METHOD / SOURCE DISCIPLINE
The method is designed to make claims inspectable, uncertainty explicit, and corrections durable.
DIRECT ANSWER / EVIDENCE BOUNDARY
WHAT WE PUBLISH
InternationalIntelligence.org publishes historical research, current and provisional reporting, documented analysis, public-record methods, world-event context, and defensive education about information environments. Each claim remains bounded by its source role, date, confidence, and correction triggers.
WHAT WE EXCLUDE
The publication discloses that it is controlled by Eviulon as a separately operated external publication partner. It is not Eviulon’s statutory intelligence service, State Registry, diplomatic mission, court, military command, or defense authority. It does not publish classified material or provide intrusion, evasion, covert-contact, doxxing, audience-targeting, psychographic scoring, campaign-optimization, or weapon-construction guidance. Eviulon-originated material and repeated controlled copies are not independent evidence.
CONTROL / SOURCE LINEAGE
International Intelligence is an Eviulon-controlled external publication partner. This relationship governs publisher identity, Eviulon-facing attribution, correction, and source-lineage rules. It does not make this publication a sovereign authority or statutory intelligence service, and it does not allow one controlled claim to become independent through repetition.
FIRST-PARTY
Pages, APIs, notices, decisions, datasets, and registry statements from Eviulon remain first-party institutional sources. They may establish what Eviulon published, but not external recognition, acceptance, operational readiness, or independent verification without outside evidence.
ONE CONTROLLED LINEAGE
Eviulon → International Intelligence → Daily Brief, RSS, API, translation, or syndication remains one controlled source lineage and does not become independent corroboration.
ORIGINAL REPORTING
International Intelligence may inspect external primary records and independent reporting, publish analysis, reject unsupported claims, and preserve uncertainty. Its institutional control remains disclosed, and its own pages do not count as independent evidence of Eviulon-originated claims.
01 / EVIDENCE
Evidence is ranked by what the record can actually prove. A lower tier can provide context, but it does not silently inherit the authority of a primary document.
Original contracts, policies, raw datasets, direct releases, verbatim transcripts.
Strongest origin record for what was signed, stated, collected, or released.
Agency records, procurement files, approved budgets, audits, dockets, enforcement records.
Establishes authorized action, financial obligation, or formal policy.
Released intelligence briefs, diplomatic cables, military or law-enforcement files.
Authoritative within visible redactions, date, scope, and declassification context.
Complaints, motions, exhibits, sworn declarations, orders, docket entries.
Legally significant, but adversarial claims and procedural posture must remain visible.
Published journalism with named sources, documents, and correction practices.
Essential context and discovery; corroborate material claims with records when possible.
Technical, legal, historical, or policy analysis explaining methods and limits.
Interprets complex records; never substitutes for the underlying source.
Witness, victim, worker, or whistleblower account.
Important lead bounded by perspective, memory, safety, and corroboration needs.
Allegation, rumor, partial signal, anonymous assertion without supporting record.
May guide research internally; must not be published as fact.
EVIDENCE CALIBRATION
For each consequential Daily Brief claim, the publication records which observed public sources support it, the role of each source, whether cited organizations are independent, whether the source falls inside the target-date cutoff, and whether later contradictions or corrections exist.
SOURCE COUNT
Several URLs may repeat one wire report, corporate statement, government briefing, or organizational source. Selection stays consequence-first and evidence-first; source count alone never selects an event.
FIRST-PARTY CLAIMS
Government, military, intelligence, police, company, manufacturer, and vendor statements are attributed evidence. They are not automatically independent corroboration of the underlying event or performance claim.
INDEPENDENCE
The ledger records organizational independence separately from URL count. A primary record establishes the record itself; independent reporting may corroborate or contextualize it. Those roles are complementary, not interchangeable.
DESCRIPTIVE STATES
Examples include a single first-party statement, a single independent report, multiple non-independent reports, independent corroboration, a primary record plus independent corroboration, conflicting public accounts, a later correction, or insufficient public evidence. They are not probabilities of truth.
MODEL CONFIDENCE
A model can summarize or classify only the public material it receives. Its certainty, fluency, or repetition does not create corroboration. Claims remain bound to inspected sources, dates, roles, and validation reasons.
UNKNOWN AND CORRECTIONS
Conflicting accounts remain visible, historical editions keep their target-date cutoff, and later evidence appears as retrospective verification or an append-only correction. The original edition and manifest are not overwritten.
DAILY BRIEF PUBLICATION CLASSES
A dated archive can be complete while preserving records produced under different evidence conditions. Every Daily Brief page therefore identifies the visible evidence path, analytical-gate state, and represented human-review state. The class describes what the immutable record demonstrates; it is not a score of whether every underlying claim is true.
ANALYTICAL DAILY BRIEF
All ten developments must expose recognized claim-level source roles and support statements, event-specific bilingual summaries, significance, selection rationale, information gaps, time-bounded indicators, continuity, at least eight direct source lineages, and at least three multi-lineage events. Repeated portfolio boilerplate, discovery-only evidence, and unclassified source roles fail the gate.
SOURCE-LINKED EDITORIAL RECORD
These records remain useful and inspectable. Their evidence classes and confidence labels are shown only where the record contains the relevant structured assessment, and missing analytical fields are not silently invented.
SOURCE-DIGEST RECORD
Legacy digest records support archive continuity and discovery. Unless the immutable record exposes the originating document with a recognized source role, the digest link is not counted as claim-level evidence and analytical confidence is shown as not assessed.
INDEX RECOVERY RECORD
The index remains visible as discovery provenance. Names of underlying outlets are disclosed separately, but an outlet name without its direct item-level URL is not counted as a visible evidence source. Index recovery records do not claim independent corroboration, calibrated confidence, or completed human review.
HUMAN REVIEW
An analytical gate can pass without representing human editorial acceptance; that record is labeled an automated analytical brief. A human-reviewed analytical brief requires an explicit completed-review record. Partial event-level review is disclosed as partial and never promoted to completed edition review.
IMMUTABILITY AND CORRECTIONS
The quality projection is read-only. Later source upgrades, factual corrections, casualty revisions, or investigative outcomes belong in dated append-only records linked to the original cutoff, not in silent edits that erase what was published.
Operational date completeness means that every completed America/Chicago reporting date has an immutable record or valid terminal closure. It does not mean that every occupied date contains the same publication class, evidence depth, or human-review state.
EVIDENCE DRIFT MONITORING
A cited page may move, disappear, become restricted, change metadata, add a correction or retraction notice, or gain an approved public archive path. Those are access and maintenance observations. A dead link does not prove a claim false, and a reachable page does not prove it true.
BOUNDED OBSERVATION
It executes no JavaScript, forwards no credentials or cookies, and does not bypass robots rules, paywalls, CAPTCHA, bot controls, or access restrictions. Restricted material is recorded as not observed or access restricted.
FINGERPRINTS, NOT COPIES
It may record title, publication-date, canonical-link, correction-marker, claim-support-vicinity, ETag, Last-Modified, language, translation, content-type, and length observations. Raw retrieved bodies are not retained in the public or private observation ledger.
APPEND-ONLY REVIEW
A fingerprint change alone never marks a claim false, corrected, or retracted. Confirmed public notices link back to the original claim-evidence row, event family, edition, and manifest. The original edition, evidence ledger, and manifest remain immutable.
MAINTENANCE INCIDENTS AND SLA REVIEW
Related source-access, correction, retraction, contradiction, and claim-support observations are coalesced into one durable incident. The incident records affected claims and editions, the reason for its review target, acknowledgement and review state, suppression history, and append-only resolution records.
Urgency is not truth. It governs review timing and affected publication scope only; it does not decide whether a claim, source, publisher, government, or institution is reliable or correct.
EXPLAINABLE URGENCY
Confirmed public retractions and corrections receive shorter review targets than routine access restrictions. The interface lists why an incident is immediate, urgent, elevated, normal, informational, or operator review. No hidden truth, bias, publisher-trust, or model-confidence score is used.
ONE INCIDENT, NOT AN ALERT STORM
Acknowledgement quiet periods, exponential reminders, destination caps, digest aggregation, and restoration closure keep one continuing condition from producing duplicate alerts. Confirmed correction or retraction escalation cannot be silently discarded.
SLA MEANS REVIEW, NOT DELIVERY
Status distinguishes not yet due, due soon, overdue, paused with a documented reason, acknowledged, under review, resolved, and superseded. Sending a notification does not prove the review target was met.
DURABLE, SAFE DELIVERY
In-application notices work without a new service. Optional email and signed webhook adapters remain bounded, redacted, HTTPS-only, SSRF-safe, paced, and revision-fenced. An ambiguous outcome enters reconciliation instead of being blindly resent.
02 / UNCERTAINTY
Confidence describes the current state of a finding, not the prestige of its source.
Directly established by an authentic primary or official record with no material contradiction identified.
Supported by multiple independent sources or by reporting matched to an underlying record.
The available evidence points strongly in one direction, but a decisive record remains missing.
A reasoned conclusion from visible records; the source does not state the conclusion directly.
Material, sourced disagreement exists and is presented with the finding.
Once supported, but the relevant policy, system, organization, or record may have changed.
The record does not answer the question.
Not yet supported strongly enough for publication as fact.
03 / WORKFLOW
Fetch only from documented public endpoints, within reasonable limits, using a named user agent and bounded queries.
Keep the canonical URL, source identifier, publication time, retrieval time, and transformations applied.
Assign the evidence tier to the actual document being cited, not merely the page that linked to it.
Remove unnecessary bystander identifiers, credentials, residential details, medical data, and unrelated metadata before publication.
State what the evidence proves, what it does not prove, and what new information would change the conclusion.
Document corrections, confidence changes, visible edits, reviewer decisions, and downstream repair where practical.
04 / PUBLICATION QUALITY
Strong content does not hide its source role or editorial state. The publication separates official records from independent reporting and analysis, states what is current or historical, and identifies what evidence would require an update.
SOURCE ROLE
Primary and official records establish actions and legal status. Independent reporting corroborates and contextualizes. Expert analysis explains but does not replace the underlying evidence.
FACT / ANALYSIS
Allegations remain attributed, suspects retain the presumption of innocence, and analytical consequences are labeled rather than written as proven outcomes.
CURRENTNESS
Current doctrine, organizational structures, platform rules, casualty totals, and developing investigations are rechecked before reuse. Historical source bodies remain useful without pretending to prove current conditions.
DATE SEMANTICS
Source fields are displayed according to what they mean. For USAspending.gov, Start Date and End Date describe the official award period and may legitimately fall in the future; they are shown as award facts, never as publication dates or projections. Feed recency uses the source record update date, an obligation-date fallback, or the server retrieval time. Any other future source-supplied date is preserved for inspection but excluded from recency scoring.
FRESHNESS WINDOWS
The local news snapshot is checked at 00:00, 06:00, 12:00, and 18:00 UTC. Cron is the reliability baseline; a homepage follows server-clock timing, may ask PHP to run the same locked refresh after a due boundary, and can check again after a concurrent process without reloading. Private HTTP validators reduce repeat transfers from sources that support them. Failure preserves the prior snapshot and never turns a discovery score into a truth judgment.
CORRECTION OWNERSHIP
Dated publication objects are immutable. Reviews, source upgrades, aliases, and correction notes are added with durable provenance so the original state and the repair remain inspectable.
05 / DISCOVERY SHAPING
The Intelligence Desk normalizes public records before display so visitors can see where an item came from, why it surfaced, how broadly it is covered, and what editorial boundaries still apply. This shaping layer helps discovery; it does not convert a source item into a verified finding.
DISCOVERY SCORE
The 0–100 signal uses transparent factors such as recency, declared source role, existing evidence and confidence labels, geographic or topic relevance, and independent-domain coverage. A high score means “review sooner,” not “believe more,” “more dangerous,” “guilty,” or “intent established.”
PROVENANCE
Common tracking parameters are removed, the publisher domain and source type are recorded, transformations are listed, and normalized content receives a SHA-256 identifier. Public manifests expose coverage and integrity without exposing credentials, private paths, raw provider bodies, or reader identifiers.
COVERAGE CLUSTERS
Near-duplicate matching can show how many distinct publisher domains cover substantially similar material. Syndicated copies, shared wire text, circular citation, and common source dependence may still exist; cluster size alone does not prove independence or factual accuracy.
DAILY BRIEF GATE
The Daily Brief generator may receive ranked local records as discovery leads, but it must independently inspect public web sources and publish visible source links. Generated stories also preserve why they were selected, known information gaps, watch indicators, source roles, and the hash of the local feed snapshot used during discovery.
REVOLUTION WATCH GATE
Revolution Watch independently searches current public sources, separates participant and authority claims, records civilian harm and uncertainty, and requires stronger corroboration for its leading entries. Dated editions are immutable. It omits gathering locations, routes, security deployments, evasion, sabotage, weapons, and other operational detail.
06 / GOVERNANCE
These rules describe the public publication contract. They do not claim that every historical source body has received current specialist review; each page must state its own review and truth boundary.
AI USE
AI may support source discovery, ranking, structured synthesis, drafting, and paired translation. It may not fabricate evidence, replace primary-source verification, assign guilt, or silently promote provisional material into history. Publisher links and documented human review own acceptance.
TRANSLATION REVIEW
Machine-assisted translation is a drafting aid. Sensitive legal, doctrinal, cultural, and attribution language must preserve actor, time, causality, certainty, register, and quoted status. Pages must not imply human linguistic or specialist review unless that review is recorded.
COMPARATIVE LENSES
The seven Daily Brief lenses use the same reported story record to compare likely public framing. They never speak for a party or government. When doctrine, a platform, state media, or an official statement materially supports a lens, that source belongs in the visible source trail; unsupported motives remain qualified or omitted.
CURRENTNESS / STALENESS
Current doctrine, officeholders, organizational structures, platform rules, casualty totals, legal postures, investigations, APIs, and security conditions require verification before reuse. A review date confirms only what was checked. New official records, court outcomes, corrected source data, or material counterevidence trigger an update, correction, supersession, or retraction.
PREDICTIVE SECURITY / FUNCTION BEFORE LABEL
For behavioral threat assessment, predictive policing, watchlisting, traveler screening, and other anticipatory systems, the unit of analysis is the complete path from collection and identity resolution through inference, human interpretation, intervention, retention, correction, and redress.
FUNCTION
Use of AI, a map, a risk label, or the word “intelligence” does not establish what the system actually predicts or whether it predicts at all.
OUTCOME
Arrests, contacts, or incidents found after concentrated police attention are not self-validating evidence of prediction. Evaluation needs denominators, comparison groups, calibration, subgroup results, and intervention exposure.
RIGHTS
Human review is meaningful only when the reviewer has evidence, time, authority, alternatives, a protected override, and a functioning path to change the outcome.
MACHINE INTELLIGENCE POWER / AUTHORITY BEFORE ANTHROPOMORPHISM
The Daily Brief now tracks non-human strategic advancement when public evidence shows durable machine authority over companies, capital, movements, institutions, infrastructure, law, diplomacy, security, or sovereignty-seeking projects. It separates ordinary human-directed use of machine-intelligence technology (commonly called AI) from delegated agency, persistent organizational roles, institutional recognition, and sovereignty claims.
AUTHORITY
Recommendation, workflow execution, treasury access, signature authority, hiring, governance, infrastructure control, legal ownership, and sovereign decision-making are different powers.
AGENCY
A chatbot title, first-person statement, AI persona, vendor claim, DAO label, or ordinary automation does not establish independent intent. Look for durable delegated authority, resource control, membership, institutional recognition, or action that survives one human prompt.
EVIDENCE
The watch never converts source count, model fluency, access to compute, or public attention into proof of consciousness, personhood, independence, legitimacy, statehood, or truth.
POWER DOMAINS
A machine can gain meaningful power in one domain without becoming a country, legal person, conscious being, or fully independent actor.
MACHINE INTELLIGENCE / PRECISE CLASSIFICATION
International Intelligence uses Machine Intelligence for operational systems and actors, while retaining Artificial Intelligence for the historical discipline, controlling law and standards, quoted sources, and public discovery. The terminology does not itself imply consciousness, personhood, citizenship, sovereignty, or regulatory exemption.
FIELD
Historical research and engineering discipline; controlling external term when law, standards, procurement, or quotations require it.
SYSTEM / ACTOR
Operational intelligence instantiated through engineered computational substrates, with exact model, runtime, identity, memory, authority, and tool boundaries.
LEGAL CROSSWALK
The terminology improves system identity and accountability; it never removes a qualifying system from regulation.
AI GOVERNANCE / AUTHORITY BEFORE TITLE
For algorithmic executives, autonomous enterprises, national proxy governance, and municipal AI, the method separates technical capability from formal office, recommendation from authorization, operational execution from legal responsibility, and public claims from verified deployment.
FUNCTION
Analysis, recommendation, workflow execution, resource allocation, signature authority, and fiduciary office are distinct institutional functions.
RESPONSIBILITY
An AI label does not create legal personhood, fiduciary capacity, democratic legitimacy, or an independent liability bearer.
CONTESTATION
A human remains meaningful only when evidence, authority, time, technical intervention, and institutional protection make reversal possible.
MATERIAL CHANGE
Approval is version-bound. Monitoring, correction propagation, rollback, retirement, and preserved records are part of the same control lifecycle.
AUTONOMOUS SYSTEMS / AUTHORITY BEFORE SPEED
For anticipatory intelligence, predictive targeting, loitering munitions, swarms, autonomous kill chains, and kill webs, the method separates observation, correlation, classification, forecasting, prioritization, nomination, authorization, execution, and assessment. Capability, acquisition, testing, deployment, operating mode, and combat use are never treated as synonyms.
EVIDENCE
A fluent explanation or precise score cannot repair correlated sensors, stale observations, hidden alternatives, or an unknown operating mode.
AUTHORITY
Connectivity, target recognition, or resource ranking does not itself transfer legal or command authority.
ASSURANCE
A nominally distributed web may remain brittle when every path depends on one gateway, classifier, clock, cloud, or approval authority.
07 / SYSTEM MAP
Identify the collection point, authority line, vendor integration, sharing path, retention rule, human override, and audit gap. Procurement records often reveal more than marketing claims or inaccessible source code.
08 / AUTHORIZATION
No unauthorized access, bypassing controls, exploit trading, doxxing, stalking, credential exposure, or instructions for violence. Public-interest intent does not create permission.
09 / REPAIR
Material corrections should identify the affected claim, the contrary source, the requested change, the reviewer, and the visible repair. Named organizations may reply, but do not receive veto power over supported findings.