Journalism made the world reportable. It built forms for turning testimony, documents, observation and argument into public accounts of events. The abundance produced by those forms has created another reportable object: the relations among the accounts themselves.
Prism calls the journalism of that object second-order news.
Aggregation is not the right name. Aggregation collects reports without describing their structure. Comparison comes closer, but often means little more than placing headlines side by side. Media criticism judges institutions and performances. Fact-checking tests propositions against evidence about the world. Summary takes many accounts and produces one more account of the event.
Second-order news does something categorically different. It treats a disclosed field of coverage as primary material and reports the structure found within it.
Second-order news is systematic reporting, based on a disclosed corpus, whose primary object is the structure of first-order coverage: the relations among published accounts, their evidentiary origins, their narrative forms and their patterns of circulation.
The definition names an object, a method and a limit. Remove any one of the three and the form collapses back into commentary or synthesis.
ORDERS OF CLAIM
The order of a claim is determined by what the claim is about.
A first-order claim concerns the event: American and Saudi ministers signed nuclear instruments; ships changed route after a Houthi warning; the House adopted a nonbinding resolution; Hugging Face detected an intrusion. Its authority depends on evidence directed towards the world—observation, testimony, documents, measurement and verification.
A second-order claim concerns the public accounts. In the nuclear map, 66 of 71 articles carried enough of the institutional sequence to require separation even while reports repeatedly called its stages “the deal.” In the Red Sea map, 57 of 74 articles carried the blockade declaration while 30 carried observed post-warning changes in shipping. In the cyber map, 19 of 20 articles' detailed mechanics descended chiefly from OpenAI's preliminary disclosure. The authority for these claims is evidence directed towards the coverage—article texts, source lineages, publication choices, timestamps and the relations among them.
The difference is simple and unforgiving. Sixty-six articles traversing a sequence establish that its shorthand circulated broadly. They do not make proposal, signature, review and entry into force the same institutional state. Nineteen articles relaying one company's mechanics establish the reach of its explanation. They do not constitute nineteen forensic reconstructions. One article carrying a contradiction establishes that the contradiction entered the observed coverage. It does not, by frequency alone, make the contradiction weak. An absent question establishes silence within a specified corpus. It does not reveal motive.
Second-order news becomes intellectually serious at precisely this boundary. It refuses to borrow authority from one order for a claim made in the other. It neither converts repetition into truth nor treats its dependence on published reporting as a reason to say nothing of consequence.
Facts about the coverage are facts. They simply have a different object.
RELATIONS ARE NEWS
The modern reader receives articles as isolated units. The most consequential properties of the field often exist between those units.
Carriage is the relation between a proposition and the articles that materially report it. It reveals reach, concentration and the shape of the common account.
Conflict is a relation of incompatibility. Two positions conflict only when they cannot both be correct under the same scope, referent and time. The category is narrower than political disagreement and more useful for being so.
Divergence is compatible difference with interpretive consequence. Accounts may choose different causes, scales, witnesses, analogies or temporal beginnings without formally contradicting one another. Divergence is where much of journalism's meaning is made.
Framing is the organising relation between selected material and the story it becomes. A signed instrument becomes a durable bargain or an agreement thrown back into uncertainty. A cyber incident becomes an autonomous escape or an operator's containment failure. A protest becomes a national movement, a crackdown, a clash or a still-fragile coalition. The frame is not merely the opinion layered on top of the facts. It governs which facts acquire narrative force.
Dependence is the relation between apparent plurality and common origin. A dozen publications may inherit the same wire copy, official briefing, press release, study or source interview. Their circulation is real. Their independence is not twelvefold.
Absence is a relation between a defined question and a defined population of articles. It is valuable only when the population is coherent and the missing material would change the reader's understanding. Otherwise omission becomes a theatre in which the analyst supplies whatever the coverage did not.
Sequence is the relation produced by the chosen clock and completion threshold. An act called retaliation, escalation or response contains an editorial decision about what came first. A noun such as deal, blockade or movement contains a decision about what stage has already been reached. Aligning clocks can resolve a false dispute or reveal a deeper one.
None of these relations belongs fully to any individual article. They emerge at field level. Once reported systematically, they provide information no isolated account can contain about itself.
THE CORPUS CREATES THE AUTHORITY
Every second-order finding is a function of its corpus.
The corpus identifies the population in which a proposition circulated, a frame recurred or a question remained absent. It determines whether a count is exhaustive or anecdotal. It reveals which languages, countries, genres and political traditions entered the comparison. It exposes the publications the analysis could not retrieve. It sets the domain over which words such as broadly, rarely and only have meaning.
A hidden corpus is not a minor methodological omission. It makes the principal claims of the form unreadable. “The media ignored” becomes an assertion without a population. “Widely reported” becomes a mood. “Consensus” becomes the analyst's impression of repetition.
A disclosed corpus converts those impressions into bounded propositions. Fifty-seven of 74 articles carried the announced blockade; 30 carried observed shipping responses. Nineteen of 20 cyber reports inherited the incident mechanics chiefly from one preliminary corporate disclosure; three materially restored operator responsibility to the account. Only one clearly distinguished the safeguard-reduced evaluation systems from public versions. The language is stronger because its reach is known.
Corpus design is therefore editorial power. Choosing outlets, time windows, article types and event boundaries determines the field that can appear. Second-order news must preserve that choice as part of the published work. The aperture is not outside the result. It is one of the conditions that produced it.
THE DISCIPLINE OF THE FORM
Second-order news is not defined by a visual interface or by the use of artificial intelligence. It is defined by a set of editorial obligations.
The corpus is disclosed. A reader can identify the population behind every claim of breadth, recurrence or absence.
Relations remain typed. Conflict is not used as a dramatic synonym for difference. Framing is not presented as factual error. Source dependence is not treated as proof of falsehood.
Circulation remains distinct from confirmation. Counts describe what travelled through the coverage. Independence and evidentiary origin are reported separately.
Attribution survives synthesis. An official estimate, an expert prediction and an observed fact do not merge into the publication's own voice merely because many articles repeated them.
Claims are inspectable. A consequential characterisation leads back to the article language, structure or source relation on which it rests. Evidence does not abolish interpretation; it gives interpretation an address.
Absence is bounded. The publication names the articles in which a question would reasonably belong and distinguishes silence in those articles from silence in journalism at large.
The map can remain unresolved. The form does not earn authority by closing every question. It earns authority by representing the structure the evidence supports, including fragmentation, asymmetry and uncertainty.
The instrument is accountable. Corrections, method changes, retrieval failures and audit results belong to the public record of the form.
These obligations amount to an editorial constitution. They distinguish second-order reporting from a digest that happens to read many sources.
BEYOND THE LEFT-RIGHT COMPARISON
Political orientation matters, but it is one explanatory variable among several. Beginning with an outlet label and reading every difference through it produces a caricature of the field.
Geography alters proximity and assumed knowledge. Genre alters what can fit: a live blog, wire dispatch, legal analysis and reported feature answer different editorial demands. Publication speed privileges official statements and inherited copy. Access determines which witnesses enter the account. Language determines which concepts survive translation. Image availability creates its own hierarchy. A newsroom's choice to reproduce a wire report can matter more to one story than its ideological reputation.
Second-order news therefore describes the accounts before assigning them an explanation. It asks what each report selected, inherited and organised. Political interpretation can follow where the pattern warrants it. It cannot substitute for the pattern.
This matters because a spectrum is often least informative where coverage is most stable. Outlets from different traditions may reproduce the same official briefing, the same agency images and the same wire chronology. A nominally plural field can narrow to one evidentiary spine. Elsewhere, two politically similar publications may diverge because one has local reporting and the other does not.
The point is not to depoliticise journalism. It is to describe power at the level where it actually enters the story.
NO VIEW FROM NOWHERE
A map of framing has a frame. Second-order news selects events, categories and relationships. It decides what counts as material and which differences deserve the reader's attention. Its prose has cadence, emphasis and argument. The form does not become objective by denying these facts.
Its claim to authority rests elsewhere: the object is explicit, the corpus is preserved, the relations are named, the evidence remains reachable and the limits are carried inside the claim.
This creates a different contract with the reader. Conventional news often asks the reader to trust an account whose reporting process remains mostly behind the finished prose. Second-order news can expose more of its construction because its evidence already consists of public accounts. The reader can inspect whether a quotation bears the assigned proposition, whether two positions truly conflict, whether a source lineage has been overstated, and whether an absence population makes sense.
The goal is not to mechanise disagreement until judgment disappears. The goal is to make editorial judgment contestable without making it disposable.
WHY THE FORM BELONGS TO THIS MOMENT
News abundance has outgrown the interfaces built to understand it.
Hundreds of accounts can appear within hours of a major event. Syndication and rapid republication create a visual impression of plurality. Official communications enter dozens of articles before independent reporting has time to develop. Search and social platforms atomise the field into links, posts and clips. Generative systems then compress the abundance into one answer, removing the relations that might have disciplined it.
The labour required to reconstruct those relations was once prohibitive at daily scale. Every article had to be read closely, its sources and claims retained, its differences compared, and every public characterisation checked against the original. A human research team could do this for a bounded study. Few publications could do it across a daily news agenda.
Machine reading changes the economics. Models can read and compare enough material for the field itself to become publishable. The same models are prone to fluent overreach, so scale must be paired with custody, independent challenge and deterministic verification. Artificial intelligence makes the form feasible. It does not define the form and it does not certify its results.
The deeper condition is cultural. Readers now inhabit an environment in which the existence of many accounts is obvious but their dependence, selection and relation are opaque. The public does not need another machine that can speak about the event. It needs institutions capable of showing how the speaking was assembled.
A FIELD LARGER THAN PRISM
Second-order news is proposed as a journalistic category, not a proprietary name for Prism's interface.
Other publications can produce it with different corpora, taxonomies and editorial traditions. They can reject Prism's distinctions, build better ones, combine field-level reporting with original investigation, or apply the form to science, law, markets and historical archives. A human newsroom can produce second-order news. A model-generated digest can fail to produce it.
The category succeeds if it gives editors and readers a precise way to demand more from comparison. Show the population. Separate circulation from confirmation. Name the relation. Preserve the origin. Attach the evidence. Publish the uncertainty at the resolution where it actually exists.
First-order journalism makes events public. Second-order news makes the structure of that public knowledge visible.
That structure is news too.