News Theory of Relativity | Frames, Sources & Reporting

One-sentence answer: the eduKateSG News Theory of Relativity proposes that a real event is not made relative by reporting, but every report reaches us through a positioned observation, limited access, source lineage, vocabulary, frame, compression and receiver—so disagreement can exist without all sides being false, while apparent agreement can be weaker than it looks when many reports descend from the same underlying source.

Reality is not relative. Our access to reality is positioned.

Imagine an explosion in a railway station.

A passenger hears the blast but cannot see its origin. A station camera sees one corridor but not the platform behind a wall. Police receive several witness reports. A hospital sees casualties, not the cause. A government spokesperson receives an operational briefing. A local journalist reaches the station twenty minutes later. A foreign correspondent watches verified and unverified video remotely. A wire service publishes the first cautious bulletin. Hundreds of websites rewrite or republish that bulletin. Social media users attach motive, blame and historical meaning before investigators have identified the device.

By the end of the first hour, there may be thousands of texts about one event.

But there were not thousands of independent observations.

That distinction is the starting point of this theory.

Why we need another model of news

Most media-literacy advice starts with sensible questions: Is the source reputable? Is the claim verified? Is the article news, analysis or opinion? Is the language loaded? Are multiple sources reporting the same thing?

Those questions matter. But modern information systems create a deeper problem.

We often confuse source count with independent evidence.

Ten outlets may look like ten confirmations while nine are downstream of the same wire report. Twenty analysts may cite ten papers that ultimately depend on the same dataset. Hundreds of posts may repeat one official claim. A search result page may show apparent consensus even though the visible pages are descendants of a small number of original observations.

This is not an argument against wire services, official sources, journalism, scholarship or consensus. It is an argument for provenance-aware evidence counting.

Research on textual reuse in journalism has demonstrated that news stories can share material through newswire copy, press releases, source-to-source copying and common-source overlap. Work on social learning similarly shows why repeated communication can cause people to double-count information that is not truly independent. In 2026, research on aggregating correlated expert judgments made the same mathematical point in another domain: a simple average can overstate evidence when experts share common information.

The News Theory of Relativity brings that problem together with framing, source criticism, VocabularyOS, NewsOS, Civilisational Relativity and World Return.

First boundary: this is not Einstein’s theory of relativity

The name is an analogy, not a claim that journalism obeys the equations of physics.

Physics uses reference frames with precise mathematical relationships. News does not provide that kind of universal transformation law. Human observers are not inertial frames. Political language is not spacetime curvature. A headline does not possess mass.

The analogy is narrower:

A measurement or description can depend on where, when and how the observer is positioned, without implying that the underlying reality does not exist.

That is the useful inheritance from the word relativity: declare the reference conditions before treating one view as the view from nowhere.

The central distinction: Event Reality is not Event Core

NewsOS already separates event from frame. The News Theory of Relativity adds a harder boundary.

Event Reality is what actually happened in the world.

Event Core is our narrowest currently defensible reconstruction of that event from available evidence.

These are not identical.

If an explosion occurred at 10:03:14, with a particular physical cause, sequence, location and set of consequences, reality had those properties whether or not any reporter knew them at 10:05.

The first news bulletin may correctly say only: “An explosion was reported at Central Station shortly after 10 a.m.; authorities have not yet announced the cause.”

That sentence is not the whole reality. It is a disciplined projection of the evidence available at that moment.

This distinction prevents a subtle mistake: imagining that careful neutral prose has somehow escaped observation limits. It has not. It is simply a better-calibrated representation.

The News Relativity Chain

A news object can be understood as a chain of transformations:

REAL EVENT → OBSERVATION → SOURCE CLAIM → VERIFICATION → SELECTION → VOCABULARY → FRAME → COMPRESSION → DISTRIBUTION → RECEIVER → MEMORY → ACTION

At every handoff, information can be preserved, added, removed, sharpened, delayed, distorted or misunderstood.

This does not mean every handoff corrupts truth. Professional journalism exists partly to make those handoffs more reliable. Reuters’ current standards emphasise accuracy, honest sourcing and correction, while the Associated Press requires attribution of reasonably disputable information and careful scrutiny when relying on anonymous material from other news organisations. Those practices are forms of calibration.

The theory asks us to inspect the entire chain instead of assigning a single trust label to the final page.

Nine reference coordinates of a news report

Before comparing two reports, pin their reference conditions.

Coordinate	Question
Position	Where was the observer relative to the event?
Time	When was the observation or report produced?
Access	What could the source actually see, know, measure or verify?
Role	Eyewitness, official, investigator, participant, analyst, journalist, institution, advocate?
Source lineage	Is this an independent observation, or downstream of another report?
Vocabulary	Which labels classify the event and actors?
Frame	Which aspects are selected, emphasised, contextualised or omitted?
Compression	What was removed to fit a headline, short bulletin, social post or summary?
Receiver	Who reads it, with what prior knowledge, language, expectations and risks?

A report is not disqualified because it has a position. Every report has a position.

The problem begins when position is hidden, incomparable positions are treated as equivalent, or position is mistaken for distortion before evidence shows distortion.

Position is not distortion

This is a major correction to older Tilt/Warp language.

A witness standing on Platform 2 is positioned. A journalist interviewing only people on Platform 2 may still be doing accurate reporting about Platform 2. A Ukrainian outlet covering a war from Ukraine is positioned. A Russian outlet is positioned. A Singapore newspaper covering regional shipping is positioned by geography, audience and institutional context. A financial newspaper is positioned toward economic consequences. A medical journal is positioned toward clinical evidence.

Position becomes distortion only when something the representation claims to preserve is materially bent.

So the upgraded Distortion Table becomes:

State	Meaning
Positioned	The report has a declared or inferable viewpoint, scope or access boundary. This is normal.
Selective	The report chooses some relevant facts, voices or contexts over others. Selection may be necessary; the question is whether it remains fit for the claim.
Tilted	Selection, wording, scale or omission pushes interpretation beyond what the evidence proportionately supports.
Warped	Multiple distortions interact so that the overall representation materially misdescribes relationships among event, cause, scale, actors or consequences.
Inverted	A label or claimed function persists while observable outcomes systematically run opposite to what the label claims to represent.

This table does not tell us automatically where an article belongs. It tells us what to test.

Framing is unavoidable—and still measurable

Communication research has studied framing for decades. Robert Entman’s influential formulation focused on selection and salience: communicators choose aspects of perceived reality and make some more prominent in a text in ways that can define problems, diagnose causes, make moral judgments or suggest remedies.

The News Theory of Relativity takes a compatible but operationally different approach. It does not try to abolish frame. It asks:

What was selected?
What was left outside the visible frame?
Which scale was chosen?
Which causal boundary was chosen?
Which actors were individuated and which were compressed into groups?
Which time horizon was used?
Which vocabulary carries moral or strategic weight?
Would the same event receive the same frame if actor identities were exchanged?

The goal is not a frame-free article. The goal is a frame-visible article.

One event can support multiple true descriptions

A common mistake is to assume that if two descriptions differ, one must be false.

Consider a factory closure.

“The company closed a factory employing 800 workers.”
“The company consolidated production into two automated facilities.”
“Eight hundred workers face redundancy after the plant closure.”
“The restructuring is expected to reduce annual operating costs.”
“The town is losing one of its largest private employers.”

All five sentences may be true.

They are not interchangeable.

Each selects a different receiver, scale, causal interest and consequence. Relativity begins by recognising that a high-dimensional event can cast several valid shadows.

The calibration problem is not “Which one is neutral?” It is “What does each projection preserve, what does it omit, and what claim is being built from it?”

The Million Photographers problem

Imagine one million photographers standing in a stadium.

If all one million stand in nearly the same seat and point in the same direction, they create one million photographs but not one million independent perspectives.

If ten photographers stand at field level, ten above the stadium, ten behind each goal, ten in the control room, ten outside in the streets and ten inspect the raw sensor data, a much smaller number of images may contain more independent information about the event.

News works the same way.

Volume of reports is not the same thing as diversity of observation.

This is why source lineage belongs near the centre of the theory.

The Source Independence Principle

Suppose fifty websites report that a minister will resign.

If forty-nine cite the same original anonymous-source story, the evidence is not equivalent to fifty separately placed reporters independently confirming the resignation from different knowledgeable sources.

AP’s standards explicitly recognise this practical problem: anonymous-source material from another news organisation requires careful scrutiny, and AP says it should make a bona fide effort to get such material on the record or at least confirm it through its own reporting. Reuters likewise emphasises honest sourcing and warns against describing one source as several.

The News Theory of Relativity generalises this into a source-lineage rule:

Corroboration strength depends not only on how many reports agree, but on how independently their evidence entered the information field.

This is not merely philosophical. Research on correlated expert judgments shows mathematically why shared information can be double-counted, while experiments on social learning show that people can inflate hearsay when they fail to account for communication pathways.

Source trees, not source lists

Traditional media literacy often encourages a source list.

The stronger approach is a source tree.

For each major claim, ask:

Who first observed or asserted this?
Who independently verified it?
Who merely repeated it?
Who transformed it through analysis or commentary?
Did several outlets use the same wire service?
Did several experts rely on the same dataset?
Did social-media accounts all originate from one video?
Are translations descendants of one original-language report?
Has the claim acquired apparent authority through repetition rather than new evidence?

A source tree reveals whether the information field has many roots or many leaves on one root.

The Wire Echo problem

News agencies are essential parts of global reporting infrastructure. Their work allows events in one location to reach thousands of publishers quickly. The problem is not the existence of wire copy.

The problem is epistemic miscounting.

Academic work has developed computational methods for detecting textual reuse and distinguishing newswire copy, public-relations material, common-source overlap and source-to-source copying. A 2025 systematic review of churnalism likewise notes the importance of newswire reuse in the contemporary news environment.

So when search produces fifty near-identical accounts, NewsOS should not conclude “fifty confirmations.” It should cluster them by provenance.

The public-facing principle is simple:

Count roots before counting headlines.

The Shared Dataset problem

The same problem appears outside journalism.

Suppose fifteen economic analyses conclude that household wealth increased. If all fifteen analyses use the same national survey, they may offer fifteen independent interpretations but not fifteen independent measurements of household wealth.

Suppose ten conflict trackers cite the same satellite-image provider. Ten analytical teams may still add genuine value, but the underlying observation channel is shared.

Suppose historians across several countries cite later works that all depend on one surviving imperial archive. Citation diversity may exceed archive diversity.

Relativity therefore distinguishes:

observer independence — different people or systems witnessed or measured separately;
data independence — different analyses use genuinely different underlying evidence;
institutional independence — sources do not share the same organisational control;
method independence — different measurement approaches can fail differently;
interpretive independence — analysts are not simply copying one explanation;
publication independence — different outlets are not merely republishing a common upstream text.

None of those forms of independence is automatically sufficient. Together they help us estimate how much genuinely new information a new report contributes.

Consensus is evidence—but consensus has geometry

This theory is not anti-consensus.

When many independent observers using strong methods converge on the same claim, that convergence can be powerful evidence.

But the strength of consensus depends on its geometry.

Ask:

How many independent observational roots exist?
How diverse are the methods?
How diverse are the archives and languages?
How much of the apparent agreement comes from citation or wire descent?
Were dissenting observations genuinely tested or merely outside the dominant retrieval field?
Did several sources converge before communicating with one another, or after one influential account became dominant?

A mature consensus usually survives provenance-aware checking better than a superficial one.

This is how we protect frontier thinking without becoming contrarian for its own sake.

The Consensus Independence Test

When a claim appears overwhelmingly accepted, run five checks.

Root count: how many independent evidence origins support the claim?
Method diversity: do different methods converge?
Frame diversity: do sources from different institutional, linguistic or ideological positions still agree on the event core?
Dependency trace: how much agreement is inherited from the same upstream source, dataset, model or authority?
Falsifier survival: have plausible counter-observations been tested strongly enough to challenge the claim?

The result is not a “consensus score.” It is a better description of what the consensus is made of.

The Observer Position problem

An eyewitness can be both extremely valuable and seriously incomplete.

A person close to an event may have high spatial detail and poor global context. A remote analyst may have broad context and no direct sensory access. A government may possess classified information but also strategic incentives. An activist may have deep community access but a declared normative commitment. A company may know its internal operations better than outsiders while having strong reputational incentives.

News relativity therefore refuses the lazy equation:

Position = bias = falsehood.

Instead:

Position defines an observation envelope.

The correct question is: What does this position make unusually visible, and what does it make difficult or impossible to see?

The Time Relativity problem

A report produced ten minutes after an event and a report produced ten months later should not be judged as though they had the same information environment.

Early reporting has one great advantage: proximity to the event.

It also suffers from severe disadvantages:

cause may be unknown;
casualty counts may be unstable;
images may be unverified;
officials may possess partial information;
motive is often speculative;
rumour spreads quickly;
the significance of the event is not yet visible.

Later reporting gains documents, investigations, comparative context and hindsight. But it can also acquire new distortions: political memory, myth, selective archiving and outcome bias.

So the theory does not assume later means truer. It says every report has a time coordinate that should travel with it.

Breaking news and matured news are different evidence objects

Breaking report	Matured report
High proximity, high uncertainty	Lower proximity, potentially stronger verification
Fast claim turnover	More stable claim ledger
Heavy reliance on first responders and officials	Broader documentary and specialist evidence
Visual verification often incomplete	More time for geolocation, metadata, comparison and forensic checks
Meaning and motive provisional	Causal interpretation may be better supported—but can still be contested

Reuters’ recent explanation of visual verification illustrates how contemporary newsrooms address this problem: source contact, metadata, public-data comparison and detection tools are used to authenticate images and videos, while final decisions still require journalistic judgment.

The Zoom Relativity problem

The same event changes meaning when the zoom changes.

A school closure can be:

a family logistics problem at household scale;
a neighbourhood access problem at local scale;
a budget decision at agency scale;
a demographic trend at national scale;
a cultural continuity problem for a minority-language community;
part of a global rural depopulation pattern at comparative scale.

None of those scales is automatically the correct one.

Distortion appears when an article changes scale asymmetrically. A local crime committed by one member of Group A is attributed to Group A as a whole, while a similar crime by Group B is treated as an individual act. A policy success is credited to a civilisation; a failure is assigned to a local official. Or the reverse.

This is scale asymmetry, and NewsOS should test it explicitly.

The Symmetry Test

Take a report and exchange actor identities while preserving the event structure.

Would the same outlet likely use the same:

verb?
headline order?
motive attribution?
scale?
historical comparison?
legal terminology?
moral adjectives?
photograph type?
amount of contextual explanation?

Failure of symmetry does not prove bias automatically. The actors may genuinely differ in legal status, history, capability or evidence.

But the test identifies where explanation is required.

Vocabulary Relativity: the same label can move between contexts

VocabularyOS adds another dimension.

Words do not arrive with one context-free political meaning. “Rebel,” “insurgent,” “militant,” “terrorist,” “resistance fighter,” “separatist,” “freedom fighter” and “armed group” can overlap while carrying very different legal, moral and strategic implications.

The theory therefore avoids the idea of an eternal zero-tilt dictionary word.

The stronger baseline is:

semantic alignment relative to a declared referent, context, legal category, time and evidence state.

A label can be accurate in one context and misleading in another. A word can be legally defined yet rhetorically loaded. A term can change meaning over decades. Different communities can use the same token with different compression.

VocabularyOS should therefore ask not merely “Is this word loaded?” but:

What referent is being named?
Which definition is active?
Which alternative labels are plausible?
Which properties does the chosen label foreground?
Which properties does it hide?
Does the same outlet use comparable labels symmetrically?
Did the meaning of the token drift over time?

Compression Relativity: every summary is a lossy projection

A 20-minute press conference becomes a 600-word story. The story becomes a 70-character push notification. The notification becomes a social-media caption. An AI system produces a three-bullet summary. A reader remembers one phrase.

Each stage compresses.

Compression is necessary. Human attention is finite. The problem is not loss itself but uncontrolled loss.

For each compression, ask:

What must survive for the summary to remain faithful?
Which uncertainty markers were dropped?
Which actor attribution disappeared?
Which caveat was removed?
Did a conditional claim become categorical?
Did “may” become “will”?
Did “officials said” become an unattributed fact?
Did a disputed number become a settled number?

This is where NewsOS intersects with the wider eduKate Lossy work.

Receiver Relativity: the article is not finished when it is published

News reaches a receiver with prior knowledge, language, culture, identity, fear, expertise and goals.

The same sentence can therefore land differently.

“The central bank tightened policy by 50 basis points” is compressed but legible to a financial professional. A general reader may understand little. “The central bank sharply raised interest rates” is more accessible but adds a relative judgement through “sharply.” “Borrowing just became more expensive” moves toward receiver consequence while losing institutional detail.

None is automatically wrong.

Receiver relativity asks what translation occurred between specialist reality and reader representation.

Narrative Relativity: not every site is trying to write a straight news article

This matters enormously.

A newspaper editorial is supposed to argue. A party website is supposed to advocate. A company press release is supposed to present the company’s position. A human-rights organisation may document harm through a normative rights framework. A military communiqué communicates operationally and strategically. A financial research note gives an investment interpretation. A satire site uses irony. A personal essay foregrounds lived experience.

Calling all of these “biased news” misses their genre.

News Theory of Relativity therefore requires a genre declaration before a neutrality test.

Genre	Primary job	Calibration question
Straight news	Report and verify	Are claims, facts, uncertainty and framing kept proportionate?
Analysis	Interpret mechanisms and implications	Are inference and evidence clearly separated?
Opinion/editorial	Argue a position	Are factual premises accurate and counterarguments represented fairly enough?
Official statement	State an institution’s position	Is it being read as a claim from an actor rather than neutral event reality?
Advocacy	Advance a cause or protect a constituency	Are evidence, norms and selected cases distinguishable?
Research	Generate or test knowledge	What methods, assumptions, sample and uncertainty govern the claim?
Eyewitness testimony	Describe experienced reality	What could the witness perceive, and what lies outside that envelope?

Neutrality is not achieved by rewriting every genre into identical prose.

Neutrality is achieved by preserving the genre and position while preventing it from being silently mistaken for another layer.

The Frontier Problem: what if the accepted frame is wrong?

This is where the theory becomes most important for research at the edge.

Consensus protects us from many errors. But all mature knowledge was once incomplete. Some true claims were initially minority claims. Some accepted explanations were later revised. Some evidence was absent because no instrument could detect it. Some archives survived unevenly. Some populations were poorly represented in the record.

The correct response is neither obedience to consensus nor automatic rebellion against it.

The theory creates a separate state:

FRONTIER — a hypothesis that is not yet established, but is sufficiently structured, evidence-aware and falsifiable to remain visible without being promoted to fact.

A proper frontier record should contain:

the observation that triggered the idea;
the proposed mechanism;
existing evidence consistent with it;
evidence that conflicts with it;
known prior work that overlaps;
competing explanations;
what has not yet been demonstrated;
the observation or experiment that could falsify it.

This protects original thinking from premature deletion while protecting readers from premature certainty.

The Four Epistemic States

State	Meaning
SUPPORTED	Evidence is strong enough for the stated scope and confidence.
CONTESTED	Substantial evidence or interpretation conflicts across credible lines.
FRONTIER	A structured hypothesis is worth preserving but remains unproven.
UNKNOWN	Evidence is insufficient to support a useful conclusion.

These states are more useful than forcing every question into true/false before the evidence has matured.

The Inversion Test becomes stronger under relativity

Older Inversion language can become political if the model begins with a morally positive word and assumes it knows the correct route.

The stronger method begins with a declared function.

Example:

Claimed function: Policy P is intended to reduce civilian exposure to Threat X over period T.

Then measure:

exposure to X;
new harms created by P;
distribution of benefits and harms;
alternative explanations;
counterfactual baseline where possible;
whether the effect persists across time.

If the observable outcome consistently runs opposite to the claimed function after plausible confounders are considered, the system may be functionally inverted.

The word does not convict the policy.

World Return does.

The News Distortion Matrix

Instead of one bias score, use a matrix.

Dimension	Low distortion	Possible distortion signal
Source lineage	Independent roots visible	Many reports descend from one unacknowledged root
Attribution	Claims remain attached to claimants	Actor claim becomes unattributed fact
Vocabulary	Labels fit declared referents	Asymmetric or euphemistic labelling
Scale	Claim stays proportional to evidence	Local event inflated to civilisation-wide essence
Time	Uncertainty matches evidence maturity	Early interpretation hardens prematurely
Omission	Relevant counter-context acknowledged	Missing context reverses likely interpretation
Visuals	Current, representative, clearly sourced	Archive or exceptional imagery presented as representative
Motive	Inference confidence stated	Motive written as fact without sufficient evidence
Genre	News/analysis/opinion role visible	Advocacy presented as neutral reporting
Receiver	Consequences and access differences visible	Aggregate output hides who receives harm or benefit

The matrix is diagnostic. It does not automatically condemn a source. It identifies where the source needs calibration.

The Relative News Packet

A mature NewsOS package should preserve several layers instead of flattening them into one paragraph.

Reality target: what event or condition is being reconstructed?
Event Core: narrowest defensible description at this time.
Verified details: independently supported observations.
Claims by actor: statements still attached to their sources.
Source tree: independent roots versus downstream repetitions.
Frame map: major serious ways the event is being interpreted.
Vocabulary map: disputed or high-load labels.
Omission map: relevant context absent from major packages.
Time state: breaking, developing, matured, historical.
Receiver map: who experiences direct consequences.
Epistemic state: supported, contested, frontier or unknown.
Attribution boundary: what larger conclusions are permitted right now.
Falsifier: what new observation would materially change the package.

This packet does not create perfect neutrality.

It creates traceable relativity.

Worked example: one protest, five reports

Suppose 20,000 people gather in a capital city after a disputed law is passed. Most remain peaceful. A smaller group clashes with police late in the evening. Police arrest 230 people. Several officers and protesters are injured.

Five outlets publish:

A: “Mass democratic protest challenges controversial law.”
B: “Violence erupts as demonstrators attack police.”
C: “Government faces largest opposition mobilisation in years.”
D: “Police restore order after unlawful gathering turns violent.”
E: “Twenty thousand gather; clashes involving a smaller group lead to 230 arrests.”

News relativity does not simply declare E neutral and the others biased.

It decomposes them.

Report	What it preserves	What it compresses or frames
A	Scale and political function of main gathering	May underweight late violence; “democratic” and “controversial” require context
B	Real clashes and police injury	Headline may make a minority action represent the whole crowd
C	Comparative political scale	Requires historical data and may imply a national opposition story
D	Legal and policing frame	“Restore order” foregrounds state function; “unlawful” depends on law and legal context
E	Event-core quantities and distinction between crowd and subgroup	Still omits why people gathered, what law was disputed and whether police action was proportionate

The best final package may use E as the event spine while preserving the legitimate questions raised by A–D in separate layers.

That is not false balance. It is structured decomposition.

Worked example: fifty headlines, three observational roots

Suppose a rumour says a major bank is facing a liquidity crisis.

Within one hour:

one financial newspaper cites two unnamed bankers;
a wire service reports that story and adds a bank statement;
thirty websites republish the wire;
ten social accounts cite the thirty websites;
five analysts discuss the market reaction;
four local reporters independently observe unusually long queues at branches.

The information field contains fifty-plus publications, but the important roots may be only:

the two unnamed bankers;
the bank statement;
independent branch observations;
market price and liquidity data.

News Theory of Relativity says the correct evidentiary object is not the headline count. It is the provenance graph.

False balance versus frame diversity

There is an obvious danger here. If we insist on “many perspectives,” do we give unsupported claims equal weight with strongly evidenced ones?

No.

Frame diversity is not vote equality.

A scientifically unsupported claim does not earn 50% of a report because someone holds it. A fabricated video does not become a “perspective.” A government denial does not cancel physical evidence. A viral rumour does not become equal to a court judgment.

The calibration rule is:

Preserve relevant frames; weight claims by evidence.

This prevents relativity from collapsing into relativism.

The Asymmetric Skepticism problem

People often demand extraordinary verification from sources they distrust while accepting favourable claims from preferred sources with little scrutiny.

News relativity therefore needs a skepticism symmetry rule:

Comparable claims should face comparable evidentiary burdens unless a relevant difference justifies the asymmetry.

This does not require equal trust in all institutions. Track records, access, methods and incentives differ. It requires us to make the reason for unequal trust explicit.

The Archive Relativity problem

News becomes history.

But what survives is not a neutral sample of everything that happened.

Powerful institutions tend to produce more records. Literate bureaucracies leave different traces from oral communities. State archives preserve different objects from family memory. Wealthy organisations retain documents that vulnerable groups may never have produced or safely stored. Digital platforms can preserve enormous volumes while also deleting or de-ranking material.

So a future historian can face a cruel inversion: the group with the greatest archival mass can appear to possess the greatest historical reality.

Civilisational Relativity already identifies archive asymmetry. News Theory of Relativity connects that insight to live reporting: today’s source inequalities become tomorrow’s evidence inequalities.

The Search Relativity problem

Modern readers rarely inspect the whole information field. Search and recommendation systems pre-select it.

A search engine, social feed or AI retrieval system changes:

which sources are visible;
which languages are represented;
which pages rank highly;
which recent reports displace older corrections;
which summaries become the next generation’s upstream sources.

This creates a second-order reference frame: not only “What did sources say?” but “Which sources did the retrieval system make available to the reader or AI?”

That means search results should not be interpreted as a referendum of reality. Retrieval is a selection process.

The AI Summary Relativity problem

AI systems introduce another compression layer.

An AI may read ten articles, but if eight descend from the same source, the model can mistake repetition for corroboration. It may flatten genre, converting editorial interpretation into apparent fact. It may merge carefully attributed claims into one synthetic sentence. It may optimise for concise helpfulness and silently remove uncertainty.

A provenance-aware AI should therefore preserve:

source lineage;
claim attribution;
independent roots;
genre;
uncertainty;
frame diversity;
time state;
disagreement.

The machine should not say “multiple reports confirm” until it has tested whether those reports are actually independent enough to count as multiple confirmations.

A compact formal model

The theory can be written conceptually without pretending to possess a universal physical equation.

Let a report R be a projection of an event reality E under a reporting frame F:

R = Projection(E | position, time, access, lineage, vocabulary, frame, compression, receiver)

The purpose of calibration is not to reconstruct E perfectly from one R. It is to combine multiple reports while estimating their dependencies and preserving uncertainty:

Best current reconstruction = calibrated synthesis of evidence roots, not a vote over surface reports.

That sentence is the mathematical heart of the theory even though the exact weighting must remain domain-specific.

What this theory can detect

echo consensus — many publications, few independent roots;
frame convergence — different sources use the same interpretation despite different evidence bases;
frame divergence — sources agree on event core but differ sharply in meaning;
source laundering — a claim loses its original attribution as it travels;
certainty inflation — tentative language becomes categorical through repetition;
scale warp — local facts become civilisation-wide claims;
genre collapse — opinion, advocacy or official statements are read as straight news;
archive gravity — high-volume record producers dominate historical reconstruction;
retrieval gravity — highly ranked pages make one frame look universally accepted;
semantic inversion — claimed function and observed receiver outcome move in opposite directions.

What this theory cannot do

A strong theory states its limits.

It cannot recover evidence that never existed or has been destroyed.
It cannot guarantee that an “independent” source is actually independent when hidden coordination exists.
It cannot infer motive reliably from language alone.
It cannot produce one universal numeric bias score for all journalism.
It cannot turn equal treatment into fairness when underlying cases differ materially.
It cannot replace specialist knowledge required to evaluate technical claims.
It cannot eliminate judgement from journalism or history.
It cannot prove an edge hypothesis merely because mainstream sources share a frame.
It cannot make a false claim true by calling it a perspective.

Most importantly, it cannot make eduKate the final authority.

The anti-capture locks

To keep News Theory of Relativity useful in the future, the following rules should remain fixed unless a stronger replacement is demonstrated.

Reality lock: the event exists independently of the report.
Representation lock: Event Core is a reconstruction, not Event Reality itself.
Position lock: position is normal and must not be confused automatically with distortion.
Genre lock: advocacy, opinion, official statements and straight news retain different jobs.
Provenance lock: count independent evidence roots before counting reports.
Attribution lock: claims remain attached to claimants until independently established.
Symmetry lock: comparable structures face comparable labels and evidentiary burdens unless relevant differences justify asymmetry.
Scale lock: local evidence does not expand automatically to a civilisation-wide claim.
Receiver lock: aggregate description cannot erase who experiences outcomes.
Frontier lock: novel hypotheses may remain visible without being promoted to established fact.
Unknown lock: missing evidence is not evidence of absence unless the search process makes that inference valid.
Owner lock: specialist domains retain authority over their mechanisms and evidence.
Falsifier lock: serious claims identify what could make them change.
World Return: reality is allowed to overrule the model.

A practical News Relativity audit

When reading an important story, ask these questions in order.

What real event or condition is being described?
What is the narrowest defensible Event Core?
Which details are independently verified?
Which statements are actor claims?
How many genuinely independent evidence roots exist?
What is the source tree?
What position and access does each root have?
What time coordinate does each report have?
Which labels carry semantic or moral load?
What frame is each outlet applying?
Which relevant contexts are omitted?
Are comparable actors treated symmetrically?
Has local evidence been inflated to a larger scale?
Is the genre being preserved?
What did compression remove?
What do receivers actually experience?
What is supported, contested, frontier or unknown?
What evidence would change the current conclusion?

How this connects to VocabularyOS

VocabularyOS watches the linguistic surface and the semantic route.

News Theory of Relativity adds the reference conditions around that language.

Together they can ask:

Is the word aligned with the declared referent?
Does the word change meaning across sources?
Is the same actor labelled differently in another frame?
Did a euphemism remove consequential information?
Did translation alter the semantic field?
Did a positive word survive while its receiver outcome inverted?

This turns the old Zero Tilt → Tilt → Warp → Inversion ladder into a more defensible process beginning with Positioned and Selective.

How this connects to NewsOS

NewsOS already separates:

Event Core → Claim Field → Frame Field → Incentive Field → Attribution Layer.

News Theory of Relativity adds four protections around that stack:

Event Reality sits outside and above Event Core.
Source lineage prevents repeated reports from masquerading as independent evidence.
Reference coordinates make observer position explicit.
Frontier states preserve unresolved ideas without forcing them into current consensus or premature fact.

How this connects to Civilisational Relativity

Civilisational Relativity asks how archives, prestige, naming, zoom and inherited standards bend civilisation-scale interpretation.

News Theory of Relativity is the live-event sibling.

One works across large historical and civilisational frames. The other works at the intake where events first become reports, reports become narratives, and narratives become future archives.

Together they create a continuous path:

EVENT → NEWS → ACCEPTED REALITY → MEMORY → ARCHIVE → HISTORY → CIVILISATIONAL INTERPRETATION

Distortion introduced early can propagate through the entire chain. Correction introduced early can propagate too.

How this changes the Civilisation rebuild

When we rebuild a Civilisation article, we should not ask only, “What does current scholarship say?”

We should ask:

Which claims are direct observations or well-supported historical reconstructions?
Which are inherited classifications?
Which archives dominate the picture?
Which societies are under-described because their records survive differently?
Are multiple modern sources genuinely independent?
Does one civilisation receive broad continuity while another is fragmented into episodes?
Are we using equal zoom?
Is an eduKate model genuinely new, or a renamed version of existing work?
If genuinely new, what evidence supports it and what would falsify it?

This makes the project both more conservative and more adventurous.

More conservative because we make fewer unsupported claims.

More adventurous because we no longer have to delete an edge hypothesis merely because the present literature has not named it.

Reason for Existence

The reason this theory exists is not to teach readers that all news is biased, all institutions are untrustworthy, or truth is merely perspective.

It exists to do almost the opposite.

It tries to preserve a hard reality underneath a soft and changing information field.

It gives us permission to say:

this happened;
this part is verified;
this is what Actor A claims;
this is how Outlet B frames it;
these twenty reports share one root;
these four observations are independent;
this interpretation is plausible but contested;
this eduKate idea is frontier, not established;
this question remains unknown.

That vocabulary is slower than certainty.

But civilisation needs the ability to remain uncertain without becoming blind, and to remain open to new ideas without becoming gullible.

World Return

Every report, model, frame and theory eventually returns to the world.

If fifty reports trace back to one false source, fifty repetitions do not save the claim.

If one unpopular witness provides verifiable evidence that overturns a comfortable consensus, popularity does not defeat the evidence.

If an eduKate frontier hypothesis repeatedly survives hostile comparison, gains independent evidence and predicts observations better than the older model, its status should rise.

If it fails, it should fall.

If the event was misdescribed, correct the Event Core.

If the frame was unfair, recalibrate the frame.

If the source tree was wrong, rebuild the provenance.

If the theory itself becomes the distortion, retire the theory.

Reality is not required to agree with the news, the consensus, the algorithm, the archive—or eduKate.

News Theory of Relativity | One Event, Many Frames, Shared Sources and the Geometry of Reporting