southern border, in Ukraine, in Congressional hearings, on the trading floor of the New York Stock Exchange. They receive accounts of those things from intermediaries. The intermediaries, taken together, are what we mean by "the media."

This means that the structure of the media system is, in a serious sense, the structure of democratic accountability. If the intermediaries do their job well — getting facts right, providing context, exposing official misconduct, presenting opposing arguments fairly, making themselves available to a broad audience at low cost — citizens have what they need to vote intelligently. If the intermediaries do their job badly, or if the system is structured in a way that incentivizes them to do it badly, the rest of the democratic apparatus runs on degraded inputs. James Madison put the principle at its strongest in 1822: "A popular Government, without popular information, or the means of acquiring it, is but a Prologue to a Farce or a Tragedy."¹ The argument has not become less true.

What this chapter argues is that the American media system in 2026 looks fundamentally different from the system that existed when most Americans alive today were born. The transformation has been more rapid and more thorough than most institutional changes in American history. It has produced a system with measurable strengths — more voices, lower entry costs, faster diffusion of accurate information when accurate information exists — and measurable pathologies. The pathologies are not equally distributed across the political spectrum, but neither are they confined to one side. Both the conservative complaint about mainstream-press unfairness and the progressive complaint about right-wing misinformation contain real empirical content. Both are also, in significant respects, incomplete and partisan-flattering.

This chapter takes those claims seriously, names the data behind them, and tries to give an honest account of what is and is not contested. The reader who finishes the chapter should understand:

• What the broadcast-era media system was, and what was lost when it ended.
• How cable, the internet, social platforms, and podcasts each changed the system.
• What the empirical evidence says about partisan lean in the major American outlets — on both sides, with sources.
• How political misinformation actually circulates, what the asymmetries are and aren't, and what the research literature actually shows.
• Why filter-bubble theory, in its strongest form, is partially supported but mostly overstated by the evidence.
• Why the collapse of local journalism since 2005 is one of the most consequential developments in American civic life.
• What the platform-regulation debate actually concerns, and where the strongest arguments lie on each side.

The chapter is organized chronologically through Section 18.5, then thematically from 18.6 through 18.13.

18.2 The Broadcast Era, 1950–1990

For about four decades after the Second World War, most Americans got most of their news from a small number of sources that operated under a particular set of rules. Three commercial television networks — ABC, CBS, and NBC — broadcast nightly news programs that, at their peak in the early 1980s, reached approximately 50 million viewers combined, or roughly 80% of households with a television set tuned to news at the dinner hour.² Local news on those same networks reached comparable audiences. Most cities had one or two daily newspapers; many smaller cities had a single locally owned paper that was read by majorities of literate adults. The Associated Press and United Press International — the wire services — supplied the syndicated content that small papers could not afford to gather themselves. National Public Radio launched in 1971 and the Public Broadcasting Service in 1970, the noncommercial complements to the commercial networks. Magazines — Time, Newsweek, U.S. News and World Report, the Saturday Evening Post, Reader's Digest, Life — reached weekly audiences in the millions.

The phrase "shared-facts baseline" describes the result. It does not mean that everyone agreed on the news. It does not mean that the news was unbiased. It means that most Americans were getting their basic information about national events from a relatively small number of overlapping sources, with overlapping editorial sensibilities, and that the disagreements that did exist tended to take place against a common background of agreed-upon facts. When CBS, NBC, and ABC reported the same casualty figures from Vietnam, the disagreement that followed was about what to do, not about whether the figures were real. When The New York Times and the Wall Street Journal — politically very different newspapers — reported the same unemployment statistic, the disagreement that followed was about what caused the unemployment and what to do about it, not about the number itself.

Two specific institutional features deserve attention.

18.2.1 The Fairness Doctrine

The Federal Communications Commission promulgated the Fairness Doctrine in 1949 and enforced it, in some form, until 1987. The doctrine required broadcasters using the public airwaves — which, the FCC reasoned, were a scarce public resource and therefore subject to public-interest regulation — to devote a reasonable amount of time to discussing controversial matters of public importance and to do so in a manner that fairly reflected differing viewpoints.³ The doctrine was imperfect; it was enforced unevenly, it produced a chilling effect on some controversial programming, and broadcasters complained that it was a content-based regulation in tension with the First Amendment. The Supreme Court upheld it nonetheless in Red Lion Broadcasting v. FCC (1969), reasoning that scarcity of broadcast spectrum justified more intensive regulation than would be permitted for print.

In 1987, under FCC Chairman Mark Fowler, the agency repealed the Fairness Doctrine. The decision was administrative, not legislative — Congress did not abolish it, the FCC did. Congress passed legislation to restore the doctrine the same year; President Reagan vetoed the legislation; the veto held. Subsequent Democratic and Republican administrations both declined to reinstate the doctrine, and the FCC formally removed the implementing regulations in 2011 under President Obama.

The repeal had a measurable effect on broadcast programming. Talk radio — which had existed before 1987 but had been constrained by the requirement to provide opposing viewpoints — expanded rapidly. The Rush Limbaugh Show nationally syndicated in 1988 and reached an audience of approximately 20 million weekly listeners by the mid-1990s, the largest commercial radio audience in American history.⁴ Other conservative talk-radio hosts — Sean Hannity, Mark Levin, Michael Savage, Glenn Beck — followed. Liberal talk radio (Air America, launched 2004) struggled commercially and largely failed; the asymmetry in the talk-radio space has remained durable. Whether the Fairness Doctrine was the cause, or whether it merely had been suppressing a market demand that would have produced this asymmetry anyway, is genuinely contested among media historians.

18.2.2 Walter Cronkite as a Type

For the broadcast era's central role, the standard exemplar is Walter Cronkite, the CBS Evening News anchor from 1962 to 1981. Cronkite was, in a famous and probably overstated phrase, "the most trusted man in America" — a description that originated as a marketing line but reflected something real. A 1972 Quayle poll found Cronkite's trustworthiness rating at 73%, higher than that of any sitting elected official, including the President.⁵

The point is not that Cronkite himself was uniquely virtuous, though by the standards of his profession he was substantial. The point is that the system that produced his role — three networks competing for a mass audience, a Fairness Doctrine constraining partisan presentation, an editorial culture inherited from print journalism's "objectivity" tradition, and the absence of any way for the audience to immediately respond at scale — produced an anchor whom most Americans, across political affiliations, treated as a basically reliable narrator of common events. When Cronkite returned from Vietnam in early 1968 and editorialized that the war had become "mired in stalemate," the editorial mattered politically because his audience was bipartisan. President Johnson reportedly told an aide, "If I've lost Cronkite, I've lost middle America." Whether the quotation is exactly accurate or apocryphal, the dynamic it describes was real.

That role does not exist anymore. There is no news figure today whom 73% of Americans trust. There is no news outlet whom 73% of Americans trust. The decline in shared trusted intermediaries is itself one of the most consequential changes of the past forty years, and the chapters that follow describe how it happened.

18.3 Cable News, 1980 to the Present

In June 1980, Ted Turner launched the Cable News Network (CNN) — a 24-hour news channel distributed via cable subscription rather than over-the-air broadcast. It reached fewer than two million homes initially. Industry insiders were skeptical that there was enough news to fill 24 hours, or enough audience that wanted news at non-traditional hours, to make the model viable. The 1991 Gulf War proved otherwise. CNN's continuous coverage of Operation Desert Storm — including live footage of the air campaign over Baghdad — created an audience habit that persisted after the war ended. By 1996, CNN reached more than 70 million American households.

That same year, two competitors launched: Fox News, founded by Rupert Murdoch and run by Roger Ailes, the longtime Republican media strategist; and MSNBC, a joint venture of NBC News and Microsoft. The three-channel cable-news landscape that emerged in 1996 has remained more or less stable for thirty years, though the audience composition and the editorial direction have shifted.

Three features of the cable-news transformation deserve emphasis.

18.3.1 The 24-Hour Cycle

A nightly broadcast had thirty minutes — actually about twenty-two minutes after commercials — to summarize a day's events. A 24-hour channel had to produce, every day, 1,440 minutes of programming. Most days do not produce 1,440 minutes' worth of genuinely new and important news. The cable channels solved this problem in three ways: by lengthening coverage of any individual story (Marcia Clark, the O.J. Simpson trial, the Lewinsky scandal, missing-children stories, hurricane coverage, court hearings); by interpolating opinion and analysis between news segments; and by hiring talent who could fill airtime extemporaneously.

The cumulative effect was a blurring of the line between news and commentary. In the broadcast era, the dividing line was usually clear: the evening news was reporting; Meet the Press was discussion; the editorial page was the editorial page. On cable, the same anchor often shifted from reporting a development to commenting on it within the same segment, often within the same sentence. Audiences, surveys later showed, often did not distinguish between the two.⁶

18.3.2 Fox News and the Conservative Cable Audience

Roger Ailes's strategic insight at Fox News was that there was a politically conservative audience that experienced the existing television-news offerings as systematically tilted against its political views, and that this audience was large, underserved, and would prove highly loyal if a channel spoke to it on its own terms. The strategy worked. Fox News surpassed CNN in total prime-time viewers in early 2002, and has led the cable-news ratings for most of the period since. Its prime-time programming features explicitly conservative hosts — Bill O'Reilly until 2017, Sean Hannity, Tucker Carlson until 2023, Laura Ingraham, Jesse Watters — alongside what the network has historically presented as straight-news programming during weekday daytime hours, anchored by figures including Bret Baier, Shepard Smith (until 2019), and Bill Hemmer.

The relationship of Fox News to the Republican Party has tightened over time. By the late 2010s the network was the single most-cited source of news among self-identified Republicans, and a 2014 Pew study found that 47% of "consistent conservatives" named Fox News as their main source of political news, with no other source named by more than 11%.⁷ The relationship has at certain moments produced friction within the network — most visibly around the 2020 election, when Fox News executives reportedly worried about losing audience share to more strongly Trump-aligned outlets like Newsmax and One America News Network — but the basic alignment has been durable.

18.3.3 MSNBC and the Liberal Cable Audience

MSNBC took longer to find its identity. Through the early 2000s the channel experimented with multiple programming approaches, including a brief and strange period featuring conservative host Tucker Carlson and another period anchored by the centrist Phil Donahue (whose program was canceled in 2003 in part, according to internal memos that later leaked, because of concern that his anti-Iraq-War views were poor fit for the post-9/11 environment). By the mid-2000s, with the rise of Keith Olbermann's Countdown and later Rachel Maddow's eponymous program, MSNBC consolidated as the explicitly liberal cable-news destination.

MSNBC's audience overlaps strongly with self-identified Democrats and progressives. The 2014 Pew study found that 12% of "consistent liberals" named MSNBC as their main source of political news, with CNN at 15% and NPR at 13%; the liberal audience is more distributed across multiple sources than the conservative audience is across Fox.⁸ The asymmetry — that conservative news consumption concentrates in one cable network and liberal news consumption is spread across several — has been consistent across subsequent surveys.

18.4 The Internet, 1995 to the Present

The internet's effect on news happened in waves. The first wave, in the mid-to-late 1990s, was the migration of existing newspapers and magazines onto the web — initially as supplements to print, then as the primary distribution channel. The second wave, around 2000 to 2005, was the rise of independent online voices that competed with traditional outlets: Matt Drudge's Drudge Report (founded 1995), the early political blogosphere (Andrew Sullivan, Glenn Reynolds at Instapundit, Kos at Daily Kos, Josh Marshall at Talking Points Memo, Ezra Klein at his own site before joining the Washington Post), and aggregators of various ideological orientations. The third wave, beginning around 2005, was the rise of the social-media platforms, which we treat in their own section.

18.4.1 The Collapse of Newspaper Revenue

The internet wave that mattered most for news institutions was none of those. It was the collapse of classified-advertising revenue — the financial foundation of the American newspaper business — to Craigslist (founded 1995, free classified listings), Monster.com, eBay, and later to specialty sites for cars (autotrader.com), real estate (Zillow), and jobs (LinkedIn, Indeed). Classified ads had historically supplied about 30% to 40% of large-paper revenue; that revenue collapsed by 75% to 90% over a fifteen-year window, according to the Newspaper Association of America's data through 2017.⁹

Display advertising — full-page ads from department stores, automakers, real-estate developers — followed. The print display-ad market shrank as advertisers moved budgets to digital, where Google and Facebook captured a dominant share of new spending. By 2020, the two companies together captured more than half of all U.S. digital advertising revenue.¹⁰ The newspaper industry's total advertising revenue fell from approximately $49 billion in 2005 to less than $9 billion in 2022 — a decline of roughly 80% in real terms.

The consequences ran through every aspect of the institution. Newsroom employment in American newspapers fell from approximately 71,000 full-time editorial employees in 2008 to approximately 31,000 in 2022, a decline of more than half.¹¹ State capitols that had been covered by full-time bureau staff from a dozen newspapers in the 1990s were covered by two or three reporters by 2015. Local-government coverage, court coverage, school-board coverage — the dense beat reporting that had been the spine of American civic information — thinned out.

We treat this in detail, with a specific case study, in Section 18.10.

18.4.2 The Drudge Phenomenon and the Rise of the Aggregator

A different consequence of the early internet was the rise of the aggregator — a website that did little or no original reporting but linked, prominently, to reporting done elsewhere, with editorial selection that itself became a form of political signal. Matt Drudge's Drudge Report, with its three-column unadorned layout and its sirens for breaking news, drove enormous referral traffic to the stories Drudge selected. By the mid-2000s, the Drudge Report was reportedly the second-largest single source of inbound traffic to The New York Times website, after Google. Editors at major papers increasingly knew that whether their story got a Drudge link could mean the difference between a quiet news cycle and a major one.

The aggregator function migrated to algorithmic platforms over the 2010s, but the underlying dynamic — that selection and amplification, not original reporting, drove the most-read content — proved durable.

18.5 Social Media, 2005 to the Present

Facebook opened to college students in 2004, to high-school students and the general public in 2006, and reached approximately 1 billion monthly active users by 2012. Twitter launched in 2006, reached substantial political-journalist adoption around 2009, and became an essential professional tool for political reporters and many politicians by the 2012 election cycle. YouTube launched in 2005 and was acquired by Google in 2006. Instagram launched in 2010 and was acquired by Facebook in 2012. TikTok (initially Musical.ly, merged with Douyin's overseas product as TikTok in 2018) reached massive U.S. adoption in the early 2020s.

The transformation those platforms produced is large enough that we devote the rest of this section to it, but a single sentence summarizes the change: the platforms replaced editorial selection with algorithmic recommendation. A reader of The New York Times in 1985 saw the stories that the Times's editors had decided were the most important. A reader of Facebook in 2015 saw the stories that Facebook's recommendation algorithm had calculated were most likely to keep the reader engaged. The two selection processes optimize for different outcomes, and the differences in outcome have been politically and culturally consequential.

18.5.1 The Algorithm and Engagement

Social-media platforms, as commercial businesses, generate revenue from advertising. Advertising rates depend on the time users spend on the platform and the actions they take while there — clicks, shares, comments, reactions. The engineering teams that build the recommendation algorithms are accordingly evaluated on a target called "engagement." Engagement-maximizing algorithms learn, over many iterations and across millions of users, what kinds of content keep particular users on the platform longest. The lesson the algorithms have learned, repeatedly and at multiple platforms, is that high-arousal content — content producing strong emotional reactions, particularly negative ones — tends to outperform low-arousal content on most engagement metrics.¹²

The political implications of this incentive structure are not subtle. Content that provokes outrage outperforms content that informs calmly. Content that depicts the other party as alien and threatening outperforms content that describes the other party's strongest arguments. Content that confirms a user's existing political identity outperforms content that challenges it. None of these patterns are absolute, and reasonable researchers disagree about their magnitude. But the direction of the algorithmic incentive — toward more emotional, more divisive, more identity-confirming content — is empirically robust across studies and across platforms.

The platforms have responded to this pattern in different ways at different times. Facebook in 2018 changed its News Feed algorithm to prioritize "meaningful social interactions," which had the unintended effect of further amplifying content that produced strong reactions; subsequent internal research (later disclosed by whistleblower Frances Haugen) showed that the change had increased divisiveness. Twitter under Jack Dorsey and later Elon Musk has experimented with various signal-boosting and signal-suppressing approaches. YouTube changed its recommendation algorithm in 2019 in ways that, by Mozilla Foundation analysis, reduced certain forms of extremist amplification but did not eliminate the underlying tendency.¹³

18.5.2 The Platform-vs.-Publisher Question

A long-running debate in U.S. law concerns whether the social-media platforms are more like publishers (which are legally responsible for the content they distribute, like newspapers) or more like common carriers (which are not responsible for the content traveling over their networks, like phone companies). The law has, since the 1996 Communications Decency Act, treated them as a third category — protected by Section 230 (which we discuss in detail in Section 18.11) from liability for user-generated content while retaining the right to moderate that content. Whether this third category is the right one, given how much editorial-style curation the platforms now perform, is one of the most contested questions in contemporary American media policy.

18.5.3 The Feed Replaces the Channel

The deeper structural change is the replacement of "channel" by "feed." A reader who chose, in 1985, to read The New York Times received a particular editorial product. A reader who chose, in 1985, to watch the CBS Evening News received a different editorial product. The reader was choosing among curated bundles. A reader who opens Facebook or X or TikTok in 2026 is not choosing a curated bundle. They are receiving an algorithmically personalized stream that is unique to them, drawn from thousands of sources, with the source-mix calibrated to keep them on the platform.

This has at least three consequences. First, two readers who think they are getting "the news" from the same platform may actually be receiving radically different information environments. Second, the reader's sense of which sources are credible is harder to develop, because they are not building a relationship with any single editorial institution — content from a Times reporter, a partisan podcaster, an anonymous troll, and an obscure blog all flow through the same interface, often visually indistinguishable. Third, the reader has limited ability to audit their own information diet, because the algorithm itself is opaque.

18.6 Podcasting and YouTube, the Long-Form Insurgency

A different transformation, partly overlapping with social media but distinct in important ways, has been the rise of long-form audio and video content distributed outside traditional editorial controls. Podcasts began in the mid-2000s as a niche format — typically distributed through RSS feeds, often produced by hobbyists with minimal editorial infrastructure. By 2020, podcasting had become a major political-media format, with the largest political podcasts reaching audiences competitive with cable-news programs.

18.6.1 The Joe Rogan Phenomenon

The Joe Rogan Experience, hosted by the comedian and former Fear Factor host, has been the largest single podcast in the United States for most of the period since 2018. By Spotify's 2024 reporting, it reaches approximately 14 million listeners per episode in the U.S. and significantly more globally; episodes are typically two to three hours long. Rogan's program has hosted guests from across the ideological spectrum — Bernie Sanders, Tulsi Gabbard, Elon Musk, Donald Trump, Jordan Peterson, Andrew Yang, Robert F. Kennedy Jr., Cornel West — typically in long-form unedited conversation. The audience skews male, younger than the cable-news audience, and politically heterogenous, with a measurable rightward tilt that has increased since 2020. The 2024 election saw Rogan endorse Trump in late October, an endorsement that political analysts on both sides of the partisan divide treated as electorally significant.

18.6.2 The Right-Leaning Podcast Ecosystem

A cluster of right-leaning long-form content has grown rapidly in the 2020s. Daily Wire — founded by Ben Shapiro, Andrew Klavan, and Jeremy Boreing in 2015 — operates a stable of podcasts (Shapiro's daily program, Klavan's, Matt Walsh's, Michael Knowles's, Candace Owens until her departure in 2024) that together reach audiences in the tens of millions. Megyn Kelly's program, launched after her departure from NBC in 2017 and redistributed via SiriusXM and YouTube, has reached top-five status in U.S. podcast rankings. Bret Weinstein's DarkHorse Podcast and his brother Eric Weinstein's The Portal operate in adjacent intellectual territory. Tim Pool, Steven Crowder, and Bannon's War Room fill different niches in the same broad ecosystem.

The ideological character of this cluster ranges from mainstream conservatism (Shapiro, Klavan) to populist right (Pool, Bannon) to heterodox-but-right-leaning (the Weinsteins). What unites them is a shared sense that mainstream-press coverage of politics is systematically unfair to conservative and heterodox views and that long-form audio/video allows arguments to be made at greater length and with less filtering than the traditional outlets permit.

18.6.3 The Left-Leaning Podcast Ecosystem

A roughly parallel cluster on the left exists, though it is smaller in audience reach. Pod Save America, founded in 2017 by former Obama White House communications staff (Jon Favreau, Jon Lovett, Tommy Vietor, and Dan Pfeiffer), has been the largest mainstream-Democratic podcast, reaching audiences in the low millions per episode. Chapo Trap House, founded in 2016 by writers in the socialist-left orbit (Will Menaker, Felix Biederman, Matt Christman, Virgil Texas until his departure in 2020, Amber A'Lee Frost), pioneered a distinctly left-of-Democratic podcast format. Bad Faith (Briahna Joy Gray and Virgil Texas), The Majority Report (Sam Seder), and the various offshoots of The Young Turks round out the left-leaning long-form audio space.

The asymmetry is notable. The largest right-leaning political podcasts have audiences several times larger than the largest left-leaning political podcasts. Whether this reflects the left's continued strength in traditional institutional outlets (NPR, mainstream cable, prestige newspapers) and consequent lower demand for alternative voices, or some other dynamic, is debated.

18.6.4 The Hard-to-Classify

A meaningful fraction of the long-form audio and video space resists easy ideological classification. Lex Fridman, an AI researcher who hosts long-form interviews with figures across politics, technology, and science, has interviewed Trump, Putin, Zelensky, Netanyahu, and a range of others; his audience and editorial sensibility cross conventional partisan lines. Sam Harris's Making Sense podcast has moved across the political spectrum on different issues — broadly liberal on questions of religion, broadly heterodox on questions of identity politics, vocally critical of Trump on character and norms. Andrew Sullivan's The Weekly Dish (Substack) and Bari Weiss's Honestly (and the Free Press) occupy adjacent terrain. Coleman Hughes, John McWhorter, and Glenn Loury operate in their own register on questions of race and identity. The size of this category should not be exaggerated — the mass audience is on the more clearly partisan products — but its existence matters for any account of the contemporary information environment.

18.7 Partisan Slant: The Empirical Lay of the Land

This is the place where many media-studies textbooks become unreadable, because they either pretend that no partisan slant exists in any of the major outlets or, less commonly, treat one side's slant as obvious and the other side's as imaginary. The empirical reality is more complicated than either, and naming it accurately requires distinguishing several different things that "partisan slant" can mean.

We distinguish four measurements:

1. Self-reported political views of working journalists. Survey data on whom journalists vote for, how they identify ideologically, and how they describe their own values.
2. Editorial-page positions and endorsements. What outlets formally take positions on, and which candidates they endorse.
3. Content analysis of news coverage. Computer-assisted and human-coded analysis of how stories are framed, which sources are cited, and which adjectives are applied to which actors.
4. Audience composition. Which voters consume which outlets.

Each measurement tells part of the story. Each can be wrong on its own. Triangulating across them produces something close to a reliable picture.

18.7.1 The Mainstream Press Leans Democratic

By all four measurements, the major mainstream-press outlets — The New York Times, The Washington Post, NPR, the broadcast networks (ABC, CBS, NBC), the major weekly newsmagazines, and the Associated Press in its hard-news reporting — tilt Democratic-leaning relative to the median American voter.

The journalist-self-identification data has been collected periodically since the 1970s, with broad consistency. Lichter, Rothman, and Lichter's 1986 study found that 81% of national-media journalists they surveyed had voted for the Democratic presidential candidate in the previous four elections. Weaver, Beam, Brownlee, Voakes, and Wilhoit's later "American Journalist" surveys (1992, 2002, 2013) found Democratic identification rates among national journalists in the 30s to 50s, Republican identification rates in the high single digits to low teens, with the balance independent.¹⁴ The 2022 update of similar surveys, conducted by Lars Willnat and David Weaver, found that American journalists who identified with a major party identified Democrat over Republican by roughly four to one, with a substantially larger group identifying as independent. Other surveys, conducted by Pew, by the Indiana University School of Journalism, and by various journalism trade publications, have produced similar magnitudes. The Democratic tilt of working journalists is not seriously disputed.

The endorsement data points the same direction. The New York Times has endorsed every Democratic presidential nominee since 1956. The Washington Post has endorsed every Democratic presidential nominee since 1976 (with a notable 2024 deviation when ownership canceled the planned Harris endorsement, prompting visible internal dissent). USA Today did not endorse for many years and has more recently issued anti-Trump rather than affirmatively pro-Democratic editorials. The Los Angeles Times's 2024 non-endorsement, ordered by owner Patrick Soon-Shiong, similarly produced internal dissent. Republican-endorsing major papers — once a category that included the Chicago Tribune, the Dallas Morning News, the Cincinnati Enquirer, and many others — have largely withdrawn from presidential endorsements or, like the Wall Street Journal editorial page, retained Republican-leaning views without endorsements.

The content-analysis literature is more contested but trends consistent. Tim Groseclose and Jeffrey Milyo's 2005 study, using citation patterns to think-tanks, ranked most major outlets as more liberal than the median voter, with the Washington Times and Fox News as the exceptions to the right.¹⁵ More recent work by computational linguists at Stanford, MIT, and elsewhere — using language-model analyses, source-selection metrics, and lexical-framing measures — has produced broadly similar ordinal rankings, with technical disagreements about magnitude. The Allsides Media Bias Chart, the Ad Fontes Media chart, and the AllSides Balanced News rating (all of which use different methodologies, and none of which are definitive) tend to place The New York Times and Washington Post in the "lean left" range, NPR slightly less so, the Wall Street Journal news pages near the center, the Wall Street Journal editorial page in the "lean right" range, and Fox News opinion programming in the "right" range.

The audience-composition data — discussed in 18.3 above — tracks the same pattern: self-identified Democrats consume the mainstream-press outlets at higher rates than do self-identified Republicans.

What this does not mean: it does not mean that The New York Times is the same as MSNBC, or that The Washington Post is the same as Pod Save America, or that NPR is propaganda for the Democratic Party. The mainstream-press outlets retain professional norms around fact-checking, source-confirmation, multiple-perspective sourcing, and editorial review that distinguish them from openly partisan outlets. They also retain financial and reputational incentives to break news that is damaging to Democrats — and they do so, as the Washington Post's reporting on the Hunter Biden plea agreement, the New York Times's reporting on the Biden classified-documents matter, and the major outlets' eventual coverage of the President's mental decline before the 2024 Democratic convention all demonstrate.

What it does mean: the cultural-political center of gravity of the working press is meaningfully to the left of the median American voter. Stories that confirm the worldview shared by that center of gravity are more likely to be covered, more likely to be foregrounded, and less likely to be skeptically interrogated than stories that contradict it. This is an institutional disposition, not a conspiracy. It has measurable effects. To deny that it exists is to argue against the data.

18.7.2 The Conservative-Media Ecosystem Leans Republican

The empirical reality on the conservative side is at least as clear, and the magnitude of the slant in the most-watched and most-read conservative outlets is generally larger than the magnitude of the slant in the most-watched mainstream-press outlets.

Fox News is the canonical case. Its prime-time programming is openly conservative; its hosts have endorsed Republican candidates, conducted interviews that function as campaign infomercials, and on multiple occasions overtly coordinated with Republican officials on messaging. The Dominion v. Fox News defamation suit (settled in 2023 for $787 million) produced internal communications showing Fox executives and on-air talent simultaneously aware that election-fraud claims they were broadcasting were false and worried that contradicting those claims would lose audience to more strongly Trump-aligned outlets. The settlement and the disclosed communications are public evidence, not interpretation.

Conservative talk radio has been Republican-aligned since the format consolidated in the late 1980s. The Wall Street Journal's editorial page, National Review, the Washington Examiner, the Federalist, the Daily Caller, Breitbart, Newsmax, OAN, the Daily Wire, and the various populist-right podcasts described in 18.6 occupy positions across the conservative-to-populist range, with substantial editorial diversity but consistent Republican alignment on most issues. The Wall Street Journal news pages are, by most content-analysis measures, near the center — substantially less editorially aligned than the editorial page.

The conservative critique is not that this ecosystem exists. It is that the existence of this ecosystem is the conservative response to a mainstream-press environment that tilts the other direction, and that the mainstream-press environment is consistently reported as if it were the unbiased default while the conservative ecosystem is reported as the deviation. There is something to that critique. There is also something to the counterclaim that, whatever its origins, the post-2000 conservative ecosystem has at certain moments been more openly indifferent to factual accuracy than the mainstream-press outlets — with the 2020 election-fraud claims as the clearest example. Both claims contain real content. Neither is a complete description.

18.7.3 Public Broadcasting

Public broadcasting — PBS for television, NPR for radio — receives a small fraction of its funding from federal appropriations (through the Corporation for Public Broadcasting), with the majority from listener and viewer contributions, foundation grants, and corporate underwriting. Content analysis and journalist-survey data suggest a modestly left-of-center editorial sensibility, though substantially less pronounced than the openly partisan outlets. NPR's senior business editor Uri Berliner published a much-discussed essay in The Free Press in April 2024 arguing that NPR's editorial culture had lost its capacity for ideological diversity and that this had distorted its coverage of significant stories; NPR's response defended its editorial practices. The episode produced congressional attention to public-broadcasting funding.

18.7.4 The Shrunken Center

The category of national-political outlets that are not visibly aligned with either party — what we might call the "contested-territory media" — is smaller than either the mainstream-press cluster or the conservative cluster. Politico and Axios operate as inside-Washington publications with reportorial cultures that emphasize political mechanics over ideology. The Atlantic runs both broadly progressive and broadly heterodox-conservative writers (Jeffrey Goldberg edits a publication that has run Conor Friedersdorf, David Frum, Jonathan Chait, Caitlin Flanagan, and Tim Alberta alongside many writers further to the left). The Bulwark operates as the institutional home of the never-Trump conservative intellectual tradition. Substack hosts an intellectually serious heterodox cluster — Andrew Sullivan, Bari Weiss (now via The Free Press), Matt Yglesias, Jonathan Haidt, Greg Lukianoff, Jesse Singal, Yascha Mounk, Glenn Loury — that resists conventional left-right classification and has captured a meaningful subscriber base. The aggregate audience of all of these publications combined remains smaller than either Fox News or The New York Times alone.

18.8 The Misinformation Question

"Misinformation" is a contested term. We use it to mean the circulation of factually false claims. We use "disinformation" for the deliberate production of false claims with intent to deceive. The boundary between them is often blurry. Both have always existed; the question is what is different about their current circulation.

18.8.1 What the Empirical Research Actually Shows

Hunt Allcott and Matthew Gentzkow's 2017 paper, the most-cited empirical study of social-media misinformation, examined the circulation of definitively false news stories during the 2016 election. They found that the average American adult saw approximately one or two false stories during the campaign, that pro-Trump false stories circulated more than pro-Clinton false stories by a factor of approximately three, and that the persuasive effect on voting behavior was probably small in aggregate.¹⁶ Subsequent work by Andrew Guess and colleagues, by the Stanford Internet Observatory, by Brendan Nyhan and Jason Reifler, and by various academic teams has produced broadly consistent findings: misinformation circulates, asymmetrically by issue and period, and its measurable effects on aggregate political behavior are typically modest rather than dramatic.

The asymmetries are real and worth naming honestly. Most of the demonstrably-false viral political content during the 2016 campaign was right-coded. Most of the demonstrably-false viral COVID-19 content during 2020–22 was right-coded (anti-vaccine claims, ivermectin/hydroxychloroquine claims, mortality-undercount claims). Most of the demonstrably-false viral content about the 2020 election outcome was right-coded (the various claims of fraud subsequently rejected by every court that examined them on the merits).

The mirror-image acknowledgments are also true and worth naming. Demonstrably-false claims about Russian-collusion specifics circulated widely on the left during 2017–19, with several specific items in the Steele dossier later substantially debunked. Demonstrably-false claims about voting machines circulated on the left in 2004 (the Kerry-vs.-Bush results in Ohio), among a small cluster of activists in 2016, and have appeared in various forms before. Demonstrably-false or substantially-misleading claims about police-shooting statistics circulate in both directions. Demonstrably-false claims about specific Trump-administration policies (such as the persistently miscounted family-separation totals from 2018, in both directions) circulated across mainstream and progressive outlets.

The net empirical picture is: meaningful asymmetry in some periods, particularly 2016–22 and on certain topics; not a complete asymmetry across the full landscape; and a system in which large amounts of false or substantially-misleading content circulate in all directions, with measurable but typically modest effects on aggregate behavior.

18.8.2 The Trump-vs.-Mainstream-Press Confrontation

Beginning in his 2016 campaign and intensifying throughout his first presidency, Donald Trump publicly characterized critical mainstream-press coverage as "fake news" and described mainstream-press institutions, on multiple occasions, as "the enemy of the people." The phrase had recognizable historical valences (it had been used by Lenin and by the Soviet press) that produced sharp pushback from press-freedom organizations and from many members of his own party. The frame did not, however, originate the conservative critique of the mainstream press; that critique had been institutionally durable since at least the 1960s, with significant intellectual lineage through William F. Buckley, Spiro Agnew's "nattering nabobs of negativism" speech in 1969, and decades of conservative-movement writing on press bias. Trump's contribution was the rhetorical intensity, the personalization, and the systematic attack on the legitimacy of the institutions themselves.

The institutional cost has been measurable. Trust in the mainstream-press institutions has fallen sharply during this period, particularly but not exclusively among Republicans; Gallup's annual trust polling shows trust in the mass media at a historical low across the population, with the partisan gap (Democrats' trust roughly 70 points higher than Republicans' on some specifications) the largest ever measured.¹⁷ Whether the trust collapse is primarily a response to Trump's framing or primarily a response to the mainstream press's actual conduct — which has at moments been openly anti-Trump in ways that, defenders argue, the unusual Trump presidency justified, and that critics argue was a self-inflicted wound — is genuinely contested.

18.8.3 Steel-Manning the Two Critiques

The right-coded critique, in its strongest form: mainstream-press institutions have allowed their professional culture to drift to the left of the median American voter; have systematically downplayed stories that complicate progressive narratives (the Hunter Biden laptop in October 2020; the COVID-origins lab-leak hypothesis in early 2020; the 2024 evidence of presidential cognitive decline in the year before his withdrawal from the race); have applied reportorial skepticism asymmetrically to the two parties' candidates; and have, through endorsement decisions, news-judgment patterns, and opinion-section composition, become substantively rather than nominally partisan in ways their professional self-image does not acknowledge. The strongest version of this critique points to specific stories where the asymmetry was visible, names the mechanism (institutional culture, hiring patterns, the social and political composition of newsroom communities), and proposes the cure (more ideological diversity in newsrooms, more skepticism about progressive pieties).

The left-coded critique, in its strongest form: the conservative-media ecosystem, particularly Fox News at its lowest moments and the populist-right talk-radio and podcast space, has actively promoted and circulated demonstrably false claims at scale — about election results, about pandemic response, about specific public-health measures — in ways the mainstream press has not. The asymmetry is not just one of slant; it is a difference in willingness to broadcast claims that the broadcasters knew or should have known were false. The strongest version of this critique points to the Dominion discovery materials, to the documented coordination between certain Fox programs and Trump White House officials, and to the empirical research on misinformation circulation, and proposes the cure (audience-side media literacy, regulatory pressure on the most clearly deceptive outlets, restoration of professional norms).

Both critiques have substantial empirical support. Both also have characteristic blind spots. The right-coded critique tends to underweight the right's own contribution to factual degradation. The left-coded critique tends to underweight the mainstream press's documented institutional asymmetries. A reader committed to actually understanding the contemporary American information environment needs to hold both critiques together and to ask, of any specific claim about a specific story, what the evidence on that specific claim actually shows.

18.9 Filter Bubbles and Echo Chambers

In 2011, Eli Pariser's The Filter Bubble argued that personalization algorithms were trapping internet users in self-confirming information environments — that each user, increasingly, was seeing only content that confirmed existing beliefs, and that this was systematically degrading public deliberation. The argument was intuitive, journalistically vivid, and widely cited. It also turned out, on subsequent empirical investigation, to be substantially overstated.

18.9.1 What the Research Found

A series of large-scale empirical studies over the 2014–2023 period have measured filter-bubble and echo-chamber effects directly. Eytan Bakshy and colleagues at Facebook (2015) found that users' personal choices about whom to follow accounted for more of their cross-cutting-content exposure than the algorithm did, and that even highly partisan Facebook users were exposed to substantial cross-cutting content.¹⁸ Andrew Guess, Brendan Nyhan, and colleagues, in multiple studies, have found that the share of Americans in extremely narrow ideological news diets is small (roughly 10% or less, depending on specification); that most Americans receive at least some exposure to cross-cutting content; and that the people most exposed to opposing views are often partisans themselves, who engage with opposing-view content combatively rather than tuning it out.¹⁹

The picture that emerges from the empirical literature is more complicated than Pariser's. Algorithmic personalization does shape content exposure; it does not, on average, shape it into the impermeable bubbles the original argument hypothesized. The bigger driver of partisan information sorting is selective exposure — the choice, made by users themselves, to consume sources they expect to confirm their existing views. Selective exposure has existed for as long as differentiated media have existed; it intensified with cable news, intensified again with the internet, and intensified again with social media, but the deeper pattern is one of human preference rather than algorithmic imposition.

18.9.2 The "Hostile Media Effect"

A related and durable empirical finding is the hostile media effect: people who are strongly partisan tend to perceive mainstream-press coverage as biased against their own side, regardless of which side they are on. Vallone, Ross, and Lepper's classic 1985 study showed both pro-Israel and pro-Palestinian viewers watching the same news segments and rating those segments as biased against their own position. The effect has been replicated repeatedly across topics and decades.²⁰

The hostile media effect has implications for the partisan-slant discussion. It means that the same coverage will be perceived as "liberal bias" by strong conservatives and "corporate centrism that protects Republicans" by strong progressives. It does not mean that there is no actual bias to measure — Section 18.7 reviewed the evidence that there is — but it does mean that the partisan perception of bias is not, by itself, a reliable signal of actual bias. Both perceptions can be present in different audiences, watching the same content, simultaneously.

18.10 The Collapse of Local Journalism

Of all the changes described in this chapter, the one with the most clearly negative civic consequences is the collapse of local journalism. The case study at the end of the chapter takes one specific newspaper closure as a worked example. Here we summarize the structural picture.

Approximately 3,000 American newspapers closed between 2005 and 2024 — about a quarter of the total in operation in 2005.²¹ Almost all of the closures were weeklies and small dailies. By 2024, approximately 1,500 American counties — roughly half — had no full-time local newspaper, and an additional substantial number had only "ghost" newspapers that retained the masthead but had eliminated most or all original local reporting. Counties without local news are concentrated in rural areas, in the South and Midwest, and in lower-income communities; the geography of the news desert is closely related to the geography of broader economic decline.

The civic effects are now well documented. Communities that lose their local newspaper see measurable reductions in voter turnout in local elections; reductions in the number of contested local elections; reductions in citizen knowledge of local-government activity; increases in municipal-bond interest costs (because investors price in higher uncertainty about how local money is being spent); increases in straight-ticket partisan voting in local races (because, absent local information, voters fall back on national partisan cues); and, by some studies, measurable corruption increases at the local level when investigative scrutiny disappears.²²

Public-trust patterns make the loss especially consequential. Local news has been the most-trusted category of American journalism in every Pew survey of news trust over the past two decades — substantially more trusted than national network news, cable news, online-only outlets, or social media. The collapse of local journalism is therefore the collapse of the most-trusted information channel, replaced by less-trusted national channels that often carry less directly relevant content.

The major chains — Gannett, Lee Enterprises, McClatchy, Tribune Publishing (until its 2021 acquisition by Alden Global Capital), and others — pursued strategies of consolidation and cost reduction through the 2010s and 2020s that, on balance, accelerated rather than slowed the decline of local reporting capacity. The case study following this chapter examines this in detail.

Rebuilding efforts exist. The American Journalism Project, Report for America, the National Trust for Local News, the Institute for Nonprofit News, and a growing nonprofit local-news sector have placed thousands of journalists in news-desert communities. State-level subsidies and tax credits for local news (in California, New York, Illinois, and other states) have been proposed and partially implemented. Whether these efforts can rebuild local reporting at the scale that has been lost is, as of 2026, an open question.

18.11 The Platform Regulation Debate

Section 230 of the Communications Decency Act of 1996 contains, in its operative provision, what has been called "the twenty-six words that created the internet": "No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider." The provision shields platforms from liability for user-generated content, while preserving the platforms' right to engage in good-faith content moderation without thereby becoming legally responsible for the content they moderate.

The provision is genuinely contested across the political spectrum. Both major parties have called for Section 230 reform, with substantially different goals.

18.11.1 The Conservative Critique of Section 230

Conservative critics argue that the platforms have moved well beyond "good-faith moderation" and now engage in systematic ideological curation that suppresses conservative speech. The 2020 Hunter Biden laptop suppression by Twitter (which Twitter executives later acknowledged was a mistake), the various deplatformings of conservative voices (including Trump after January 6), the documented internal communications produced by the Twitter Files releases in 2022–23, and the documented contacts between platforms and federal officials regarding content decisions are the principal exhibits. The argument: Section 230's protection was premised on platforms acting as neutral conduits, and platforms that act as ideological editors should bear ideological-editor responsibilities. The proposed cure: narrow the Section 230 shield to apply only to platforms that demonstrate genuine viewpoint neutrality.

Texas's HB 20 (2021) and Florida's SB 7072 (2021) attempted to enact this reasoning at the state level, prohibiting large social-media platforms from "viewpoint discrimination" and granting users private rights of action against suppressive moderation. The two laws were challenged and reached the Supreme Court as NetChoice v. Paxton (Texas) and NetChoice v. Moody (Florida). The Court's 2024 decisions vacated the lower-court rulings on procedural grounds and remanded for further analysis, while signaling that platforms have substantial First Amendment editorial-discretion claims of their own. The Texas and Florida laws remain partially blocked.

18.11.2 The Progressive Critique of Section 230

Progressive critics argue that Section 230's broad immunity has allowed platforms to host harmful content — disinformation, harassment, child-sexual-abuse material, terrorism recruitment, election-undermining claims — at scale, with insufficient accountability. The platforms, the argument runs, have the technical capacity to moderate at much higher levels of effectiveness; they decline to do so because moderation is expensive and reduces engagement; the legal shield removes the financial incentive that would otherwise force them to moderate well. The proposed cure: condition Section 230 protection on demonstrated investment in content moderation, on transparency about algorithmic ranking, and on user redress when moderation decisions are wrong.

Senator Mark Warner's bipartisan SAFE TECH Act and the bipartisan EARN IT Act represent attempts in this direction. The Biden administration in 2021 endorsed Section 230 reform without specifying details. The Trump administration in both terms has called for Section 230 reform from the conservative direction.

18.11.3 The Murthy v. Missouri Litigation

A separate but related strand concerns government coordination with platforms on content moderation. The states of Missouri and Louisiana brought suit alleging that federal-government officials had pressured social-media companies to remove or suppress speech in violation of the First Amendment. A district court issued a sweeping injunction limiting federal contact with platforms; the Fifth Circuit narrowed the injunction; the Supreme Court in Murthy v. Missouri (2024) reversed the lower courts on standing grounds, finding that the plaintiffs had not established a sufficiently direct injury. The merits question — when does government communication with platforms cross from permissible engagement to unconstitutional coercion — remains unresolved.

18.11.4 The International Frame

The European Union's Digital Services Act (DSA), which came into full force in 2024, takes a more interventionist approach than U.S. law: it imposes specific transparency, risk-assessment, and content-moderation obligations on large platforms, with substantial fines for noncompliance. The United Kingdom's Online Safety Act (2023) takes a similar approach. American platforms operating in those jurisdictions are subject to the DSA's requirements; the regulatory divergence between U.S. and European approaches is now the largest in the history of internet regulation. Whether the European model is exportable to the United States, and whether the U.S. First Amendment forecloses key provisions of it, are open questions.

18.12 Press Freedom

The First Amendment's protection of the press is older and broader than any other major democracy's. The Founders' debates show explicit concern with preventing the prior-restraint and seditious-libel suppression that had characterized the British system; the constitutional text and the early case law (Near v. Minnesota, 1931, on prior restraint; NYT v. United States, 1971, on the Pentagon Papers) developed protections for press activity that remain among the world's strongest.

18.12.1 NYT v. Sullivan

New York Times v. Sullivan (1964) established the actual-malice standard for defamation claims by public officials: such plaintiffs must show that the defendant either knew the statement was false or acted with reckless disregard for its truth or falsity. The standard was extended to public figures in Curtis Publishing v. Butts (1967). The doctrine has substantially shielded American journalism from defamation liability and is widely credited with enabling investigative reporting that would not have been possible under traditional common-law libel rules.

In recent years, several Supreme Court justices — Justice Clarence Thomas (most prominently, in his 2019 concurrence in McKee v. Cosby) and Justice Neil Gorsuch (in his 2021 concurrence in Berisha v. Lawson) — have called for the Court to revisit Sullivan. Their argument: the original-meaning case for Sullivan's actual-malice rule is weak; the doctrine has produced a press culture protected from accountability for genuine recklessness; and the changed media environment (with social-media defamation now reaching audiences vastly larger than the print era) requires reconsideration. Defenders of Sullivan argue that the doctrine remains essential to investigative journalism and that any narrowing would chill reporting on powerful figures.

The strongest version of the retention argument: investigative journalism necessarily reports on incomplete information and sometimes gets details wrong; without the actual-malice standard, every error becomes a tort claim, and the cost of defending against claims (even meritless ones) crushes outlets without deep pockets. The strongest version of the reform argument: the modern environment, in which large outlets and individual social-media users alike can publish to vast audiences, allows recklessness in ways the Founders could not have imagined; some intermediate standard between strict liability and actual-malice may better balance speech and reputational interests.

18.12.2 Reporter's Privilege and Shield Laws

Approximately forty states and the District of Columbia have shield laws that protect journalists from being compelled to identify confidential sources. The protections vary in scope. There is no federal shield law, despite repeated proposals (most recently the bipartisan PRESS Act, which has passed the House multiple times and never the Senate). Federal courts have limited authority to recognize a reporter's privilege under common law, with the Supreme Court's Branzburg v. Hayes (1972) refusing to recognize a constitutional privilege.

The absence of a federal shield law has produced periodic confrontations between federal prosecutors and journalists. Judith Miller v. United States (2005, in which Miller was jailed for refusing to identify her source in the Plame investigation) is a canonical case. The Trump and Biden administrations both pursued leak investigations that involved journalists; both administrations also drew criticism for the reach of those investigations. The PRESS Act remains, as of 2026, a perennially-proposed but never-enacted measure.

18.12.3 International Press-Freedom Rankings

The United States's position in international press-freedom rankings has slipped. Reporters Without Borders' World Press Freedom Index ranked the U.S. 17th in 2002, 27th in 2010, 41st in 2020, and 55th in 2024.²³ The slip reflects multiple factors: the Trump-era hostility to mainstream press; the Biden-era leak-investigation activity; the consolidation of media ownership; the financial precarity of local journalism; and several specific incidents (police arrests of journalists during 2020 protests, journalist-related violence). The U.S. score remains substantially above the global median; the trend, however, is downward.

18.13 The New Media Business Models

The financial wreckage of the 2005–2020 period left an industry in active reconstruction. Several distinct models have emerged.

Subscription-supported institutional media. The New York Times now exceeds 10 million paying digital subscribers globally — a transformation, given that in 2010 it had a few hundred thousand. The Washington Post, the Wall Street Journal, The Atlantic, The New Yorker, the Financial Times, and a growing number of international and specialty outlets have built subscription-supported business models that sustain substantial professional newsrooms. The model works best for outlets with strong national or specialty brands; it has been less successful at the local level.

Ad-supported digital. The vast majority of digital news traffic is monetized through advertising, with the platform companies (Google, Meta) capturing a substantial share of new ad spending. The economics for ad-supported pure-digital news outlets remain difficult. BuzzFeed News (closed 2023), Vice News (parent company bankruptcy 2023), and many others have failed at this model; HuffPost and a small set of others continue with reduced staffs.

Substack and the creator-economy newsletter. Substack, founded 2017, allowed individual writers to operate paid-subscription newsletters with platform-provided distribution and payment infrastructure. By 2024, Substack hosted hundreds of writers with paid-subscriber bases in the thousands or tens of thousands, and a smaller number with bases in the hundreds of thousands. The aggregate revenue is meaningful but small relative to the institutional press; the cultural footprint is larger. The creator-economy model has captured a significant fraction of the heterodox, contrarian, and ideologically-difficult-to-place commentary space.

Nonprofit local journalism. Outlets like ProPublica (national), the Texas Tribune, Voice of San Diego, MinnPost, CalMatters, VTDigger, the Connecticut Mirror, and a growing number of state and regional newsrooms operate on philanthropic-and-membership funding. The American Journalism Project, founded 2019, has invested approximately $200 million in local nonprofit newsrooms. Whether the nonprofit-local model can scale to replace the lost commercial-local capacity is an open question; the model is now established as part of the long-term media ecology.

Public broadcasting. PBS and NPR retain federal funding (through CPB), supplemented by listener and viewer contributions and underwriting. Public-broadcasting funding has been a recurring partisan flashpoint, and the 2025 Trump administration proposed substantial cuts to CPB appropriations. Whether the public-broadcasting infrastructure survives the next funding cycle is, as of 2026, contested.

18.14 What This Chapter Asks You to Hold

The American media system in 2026 is not the same system that existed in 1985, or in 1995, or in 2005. The transformation has been more rapid than the institutional capacity of the country to adapt to it, and the consequences are still working themselves out. The exercises and case studies that follow ask you to apply the chapter's frameworks to specific stories, specific outlets, and your own information environment.

The honest summary statements are:

• The mainstream press's editorial culture leans Democratic-leaning. The conservative-media ecosystem is Republican-leaning. Both are real. Both have measurable effects. Neither implies the other does not exist.
• Misinformation circulates at scale on social media and elsewhere. The asymmetries are real but partial: significant on certain topics in certain periods, more symmetric on others, never zero in either direction.
• Filter bubbles exist but are weaker than the most popular versions of the theory imply. Selective exposure, by users themselves, drives more partisan information sorting than algorithmic personalization does.
• Local journalism has collapsed. Its collapse has measurable civic costs. Rebuilding efforts exist but have not scaled to the loss.
• Section 230 reform is genuinely contested across the political spectrum, with serious arguments on multiple sides and very different proposed cures.
• NYT v. Sullivan protects American press freedom in ways most other countries' law does not, but its retention is no longer politically uncontested.
• The American press-freedom ranking has slipped. The slippage reflects real trends, not measurement noise.

A citizen is not obligated to draw any particular normative conclusion from any of these facts. A citizen is, however, badly served by an account of the media system that pretends any one of them away. The reader who finishes this chapter should be in a position to evaluate, on their own and on the evidence, what the contemporary information environment is doing to the country's capacity for democratic self-government. That evaluation is one of the central tasks of democratic citizenship in 2026.