Almost every public quarrel about the news media begins with a name. A proprietor is identified, a headline is quoted against him, an editor’s past employment is rehearsed, and the argument closes as if a motive had been demonstrated and a mechanism explained. The quarrel feels satisfying because it has a villain, a document, and a verdict. It is also, in the judgement of the strongest Marxist writing on the press and broadcasting, an argument built on the weakest available version of the theory it claims to use. Marxist analysis of news contains at least three distinct accounts of how coverage becomes ideological, and they are not three strengths of one claim. They locate the cause in different places, they predict different patterns in the evidence, they fail in different ways, and an argument that slides between them without noticing will be refuted in the form in which it happens to be standing when the refutation arrives.

The three accounts can be named before they are examined. The instrumentalist account holds that news serves the interests of capital because the organisations that produce it are owned by capitalists, directed by their appointees, and disciplined by their commercial purposes. The structuralist account holds that news would lean in much the same direction whoever owned it, because the routines of gathering, the economics of the format, and the dependence on official sources are built into the production process itself. The hegemonic account, developed most influentially by Stuart Hall, holds that news is a field of contest in which a dominant framing is offered, defended, and sometimes refused, so that consent is continually won rather than simply imposed. Each account has a founding literature, a body of evidence it handles well, and a class of facts it handles badly. The purpose of separating them is not taxonomic tidiness. It is that nearly every real dispute about media bias is lost or won at the moment the disputant, usually without knowing it, selects one of the three and is then attacked, fairly or not, as though the selection were the whole theory.

The argument about news media bias usually starts in the wrong place

Begin with the shape of the ordinary accusation. A reader finds a story slanted: a strike reported as a disruption rather than as a claim, a budget described through its effects on markets before its effects on households, a war framed in the vocabulary of the government conducting it. The reader then asks who benefits, answers with the name of an owner or an advertiser, and treats the answer as the explanation. The move from slant to beneficiary to intention is so familiar that it rarely gets examined, yet each step conceals a difficulty. Slant can be established by comparison and measurement. Benefit is harder, since a framing can serve an interest nobody pursued. Intention is hardest of all, because the inner purposes of proprietors and editors are precisely what an outside critic cannot inspect, and any theory that requires them has placed its central claim where evidence cannot reach.

The Marxist tradition’s own history explains why the intention centred version dominates public argument despite its weakness. Of the three accounts, instrumentalism is the only one that fits the grammar of scandal. It has persons, decisions, and a smoking memorandum, and it therefore travels well through the very media it describes, which select for conflict and character in every other field and naturally select for them here too. Structuralism, by contrast, offers deadlines, source lists, advertising ratios, and the price of newsprint: causes without faces. Hegemony offers a process without a guaranteed outcome, which is the least quotable commodity of all. The result is a systematic distortion in the reception of the theory, parallel to the distortion the theory describes in its object. The account of media ideology that circulates most widely is the one the media’s own selection pressures favour, and it is the one least able to survive an informed opponent.

That opponent has an easy victory available, and it is worth watching the victory being won, because the manner of it teaches the central lesson of this article. The critic of instrumentalism needs only one honest counter example: a proprietor who published an investigation damaging to his own commercial interests, a newspaper that campaigned against a policy its advertisers favoured, an editor who resigned rather than bend a story. Each instance is real, and against a theory of command they are decisive. The defender then retreats, usually to the claim that the exceptions prove a general rule, or that the owner’s deeper interest was served by a display of independence, and at that moment the theory has become unfalsifiable in the mouth of its own defender, since every outcome, obedience and defiance alike, is absorbed as confirmation. A theory that began as an accusation ends as a mood.

The name this article gives to that whole sequence is the intention trap. The trap is not the claim that owners sometimes intervene, which is documented and true. The trap is the inference structure that makes intention the load bearing element of the explanation: find the slant, name the beneficiary, presume the instruction, and treat the presumption as proven by the slant it was summoned to explain. Once inside the trap, the analyst cannot lose an argument and cannot win one either, because the opposing instances are never allowed to count. The structuralist and hegemonic accounts are stronger precisely where the trap is weakest. Neither requires a villain. Each specifies a mechanism that operates through ordinary professional conduct, and each therefore generates claims that evidence can support or damage. Distinguishing the three accounts, and knowing which one any given argument is making, is the skill the rest of this article builds, section by section, before applying it to the hardest cases: the propaganda model, the Glasgow content analyses and their critics, the platform reordering of news distribution, and the stubborn empirical question of filter bubbles.

Is media bias deliberate or structural?

Marxist accounts divide on this. Instrumentalist versions locate bias in ownership and editorial direction. Structuralist versions locate it in routines, sourcing, and commercial format, requiring no intention. Hegemonic versions treat news as a contested field where consent is won and can be refused by audiences.

The question inside that heading deserves a moment of respect before the accounts are separated, because it is the question readers actually bring. Deliberate or structural is not a request for a label. It is a request to know where to look: at persons or at processes, at the proprietor’s office or at the reporter’s Monday morning, at what was ordered or at what nobody needed to order. The three accounts are three different answers to the question of where to look, and they send the investigator to different archives. Instrumentalism sends her to share registers, appointment columns, and memoirs of dismissed editors. Structuralism sends her to the newsroom itself, to observe the diary meeting, the agency wire, the source round, and the advertising department’s quiet knowledge of which subjects sit badly beside which products. Hegemony sends her to the audience as well as the newsroom, to the letters page, the focus group, and the stubborn comedy with which viewers remake the meanings they are sent. An investigator with limited time, which is every investigator, must choose an archive before she can gather a fact, and the choice is already theoretical.

One further preliminary matters, because confusion about it poisons the whole field. Bias and ideology are not synonyms in this literature, and the difference carries much of the argument. Bias, in ordinary usage, names a deviation from a standard of fair or accurate reporting that the speaker believes could in principle be met: the story was biased, a better journalist would have written it straight. Ideology, in the Marxist usage reconstructed in the marxist theory of ideology guide elsewhere in this series, names something prior to individual performance: a framework of categories, saliences, and taken for granted connections within which events become reportable at all. A newsroom can be scrupulously unbiased by professional standards, quoting both sides, checking every figure, correcting every error, and still produce ideology in this second sense, because the decision about what counts as an event, whose statements count as the definition of the situation, and which questions structure the interview was made before any question of fairness arose. Most public argument uses the first word while the theory is about the second, and the mismatch explains why journalists, who know their professional conduct from inside, so often experience Marxist media criticism as a personal accusation it need not be.

Three accounts of how news becomes ideological

Set the three accounts side by side and their incompatibility becomes visible, which is the point of the exercise. Instrumentalism and structuralism disagree about the causal weight of ownership: the first makes it the engine, the second makes it one condition among several and predicts that a change of owner, leaving routines and markets untouched, would change less than the first expects. Structuralism and hegemony disagree about closure: the structuralist mechanism grinds on with a certain indifference to its content, while the hegemonic account insists the content is a stake in a continuing struggle whose outcome is not written into the machinery. Hegemony and instrumentalism disagree about audiences most sharply of all. For the crude instrumentalist the audience is a recipient; for Hall it is a producer of meanings, capable of refusing the preferred reading wholesale. The accounts can be combined in a sophisticated analysis, and often are, but the combination has to be argued joint by joint, because each joint conceals a place where the mechanisms pull against each other.

The three accounts grid below states the comparison in the form the rest of the article uses. It is the findable artifact for this page, and the sections that follow unpack each row: the mechanism in its strongest form, the evidence that has actually been offered, and the counter evidence that its defenders are obliged to answer.

Account	Mechanism	What it predicts	Evidence that would support it	Evidence that would count against it	Familiar argument in this form
Instrumentalist	Owners and their appointees direct coverage toward the interests of capital	Coverage shifts when ownership shifts, and proprietorial interventions cluster on commercially sensitive stories	Documented interventions, dismissals of editors, suppressed stories with commercial stakes	Owners repeatedly publishing against interest with no later penalty or correction	The proprietor ordered it
Structuralist	Routines, sourcing dependence, and commercial format select and frame stories regardless of intention	Similar framing across outlets with different owners, strongest where deadlines and official sources dominate	Convergent framing in comparative content analysis, reliance on official definers, advertiser sensitive topic avoidance	Framing that tracks ownership rather than format, or routines that systematically amplify subordinate voices	The system produces it whoever is in charge
Hegemonic	Dominant framings are offered within a contested field and must win audience consent	Preferred readings dominate without uniformity, negotiated and oppositional decodings appear at predictable social locations	Audience research showing patterned decoding, framings shifting after public contest, concessions inside coverage	Uniform passive acceptance across audiences, or framings that never move under pressure	The coverage manufactures consent, and consent can fail

Read the rows as wagers. Each account bets on where the pattern will show up if an investigator goes looking, and the bets differ. The instrumentalist bets on variance between outlets explained by their owners. The structuralist bets on invariance between outlets explained by their shared production conditions. The hegemonic account bets on variance within a single text’s reception explained by the social position of its audiences. Because the bets differ, the same body of research can confirm one account while embarrassing another, and the lazy synthesis in which all three are somehow true at once dissolves on inspection. What a mature Marxist media analysis usually contains is a hierarchy: an account of which mechanism dominates in a given medium, period, or story type, with the others assigned their proper and smaller provinces. The propaganda model, examined below, is best read as exactly such a hierarchy, a structuralist engine with instrumentalist components bolted on at specified points, which is one reason careless summaries of it go wrong.

A word about ancestry will prevent a common confusion. The three accounts are not three historical stages in which a crude theory matured into a subtle one, though textbooks often arrange them that way. Instrumentalist claims appear in the earliest socialist press criticism, in complaints about the commercial press of the nineteenth century, and they retain serious defenders. Structuralism descends from the sociology of news production that grew up in the 1970s, when researchers began sitting in newsrooms rather than only counting column inches, and from the older Althusserian claim that ideology works through apparatuses and practices rather than through sermons. Hegemony descends from Gramsci’s prison reflections on consent, carried into media study by the Birmingham Centre for Contemporary Cultural Studies and above all by Hall. The accounts are rivals with genealogies, not rungs on a ladder, and the marxist media theory explained survey in this series traces the wider tradition from which all three were drawn. What follows takes each in turn at full strength, because a comparison among weakened versions would teach nothing.

The instrumentalist account: ownership, direction and command

Instrumentalism begins from a fact nobody disputes and draws from it the strongest conclusion the fact will bear. News organisations of any scale are expensive. Printing plants, broadcast licences, distribution networks, and legal departments require capital in quantities that place ownership in few hands, and the owners of an asset ordinarily expect the asset to serve their interests, or at minimum not to injure them. The instrumentalist argument adds a premise: that in the case of media assets the expectation is routinely enforced, through the appointment of sympathetic editors, the signalling of preferences, the occasional direct intervention, and the quieter discipline by which ambitious journalists learn the boundaries of the house without ever being shown a written rule. Ralph Miliband gave the account its most cited scholarly form in The State in Capitalist Society, published in 1969, where the mass media appear as part of the ideological apparatus of a capitalist order, staffed disproportionately by people drawn from and sympathetic to the dominant class, and reliably hostile to the ideas that order cannot accommodate.

The strongest evidence for the account is not statistical. It is documentary, and it accumulates at the level of the individual firm. Press history is dense with proprietors who treated their newspapers as instruments and said so. The great British press barons of the early twentieth century, Northcliffe and Beaverbrook among them, built political campaigns out of their titles and boasted of the power involved. Beaverbrook, asked about his purposes, described running a paper for propaganda with a candour later proprietors learned to avoid. Editors have been dismissed after displeasing owners often enough that the pattern has a literature of its own, and reporters in many countries can recount the story that was softened, delayed, or killed when it touched the owner’s other interests: a property development, a regulatory application, a political friendship. The instrumentalist reads these episodes not as scandals interrupting a neutral norm but as the norm becoming visible at its edges, the moments when a discipline normally exercised through anticipation has to be exercised in person.

Selection does much of the account’s work before any intervention is needed, and this is its most defensible component. Owners appoint editors, editors appoint senior staff, and senior staff set the terms on which juniors are judged. Nobody needs to instruct a newsroom whose leadership was chosen for its instincts. The process resembles the appointment of judges or central bankers: formal independence coexists with a selection system that reliably produces a range of acceptable views and excludes others in advance. A proprietor who never telephones the newsroom may still govern its output more completely than a censor who telephones daily, because the censor fights the text while the appointment system has already shaped the hands that write it. Critics who demand the telephone call as the only admissible evidence of control are demanding the crudest mechanism as proof of the subtlest one, and the instrumentalist is entitled to that reply.

What does an instrumentalist explanation actually claim?

An instrumentalist explanation claims that coverage serves the interests of capital because owners direct it, through appointments, signals, and interventions, and that slant therefore traces back to decisions by identifiable people. Its evidence is documentary: interventions, dismissals, and suppressed stories, and it weakens wherever owners demonstrably publish against their own interests.

The account also has a political economy of attention working for it. A media firm is rarely only a media firm in the portfolio of its owner. Conglomerate ownership, which spread through the industry in the later twentieth century, ties a newspaper or a channel to interests in energy, defence contracting, telecommunications, property, and finance, each with regulatory needs before the state. Coverage of those sectors then has a balance sheet significance beyond the newsroom, and the temptation to align editorial judgement with the parent company’s petitions does not depend on any owner’s ideology. It depends on arithmetic. The instrumentalist’s best modern form is therefore not the cigar smoking baron but the diversified corporation, for which a news division is simultaneously a business, a lobbying asset, and a reputational shield, and whose managers are rewarded for treating it as all three.

Where the account breaks down is equally specific, and its defenders do the theory no service by refusing the inventory. The first failure is variance. Outlets under the same ownership regularly take opposed lines on the questions their owner is supposed to care about, because audiences differ and because a stable of titles is often managed as a portfolio of brands rather than as a single voice. The second failure is the published counter example of the damaging kind: investigations that cost a proprietor money, allies, or legal peace, pursued to publication by his own journalists. The thalidomide campaign conducted by the Sunday Times in Britain, a long and expensive fight against a major company on behalf of injured children, is hard to file under service to capital in general, whatever is said about the distinction between fractions of capital. The third failure is professional: the account has no serious room for the occupational pride, rivalry, and scepticism of journalists themselves, who are treated as transmission belts in a model that their own memoirs, including the disaffected ones, rarely confirm.

Miliband himself was subtler than the caricature, and the debate his book provoked is instructive. Nicos Poulantzas, reviewing the argument, objected that it located power in the persons of the dominant class rather than in the structures that reproduce their dominance regardless of personnel: replace the editors, on the structural view, and the machine replaces their output too. The exchange, continued across the 1970s, is the seed of the division this article is organised around, and it matters because the two sides imply different politics and different journalism. If instrumentalism is right, the remedy for ideological news is a change of ownership and the exposure of interventions. If structuralism is right, a change of ownership that leaves the routines, the source system, and the advertising market intact will disappoint its champions within a single news cycle. Readers who want the theoretical root of the distinction can follow it into the ideology critique method article in this series, which sets out how a claim about a text is converted into a claim about the process that produced it.

A last refinement keeps the account honest. Instrumentalism works best as a theory of the boundary rather than of the content: of what a given outlet finds unpublishable, unsayable, or unassignable, rather than of the thousand daily decisions within the boundary, which are governed by routine and professional judgement. Every newsroom veteran recognises the boundary phenomenon. Certain stories are simply not proposed, certain comparisons not drawn, certain sources not telephoned, and the absence leaves no documentary trace for the historian of interventions. The boundary is where ownership bites hardest and evidence thins fastest, which is why arguments about instrumentalism so often talk past each other: the defender points at the shaped silence, the critic demands the memorandum, and the silence, being silence, produces nothing to exhibit. A structuralist will add that the boundary is set by markets and sources as much as by owners, and that addition is the doorway into the second account.

The structuralist account: routines, sourcing and the shape of the day

Structuralism starts the analysis one floor below ownership, in the working week of the ordinary newsroom, and its founding gesture is a change of question. Instead of asking whose interests a story serves, it asks what a story has to be in order to exist at all by six in the evening. News is manufactured under deadline from materials that arrive unevenly, and the sociology of news production that developed in the 1970s, in studies of newsrooms in Britain and the United States, catalogued the devices by which an unpredictable world is rendered processable. There is the diary, which fills the bulletin in advance with scheduled events: court sittings, council meetings, press conferences, anniversaries, trials. There is the agency wire, which supplies a pre selected flow of the world’s happenings already ranked by somebody else’s judgement. There are the beats, the standing rounds of institutions, police, courts, parliament, hospitals, unions, each generating a dependable ration of occurrences in a form the desk can use. And there is the format, the fixed number of pages or minutes, the hierarchy of running order, the requirement that the leading item be graspable in a sentence. None of these devices was designed to serve an ideology. All of them together determine, before any political question is posed, what sort of thing will count as the news.

The consequences for content are systematic, and they lean in a direction a Marxist recognises. Institutions generate more reportable events than crowds do, and office holders generate more than the officeless, so the diary and the beat system deliver a world pre populated by the already powerful. A minister can summon cameras by announcing an initiative; a tenants’ association must manufacture an occurrence, a stunt, a delegation, a fire, before the same cameras consider the journey. The asymmetry is not a conspiracy and requires no instruction. It is the arithmetic of subsidised information. Sociologists of the source relationship gave the mechanism its sharpest formulation in the concept of the primary definer, developed by Hall and his colleagues in Policing the Crisis, published in 1978. On their account, accredited institutional sources, police, courts, government departments, enjoy structured access to the media as the first definers of events, and journalists, working under time pressure and professional rules of impartiality that privilege attributed statement over unattributed assertion, reproduce those first definitions as the framework within which later challenge must occur. The challenger is heard, but heard as a reaction, inside a story whose terms were set elsewhere.

How do newsroom routines produce bias without orders?

Newsroom routines produce bias without orders because the diary, the agency wire, the beat system, and the deadline reward institutional sources and pre formatted events. Accredited definers speak first, challengers answer inside a frame already set, and selection repeats daily until its direction looks like the world’s own shape rather than a product of the process.

News values complete the mechanism on the side of selection. The classic study by Johan Galtung and Mari Ruge, published in 1965, asked what makes an occurrence cross the threshold into foreign news and found a logic of consonance: events are selected for negativity, for proximity, for the involvement of elite persons and elite nations, for fit with an existing narrative, and for the availability of pictures. Later newsroom research translated the list into the vernacular of the desk, where a story is strong if it is new, close, consequential, conflictual, and tellable. Notice what the criteria exclude by design rather than by malice. Processes without events, a slow decline in a region’s employment base, the gradual conversion of secure work into insecure work, generate no occurrences and therefore no news until they erupt into the event form: a closure, a riot, a suicide, a statistic released at a press conference. The structuralist conclusion follows directly. News is constitutionally better at reporting the actions of institutions and the eruptions of processes than at reporting the processes themselves, and since the actions of institutions are usually the management of an existing order, the accumulated output of an honest newsroom tilts toward the order’s own self description without a single dishonest act anywhere in the chain.

Impartiality rules, paradoxically, deepen the tilt, and this is the structuralist argument at its most counter intuitive and most persuasive. The professional norm that a contested claim must be attributed sends the journalist to the sources entitled to be quoted, which are the institutions again. The norm that both sides be heard defines the sides by the recognised parties to a dispute, which excludes positions no institution carries. The economist W. Lance Bennett later named a version of the mechanism indexing: journalists calibrate the range of voices in a story to the range of debate among officials, so that when officialdom is united the story is closed and when officialdom divides the story opens, with the press functioning as an index of elite disagreement rather than as an independent forum. A journalist can execute every one of these norms honourably and still deliver, day after day, a world in which the significant actors are governments, markets, courts, and police forces, and in which everybody else appears mainly when something has gone wrong.

Commercial format binds the whole apparatus together. A newspaper is sold twice, once to its readers and once, at a higher price, to its advertisers, and Dallas Smythe’s old argument that the audience itself is the commodity produced by media firms states the relation at its starkest. The structuralist does not need the owner to intervene, because the advertising market intervenes continuously and impersonally. Subjects that sit badly beside consumer advertising, investigations that threaten major advertisers, and treatments that depress the affluent demographics advertisers pay to reach all face a headwind that nobody ordered and everybody in the building can feel. Programme makers in commercial broadcasting describe the same pressure from the other side: the content is the wrapper that delivers an audience in a buying mood to the advertisements, and material that shatters the mood is expensive in a currency the accounts department counts precisely. Public service broadcasting, funded otherwise, escapes this particular headwind and meets others instead, the charter renewal, the licence settlement, the ministerial displeasure, which is why the structuralist expects its output to differ in degree and in vulnerability rather than in kind.

The structuralist account’s great evidentiary triumph is that its central phenomenon was first demonstrated on a public broadcaster, where the ownership mechanism was weakest. The Glasgow research examined below found patterned slant in the coverage of the British Broadcasting Corporation and the independent television news of the period, organisations with no proprietor to telephone the desk. If slant appears where command is absent, command cannot be the whole explanation, and some combination of routine, sourcing, and format must be doing the work. That is the structuralist wager in its purest experimental form, and it is why the Glasgow studies, rather than any memoir of a bullied editor, became the tradition’s most fought over evidence.

Structuralism has its own characteristic failure, and honesty requires it in the same detail. A mechanism that explains everything by process struggles to explain change, dissent, and the moments when coverage genuinely breaks ranks: the investigation that runs for months against an institution the desk depends on, the correspondent who files against the frame, the editor who leads with the process story that has no event. The account can absorb these as turbulence inside a stable system, and sometimes that absorption is correct, but a theory of reproduction that cannot specify its own failure conditions drifts toward the same unfalsifiability that discredits the intention trap, approached from the opposite side. The hegemonic account enters at exactly this seam, accepting the structuralist machinery and denying its closure.

The propaganda model: five filters and a misunderstanding

In 1988 Edward Herman and Noam Chomsky published Manufacturing Consent, and its central device, the propaganda model, became the most widely discussed Marxist adjacent account of news ever produced, as well as the most widely misdescribed. The misdescription is specific and worth correcting at the root: the model is routinely summarised, by admirers and opponents alike, as the claim that journalists conspire, or are instructed, to deceive the public on behalf of the powerful. Its authors state the opposite at length. The model proposes that a set of filters, operating through market structures and professional incentives, sifts the raw material of events before it reaches the public, so that the surviving output is systematically skewed without any central direction and without most participants experiencing themselves as propagandists. Herman and Chomsky describe the process as a guided market system, in which the guidance is exercised by the constraints under which media firms compete. In the taxonomy of this article the propaganda model is a structuralist account with an instrumentalist first filter, and reading it as anything else guarantees a sterile argument.

The first filter is ownership: the size of media firms, their concentration, the wealth of their owners, and their orientation to profit. Large firms require large capital, large capital invites interlocking interests, and the interlocks, with finance, with advertisers, with government as licenser and customer, set outer boundaries on what the firm will find congenial. The second filter is advertising, which the model treats not as a subsidy but as a licensing system. Media that depend on advertising must deliver audiences advertisers want, in environments advertisers approve, and history supplies the model’s cleanest illustration: the British radical press of the nineteenth and early twentieth centuries died less often by prosecution than by the withdrawal and refusal of advertising, while tame competitors flourished on revenue their politics never earned them at the newsstand. A working class newspaper could win readers and still lose the economic contest, because readers were never the customer who paid the larger bill.

The third filter is sourcing. Media organisations need a dependable, economical flow of material, and the institutions of government and business maintain large bureaus precisely to supply it: press offices, briefings, embargoed reports, expert spokespeople available at deadline. The subsidy is rational from both sides and ruinous for symmetry, because groups without such bureaus must buy their way into the news with occurrences, while the subsidised sources walk in through the diary. The model draws the consequence that journalists become dependent on the sources they are meant to scrutinise, and that dependence disciplines criticism more reliably than fear: a correspondent who alienates the ministry loses the flow on which her beat depends. The fourth filter is flak, the negative feedback that disciplined coverage attracts: complaints, lawsuits, letter campaigns, think tank rebuttals, and political denunciation, produced most efficiently by actors with resources. Flak raises the cost of certain lines of inquiry and lowers the cost of others, and it works even when every individual complaint fails. The fifth filter in the original formulation was anti communism, described as a national religion and control mechanism of the Cold War order: a master frame that sorted the world’s victims and villains in advance and relieved journalists of the burden of fresh judgement. After the Cold War the slot did not stay empty, and later discussion by the authors and their critics has treated the fifth filter as a variable position occupied by whatever overarching enemy image a period supplies.

What the model predicts matters more than the inventory, and the predictions are of a kind content analysis can test. The model does not predict uniform praise of the powerful. It predicts asymmetry: paired events of similar character will receive dissimilar treatment depending on their relation to dominant interests. Herman and Chomsky built their case on such pairings. Enemy atrocities and client atrocities, a murdered priest in an adversary state against murdered clergy in a client state, elections held under governments of opposed alignments, are predicted to differ in volume, placement, vocabulary, and moral temperature, and the authors presented column counts and tone comparisons in support. The paired example method is the model’s real methodological contribution, because it converts a diffuse suspicion of slant into a comparison with a control, and because it explains the model’s most misunderstood implication: the operative bias is expected to appear less in fabricated stories than in selection, emphasis, and omission, in the story never assigned and the victim never named. Fabrication is risky and rare in professional media. Selection is free, invisible, and total, and a model of the media that hunts for lies will miss the mechanism that matters.

The model’s treatment of journalists deserves emphasis, since it is here that the conspiracy misreading does most damage. On the model’s own account, most journalists are sincere, many are sceptical of power, and the filters work upstream of their sincerity by setting the premises within which assignments are given and copy is judged. Chomsky’s repeated formulation is that the system selects for conformity the way an educational system does: people who cannot internalise the premises rarely rise, and those who rise rarely experience their premises as premises. The claim is about careers and institutions, not about hidden loyalties, and it is either true or false as a piece of sociology, testable in principle against the career patterns of journalists who broke ranks. That some did break ranks, and paid for it, while others broke ranks and were celebrated, is data the model’s defenders and critics both cite, which suggests the honest verdict is probabilistic rather than mechanical: the filters raise the price of some stories and lower the price of others, and prices influence output without dictating it.

Criticism of the model, at its strongest, attacks three joints. The first is functionalism: the model can read like a machine whose output, service to power, is guaranteed by its design, so that every apparent counter example, the hostile investigation, the exposed scandal, is reclassified as a safety valve or a quarrel among elites, and the theory protects itself exactly as the intention trap does. Defenders reply that the model’s predictions are comparative and were stated in advance of the counts, which is true of the paired studies, though less true of the model’s looser journalistic afterlife. The second joint is the audience, which appears in the model mainly as the object of manufacture, a silence the hegemonic account fills. The third joint is change: a model assembled from the political economy of the late twentieth century must be reassembled when the advertising market collapses, the audience fragments, and distribution passes to platforms, a reassembly taken up later in this article. None of the three criticisms restores the conspiracy reading. They locate the model’s genuine limits, in closure, in reception, and in period, and a reader who can state them owns the model rather than renting it. The commercial mechanism at the model’s centre, advertising as the purchaser of audiences, is treated from the consumer’s side in the companion analysis of advertising consumption ideology, and the two articles’ claims stand or fall together.

The Glasgow studies: measurement, and the fight over the measurements

Theory in this field often floats free of evidence, and the Glasgow Media Group is the standing exception, which is why its work attracted a fight of unusual ferocity. The group, based at the University of Glasgow and led in its best known phase by Greg Philo and John Eldridge among others, set out in the 1970s to study British television news by the laborious method of recording bulletins, transcribing them, and analysing the language and structure of whole sequences of coverage. Its first volume, Bad News, appeared in 1976, followed by More Bad News in 1980 and Really Bad News in 1982. The chosen territory was industrial relations, the subject on which the claim of television bias was most often traded as anecdote, and the group’s ambition was to replace anecdote with a corpus: every bulletin in a defined period, examined for whose definitions led, whose vocabulary was adopted, and which explanations were present or absent.

The findings, stated at their most careful, concerned framing rather than falsehood. Coverage of strikes, the group argued, organised the event around disruption: the effects on the public, the inconvenience, the spectacle of picket lines, the language of industrial anarchy and held to ransom. The causes of disputes, wages eroded by inflation, staffing, safety, the history of a negotiation, appeared rarely and late, and the strikers’ own case was typically rendered in the vocabulary of their opponents or summarised rather than spoken. Management and government sources appeared as the bearers of explanation and of proposals, the reasonable parties responding to a problem, while unions appeared as the problem to which response was made. Even the apparently neutral grammar of the bulletin did work: a dispute was something that happened to production and to the public, and its origins in decisions taken in boardrooms and ministries receded into an unexamined background. The group was explicit that it was not alleging fabrication, and often not alleging individual bias at all. Its claim was structural in the sense this article has defined: that the professional routines of balance, actuality footage, and official sourcing, applied conscientiously, produced a partial world.

What did the Glasgow researchers actually measure?

The Glasgow researchers recorded and transcribed television bulletins across defined periods and measured framing: which sources defined disputes, whose vocabulary reports adopted, whether causes were explained, and how strikers and managers were positioned grammatically. Their findings concerned selection and framing in industrial coverage, not invented facts, which is why the subsequent dispute turned on method.

The rebuttal arrived in the middle of the following decade and was as systematic as the original work. Martin Harrison’s study TV News: Whose Bias?, published in 1985, re examined the group’s methods and challenged them at the level of the sample, the coding, and the inference. Harrison argued that the periods chosen were unrepresentative, that industrial coverage in the corpus was selected and summarised in ways that flattered the thesis, and that some celebrated claims, about the treatment of particular disputes and the assumptions attributed to industrial correspondents, did not survive contact with the full record. Underlying the technical objections stood a conceptual one, pressed by several critics: to demonstrate bias one needs a defensible account of the unbiased version, the true shape of the dispute against which the broadcast version deviates, and the Glasgow group’s benchmark sometimes seemed to be supplied by its own politics. The exchange hardened into one of the defining methodological quarrels of British media sociology, and the broadcasters, stung, commissioned and cited research of their own.

A fair balance sheet, four decades of scholarship later, credits both sides with permanent points. The group’s demonstration that slant can be a property of structure rather than of statement, measurable across a corpus and reproducible by other coders in its broad direction, moved the whole field: framing analysis as practised across media studies descends from this work, and the concepts of primary definition and agenda setting absorbed its influence even where the group’s name went unmentioned. Harrison’s insistence that sampling decisions and the counterfactual benchmark be defended in the open also moved the field, and later content analysis is more careful about both because he was unforgiving about each. What the exchange did not produce is a settled number for the bias of a bulletin, and the mature lesson is that it cannot. Bias of the structural kind is a claim about patterned selection relative to an arguable alternative, and its proof will always be partly an argument about the alternative. The quarrel is therefore best read as the tradition’s credentials: the Marxist account of news, in its empirical branch, generated hypotheses definite enough to be attacked on method, survived in revised form, and left its attackers’ standards inside its own practice. Few theories of media can show a comparable scar tissue.

Two cautions belong to any honest use of the Glasgow inheritance. The first is that the studies belong to a media order that has partly vanished: a small number of channels, mass simultaneous audiences, and industrial conflict as the central domestic story. The mechanisms the group identified, primary definition, event centred coverage, the grammar of disruption, are portable; the magnitudes are not, and citing the findings as a description of a fragmented, platform distributed news ecology is an anachronism the group itself would be the first to flag. The second caution runs the other way. The studies are sometimes treated, including by sympathetic lecturers, as having proved the whole Marxist theory of media, when what they examined was the output of two British broadcasters on one subject in one decade. Generality has to be earned case by case, and the accounts separated in this article make different predictions for other subjects: defence coverage, where official sourcing is tighter still; health coverage, where professional definers compete; celebrity and sport, where the commercial format rules almost alone. The theory that survives is the one that tells the investigator in advance how the pattern should vary, and why.

The hegemonic account: Hall, encoding and the contested field

The third account changes the subject in a way its rivals initially resisted. Instrumentalism and structuralism both study production: owners, routines, filters, outputs. The hegemonic account, built by Stuart Hall and the Birmingham school out of Gramsci’s theory of hegemony, insists that an output is not yet an effect, and that between the bulletin and the belief stands an audience doing work. Hall’s essay on encoding and decoding, drafted in 1973 and published in its canonical form in 1980, proposed that media messages are encoded within frameworks of meaning that prefer a particular reading, the dominant or preferred reading, but that decoding is a distinct moment with its own conditions. Audiences occupying different social positions may take the preferred reading whole, may negotiate it, accepting the general frame while exempting their own situation, or may decode oppositionally, understanding exactly what the message intends and reversing its valuations entire. The full reconstruction of the parent theory is given in the gramsci hegemony complete guide, and only its media consequences need detain this article.

Three consequences follow, and each corrects a habit of the other accounts. The first is that ideology in news is a claim about an offer, not about an injection. Coverage proposes a definition of the situation, stocks it with heroes and problems, and invites identification, and the proposal can fail, as every campaign that bemused its target audience demonstrates. The second is that failure and success are patterned rather than random. Hall’s hypothesis, pursued by David Morley in audience research with the Nationwide programme at the end of the 1970s, was that decoding positions correlate, imperfectly but detectably, with social position: groups whose experience gives them an alternative framework ready made, trade unionists watching industrial coverage, communities with their own institutions and memories, decode against the grain more readily than groups for whom the preferred reading matches the grain of daily life. Morley’s findings were messier than the hypothesis, decodings varied within groups and some readings crossed class lines entirely, and the messiness is itself part of the account’s credibility: hegemony predicts contest with a home advantage for the dominant frame, not a league table of immune and infected classes. The third consequence is that framings must be maintained. Because consent can be withdrawn, coverage of a settled question must be renewed, concessions must be absorbed and displayed, and genuine grievances must be given partial voice if the frame is to stay plausible. The presence of critics inside the bulletin, so often cited against the ideological character of news, is on this account one of the mechanisms by which a hegemonic frame stays alive: the objection is heard, answered, and folded into a story whose terms survive the encounter.

How does encoding and decoding change the audience question?

Encoding and decoding separates the meaning built into a message from the meaning an audience makes of it. Producers encode a preferred reading inside a dominant frame; audiences may accept that reading, negotiate it against their own experience, or reverse it entirely, so effects become an empirical question rather than an assumption built into the text.

Moral panic research supplied the hegemonic account with its most durable case study of framing in motion. Stanley Cohen’s Folk Devils and Moral Panics, published in 1972, examined how British seaside disturbances between the mods and rockers were amplified by press treatment into a national symbol of youth disorder, with each stage of reaction, police, courts, editorial, feeding the next. Hall and his colleagues then took the apparatus into harder territory in Policing the Crisis, arguing that the panic over mugging in the 1970s did ideological work for a state moving toward a more authoritarian settlement: a street crime, redefined through police and court definitions relayed by the press, became the lens through which a general crisis of order could be seen and a law and order programme consented to. Critics have disputed the displacement thesis and the readings of motive, and the dispute is healthy, but the analytical template, primary definers set terms, media amplify through their values and formats, public anxiety reorganises politics, survives inside journalism studies whatever verdict is reached on the British case.

Notice how the hegemonic account repositions the journalist. In the instrumentalist story she is an instrument; in the structuralist story she is a function of her routines; in Hall’s story she is a professional working inside a framework of taken for granted definitions whose power lies in their obviousness, and who may herself negotiate or oppose in the margins of the copy. The account also repositions the owner, who becomes the guarantor of a field rather than the author of a line, and the audience, which stops being the theory’s embarrassment. Where the propaganda model’s readers complained that real audiences laugh at the news, heckle it, and believe it selectively, the hegemonic account has an answer that is not a patch: selective belief, laughter included, is what the theory predicts, since hegemony in Gramsci’s original sense never meant unanimity. It meant leadership, the organisation of consent across a population that retains its own experience and can, under changed conditions, transfer its allegiance. A media order that has to persuade is different from one that merely instructs, and confusing the two leads critics to declare ideology dead every time an audience talks back.

The account’s weakness is the mirror of its strength. Because it expects contest, negotiation, and partial failure everywhere, it can be hard to imagine the evidence that would embarrass it, and unsympathetic readers have asked whether hegemony names a finding or merely a mood of sophisticated ambivalence. The fair reply is that the account does make risky claims: that preferred readings cluster measurably in texts, that decoding positions correlate with social resources for counter definition, and that framings move under sustained public pressure rather than holding forever. Audience research programmes have tested versions of all three with mixed and therefore informative results. A theory that can be partly wrong in three directions is in better shape than a mood, even if it will never produce the clean paired counts the propaganda model trades in.

Audiences: scepticism, selection and the limits of effects

Every account of media ideology must eventually face the person in the armchair, and the empirical record on that person is inconvenient for all simple versions of the theory. Audience research across several decades finds a public that is neither the empty vessel of crude propaganda fears nor the sovereign consumer of advertising copy. People select outlets congenial to views they already hold, which limits conversion and complicates any claim that coverage manufactures opinion from nothing. They discount sources they distrust, sometimes so thoroughly that a true story fails to move them because of its letterhead. They discuss, joke, and recruit the news into arguments of their own, and their recollection of a bulletin a week later is a reconstruction coloured by everything that has happened since. Effects research in its careful form finds influence concentrated in agenda and framing rather than in verdict: coverage is better at deciding what audiences think about, and in which vocabulary, than at deciding what they finally conclude. That distribution of effects is exactly what the structuralist and hegemonic accounts, properly stated, predict, since both locate ideology in selection and definition rather than in persuasion to order. It is the crude instrumentalist picture, of a public written upon by owners, that the audience evidence embarrasses most.

Selection by audiences also feeds back into production, and the feedback loop is a structural mechanism in its own right. Editors read ratings and sales figures as verdicts, and a framing that empties the audience is revised while a framing that grows it is repeated, so the convictions of the public, including its prejudices and its appetite for reassurance, are inscribed into output by commercial discipline. Marxist analysis has sometimes treated this loop as a refutation, as though audience demand dissolved the question of power. It does not. It relocates part of the mechanism. A market that rewards the confirmation of existing common sense will underproduce challenges to that common sense even when no proprietor cares, and the audience’s own preferences were themselves formed inside a culture the theory is trying to explain. The loop makes the system harder to blame on anybody, which is precisely the structuralist point restated at the level of consumption.

One distinction keeps this section honest. Scepticism is unevenly distributed across subjects. The same viewer who discounts a foreign ministry’s claims about a distant war may absorb, unexamined, the framing of a dispute in his own town, because distant scepticism costs nothing and local frames are load bearing for daily life. Ideology, on the account being built here, does its deepest work not where audiences are most gullible but where the frame is most useful, sparing its holders the effort of rethinking arrangements they must live inside tomorrow morning. That is why the tradition keeps returning to the categories of ordinary economic life, wages, prices, strikes, rents, taxes, as its preferred evidence: those are the subjects on which a supplied definition saves the most labour and is therefore least often audited.

The state, the law and the boundaries of the sayable

A Marxist account of news that confined itself to owners and markets would omit the actor that sets the outer walls of the field: the state, working through law. Libel law disciplines investigation by price, since the cost of defending a true story can exceed the budget of the desk that published it, and the chilling effect is distributed by wealth in both directions, protecting rich subjects from scrutiny and exposing poor publishers to ruin. Official secrecy legislation and the informal systems around it, the lobby briefing, the unattributable source, the prosecution of leakers, govern the flow of the state’s own information and reward the correspondents who stay inside the arrangements with the access the structuralist account describes. Licensing and charter systems make broadcasters periodically answerable to governments for their survival, a structural fact about their position that requires no telephone call in any particular week to shape a decade of caution. Spectrum allocation, press regulation, tax treatment of advertising, and the rules of political broadcasting all stand upstream of content, and none of them appears in a content analysis.

The law’s role is genuinely double edged, and the article’s neutrality discipline requires the other edge in full. Legal frameworks also protect the press against the state itself: shield provisions, freedom of information statutes, constitutional guarantees, and courts that have repeatedly ruled for publication against governments are not decorations on the system but working parts of it, and the history of twentieth century journalism includes courts defending newspapers against the very executives those newspapers were embarrassing. A structuralist analysis can absorb this without strain, since its claim is about patterned pressures rather than total control, and a legal order that simultaneously protects publication and prices investigation is a pressure system with counterweights built in. What the Marxist account adds is the observation that the counterweights are themselves unevenly available: the constitutional guarantee belongs to every citizen, while the libel defence fund, the specialist counsel, and the insurer belong to organisations, and the freedom to publish means most in the hands of those who can afford to be sued. Formal equality of rights over a material inequality of capacity is, the tradition notes, a pattern its theory was built to notice, in media as in every other market.

The pluralist objection, stated at full strength

The counter case deserves its strongest form, not the summary usually fed to it. It begins with competition. News markets in open societies contain multitudes, and among them, at almost any moment, are outlets whose editorial identity is hostility to concentrations of wealth and power: investigative units that have exposed corporate fraud and ministerial corruption, satirical publications with mass circulations, foreign broadcasters beyond the reach of domestic owners, and a long tradition of dissenting journalism that the Marxist account itself relies on for much of its evidence. A system manufacturing consent with any efficiency would presumably suppress the tradition its critics cite, yet the critics’ footnotes are full of the system’s own products. The objection continues with professional norms. Journalism maintains an occupational ideology of its own, verification, attribution, the separation of news and comment, the correction, and the norm is not a fiction: reporters have gone to prison rather than reveal sources, editors have published stories that destroyed their own commercial relationships, and the profession’s internal honours reward precisely the exposure of the powerful. Norms that cost their holders money and liberty are not plausibly described as a mask for the interests they damage.

The objection then turns to the audience and to the owner together. Audiences in a plural market punish perceived servility and reward perceived independence, so that credibility functions as a commercial asset a proprietor squanders at his peril; the history of newspapers that became mere bulletins for their owners is a history of declining circulations, which means the market supplies its own corrective to instrumental control. And proprietors themselves keep failing to behave as the simple theory requires. Owners have bankrolled loss making titles for prestige, for political adventure, for the pleasure of the game, and have let editors savage policies the owner’s other businesses were lobbying to obtain, because an empire of media is also a status economy whose coin is influence and whose influence depends on being believed. Add the fragmentation of the object, channels multiplying, then unbundling into feeds, and the objection concludes that whatever unified apparatus the theory described has dissolved into a noisy bazaar in which power certainly speaks, but so does everybody else, and louder every decade.

Why do proprietors sometimes publish against their own interests?

Proprietors sometimes publish against their interests because credibility is a commercial asset, because prestige and influence reward visible independence, because editors and journalists defend professional norms at real cost, and because a portfolio of titles serves different audiences. Such cases refute command theories of control while leaving structural accounts of selection largely untouched.

The reply must begin with concessions, and they are not small. Against the instrumentalist account in its crude form, the objection succeeds. Owners do publish against interest, competition does produce hostile outlets, professional norms do bite, and any Marxist media theory that cannot say so plainly has confused itself with the propaganda it studies. The objection also succeeds against every version of the theory that predicts uniformity. Coverage is not uniform, dissent is not token in any simple sense, and the fragmentation of the audience is a real transformation that later sections of this article take seriously rather than explaining away. A defence that begins anywhere else will not be believed, and should not be.

What survives the objection is the claim the empirical tradition actually made, which concerns asymmetry rather than uniformity. The question is not whether critical coverage exists but how the costs and rewards are distributed across kinds of criticism. Exposing a minister’s affair and exposing the structure of an industry are both called investigative journalism, yet they differ in their legal exposure, their advertising consequences, their source costs, and their career rewards, and a system can celebrate the first while taxing the second without any contradiction or conspiracy. Similarly, the bazaar of voices is real, and the question the structuralist asks about it is distributive: which voices arrive with bureaus, budgets, legal departments, and standing invitations, and which arrive as occurrences that must fight for a hearing each time. Pluralism of utterance is compatible with steep inequality of amplification, and the inequality, being measurable in access, airtime, and the order of definition, is where the Marxist claim now lives. The verdict on the objection is therefore divided, as the evidence is. It dismantles the theory of the proprietor’s telephone. It leaves standing, and arguably sharpens, the theory of the tilted field, in which nobody needs to telephone because the tilt is built into who can speak first, loudest, and at least personal risk.

The categories of economic reporting: the economy as weather

The tradition’s most instructive evidence is often hiding in the business pages, where ideology appears not as slant on a controversy but as the furniture of an uncontroversial genre. Consider the standard grammar of market reporting. Indices rise and fall with the agency of weather; markets are nervous, confident, relieved, or punishing, endowed with moods to which governments and populations must adjust. The sentence structure does quiet work. When the subject of the sentence is the market and the object is everybody else, a particular theory of the economy has been installed without being argued: that the judgements of asset holders constitute a force of nature to which policy responds, rather than a set of decisions by identifiable institutions that policy might address, tax, or overrule. Reporters who write the grammar are not deceiving anybody. They are writing for an audience of investors and managers whose standpoint the genre adopts as the standpoint of the reader, and the business section’s reader is imagined holding a portfolio even when the newspaper’s actual reader holds a bus pass. Genre, not instruction, fixes the point of view, and the point of view is the ideology.

The same grammar organises the vocabulary of industrial life. One side of industry has leaders, chiefs, and bosses whose decisions are news; the other side has a workforce that appears in the aggregate, as a cost, a claim, or a problem of flexibility. Wage increases are reported through their effects on inflation and competitiveness, price increases through their effects on margins and strategy, and the asymmetry is a matter of which consequences the desk treats as the story’s natural sequel. None of this requires a proprietor with opinions about wages. It requires a beat structure in which the company’s results are released at a press conference with documents and the workforce’s experience is not, a source system in which the finance director is quotable by name and the shop steward is a reaction shot, and a format in which the economy must be narrated daily whether or not anything has happened to it. The structuralist account was built for exactly this stratum of evidence, the level at which reporting can be simultaneously accurate in every particular and ideological in its total shape.

Against the whole genre stands one stubborn complication that a fair analysis must register. Business journalism also produces the most relentless adversarial coverage of individual companies anywhere in the press, because its readers pay for warning. Fraud is exposed, chief executives are hounded from office, and the genre’s heroes include reporters who ruined the reputations of firms their newspapers courted for advertising. The Marxist reply distinguishes levels again: scrutiny of firms is the genre’s function and is compatible with reverence for the framework, in the way that a racing page can savage every jockey while never questioning the race. Whether that distinction persuades is a judgement the reader should make with both facts in view, the genuine ferocity toward companies and the near total absence, in the same pages, of the standpoint from which a wage is not a cost but a living.

Foreign news and war: where the three accounts converge

Foreign reporting is the territory where the three accounts come closest to agreement, because the production conditions are at their most extreme. The foreign desk depends on a handful of agencies, on official briefings, on embassy and ministry sources, and on the logistics of access granted by the governments being reported on. Costs confine permanent bureaus to a shrinking list of capitals, so vast regions enter the news only through the event forms the structuralist catalogue predicts: the coup, the disaster, the atrocity, the summit. Galtung and Ruge’s original study was of foreign news, and its findings about elite nations and negative events describe a map of the world in which some countries are protagonists and others are scenery that occasionally catches fire. No proprietor needs to draw that map. It is drawn by the cost of the bureau, the availability of the feed, and the audience’s assumed geography of concern.

War tightens every filter at once. Access becomes a military gift, accreditation a condition of movement, and the primary definers wear uniform. The indexing mechanism reaches its purest form: coverage opens to dissent roughly when officialdom at home divides, and closes when it unites, regardless of the merits visible on the ground. The propaganda model’s paired comparisons were drawn largely from this territory, from the differential treatment of atrocities according to the alignment of the perpetrator, and whatever verdict is reached on the model’s counts, the phenomenon it named, the sorting of victims into the worthy and the unworthy by their propaganda value to the home state, remains the single most replicated observation in the critical literature on war coverage. Instrumentalism contributes least here, since owners rarely need to intervene in a system whose patriotism is supplied by sourcing and access, and hegemony contributes the necessary reminder that war frames have repeatedly collapsed under the pressure of events and images, that consent for wars is won late and lost early, and that the field, even at its most managed, is a field.

From gatekeepers to algorithms: the platform reordering

The media order the classic accounts described has been structurally rebuilt, and an honest application of the theory must rebuild with it. The advertising model that subsidised mass journalism for a century collapsed as advertising migrated to platforms that sell attention with a precision newspapers could not match, and classified and recruitment advertising, once the quiet foundation of local papers, moved away almost entire. News organisations became dependent twice over: on platforms for distribution, since a large share of their readers now arrive through feeds and search rather than through any front page, and on the platforms’ rules, which are revised without consultation and enforced without appeal. The editor’s gatekeeping function, the selection of the front page and the running order, did not disappear. It was joined, and in reach surpassed, by a second selection system operating on different principles: the ranking algorithm, which orders the world for each user by predicted engagement.

What replaced the gatekeeper when distribution moved to platforms?

The gatekeeper was not abolished but joined by a ranking system that selects for each user by predicted engagement rather than by editorial judgement. Editors still commission and frame stories, while algorithms decide their reach, so selection now operates twice, once in the newsroom and once in the feed, under different incentives.

The structuralist account survives the transition with its mechanism translated rather than refuted, and the translation is worth doing carefully. The old structuralism said: follow the routines, the sources, and the money, and the slant will explain itself. The money has moved. A newsroom optimising for platform distribution watches referral data the way it once watched the proprietor’s moods, and the metrics reward the emotional, the conflictual, the identity confirming, and the instantly graspable with a discipline no editor needs to impose, because the story that fails the feed simply fails to exist for most of its potential audience. Outrage and reassurance, the two cheapest engagements, acquire a structural advantage that resembles the old advantage of the official source: both are subsidies, one of information and one of attention, and both tilt output without instructing anybody. The primary definer system persists inside the new machinery, since official accounts and established brands carry verification, follower mass, and the archive of engagement that ranking systems read as authority. What has genuinely weakened is the shared agenda. The mass bulletin addressed a public that had, in some measure, seen the same world; the feed addresses segments whose worlds overlap only at the largest events, and hegemony, which presupposed a common field in which consent could be organised, must now explain how leadership works across fields that never meet.

Instrumentalism, meanwhile, has enjoyed an unexpected revival at the platform level, and the revival should be recorded without gloating. The platforms themselves are owned, their ranking systems are designed by identifiable firms pursuing commercial and sometimes openly political purposes, and the opacity of the algorithm gives the old instrumentalist question a new address: not who telephoned the editor, but who set the objective function, and in whose interest the engagement metric was chosen over the available alternatives, civic information, diversity of exposure, or declared user preference. The question is a good one and the structuralist has no reason to fear it, because the answers, when they come, describe design decisions whose effects then run structurally, at scale, without further intention. The two accounts meet in the algorithm as they once met in the appointment of an editor: a decision at the top whose consequences are produced by a machine that does not need the decider in the room.

The filter bubble thesis is the point at which this article’s discipline about evidence matters most, because the thesis is the rare media claim that passed into common speech before it passed through the literature. In the form popularised by Eli Pariser’s book of 2011, the thesis holds that personalised algorithms seal each user inside a unique information universe, invisible, involuntary, and increasingly sealed against dissonant fact. The Marxist temptation is to adopt the thesis wholesale, as the structuralist mechanism perfected: selection without an editor, ideology without an ideologue. The empirical literature counsels restraint. Study after study of actual browsing and exposure has found the picture weaker and stranger than the metaphor. Users of search and social media typically encounter a wider range of sources than the metaphor suggests, personalisation explains less of the variance in exposure than users’ own choices do, and the audiences most sealed against opposing views are often heavy partisans whose sealing is self administered and predates the algorithm. Researchers such as Axel Bruns have argued that the filter bubble functions in public debate as a displacement, blaming a technology for polarisation whose roots lie in politics and in the audiences’ own motivated reasoning. The honest summary is directional rather than numerical: algorithmic curation does narrow some pathways and widen others, its effects are heterogeneous across users and platforms, and the strong thesis of sealed, invisible universes is not an established finding. Including that correction is not a concession wrenched from this article. It is the article’s method applied to its own side, since a theory of ideology that cannot grade the evidence for a congenial claim will grade nothing.

Selection rather than invention: the anatomy of an omission

Ask where ideology is most powerful in news and the tradition’s mature answer is unfashionable in its modesty: in the story that does not appear. Fabrication is rare, because fabrication is detectable, punishable, and professionally shameful. Emphasis is common but contestable, since the reader can see the emphasis and discount it. Omission is neither rare nor visible. The unassigned story leaves no text to criticise, no correction to demand, no byline to blame, and its absence is indistinguishable, from the reader’s chair, from the absence of the event itself. This is why the propaganda model insists that its subject is the boundaries of the expressible rather than the content of the expressed, and why the Glasgow group’s most telling findings concerned the explanations that were not offered and the questions that were not put. An analysis of news ideology that confines itself to the printed page has accepted the field’s own first selection, the decision about what there would be a page about.

Omission has mechanics, and they are worth naming because naming them converts a suspicion into a research programme. Stories die at identifiable gates: the pitch that sounds like a campaign rather than an event, the subject whose sources are all expensive and whose documents are all contested, the investigation whose legal risk is priced before its public value is weighed, the follow up that no desk owns because the original story belonged to a rival. They also die of the calendar, crowded out by the scheduled events the diary guarantees will fill the bulletin, so that the unreproducible process story is always the first casualty of a busy day. Each death is defensible in the language of the desk, and the cumulative pattern is legible only from above the desk, across months of front pages compared against an independent inventory of what the period contained. The methodological moral generalises beyond this article’s subject. Wherever an institution’s output is a selection from a larger possible output, the ideology of the institution lives in the selection function, and the selection function is studied by reconstructing what was available and refused, not only by annotating what was chosen.

Can a structural account be falsified?

The standing objection to everything argued so far must now be faced in its precise form: that structural accounts of media ideology are unfalsifiable, since every outcome can be absorbed, conformity as proof of the mechanism and dissent as proof of its subtlety. The objection is the mirror image of the intention trap, and it would be fatal if the tradition had never risked a prediction. The record shows otherwise. The Glasgow group published coding schemes and counts that a hostile researcher could and did re examine, and lost some of the ensuing arguments, which is what falsifiability looks like in practice. The propaganda model’s paired comparisons state in advance which cases must differ and by what signs; a literature of counter counting exists and has damaged particular claims while leaving the method standing. Hall’s decoding hypothesis generated audience studies whose results partly refused the hypothesis, and the refusal is cited inside the tradition rather than buried by it. The falsifiable core shared by the three accounts is comparative: slant must covary with the variable each account names, ownership for one, production routine for another, audience resources for the third, and where covariance fails, the account shrinks to fit its surviving territory. That the broader worldview survives its local defeats is true of every research tradition in the social sciences, and is a reason for vigilance rather than a disqualification.

A practical test battery follows for any specific bias claim, and students of the subject are advised to run claims through it before believing their own. First, is the claim about content, selection, or reception, and does the evidence offered match the level of the claim? Second, is there a paired comparison, a like case treated unlike, or only a single case annotated with disapproval? Third, does the claimed mechanism generate the pattern more cheaply than the innocent explanations, deadline, scarcity, professional habit, audience taste, that compete with it? Fourth, what observation would the claimant accept as damaging, and has it been looked for? A claim that passes the battery may still be wrong, but it is wrong in the way of an empirical claim, exposed to the record. A claim that fails the battery is not false either; it is unfinished, a suspicion wearing the uniform of a conclusion, and most public argument about media bias, on every side, is unfinished in exactly this way.

A case in full: the miners, the camera and the order of events

The British miners’ strike of 1984 and 1985 remains the most intensively studied episode in the critical literature on British news, and it earns the position because every mechanism in this article can be watched operating inside it at once. Begin with the structural facts. The dispute was covered from a diary dominated by official sources: the government, the coal board, and the police spoke through daily briefings and statements, while the union’s case reached the bulletin through the narrower channel of the press conference and the reaction quote. The event form favoured confrontation, so the strike entered most homes as a sequence of picket line clashes, arrests, and returns to work, counted and dramatised, while the underlying process, the economics of the coal industry and the future of the communities built on it, appeared mainly as background paragraphs supplied by specialists. The vocabulary of definition was set early and from above: the strike was framed as a challenge to law and to the electorate’s verdict, its leader cast as the central character in a drama of personality, and the striking miners appeared in coverage far more often as participants in disorder than as speakers explaining a case.

Inside that structure sits one small, documented scandal that students of the period meet in every textbook, the coverage of the confrontation at the Orgreave coking plant in June 1984. Television news showed sequences in which mounted police charged after being pelted by miners, and the editing placed the stone throwing before the charge in a way that reversed the order of the confrontation as many present recorded it; the broadcaster later acknowledged that the transmitted sequence had events in the wrong order, while maintaining the error was made in the pressure of the edit rather than by design. The episode is a gift to careful teaching precisely because it refuses every simple moral. For the instrumentalist it looks like a smoking gun until the explanation, haste in a chaotic edit suite, is examined, at which point it becomes something more disturbing for all three accounts to contemplate: a reversal that nobody ordered and that everybody in the chain allowed to pass, because it matched the story the day had already told them. The frame did not need to be imposed on the footage. The footage was intelligible, to professional eyes working at speed, only inside the frame.

Set beside Orgreave the strike’s quieter asymmetries and the hegemonic dimension completes the picture. Working miners were individuated, named, filmed at their gates with their families, granted the full grammar of personal testimony; striking miners were collectivised, numbered, and filmed from behind police lines. The distinction is not fabrication and it is not instruction. It is the accumulated output of access, sympathy gradients in the source system, and the narrative need for a protagonist whose return to work can carry the story’s resolution. And the frame, dominant as it was, never achieved closure. The strike retained substantial public support for a year against the grain of the coverage, sustained by union and community institutions that supplied an alternative definition with its own media, its own meetings, its own memory, which is the hegemonic account’s central claim exhibited in a single case: consent was contested throughout, on unequal ground, and the inequality of the ground is the part of the sentence the word contested must never be allowed to hide.

Objectivity: the commercial childhood of a professional norm

No concept in journalism carries more moral weight than objectivity, and the Marxist literature’s most disquieting contribution to the subject is a piece of history: the norm was born as a business strategy before it was canonised as an ethics. The wire agencies of the nineteenth century sold the same report to newspapers of opposed parties, and a report usable by all buyers had to be stripped of the partisan colouring that the openly political press of the period wore as a badge. Inverted pyramid writing, attribution, and the separation of fact from comment were technologies of that market, allowing one commodity to serve many masters. Mass advertising completed the education, since a newspaper financed by selling a broad audience to advertisers had commercial reasons to offend no large segment of it. None of this history makes objectivity a fraud. Norms outgrow their origins, and a discipline learned for commercial reasons can be practised for honourable ones, defended at real cost, and turned against the interests that first rewarded it. But the history does dissolve the innocence of the norm’s self description, in which objectivity appears as the natural ethics of a free profession rather than as a form with a genealogy inside a market. The structuralist reads the genealogy as the argument in miniature: the form of the news was set by the economics of its sale long before any individual journalist chose a value, and the values then grew, sincerely, inside the form.

The tradition began in a newsroom

Marxist media criticism did not begin as an academic theory applied to journalism from outside. It began as journalism, under censorship, and the biographical root explains some of the tradition’s permanent concerns. Marx’s first published work was a defence of press freedom written in 1842 for the Rheinische Zeitung, against the Prussian censorship, and its argument already contains the seed of the structural account: that a censored press and a press deformed by its conditions produce a distinctive falseness, not merely by suppressing truths but by shaping the character of what can be written at all, training writers in euphemism and readers in suspicion. The newspaper he edited was suppressed within the year. He spent much of the following decade as a working correspondent, most durably for the New-York Daily Tribune, filing on deadlines about wars, budgets, and colonial policy for a bourgeois newspaper whose editors cut, delayed, and occasionally rewrote his copy, an experience of the commercial news process from the contributor’s desk that no theorist of the press should be without. Engels wrote on military affairs for papers he despised. The tradition’s founders knew the diary, the agency report, and the proprietor’s pencil as working conditions before they knew them as categories.

The nineteenth century also supplied the tradition with its great material lesson in the political economy of the press, the British campaign against the taxes on knowledge. Newspaper stamp duties, paper duties, and advertisement duties priced the radical press toward extinction for decades, and their repeal across the middle of the century was understood by contemporaries of every party as a constitutional event, because it decided who could afford to address the working class in print. What followed was the education the propaganda model later formalised. Freed from the stamp, the cheap press flourished, and then the economics of advertising slowly selected among the flourishing: titles dependent on working class readers alone found their advertising revenue thin, titles that delivered affluent readers prospered regardless of their politics, and the radical press of mass circulation withered inside a formally free market. No censor closed those newspapers. Their readers did not desert them in a body. They were unbuilt by a revenue structure, and the distinction between being suppressed and being unbuilt is the distinction between the censor’s office and the structuralist account, stated in the industry’s own history a century before the account had a name.

The audience commodity in the age of the feed

Smythe’s claim that media sell audiences to advertisers was, for decades, the tradition’s most elegant piece of political economy and its most contested, since it seemed to demote journalism to packaging. The platform economy has performed the experiment that settles much of the argument. The large platforms are, by revenue, audience selling machines of a purity Smythe’s broadcasters never achieved: their product is segmented attention, priced by prediction, and their content is supplied substantially by others. News organisations inside that economy experience the audience commodity from the supplier’s side. Their journalism is raw material for an attention market whose prices they do not set, and the metrics returned to the newsroom, the click, the dwell, the share, are the voice of the customer speaking in the only language the supplier is equipped to hear. A structuralist analysis of news ideology in this order therefore studies two selection systems in series: the newsroom’s, still governed by sources and routines, and the platform’s, governed by predicted engagement, with the second system’s preferences flowing backward into the first through the analytics dashboard on the news editor’s screen. The ideological consequences most worth investigating follow from the backward flow: the headline written for the share rather than the reader, the story’s first paragraph rebuilt as a hook, the slow demotion of subjects whose audiences are loyal but small and unprofitable to advertisers. The mechanisms are new in their instrumentation. In their logic they are the oldest thing in this article, the shape of the commodity deciding, in advance and without malice, what the news will mostly be.

Who becomes a journalist, and why the theory cares

One mechanism sits upstream of every other in this article and is the easiest to document: recruitment. Journalism in its prestigious reaches draws disproportionately from the educated metropolitan middle class, a pattern produced by mechanisms as unglamorous as unpaid internships, which filter for family subsidy, by the concentration of media employment in expensive capital cities, and by credential requirements that convert a once open trade into a graduate profession. Surveys of the profession in Britain and the United States have repeatedly found journalists markedly more educated, more metropolitan, and less likely to come from manual working class homes than the populations they report on, and the pattern has intensified as the industry’s economics have raised the price of entry. No conspiracy assembles this workforce. A sequence of individually rational filters does, and the structuralist files the result beside the source system and the diary: the people least likely to find the dominant definitions strange are the people most likely to be holding the notebook. The point must be stated with care, because it is about standpoint and not about virtue. Working class entrants to journalism are fully capable of reproducing the dominant frame, and privileged entrants are fully capable of breaking it, as the profession’s own heroes and villains demonstrate in both directions. Recruitment tilts the probabilities, supplying the newsroom with a common sense whose limits feel like the limits of the reasonable, and tilting probabilities is all this article has ever claimed the deep mechanisms do.

The metropolitan concentration adds a geography to the sociology. A national press produced overwhelmingly in one capital city will systematically experience the rest of the country as a destination for expeditions: the region visited after a disaster, a closure, or an election surprise, interviewed, and left. The resulting coverage has a recognisable ethnographic tone, the discovery of poverty or anger as though of a foreign custom, and regional readers have complained of it for generations in terms the theory can only endorse. Decentralisation schemes and regional bureaus fight the pattern with partial success, because the diary and the source system are centralised even when the desks are not: parliament, the courts of record, the markets, and the ministries all sit in the capital, and the beat system drags the map back to the centre however the offices are arranged.

The interview: adversarial theatre inside an agreed frame

The broadcast interview is the genre in which modern news most visibly performs its independence, and the performance deserves the same double reading as objectivity itself. The adversarial interview, in which a presenter interrupts, presses, and refuses the evasion, is a genuine discipline: ministers emerge from it damaged, and the genre’s celebrated practitioners have ended careers with questions. Audiences who cite the interview as proof that media serve no master are pointing at something real. Yet the theatre has a frame, and the frame is set before the first interruption. The choice of subject, the identity of the guest, and above all the premise carried by the opening question determine the territory across which the combat will range. An interviewer may savage a minister for failing to meet a target while the target itself, its construction, its casualties, its alternatives, sits outside the exchange as the shared assumption of both combatants. The soundbite economy then selects from the combat the twelve seconds that can travel, which are rarely the seconds in which the premise trembled. Hegemony theory was designed for this exact configuration, dominance exercised not by silencing the argument but by owning the ground on which arguments are held, and it predicts the paradox audiences register without a vocabulary for it: news that feels combative and arrives, season after season, at remarkably stable conclusions about what the questions are.

One closure, three readings: the accounts at work

Abstractions earn their keep on a single case, so take the most standard story in the industrial repertoire: a manufacturer announces the closure of a plant in a town built around it, with the loss of several hundred jobs, production to move abroad. Watch the story happen. The announcement arrives as a company statement, timed, documented, with a quotable executive and a rationale in the vocabulary of competitiveness. By the evening bulletin the closure has a shape: the company’s reasons lead, the union’s response follows inside the company’s framing, the town’s grief supplies colour, an economist supplies context, and a minister expresses regret and points to retraining schemes. Nothing in the bulletin is false. Every element of the shape, however, was selected, and the three accounts read the selection differently, which is the exercise this whole article has been preparing.

The instrumentalist reads the shape from the ownership page. Is the outlet’s parent exposed to the same pressures as the manufacturer, does it hold interests the closure story might disturb, has the desk been guided away from the comparison that would embarrass a sister company? She predicts that outlets owned by interests aligned with the manufacturer will soften the story’s edges, shorten it, and lose it faster, and she checks the record across outlets for exactly that variance. The structuralist barely glances at the register. He predicts the shape from the production process alone: the company statement is a subsidy, the union’s grief is an occurrence, the workers’ own analysis of the plant’s order book is not an event and will not be gathered, the minister’s regret is diary material, and the slow question, why a profitable plant closes, requires documents, time, and a specialist the desk may not employ. He predicts the same shape in the public broadcaster and the cooperative paper, and he is usually right. The hegemonic reader watches a different transaction. The bulletin offers a preferred reading, regrettable necessity, global forces, no alternative, and stocks it with the executive’s sorrow and the minister’s schemes. She then asks what the town does with the offer: whether the local paper, the union’s meetings, and the town’s own memory of the plant’s profits supply a counter definition strong enough to sustain an oppositional decoding, and whether the national frame bends, over the following weeks, to absorb the protest as a story about a community’s spirit rather than about a decision’s merits. Three investigators, one bulletin, three different notebooks. The mature analysis keeps all three notebooks open and says, with evidence, which mechanism did most of the work in this case. The immature one picks the notebook that flatters its owner’s politics and calls the choice a theory.

Run the same exercise on a contrasting subject and the weights shift, which is the real test of the method. Take a hospital story: a health authority announces a reorganisation, wards to close at one site and services to concentrate at another, presented in the language of clinical excellence and safety. The structuralist machinery engages at once, since the authority’s documents, spokespeople, and guided tour supply the subsidised account, while the affected patients and junior staff must manufacture occurrences to be heard, and the process beneath the event, the funding settlement that made some reorganisation unavoidable, has no press conference of its own. The instrumentalist finds less to grip, unless the outlet’s owners hold interests in the private providers who circle every public reorganisation. The hegemonic reader finds the richest field of the three, because a hospital is one of the few institutions whose users possess direct, embodied counter knowledge: nearly every viewer has been a patient, and a frame that contradicts the remembered corridor and the cancelled appointment meets an oppositional decoding ready made in the audience’s own biography. The weights differ by subject, by medium, and by period, and a theory of media ideology that cannot redistribute its weights from case to case is not a theory but a slogan with three verses.

The division of labour among the three accounts

The comparison can now be settled into the working rule this article recommends. Instrumentalism is the correct instrument for the boundary and the intervention: for documented direction, for the anatomy of appointments, for the conglomerate’s balance sheet reasons, and for the platform owner’s hand on the objective function. It fails as a general theory of content and should never be asked to carry one. Structuralism is the correct instrument for the ordinary output: for framing convergence across unlike owners, for the source system and the diary, for the economics of format and the metrics of the feed. It fails at change, at rupture, and at the audience, and should be supplemented exactly there. Hegemony is the correct instrument for the life of framings in public: for consent won and lost, for moral panics and their decay, for decodings patterned by social position, and for the absorption of dissent that keeps a frame alive. It fails when asked for the precision of the content analysts, and its users should borrow their counts rather than affect to despise them. The propaganda model stands inside this settlement as structuralism’s most developed engine, with the Glasgow tradition as its measurement wing and Hall’s work as the standing correction to both on the question of reception. A reader equipped with the settlement can classify any new argument about media bias within a paragraph of hearing it, identify which evidence would strengthen it and which would kill it, and detect the moment an arguer abandons a dying account and continues the quarrel from inside another one without announcing the move. That detection, practised until it is reflex, is the transferable skill. The theories will be revised by media not yet invented; the discipline of asking which mechanism, at which level, with what evidence, will survive every revision, because it is not a doctrine about the press. It is the refusal to let an accusation substitute for an explanation, applied to the industry whose business is accusation and explanation alike.

That discipline also protects the theory’s friends from its worst habit, which is triumphalism at the moment of any scandal in the press. A proprietor caught intervening has confirmed the boundary mechanism, not the whole edifice, and a newsroom caught converging has confirmed the routines, not a conspiracy. Each finding buys only the territory its mechanism covers, and the map of territories, revised case by case, is the closest thing the Marxist study of news possesses to a settled result.

Scandal, exposure and the limits of self repair

The strongest exhibit in the pluralist case is the press investigating power at genuine cost, and the hardest version of that exhibit is the press investigating itself. The British phone hacking scandal, in which a national newspaper’s journalists were shown to have intercepted the voicemail of public figures and of a murdered schoolgirl, was exposed principally by another newspaper, the Guardian, whose reporters pursued the story for years against denials, legal threats, and the indifference of much of the rest of the industry, until the evidence forced a police reinvestigation, the closure of the offending title, and a public inquiry into press conduct. Every element of that history resists the crude instrumentalist account: a proprietor’s paper destroyed a rival proprietor’s asset, journalists defied flak of the most organised kind, and the system’s own products supplied the tradition’s evidence once again. A Marxist analysis that cannot honour the episode has no business lecturing journalists about evidence.

Honour it, then, and observe what the episode also shows, because the structuralist reading of the same facts is not a refusal but a completion. The exposure depended on one newsroom whose commercial position allowed the investment: the story took years, earned nothing measurable, and was carried by an organisation insulated by a trust ownership structure from the pressures this article has catalogued, which is to say the exception was itself structurally produced. The rest of the industry, including titles that had competed for the same celebrity material by the same methods, was slow to follow a story that threatened shared arrangements with police and political sources, and several attacked the investigators before joining them. The inquiry that followed then became a demonstration of primary definition in reverse, as the industry organised, through its associations and its access, to define the scandal as the crime of a few and the remedy as a danger to freedom, a framing whose success can be measured in the modesty of the remedies that survive the argument. Scandal, on this reading, is the system’s storm rather than its refutation: real damage is done, real villains fall, the public is genuinely served, and the source relations, revenue structures, and appointment systems that produced the conditions resume their work in calmer weather. Both halves of that sentence are required. An account that keeps only the second is as dishonest as the celebration that keeps only the first.

The alternative press: an existence proof with an economic autopsy

Dissenting media are the standing refutation of any claim that the ideological field is closed, and their history is simultaneously the standing demonstration of the field’s economics. From the unstamped radical papers of the nineteenth century through the underground press of the 1960s to the independent media centres of the anti globalisation movement and the blog networks of the following decade, alternative outlets have repeatedly shown that the dominant framing can be refused in print, at length, with talent. They have also repeatedly died of the same causes, and the causes are the filters of the propaganda model read from the victim’s side: advertising withheld or never offered, distribution controlled by wholesalers and platforms with no obligation to carry them, libel exposure uninsurable at their scale, sources and access denied, and audiences whose poverty is the very reason for the publication. The pattern deserves a name of its own inside the theory. Call it the asymmetry of survival: the cost of entry to public speech has fallen across the whole period this article covers, while the cost of endurance, of reaching a mass public daily for decades, has remained a function of exactly the resources whose distribution the theory is about. A media ecology can therefore be formally open and materially stratified at once, and most of the quarrel between pluralists and Marxists dissolves once the two adjectives are kept apart. The pluralist counts voices, and the count rises. The Marxist weighs amplification and endurance, and the weights barely move. Both measurements are real, and a serious analysis reports both.

The words a bulletin cannot help using

Content analysis earns its authority at the level of the single word, where framing stops being an interpretation and becomes a list. News organisations maintain style rules for contested vocabulary, and the rules are a public record of ideology being administered: when a fighter is a terrorist, a militant, a gunman, or a soldier is decided by guidance, by convention, and by the alignment of governments, and large agencies have tied themselves in famous knots attempting a neutral rule, restricting the word terrorist to attributed quotation while their clients’ headlines restored it. Governments are regimes when they are enemies and administrations when they are friends. Economies suffer disruption from strikes and enjoy flexibility from closures. Immigration arrives as a flow, a wave, a flood, in the vocabulary of natural disaster, or as a contribution, in the vocabulary of accountancy, and the choice between the hydraulics and the ledger is made by subeditors at speed, from habit and house style, in the last minutes before transmission. None of these choices is an order from an owner. All of them are reproducible across outlets that share a culture, which is why word level analysis persuaded so many sceptics where manifesto level claims had failed: the pattern is too fine grained, too consistent, and too obviously unplanned to be a conspiracy, and it leans, in case after case, the way the structural accounts predict. The lesson cuts the other way as well, and the tradition’s honest practitioners state it: where the vocabulary runs against the predicted lean, as it does on subjects where professional or public counter pressure has organised itself, the exception is data too, evidence of a frame being contested at the level of the word list, which is hegemony’s home ground.

Corrections, complaints and managed participation

Every mature news system operates machinery for absorbing criticism of itself, and the machinery is a small hegemonic system in working order. The correction column concedes the error of fact, a date, a name, a figure, and by conceding it certifies the larger fairness of the whole, though the correction never runs at the size of the original claim and the framing that carried the error is not itself a correctable item. The readers’ editor and the ombudsman institutionalise complaint, hearing the reader on accuracy and taste while the structural questions, why this story, why this source, why this word, arrive as letters and depart as courtesies. Participation formats complete the circuit. The phone in and the comment thread admit the public’s voice into the product under conditions set by the host, the topic chosen by the desk, the callers screened, the clock absolute, so that dissent appears inside the programme as one of its features and departs without residue. It would be cynical to call these mechanisms shams; corrections correct, ombudsmen sometimes rule against their own editors in public, and phone ins have ambushed presenters with evidence no newsroom would have gathered. The hegemonic point is finer. A frame survives not by excluding its critics but by giving criticism a scheduled room with the furniture arranged, and the test of the room is what can be changed from inside it. Errors of fact can be. The selection system that generates the errors’ direction cannot be reached from the corrections column, and the distance between those two powers is the exact size of the concept of hegemony.

Three sightings: the vocabulary at work in ordinary argument

The accounts separated in this article are not confined to seminar rooms, and catching them in the wild is the quickest training in their use. A minister, accused of leaning on a broadcaster, replies that the editor is independent and no instruction was given, and the reply is an instrumentalist acquittal: no telephone call, no offence. A press critic, reviewing a week of coverage of a public sector dispute, notes that every outlet led with the inconvenience to commuters and none with the staffing levels the dispute concerned, and the observation is structuralist whatever the critic’s politics, since its evidence is convergence and its mechanism is the source system. A media scholar, asked whether coverage of a protest movement was biased, answers that the movement’s own media changed what the bulletins could say within a month, and the answer is hegemonic, a claim about a frame under negotiation. Public argument about the media fails most often at the joints between these sightings, when an instrumentalist acquittal is allowed to answer a structuralist charge, or a structuralist observation is inflated into an instrumentalist accusation it cannot support. The reader who can name the account in play, and refuse the substitution, has acquired the skill the series exists to teach, and can apply it to arguments made by governments, by proprietors, and, with the most profit, by the reader’s own side.

Rolling news: the diary at double speed

The arrival of continuous news channels altered the structural equation in a direction the classic accounts understate, and the alteration rewards a moment’s analysis. A bulletin that must fill twenty four hours cannot wait for events at the pace events occur, so the diary’s share of the output grows, the threshold for leading a story falls, and a new genre colonises the gaps: anticipation and reaction, the preview of the announcement and the harvesting of responses to it, in which the primary definers are served twice, once for the event and once for the choreography around it. Speculation, supplied by correspondents required to speak before they know, draws on the safest stock available, the conventional reading of the institution’s intentions, so that uncertainty itself is reported inside the dominant frame. The rolling format also intensifies the economics of the picture: a story with live images outranks a more consequential story without them for hour after hour, entrenching the event bias the Glasgow group found in a slower system and adding a bias of sheer visibility. None of this was chosen by owners as ideology. It is the arithmetic of the empty hour, and it is the purest available demonstration that the structuralist mechanism operates independently of anybody’s politics, including the politics of the political scientists who first described it.

The picture desk: ideology at the speed of the eye

Text has dominated this article because text dominates the literature, but the photograph and the moving image carry framing at a speed no paragraph can match, and the picture desk is a selection system with its own routines. The choices are granular: which frame of the minister, caught mid sentence or composed; which image of the protest, the banner or the broken window; whether the flooded street is photographed with its residents wading or from the helicopter that makes it scenery. Cropping is an argument conducted with a blade, and the archive is an ideology in storage, since the file photograph chosen to illustrate a recurring subject, the union leader, the benefit claimant, the banker, accumulates into a visual definition no single selection intended. Picture editors describe their judgements in the profession’s ordinary language of impact, clarity, and story, and the description is honest. The structural observation stands beside it undisturbed: impact is measured against an audience presumed to recognise the story already, so the picture that confirms the existing definition is reliably the clearest, and clarity, compounded across a million small decisions, is how a visual common sense is manufactured without a single falsified image.

Satire, entertainment and the solvent of ridicule

Entertainment is the territory where the theory of media ideology meets its most ambiguous evidence, and the ambiguity should be preserved rather than resolved. Satire has plainly done oppositional work in every media order that permitted it: the cartoon, the revue, and the comedy panel show have punctured reputations that news treated with deference, and regimes of many kinds have feared ridicule enough to censor comedians before journalists. The hegemonic account claims the phenomenon without strain, since laughter is one of the forms in which consent quietly lapses, and a leader who has become a joke has lost something no editorial can restore. Yet the same formats operate inside the commercial structures this article has described, scheduled, formatted, and sold, and a cooler tradition of criticism reads licensed ridicule as the frame’s safety valve, dissent discharged as laughter in a sealed half hour after which the bulletin resumes. The empirical question is what audiences carry out of the theatre: a suspicion that survives into the voting booth and the picket line, or a catharsis that substitutes for both. Research on the question is thinner than the confidence of either camp, and the honest article reports the dispute rather than settling it, noting only that the two readings correspond, once again, to the two tempers within the theory itself, the structuralist suspicion of every permitted thing and the Gramscian insistence that culture is a battlefield precisely because its outcomes are not arranged in advance.

The measured public: ratings and the definition of the audience

Underneath every mechanism described so far sits a representation of the audience itself, and representations of audiences are ideology in its most consequential modern form. The audience that advertisers buy, that schedulers serve, and that editors imagine is not the public but a measurement of it: panels, meters, and metrics that count some behaviours and not others, weight some households and not others, and define value in the currency of purchasable attention. Journalism addressed to the measured public is addressed, in part, to an artifact of the measurement industry, affluent where advertisers pay for affluence, impatient where the meter punishes patience, and absent altogether where no meter is installed. Public service media were founded partly in flight from this definition, and their long argument with their funders has been an argument about which representation of the audience shall govern: the citizen, the consumer, or the ratings point. The Marxist contribution to that argument is to insist that the choice is not technical. Each definition of the audience smuggles in a definition of what news is for, and the ratings definition, victorious across the commercial system and advancing inside the public one, answers the question in advance, in the language of the market, before any journalist has decided anything at all.

Media systems compared: the theory tested across borders

Comparison across countries is the closest approach the social sciences offer to an experiment, and media research possesses a comparative literature strong enough to discipline every account in this article. Daniel Hallin and Paolo Mancini’s Comparing Media Systems, published in 2004, sorted the democracies of Europe and North America into three broad models: a liberal model, market dominated with strong professionalisation; a democratic corporatist model, in which strong public broadcasting and organised social groups share the field; and a polarised pluralist model, in which the press remains entangled with parties and the state. The taxonomy matters here because it converts the theory’s mechanisms into variables. If structuralism is right, then systems differing in ownership, funding, and source relations should differ measurably in output and in what their citizens know, while systems sharing those structures should converge despite different political cultures. A body of subsequent research, including a widely cited comparison by James Curran and colleagues of knowledge levels across media systems, found citizens of strong public service systems better informed about public affairs than citizens of the most commercialised system, with the difference concentrated among the less advantaged, exactly the distribution a structural account predicts and a pure audience sovereignty account does not. The comparative record does not crown a winner among the three accounts. It does something better: it shows the mechanisms operating at national scale, visible in the architecture of systems rather than in the intentions of any editor, and it gives the student a literature in which the theory’s claims arrive with standard errors attached.

When the money runs out: ownership returns through the back door

The economics of news have entered a period in which the old questions return wearing new clothes, and the transformation deserves an account of its own because it rearranges the three theories’ territories. The collapse of advertising revenue has closed newspapers by the thousand in the United States and thinned them everywhere, leaving districts without any professional observer of their courts, councils, and planning committees, the news deserts of the literature. Research on the consequences is young but consistent in direction: where local journalism dies, civic participation falls, municipal borrowing grows more expensive under reduced scrutiny, and the remaining information space fills with national partisan content and rumour. Into the vacated field have stepped new proprietors of recognisable kinds: chains assembled by investment funds, for which the titles are assets to be harvested, their staffs cut to the bone and their buildings sold, and wealthy individuals purchasing prestige newspapers as a citizen buys a vineyard, for standing as much as for return. The fund owned chain is instrumentalism’s nightmare made ordinary, ownership exercising its power not by dictating lines but by abolishing the capacity to report at all, an ideological effect achieved through headcount rather than through any editorial decision, and invisible to content analysis because the content has ceased to exist. The omission principle reaches its limit case here: the most complete selection is the newsroom that is no longer there, and a theory of media ideology for the coming period will need an account of absence as its first chapter, with the economics of attention and the politics of subsidy, who pays for the diary to be kept, as its second.

The local paper: ideology at parish scale

The theory in this article was built mostly from national media, and testing it against the local press, where the sociology inverts several of its assumptions, is a useful exercise in what the mechanisms really depend on. The local newspaper’s journalist knows her sources socially, meets the councillor in the supermarket, and depends for copy on a small circuit of institutions, the council, the courts, the police, the hospital, whose press officers she cannot afford to alienate because there is no second circuit. Primary definition at parish scale is not a tendency but a description of the working week. Advertising dependence is likewise unmediated: the local paper historically lived from the classified columns, the property pages, and the announcements of the very council and developers whose decisions its news pages existed to scrutinise, and the structural conflict required no proprietor’s instruction to operate, only the knowledge, shared by everyone in the building, of which campaigns the revenue could survive. The result is a recognisable local genre, boosterish about the town’s prospects, tender toward its major employers, ferocious about litter and parking, in which the ideological work is done by the economics of intimacy rather than by any politics the staff would recognise as their own.

And yet the local paper also embarrasses the theory in one respect the national media do not, because its readers possess the means of verification. A national audience cannot check the framing of a foreign war against experience; a local audience walks the street the paper describes, knows the family in the court report, and punishes falsity of atmosphere with a speed no ombudsman can match. Local journalists therefore work under a discipline of proximity that substitutes, partially, for the disciplines of competition the market has withdrawn, as local titles consolidated into chains and monopolies across the period. The hegemonic reading fits the parish best: the local paper’s frame is negotiated daily with a readership that is also its subject, its source base, and its social world, and consent there is neither manufactured nor free but transacted, in public, at close quarters, by people who will meet again. The lesson for the general theory is that the strength of every mechanism in this article varies with the audience’s independent access to the facts, which is why the tradition’s claims grow stronger as the subject grows more distant from the reader’s own life, and why foreign and economic news, where verification is scarcest, remain its most persuasive exhibits.

Competition: the scoop, the pack and the herd

Competition is the pluralist objection’s favourite mechanism, and it deserves a section of its own because the reality is double in a way neither camp’s slogan captures. Rivalry between outlets unquestionably produces exposure: the scoop is the profession’s supreme currency, and the desire to beat the rival across the street has dragged more secrets into print than any charter of press freedom. Watergate, the type case of adversarial journalism in the American memory, was a circulation and prestige war as much as a civic mission, pursued by a newspaper that could measure the value of the story in the rivals it humiliated. Any theory that treats competition as a mere disguise for collusion will fail against that record, and the Marxist tradition at its best has never tried: its claim begins where the scoop’s limits begin.

For the scoop has a sociology, and Timothy Crouse named its pathology in The Boys on the Bus, his 1973 account of the press corps on the American campaign trail: pack journalism. Correspondents travelling together, filing together, and reading one another’s copy converge on a common story with a common judgement, because the professional risk of being wrong alone vastly exceeds the risk of being wrong together. An editor forgives the reporter who missed what everybody missed; the reporter who alone dissents from the pack’s reading and errs is finished. The incentive structure of competition therefore selects for herd behaviour at the level of interpretation even while it selects for enterprise at the level of fact gathering, and the two selections coexist in the same newsroom on the same day. Desks race to be first with the event and last to depart from its accepted meaning. The phenomenon generalises far beyond campaign buses. Financial journalism in the run up to the crisis of 2008 displayed the pack structure in its purest form, a small corps of specialists sharing sources, sharing assumptions, and sharing the fate of having missed, together, the largest story of their careers, while the handful of dissenting voices were filed, by the pack’s own taxonomy, as eccentrics. Being wrong together is not a conspiracy. It is a labour market.

The herd mechanism also explains one of the field’s most reliable rhythms, the feeding frenzy and its opposite, the story that cannot be placed. A subject anointed by the pack receives coverage out of all proportion to its weight, as each desk defends itself against the charge of having missed the story of the week, while a subject the pack has not anointed cannot buy attention at any evidentiary price, since the first desk to move bears the whole risk of eccentricity. Agenda setting research formalised the rhythm as intermedia influence: elite outlets set the agenda for the rest, and the rest set it for the broadcasters, so that pluralism of ownership coexists with a hierarchy of definition in which a very small number of newsrooms decide what the others will treat as the day’s reality. The structuralist reads the hierarchy as the primary definer system operating inside the profession itself. The hegemonic reader adds the temporal dimension: the pack’s anointing can be broken, by an event too large to file under the existing reading, by a rival institution with its own megaphone, or by an audience that refuses the offered frame in numbers the metrics cannot ignore, and the history of coverage is punctuated by such breaks, after which a new common reading assembles itself with remarkable speed. Competition, in sum, is neither the pluralist’s solvent nor the crude theorist’s fiction. It is a real force with a bias of its own, toward enterprise in facts and conformity in meanings, and a mature account of media ideology assigns it that precise and limited role.

The agency wire: the wholesaler nobody reads

One institution in the news system is almost never discussed by its audience, because its product reaches the public only inside other organisations’ pages, and it may be the most consequential selection mechanism of all. The international news agencies gather the raw material of the world’s news and wholesale it to subscribers, and for most outlets, most of the time, the agency file is not a supplement to the news but its outer boundary: what the agencies do not cover, the subscriber does not know, and what they cover in three paragraphs will rarely justify the cost of a correspondent’s independent inquiry. Agency journalism is professional, fast, and by the standards of this article remarkably disciplined, since the wire must serve clients of every politics and therefore perfected the factual, attributed, colourless style from which the objectivity norm descended. The structural consequences are nonetheless profound. The agencies’ bureau map is the world’s news map, their ranking of the file is the first draft of every front page, and their house vocabulary, settled for a global market, settles with it the default terms in which events arrive at a thousand desks. An ideology analysis that begins at the newspaper has begun one step too late. The first selection was made upstream, by an organisation the reader has never heard of, working to a specification the market wrote, and the finding generalises the article’s whole argument in a single institution: the deeper a mechanism sits in the production chain, the less visible it is, the less it is argued about, and the more completely its output is mistaken for the world itself.

For students and exam candidates

Any media question on a syllabus can be organised by the three accounts, and examiners reward the candidate who announces the organisation. Open by separating the levels: a claim about bias may concern ownership, production routines, or audience reception, and evidence for one level is not evidence for another. Deploy the intention trap by name as your critical instrument: the weak essay presumes an instruction behind every slant, the strong one tests mechanisms. For the propaganda model, learn the five filters as a system with a denial of conspiracy attached, and prepare the paired example method as your evidence, since a comparison with a control case is worth more than a column of annotated grievances. For the Glasgow studies, hold both ends of the exchange, the framing findings and Harrison’s methodological rebuttal, because the marks live in the quarrel rather than in either side of it. Keep Hall’s three decoding positions attached to Morley’s audience work so the hegemonic account arrives with evidence rather than as an attitude. Use a structured notebook to file each theorist under mechanism, prediction, and damaging evidence; the Marxism study notes companion in this series is arranged for exactly that filing, and a grid learned by heart will reconstruct under pressure when memorised paragraphs dissolve.

For teachers and seminar leaders

Teach this topic as an exercise in distinguishing explanations before it becomes a contest of loyalties. The content analysis exercise named in the brief repays its preparation reliably: give the class two outlets covering one industrial dispute or one closure, and require three separate inventories, one of facts asserted, one of sources and their order, and one of vocabulary, before any discussion of bias is permitted. The inventories teach the structural claim experientially, because students discover the convergence of framing in their own columns of notes, and the argument that follows is about their evidence rather than about their allegiances. Follow with the paired example discipline: for every claim of slant, require the like case treated otherwise, and let claims that cannot find their pair be reclassified by the class as suspicions. On the propaganda model, set the conspiracy misreading as a trap in the first session by asking students to find the passage in which the authors deny it, and on filter bubbles set the empirical correction beside the popular thesis so that the grading of evidence is practised on a claim the room wants to believe. Close with the Orgreave sequence as the capstone case, since it lets every account score one point and concede one, which is the intellectual posture the whole topic exists to teach.

For researchers, historians, and journalists

For researchers, the live problems are measurement problems. The selection function of a platform distributed news system is partly inside proprietary ranking systems, and the field needs exposure data gathered independently of the platforms’ own reports if the structural account is to be tested rather than asserted for the new order. Audience research must likewise resist its own convenience samples, since the decoding claims inherited from Hall stand or fall on representative evidence about who refuses which frames. Historians of the press hold an underused asset: the archive of the unprinted, the killed story, the spiked investigation, the proprietor’s correspondence, is the only direct record of the boundary this article describes, and every memoir quarrel over an editor’s dismissal is a source to be triangulated rather than a verdict to be quoted. For journalists, the professional application is set out in the companion piece on marxism for journalists, and its core discipline is brief enough to state here. Before calling coverage biased in the Marxist sense, name the level of the claim, ownership, routine, or reception; produce the paired comparison; state the mechanism; and record what would count against the claim. A newsroom that ran that checklist on its own output, subject by subject, would learn more about its ideology in a month than a decade of complaints from outside can teach it, because the mechanisms described in this article operate through honourable routines and can therefore be audited only from inside them. The audit is uncomfortable precisely where it is fair: it accuses nobody of lying, and it excuses nobody on that ground.

One day in the diary: the mechanism at walking pace

Abstractions about routines become convincing when they are timed, so follow a single bulletin from its morning conference to its transmission and mark the points at which the world was selected. The conference opens with the overnight wire and the diary: the ministerial announcement scheduled for midday, the verdict expected in the trial, the report launch with its embargoed summary already in the building. Three of the evening’s five likely items are therefore institutional before a reporter has left the office, and they arrive with spokespeople, documents, and pictures arranged. The fourth item is contested ground where the desk commissions a package, and the producer, choosing between the specialist who files rarely and the generalist who files by four, chooses under a deadline that is itself a theory of the audience’s patience. By mid afternoon the running order meets reality: a fire with dramatic pictures displaces a process story about the fire service’s budget, not by decision but by the physics of the format, since the bulletin cannot open with an absence. The minister’s announcement has meanwhile been pre briefed to selected correspondents, so the package leads with the ministry’s definition and carries the opposition’s reply in the position of reply, and the union whose members the policy concerns is telephoned at five for a reaction it must compose against a text it has only just seen. At transmission the bulletin is balanced, accurate, and vigorously presented. Every fact in it can be defended at a complaints hearing. And the day it describes is a day in which institutions acted, reacted, and explained, while the unannounced world, the ward quietly short staffed, the rent quietly raised, the process quietly completing itself, waited for an eruption to make it legible. Repeat the day two hundred times and the accumulation is a picture of society. The picture was nobody’s intention. That is the whole structuralist argument, and it is also why the argument is so hard to refute at the level of the individual bulletin: it concedes every local innocence and indicts only the sum.

Where command is real: state media and the boundary of the theory

The three accounts were built for media systems in which control is indirect, and their boundary is instructive. In state media systems of the classic kind, where the ministry appoints the director, the censor reads the copy, and the line arrives by instruction, instrumentalism is not a fallacy but a description, and the intention trap never springs because the intentions are documented in the system’s own regulations. The comparison disciplines the theory in two directions. It shows what genuine command looks like, and thereby sets the standard that command claims inside open systems must meet: an instruction, an appointment pattern with teeth, a sanction observed in operation, rather than a slant with a beneficiary attached. And it explains a paradox that observers of both systems notice, that audiences under open command often discount their media more thoroughly than audiences under indirect influence, since a known propagandist is a discounted one, while the framing produced by routines and markets arrives wearing the credentials of independence. The tradition has sometimes drawn from this paradox the cynical moral that indirect control is simply more efficient. The better moral is analytical. Effects depend on the audience’s theory of the source as much as on the source’s machinery, which returns the analysis to Hall’s territory, the contested field in which a message’s fate is decided jointly by its encoding, its route, and the resources of suspicion its audience brings to the meeting.

Frequently Asked Questions

Q: Is the news media ideological?

In the Marxist sense used in this article, yes, though the claim is narrower than it sounds. The claim is not that journalists invent a false world or take instructions from a committee. It is that the categories through which events become news, the sources entitled to define them, the vocabulary available to describe them, and the commercial and professional routines that select among them together form a framework that systematically favours some descriptions of society over others. That framework is ideology in the relevant sense: a way of seeing whose partiality is built into its procedures rather than announced in its conclusions. A bulletin can be accurate in every particular and ideological in its total shape, because the ideology sits in the selection, the order of definition, and the questions asked, all of which precede accuracy as an issue.

Q: What is the propaganda model?

The propaganda model is the account of news production set out by Edward Herman and Noam Chomsky in Manufacturing Consent, published in 1988. It proposes that five filters act on raw events before they reach the public as news: the size, concentrated ownership, and profit orientation of media firms; advertising as the dominant revenue and therefore a licensing system; dependence on government, business, and expert sources that subsidise the flow of material; flak, the costly negative feedback that disciplined coverage attracts; and an overarching enemy ideology, originally anti communism, that sorts the world’s events in advance. The model predicts patterned asymmetry rather than uniform propaganda: paired cases of similar character will be treated dissimilarly according to their relation to dominant interests. Its authors present it as a model of market forces operating through institutions, and they test it by comparative content analysis rather than by claims about journalists’ motives.

Q: Are the five filters still valid?

As an architecture, largely yes; as an inventory, it needs revision for a platform distributed system. Ownership concentration has intensified in some sectors and mutated in others. Advertising remains a licensing force but now operates substantially through platforms that intercept the revenue and return only metrics to publishers. Sourcing dependence survives intact, since official definers still subsidise the news with material and access. Flak has been industrialised by social media pile ons and organised campaigns, which lower its cost and raise its speed. The fifth filter’s original content, anti communism, expired with the Cold War, and the slot is better treated as a variable occupied by whichever enemy image a period supplies. What has changed most is distribution: an algorithmic selection layer now sits between the newsroom and the audience, and any restatement of the model must add it as a mechanism in its own right rather than as a detail.

Q: Did the Glasgow Media Group prove media bias?

The group proved something more specific and more durable than the slogan. Recording and transcribing British television bulletins across defined periods in the 1970s, it demonstrated patterned framing in industrial coverage: disputes organised around disruption, causes rarely explained, management and government sources positioned as the reasonable definers, and strikers rendered largely as a problem to be managed. That demonstration shifted the field’s question from whether individual reports were fair to how structure produced partiality. What the group did not do is settle the matter beyond dispute. Martin Harrison’s methodological rebuttal of 1985 attacked its sampling, its coding, and the benchmark of unbiased coverage it implied, and parts of that rebuttal landed. The mature verdict is that the Glasgow studies established framing as a measurable property of bulletins and left a permanent argument about measurement, which is what a scientific contribution looks like in a contested field.

Q: Is media bias deliberate or structural?

The evidence supports structural as the general answer and deliberate as a local one. Deliberate bias exists: owners have intervened, editors have been dismissed, and stories have been suppressed for commercial and political reasons, all documented. But the largest and most consistent patterns in the research literature, convergent framing across outlets with unlike owners, slant inside a public broadcaster with no proprietor, the grammar of business reporting, the order in which sources define events, arise from routines, sourcing, and format, and would survive a complete change of personnel. The deliberate cases are therefore best treated as the visible peaks of a mostly submerged structure. An analysis that hunts only for instructions will miss the structure, and an analysis that denies instructions on principle will miss the peaks. Naming which phenomenon a given dispute concerns is the first discipline this article recommends.

Q: What is the difference between instrumentalist and structuralist media theory?

The two accounts locate the cause of media ideology in different places and therefore predict different evidence. Instrumentalism locates it in ownership and direction: media serve the interests of capital because capitalists own the organisations, appoint their editors, and intervene at the boundary of the publishable. Its characteristic evidence is documentary, interventions, dismissals, suppressed stories, and its characteristic failure is the proprietor publishing against interest. Structuralism locates the cause in the production process itself: the diary, the beat system, dependence on official sources, advertising finance, and format, which together select and frame the news whoever owns the outlet. Its characteristic evidence is convergence, similar framing under different owners, and its characteristic failure is rupture, the moments when coverage breaks its own routines. The practical test between them is variance: slant that tracks ownership supports the first, slant that survives changes of ownership supports the second.

Q: Do audiences simply absorb media messages?

No, and the tradition’s most influential audience theory was built on that refusal. Stuart Hall’s encoding and decoding model proposes that producers build a preferred reading into a message, but audiences decode from their own social positions, accepting the preferred reading, negotiating it against their experience, or reversing it in an oppositional decoding. Audience research inspired by the model, including David Morley’s work on the Nationwide programme, found decodings patterned by social resources rather than dictated by texts, though messier than the original hypothesis. The finding cuts in both directions. It refutes the picture of a passive public written upon by owners, and it also refutes the complacent opposite, since decoding is not free invention: the preferred reading arrives with institutional amplification, and oppositional decoding requires alternative frameworks that are themselves unevenly distributed across society. Absorption is rare, surrender is rare, and negotiation, on unequal terms, is the norm.

Q: Are algorithms the new gatekeepers?

They are an additional gate, operating on different principles from the editorial gate they joined. Editors gatekeep by judgement under professional norms, selecting what is made and what leads. Ranking algorithms gatekeep by predicted engagement, selecting what each user actually encounters from the far larger pool of what was published, and they do so continuously, personally, and without publishing their criteria. The consequences differ from the editorial regime in ways that matter for ideology: the shared agenda weakens, since no two users meet the same front page; emotional and identity confirming material gains a structural subsidy; and the metrics the algorithms return reshape newsroom decisions upstream, so that algorithmic selection indirectly commissions as well as distributes. Calling algorithms the new gatekeepers is accurate about reach and misleading about intention. They have no editorial purpose to interrogate, only an objective function set by their owners, which is why the ownership question returns, in this layer, with full force.

Q: Do filter bubbles actually exist?

In the strong popular form, the evidence does not support them. The thesis, popularised by Eli Pariser in 2011, holds that personalisation seals users inside unique, invisible information universes cut off from dissonant views. Empirical studies of browsing and exposure have generally found something weaker and more heterogeneous: algorithms explain less of the variance in what people see than their own choices do, many users encounter cross cutting material through search and social feeds, and the most sealed audiences tend to be committed partisans whose insulation is self chosen and older than the platforms. Some narrowing effects are real, particularly in recommendation driven environments, and effects vary sharply by platform and user. The responsible conclusion is that filter bubble names a genuine mechanism of modest and uneven size, inflated by public debate into a master explanation for polarisation whose causes lie substantially elsewhere, in politics, identity, and the economics of attention.

Q: Is Manufacturing Consent a conspiracy theory?

No, and its authors anticipated the charge by denying it in the text. A conspiracy theory explains an outcome by the secret intentions of a coordinating group. The propaganda model explains outcomes by the public operation of market institutions: ownership structures, advertising dependence, source subsidies, organised flak, and a prevailing enemy ideology, acting on journalists who are assumed to be sincere professionals working inside incentives they did not choose. No secret meetings are required, no orders are postulated, and the model’s evidence is comparative content analysis that any sceptic can recount. The confusion arises because the model’s conclusions resemble those of cruder theories, and because its title is a provocation. Readers who classify arguments by their evidence rather than by their conclusions will find the model a piece of institutional sociology, open to the ordinary objections against functionalist explanation, but not a member of the conspiracy family at all.

Q: How can news selection be ideological without any instruction from an owner?

Because selection is governed by routines that have direction built into them before any individual decides anything. The diary fills the bulletin with institutional events, which are announced by the powerful. The beat system stations reporters at the doors of organisations, which supply statements, documents, and access, while unorganised experience must manufacture an occurrence to be heard at all. Deadlines reward sources that deliver on time in usable form, which is a description of a ministry press office and of almost nothing else. Format rewards the event over the process, the conflict over the condition, and the picture over the statistic. Advertising finance penalises subjects that sit badly beside consumer marketing. Each routine is defensible as professional practice, and their sum is a selection function with a systematic lean, which reproduces itself daily through the honest work of people who have never received an instruction in their lives.

Q: What should a journalist check before calling coverage biased in the Marxist sense?

Four checks, in order. First, the level of the claim: is the charge about ownership and intervention, about production routines, or about audience reception, and does the evidence offered belong to that level? Second, the pair: can a comparable case be produced, similar in character and treated dissimilarly, since a single annotated example proves grievance rather than pattern? Third, the mechanism: what process, a source subsidy, a deadline, an advertising sensitivity, an appointment, would generate the pattern, and is that process observable rather than merely imaginable? Fourth, the counter evidence: what would count against the claim, and has it been looked for, including the outlet’s own record of publishing against the interest alleged? A claim that survives the four checks deserves print and names its level. A claim that fails them may still be a legitimate complaint about accuracy or taste, but it has not yet become an analysis of ideology, and calling it one borrows authority the work has not earned.

Q: Why does the propaganda model predict omissions better than invented stories?

Because invention is expensive and omission is free. A fabricated story can be checked, denied, litigated, and corrected, and it exposes the fabricator to professional ruin, so rational institutions rarely need it and the model does not predict it. Omission carries none of those costs. The unassigned story, the unasked question, the victim whose death is never counted, leave no text to refute and no byline to blame, and their absence is invisible to an audience that cannot know what the day contained. The model’s filters all operate as selection pressures, on what is gathered, commissioned, promoted, and repeated, so its predictions naturally concern the boundaries of coverage: which events cross them, in what volume, with what vocabulary, and which never approach. That is why the model’s strongest evidence is comparative, counts of attention given to paired atrocities, paired elections, and paired dissidents, rather than exhibits of falsehood. Selection is where the power is, and omission is selection in its purest form.

Q: How do audience markets shape news ideology without direct censorship?

Through the price system rather than through prohibitions. A commercial news organisation sells its audience to advertisers, and the sale disciplines content in advance: subjects that depress the audience’s buying mood, investigations that threaten major advertisers, and treatments that alienate the affluent demographics advertisers pay to reach all become expensive in a currency the organisation counts daily. Editors internalise the discipline as professional judgement about tone, placement, and pursuit, so no censor is required and none can be produced for inspection. Audience measurement deepens the mechanism by defining which public counts: the measured, affluent, attentive audience is served, and unmeasured publics find their concerns systematically underproduced. The platform economy has intensified the same logic into metrics, with engagement data returning to the newsroom as a continuous verdict on every headline. In each form the shaping is real, directional, and accomplished entirely by incentives, which is why the tradition treats markets not as the alternative to control but as one of its instruments.

Q: What separates an ideology account of news from a complaint about inaccuracy?

The level of the claim and the status of the remedy. A complaint about inaccuracy identifies a false statement and is satisfied by a correction, after which the record stands straight. An ideology account is about statements that may all be true. It concerns the selection of which truths are gathered, the order in which sources are allowed to define them, the vocabulary in which they are framed, and the questions that structure the telling, and no correction column can reach it because nothing in it is, individually, an error. The two accounts also differ in what would refute them. Inaccuracy is refuted case by case, by documents. An ideology claim is refuted comparatively, by showing that selection does not lean as predicted across a corpus, or that the lean tracks an innocent variable better than the mechanism proposed. Confusing the two produces the commonest sterile exchange in the field, the critic alleging ideology, the newsroom producing its corrections record, and both leaving persuaded that the other has not understood the question.