Editorial contract
A publication that can be inspected, challenged, and corrected.
The site serves both people and machine readers. Clear prose, visible metadata, stable links, structured data, source manifests, checksums, and active memory records support discovery, but none of them replace evidence review or human judgment.
01 / Authority
Source hierarchy and truth boundaries
When records disagree, iarpg.com uses a declared order of authority. The order prevents an older report, generated artifact, or exploratory source from silently steering current implementation.
| Priority | Source | What it establishes |
|---|---|---|
| 1 | Repository and AGENTS instructions | Non-negotiable operating constraints and required release process. |
| 2 | Current verified implementation and tests | What the package actually does under the tested environment. |
| 3 | Active .uai memory | Compact current state, decisions, blockers, architecture, and operating guidance. |
| 4 | Current authoritative durable documents | Reviewed detail, rationale, implementation history, and evidence. |
| 5 | Machine-readable manifests and configuration | Enumerated records, routes, schemas, checksums, and generated metadata. |
| 6 | Historical or exploratory reports | Research context and preserved history, not automatic current truth. |
Truth boundary: a passing local test establishes behavior in the tested environment. It does not establish production-host behavior, native assistive-technology speech output, every browser, search ranking, answer-engine inclusion, or a shipped worldwide MMO.
02 / AI disclosure
AI assistance and automated generation
AI systems may assist with research organization, drafting, code generation, schema generation, test scaffolding, summarization, and interface iteration. AI assistance is not treated as evidence by itself. Claims must be supported by repository evidence, authoritative first-party sources, locally preserved reports, or clearly labeled inference.
Machine-generated dialogue inside the game concept may express claims, uncertainty, or deception, but it never owns authoritative mission truth, permissions, evidence state, rewards, warrants, completion, or player sanctions. Deterministic server or package state remains authoritative.
Automation safeguards
- Generated structured data must match visible page content.
- Generated indexes and manifests are validated against the files they describe.
- Local canonicalization and checksums make package comparisons reproducible.
- Unknown imported fields are preserved rather than silently discarded.
- No crawler, answer engine, or AI assistant is promised visibility merely because a discovery file exists.
03 / Global method
International fairness without false equivalence
International coverage applies the same evidence burden to every country, authority, institution, and community. Equal treatment does not require pretending that all systems have identical powers, conduct, safeguards, or consequences. It requires comparable sourcing, uncertainty, rights review, oversight review, and correction standards.
- Distinguish formal authority from authority exercised in practice.
- Seek locally authored or local-language sources where feasible and record limitations when they are unavailable.
- Use endonyms and naming context where useful without erasing widely recognized alternatives.
- Include affected-community evidence, rights, remedies, oversight, reforms, and multinational relationships.
- Avoid permanent national, cultural, religious, institutional, or role-based hero and villain framing.
- Compare adjective intensity, threat framing, uncertainty, rights coverage, and gameplay consequences across profiles.
04 / Corrections
Corrections, supersession, and change history
Corrections should change the smallest authoritative layer that can resolve the problem, then regenerate dependent pages, indexes, manifests, mirrors, checksums, and release notes. Material corrections are recorded in the next versioned release rather than silently rewriting the downloadable package after publication.
Correction sequence
- Identify the affected claim, record, page, report section, or generated artifact.
- Check implementation, tests, active memory, durable evidence, and source lineage.
- Mark unresolved conflict explicitly instead of selecting a convenient answer.
- Update current truth, preserve useful superseded history, and add supersession links.
- Regenerate discovery files and validate every affected link, anchor, checksum, schema, and mirror.
- Document the change and remaining limitations in release notes and validation evidence.
05 / Safety
Fictional design, not real-world operational instruction
IARPG is a fictional game and standards-design publication. It may discuss high-level concepts such as tasking, cover, collection, evidence, deception, counterintelligence, extraction, justice, and review. It does not provide actionable instructions for real-world intrusion, targeting, surveillance abuse, weapons use, evasion, coercion, evidence destruction, doxxing, or illegal conduct.
Clinical and mental-health research in the archive is contextual material, not diagnosis or treatment for an individual. Public gameplay language must not stigmatize mental illness, turn real clinical conditions into enemy identities, or imply that a country or culture is inherently dangerous.
06 / Versioning
Release status and promotion evidence
The current publication is 2.0.12-wip and remains work in progress. A WIP package may contain substantial working functionality and validation, but it cannot be called stable until every promotion gate has current direct evidence.
Known unresolved gates include moderated human comprehension and task-completion evidence, native Firefox runtime evidence, native assistive-technology speech-output evidence, real target-host behavior, production search-console inspection, and field Core Web Vitals. Search, answer-engine, and generative-engine visibility also depend on external crawling, indexing, ranking, and product behavior outside this repository.
