---
uai: "1.0"
type: evidence-attribution-method
status: active
stable_id: iarpg-evidence-attribution
project: iarpg.com
release: IARPG-OPS-2
version: "2.0.30-wip"
review_state: current-until-superseded
---

# Evidence Attribution, Oversight, and Leaderless-Network Method

## Current Publication Contract

- Public route: [`/evidence-attribution`](../evidence-attribution.php#attribution-ladder).
- Source collection: 12 byte-preserved Markdown reports under [`docs/source-files/evidence-attribution-oversight-and-leaderless-networks/`](../docs/source-files/evidence-attribution-oversight-and-leaderless-networks/).
- Canonical wrappers: 12 reviewed-with-limitations reports under `docs/long-term-memory/reports/`.
- Canonical integration findings: [Evidence Attribution, Oversight, and Leaderless-Network Research Integration Report](../docs/long-term-memory/reports/evidence-attribution-oversight-and-leaderless-network-research-integration-report.md#findings).
- Machine publications: [source index](../evidence-attribution-source-index.json), [72-disposition review register](../evidence-attribution-research-review-register.json), [attribution method](../evidence-attribution-method.json), [leaderless-network taxonomy](../leaderless-network-taxonomy.json), and [oversight/remedy capability model](../oversight-remedy-capability-model.json).
- Review state: structural synthesis only. Source-authored “Final” wording, internal confidence labels, citations, translations, legal conclusions, current incidents, officeholders, platform policies, and technical attribution remain source claims until independently reverified.

## Eight-Rung Attribution Ladder

1. **Observed content or event** — establish that an artifact, statement, outage, filing, or event exists.
2. **Content status** — test a discrete factual proposition for support, contradiction, misleading context, or unresolved interpretation.
3. **Coordination** — require evidence of shared administration, infrastructure, timing, tasking, or planning; similarity and virality are insufficient.
4. **Actor identity** — require direct legal, technical, documentary, or authenticated evidence for the operator responsible for the bounded conduct.
5. **Sponsor or principal** — require evidence stronger than geography, language, affinity, funding alone, shared adversary, or post-event praise.
6. **Intent** — distinguish direct planning or tasking evidence from inferred motive.
7. **Reach** — use verified exposure, distribution, audience, or service-impact evidence.
8. **Effect** — require an explicit causal design before claiming belief, behavior, policy, or strategic change.

Stop public wording at the highest supported rung. Evidence for one rung never silently establishes the next.

## Three Separate Evidence Axes

- **Claim status:** claimed, reported, corroborated, substantiated, proven, refuted, unresolved, or superseded.
- **Analytic confidence:** high, moderate, low, or undetermined.
- **Source reliability:** established, mixed, unestablished, or deceptive-history.

Do not merge these axes. Reliability describes a source record; confidence describes the stability of a judgment; claim status describes what evidence currently establishes.

## Source-Chain Independence

- Trace every item back to the earliest accessible origin.
- Count independent acquisition paths, direct records, witnesses, telemetry, legal processes, or separate investigations—not headlines, reposts, or citations that resolve to the same statement.
- Treat actor-controlled channels and self-published datasets as evidence of what the actor claimed or released, not automatic proof of content, identity, sponsorship, intent, reach, or effect.
- Treat provenance metadata and content credentials as origin and transformation signals, not truth verdicts. Missing metadata is not proof of falsity.
- Preserve original bytes, hashes, dates, transformations, custody, contradictions, corrections, and supersession.

## Leaderless and Pseudonymous Network Boundary

Use role-specific evidence for label or brand, collective identity, audience, participant, amplifier, infrastructure administrator, campaign organizer, technical actor, splinter group, imitator or false flag, and externally assigned member. A shared symbol, slogan, hashtag, language, community, or adversary does not establish membership, command, sponsor, intent, dangerousness, or guilt.

Leaderless does not mean structureless: temporary influence can concentrate around access, expertise, reputation, resources, or administration. Record the time period and bounded role instead of inventing a permanent command chart.

## Oversight and Remedy Capability Boundary

Compare actual capabilities, not institutional names:

- ex ante authorization;
- records and system access;
- compelled cooperation;
- inspection and audit;
- complaint intake;
- independent adjudication;
- binding remedy;
- public reporting;
- appeal and correction.

Formal authority, practical access, timeliness, enforceability, resource adequacy, secrecy, notice, language and disability access, publication, appeal, and affected-person remedy remain separate. Never add capabilities into a country, legitimacy, effectiveness, rights, or morality score.

## Metadata and Identity Inference Boundary

Keep four layers distinct: observed data, derived attribute, identity or relationship hypothesis, and consequential decision use. Every transformation needs purpose, source, assumptions, model or rule version, uncertainty, missing-data limits, privacy review, notice, correction, challenge, and remedy. Do not identify private people, infer protected characteristics, create target lists, or turn metadata analysis into surveillance or evasion guidance.

## Correction and Supersession

Preserve the original claim and evidence state. Add the new evidence as a dated record, state the smallest bounded disposition changed, identify the wording and publication effect, retain unresolved disagreement, and expose appeal or further correction routes. Never silently rewrite old certainty into new certainty.

## Website Integration

- [Evidence Lab](../evidence-lab.php#attribution-guidance) exposes the distinction between observation, assessment, and attribution.
- [International Fairness Methodology](../global-methodology.php#evidence-attribution) applies equal attribution and source-chain burdens across jurisdictions.
- [Network Influence](../network-influence.php#leaderless-role-boundary) separates labels, audiences, amplifiers, administrators, organizers, and technical actors.
- [Faction Directory](../factions.php#leaderless-structure) records temporary power and internal disagreement without inventing permanent hierarchy.
- [Alliance Observatory](../alliances.php#attribution-and-control) distinguishes partnership, support, sponsorship, direction, command, reach, and effect.
- [World Systems](../world-systems.php#evidence-attribution-oversight) links attribution discipline to evidence reliability, oversight, remedy, and correction.
- [Scenario Modeling](../scenario-modeling.php#reach-effect-boundary) separates exposure and disruption from causal effect and model output.
- [Authority Directory](../authorities.php#oversight-capability) separates authorizing institutions from inspection, complaint, adjudication, and remedy.
- [Editorial Policy](../editorial.php#attribution-and-source-chains) makes the method part of current publication doctrine.
- [Glossary](../glossary.php#attribution-rung) defines the new vocabulary with permanent fragments.

## Safety, Fairness, and Publication Boundary

Do not publish intrusion, denial-of-service, credential discovery, private-person identification, doxxing, surveillance, evasion, covert contact, targeting, influence-execution, weapons, recruitment, or operational campaign instructions. Do not create country, community, reviewer, organization, source, susceptibility, dangerousness, legitimacy, morality, loyalty, protected-characteristic, recruitability, or targetability scores.

Automated source custody, schema validation, link checks, and reviewed synthesis do not establish factual correctness, legal validity, technical attribution, human comprehension, locally informed review, consultation, geopolitical neutrality, native Firefox behavior, native assistive-technology speech output, production readiness, field performance, or live search and crawler outcomes.

## Durable References

- [Integration findings](../docs/long-term-memory/reports/evidence-attribution-oversight-and-leaderless-network-research-integration-report.md#findings)
- [Information operations reviewed synthesis](../docs/long-term-memory/reports/information-operations-disinformation-and-neutral-evidence-presentation.md#reviewed-synthesis)
- [Comparative oversight reviewed synthesis](../docs/long-term-memory/reports/comparative-intelligence-oversight-transparency-and-remedy.md#reviewed-synthesis)
- [OSINT verification reviewed synthesis](../docs/long-term-memory/reports/osint-verification-and-attribution-under-uncertainty.md#reviewed-synthesis)
- [Leaderless-movement comparison](../docs/long-term-memory/reports/anonymous-and-other-leaderless-or-pseudonymous-movements.md#reviewed-synthesis)
- [Metadata-inference reviewed synthesis](../docs/long-term-memory/reports/surveillance-metadata-data-brokers-and-identity-inference.md#reviewed-synthesis)
- [Durable pointer ledger](long-term-memory.uai#evidence-attribution-oversight-and-leaderless-network-research)
## Human-Review Readiness

The current method is prepared for later human review through [twelve source packets](../evidence-attribution/review-readiness/index.json) and the [review-readiness protocol](../docs/long-term-memory/reports/evidence-attribution-human-review-readiness-and-translation-quality-protocol.md#findings). Structural readiness does not establish approval.