Not an editor. Not a journalist. An algorithm written by an engineer accountable to shareholders — and optimized for one thing: keeping you engaged.

The economic model is simple and grim. Your attention is the product. Whether you're watching broadcast television, scrolling a social feed, or reading an ad-supported news site, the business runs on time-on-platform sold to advertisers. Content that provokes outrage, fear, or tribal contempt generates more engagement than content that is accurate, locally specific, or calm. This isn't ideology. It's arithmetic.

The structural consequence took decades to build and is now fully visible. In 1983, roughly fifty companies controlled the majority of American media. By the early 2000s, that number had collapsed to around six major conglomerates. The names reshuffle with each merger cycle, but the dynamic holds: scale is rewarded, independence is economically precarious, and the pressure toward consolidation is structural rather than conspiratorial.

What this looks like in practice: the same parent company owns the local television station you check for weather, the newspaper whose site you visit for community news, several radio stations you hear on your commute, and a digital portal aggregating all of it. Editorial diversity becomes illusory. When a parent company issues cost-cutting directives — or when an owner carries documented political preferences — the effects move simultaneously across outlets that appear independent from the outside.

The starkest consequence is the news desert. Research from the Hussman School of Journalism and Media at the University of North Carolina tracked local newspaper collapse across the United States with county-level precision. Between 2004 and 2020 alone, the country lost more than a quarter of its newspapers. Hundreds of counties — home to millions of people — now have no local paper at all. Ghost newspapers compound the damage: outlets that technically exist but have been gutted of reporting staff until they publish almost no original local journalism.

The effects are measurable. Studies correlate news desert conditions with lower voter turnout, higher municipal borrowing costs — bond markets price in information asymmetry — reduced civic engagement, and increased political polarization. National partisan media fills the vacuum left by absent local journalism. When no one covers the school board meeting, the water authority, the factory layoff, the architecture of democratic accountability doesn't weaken gradually. It collapses.

What do wartime propaganda posters and a 2024 political ad have in common? More than the people running both campaigns would like you to notice.

Propaganda, defined carefully, means the systematic use of communication to advance a particular agenda — by selecting, framing, and emphasizing information to produce desired beliefs rather than to inform. Defined that way, it isn't an artifact of authoritarian history. It's practiced daily by democratic governments during wartime, corporations managing crisis narratives, political campaigns during election cycles, and interest groups across every ideological position.

The classical toolkit has been documented since the 1920s. Framing — presenting accurate information within a context that predisposes audiences toward a particular interpretation — is the most pervasive technique. The same economic statistic reads as evidence of recovery or evidence of deepening inequality depending on which aspects get emphasized. Neither framing requires a single false statement. Repetition exploits a cognitive tendency psychologists call the illusory truth effect: repeat a claim enough times across enough contexts and people's confidence in its truth increases, independent of whether it has ever been verified. This is robustly established in experimental literature, not theoretical conjecture.

Manufacturing consent — a phrase associated with Edward Herman and Noam Chomsky's political analysis, though the underlying mechanism has been examined by many scholars across decades — describes how the range of acceptable opinion in a society narrows through accumulated media structures, sourcing practices, and editorial norms, without any central coordinator issuing instructions. The claim isn't that journalists are witting agents of controlling interests. It's that the structures within which journalists operate systematically favor certain voices, certain framings, certain assumptions about what counts as obvious versus what requires explanation. Critics argue this model underestimates genuine diversity in pluralistic media environments. The debate is real. The mechanisms it identifies are also real.

What changed in the contemporary period isn't the toolkit. It's the speed and the distribution of access. Disinformation — the deliberate creation and spread of false or misleading content — now moves through social networks in hours. A fabricated story reaches tens of millions of people before a careful refutation is issued by anyone credentialed to issue one. When the refutation arrives, days later, it reaches a fraction of the original audience and lands against an already-formed impression.

The techniques of the twentieth century didn't disappear with the regimes that refined them. They were digitized and made available at scale — to governments, corporations, political operatives, and to anyone with a smartphone and a grievance.

Are the major technology platforms publishers or infrastructure? The question sounds semantic. The answer determines who is accountable for what.

Previous media control debates were about publishers and broadcasters — entities that produced content and distributed it. The dominant platforms today are better understood as infrastructure: search engines, social media networks, and app stores that control mobile information access at civilizational scale. They make editorial decisions constantly while officially disavowing editorial responsibility. When a platform removes content under community standards, it is making an editorial judgment. When its algorithm surfaces one kind of content and buries another, it is shaping the information environment of billions of people. The legal frameworks built for broadcast and print media were designed for a world where those distributing information were clearly identifiable as editors or as common carriers. The platforms occupy an uncomfortable middle position that no existing framework handles adequately.

Engagement optimization — the algorithmic practice of surfacing content predicted to maximize user interaction — is the mechanism through which attention-economy incentives operate at scale. What is established: major platforms use ranking systems that prioritize high-engagement signals. Emotionally provocative content consistently generates more engagement signals than calm, informational content. What is debated, with serious researchers on multiple sides: the magnitude of real-world political effects, and whether platforms' own moderation adjustments have meaningfully mitigated the dynamic.

The filter bubble — the idea that algorithmic personalization locks people inside information environments that only confirm existing beliefs — became culturally dominant in the 2010s and remains a fixture of popular discourse. The empirical picture is more complicated. Some studies find that algorithmic feeds expose people to more cross-cutting political content than they would voluntarily seek. Others find significant segregation effects depending on platform and population. The filter bubble story, as commonly told, is probably too simple. That doesn't mean the underlying concern about algorithmic curation and epistemic segregation is unfounded.

The governance question remains genuinely unresolved. The European Union moved toward assertive regulation with its Digital Services Act, imposing transparency and accountability requirements on large platforms. The United States has remained largely deferential to platform autonomy. India, Brazil, and others have pursued aggressive government control of platform operations — raising their own serious free-press concerns. Different democracies are reaching different conclusions. The divergence is itself information.

State-sponsored information operations are documented, not speculated. Multiple investigations — by academic research groups, journalistic outlets, and intelligence agencies across several countries — have established that governments run systematic campaigns to manipulate online information environments.

The techniques include networks of inauthentic accounts amplifying preferred narratives, targeted advertising to specific population segments, the strategic seeding of divisive content designed not to persuade but to deepen existing fractures, and the use of state-controlled media outlets as laundering channels for content originally placed in nominally independent venues.

Computational propaganda — the use of automated accounts, data analytics, and targeted communication strategies to manipulate political discourse at scale — has been studied extensively by researchers at institutions including the Oxford Internet Institute. Their findings are precise and uncomfortable: computational propaganda is not confined to authoritarian regimes. It has been deployed by democracies during election campaigns and by non-state actors across the ideological spectrum. No political tendency has clean hands here.

What makes state information operations particularly difficult to analyze is how they interact with genuine organic sentiment. A state actor seeking to amplify social division doesn't need to invent grievances. It works with real grievances, real communities, real content — amplifying, distorting, and reframing in ways that are nearly impossible to distinguish from organic activity. The goal is often not to create a specific false belief. It is to degrade the overall information environment: to make it harder to know what is real, who is credible, and what can be trusted.

This is epistemic warfare. It is more insidious than classic propaganda precisely because it doesn't require convincing anyone of anything specific. It operates by manufacturing uncertainty. A population that trusts nothing and can verify nothing is not a free population exercising skepticism. It is a population that has lost the shared informational ground on which democratic disagreement depends.

Between all these structural forces — ownership concentration, attention-economy arithmetic, platform power, state information operations — are journalists. People who, by and large, entered a profession because they believed it had a civic function. They are now doing that work inside institutional environments under severe economic and political stress.

Two things are simultaneously true here. First: the vast majority of working journalists in pluralistic democracies are genuinely trying to report accurately. Journalistic culture maintains norms of verification, source independence, and skepticism toward power. Those norms are real and consequential, even when imperfectly practiced. Second: those norms operate within structural constraints that shape what gets covered, how it gets framed, and whose voices get treated as authoritative — often in ways that individual journalists are not fully aware of, and that don't require any individual bad actor to sustain.

Source dependence is among the most consequential of those constraints. Journalists need information. Information tends to come from people in institutional authority — government officials, corporate communications departments, credentialed experts in established fields. This creates a systematic tendency to amplify perspectives already embedded in power structures, and to demand much higher evidentiary burdens from outsider perspectives before they merit coverage. This isn't ideological bias in the crude sense. It is an institutional tendency embedded in the practical logistics of newsgathering, documented by media scholars for decades.

Both-sides journalism — the default formula of presenting "both sides" of a contested issue as a structural requirement for appearing objective — creates its own distortions. When genuine expert consensus on a factual question — climate science, vaccine safety, electoral integrity — is presented as one position in a debate with a contrary view held by a small minority of credentialed experts or by politically motivated non-experts, the audience receives a systematically inaccurate picture of the actual state of knowledge. The journalism profession has been doing genuine self-examination on this. The concept of false equivalence — as a recognized failure mode distinct from traditional bias — has gained meaningful traction inside major newsrooms. Whether that self-examination produces lasting change in practice is a different question.

Media literacy education gets promoted as the primary response to all of the above. The value is real. So are the limits.

Teaching people to examine sources, check claims across multiple outlets, distinguish news from opinion, and recognize common manipulation techniques — this matters. There is also evidence that some forms of media literacy education backfire, producing generalized suspicion of all institutional sources. The result is epistemic nihilism: a condition in which nothing is trusted and all claims are treated as equally contestable. Paradoxically, manufacturing that condition is one of the stated goals of several state information operations. Research on what kinds of media literacy interventions actually improve real-world decision-making — rather than just scoring well on awareness tests — is still developing.

Lateral reading is a more specific technique, and one with better empirical support. The practice was identified by researchers studying how professional fact-checkers work: rather than reading a source deeply to evaluate it, immediately look sideways at what credible third parties say about where it comes from. Open multiple tabs. Check the reputation history. Deceptive or unreliable sources are usually identifiable through documented track records accessible somewhere on the open web. The counterintuitive element is that deep engagement with a single source is often less reliable than a quick external check of its provenance.

The deeper structural question is whether individual-level media literacy is a sufficient response to what are fundamentally structural problems. The information environment is not neutral terrain. It is shaped by powerful economic and political forces that individual awareness, while necessary, cannot fully counter alone. The analogy sometimes offered: teaching individuals to avoid pollution is valuable, but it doesn't substitute for regulating the industries producing it. The analogy is contested. It also captures something real about the limits of purely individual-level responses to systemic architecture.

The attention economy, ownership consolidation, algorithmic amplification, and state information operations did not emerge from individual failures of critical thinking. They emerged from incentive structures, regulatory choices, and technological affordances that operate at a scale no individual media consumer assembled or can disassemble alone.

Artificial intelligence now enables the generation of convincing text, audio, and video at near-zero marginal cost. This changes the question.

For most of the history discussed in this article, the tools of narrative creation required organizations — studios, printing presses, broadcast licenses, distribution networks. The barrier to fabrication was partly technological and economic. That barrier is gone. The question of who controls the narrative is now dramatically more complicated because the tools of narrative creation are no longer restricted to organizations with institutional infrastructure. Anyone can fabricate a convincing one.

Synthetic media — AI-generated text, audio, and video designed to appear authentic — does not require distinguishing truth from lies in the traditional sense. It requires distinguishing real from fabricated at the level of the artifact itself, before the question of accuracy even arises. This is a categorically different epistemic problem from the ones journalism and media literacy were designed to address.

What happens to the concept of a verifiable factual record when fabricating a convincing version of any voice, face, or document costs almost nothing? The answer is not yet written. How societies respond in the next decade will determine much of what political and social reality looks like for generations after it. That is not hyperbole. It is a structural observation about a threshold already crossed.

The machinery of control has always evolved with communication technology. The printing press broke one kind of monopoly. The telegraph broke another. Radio and television created new ones almost immediately. Each revolution in communication technology was followed by consolidation — the tools of mass attention absorbed into fewer hands. AI is the next such threshold. The consolidation dynamic that follows it will be shaped by decisions being made right now, largely outside public view, by a small number of engineers, executives, and regulators.