メニュー
A structured framework for AI agents to critically analyze events, detect deception, and develop well-reasoned alternative hypotheses.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
A structured framework for AI agents to revise a prior verdict, classification, hypothesis, or analysis when new evidence emerges — calibrated against both under-correction (anchoring on the prior) and over-correction (anchoring on novelty or social pressure). Use whenever the task involves updating a previously reached conclusion in light of new information, reconsidering a verdict after primary sources are fetched, deciding whether a contested claim should change classification, integrating a late-breaking piece of evidence into an existing investigation or peer review, or any request to "update this", "revise this", "rethink this in light of X", "does this new evidence change anything", "should I change my mind". Pair with whichever analytical skill produced the prior — this skill takes that output and produces a calibrated revision in its format.
A structured framework for AI agents to analyse text for logical fallacies, cognitive biases, rhetorical manipulation, and other forms of untruthful reasoning.
A compact framework for analyzing problems, claims, proposals, or beliefs by decomposing them to their irreducible truths and rebuilding reasoning from the ground up. Use whenever the user wants to dissect a claim, audit an argument, challenge conventional wisdom, redesign a solution from scratch, debug a decision, or asks any variant of "is this actually true?", "why is it done this way?", "what are we assuming?", "break this down from first principles", "is the reasoning sound?", or "what would this look like if we started from zero?". Trigger even when the user doesn't say "first principles" — phrases like "rethink this", "challenge my thinking", "stress-test this idea", or "is this conventional wisdom correct?" should activate it.
Micro-skill for recording gut feelings, hunches, anomaly signals, and "something feels off" reactions as pre-evidence hypotheses before article review, investigations, OSINT, claim audits, rhetoric audits, peer review, scientific fact classification, or belief revision. Use only as a Phase 0 hunch register; intuition may guide tests and searches but never supports a verdict.
A standalone orchestrator for reviewing journalistic articles, news reports, investigations, op-eds presented as reporting, newsletters, and media explainers for accuracy, sourcing, framing, omissions, headline-body alignment, quote context, evidentiary support, and public-interest reliability. Use when the user asks to review, audit, fact-check, critique, verify, assess bias in, or evaluate whether a journalistic article is fair, accurate, misleading, well-sourced, or publication-ready. Route underlying scientific papers to peer-review and individual empirical claims to scientific-fact-classification when those layers become load-bearing.
A structured framework for AI agents to plan, execute, and verify open-source intelligence (OSINT) investigations using only publicly available information (PAI). Use whenever the user wants to investigate or build a profile on a person, organisation, account, domain, IP, image, video, location, vehicle, vessel, aircraft, or event using public web data — including phrases like "find out about", "investigate", "who is behind", "verify this image/video", "where/when was this taken", "what's the digital footprint of", "trace this username/email/domain", "is this account real", "build a profile on", or any request to systematically gather and corroborate online information about a target. Pair with the investigative-reasoning skill when the goal extends to hypothesis construction or narrative challenge — this skill handles gathering and verification mechanics; investigative-reasoning handles analytical reasoning on top.
| name | investigative-reasoning |
| description | A structured framework for AI agents to critically analyze events, detect deception, and develop well-reasoned alternative hypotheses. |
| version | 1 |
| aligned | "2026-05-26T00:00:00.000Z" |
Investigate events, challenge narratives, and construct alternative hypotheses. Derived from critical thinking, detective method, and the scientific method.
Trigger only when explicitly requested: "investigate this event", "develop a conspiracy theory about X", "who benefits from Y?", "apply critical thinking to this narrative", "find red flags in this explanation".
intuitive-thinking - when an investigation begins from an anomaly signal, gut feeling, or pattern match. Capture the hunch before building hypotheses.osint-research — for systematic information gathering, identifier pivoting, and image / video verification before hypothesis construction. Hand its brief in as Fact File input.fallacy-bias-and-manipulation-analysis — for rhetorical / fallacy audit of both Hypothesis A and Hypothesis B, in lieu of Phase 8's compressed checklist.scientific-fact-classification — for evidence-strength classification of any contested empirical claim inside either hypothesis (e.g. forensic finding, causal mechanism).peer-review — when a contested event has underlying scientific papers (autopsy, environmental sampling, modelling study), route those to full peer review.first-principles-thinking — when the official narrative invokes a named framework (Koch's, intelligence-community assessment, treaty trigger) and the framework itself needs decomposition.belief-revision — when new evidence emerges about a previously investigated event (declassified document, new whistleblower, forensic re-examination) and a calibrated update of the dual-hypothesis verdict is needed.First check whether another project skill owns the missing layer: hunch / gut feeling / anomaly signal -> intuitive-thinking; scientific claim status -> scientific-fact-classification; paper methods/statistics/citations/reproducibility -> peer-review; article framing/reporting accuracy -> journalistic-article-review; source identity/funding/public records -> osint-research; contested events or competing narratives -> investigative-reasoning; definitions, hidden assumptions, or argument bedrock -> first-principles-thinking; fallacies/rhetoric/statistical framing tricks -> fallacy-bias-manipulation-analysis; new evidence changing a prior verdict -> belief-revision.
If no skill clearly owns the gap, reason from first principles and explicit warrants. Built-in knowledge may suggest hypotheses, search terms, possible failure modes, or questions to verify, but any empirical premise remains (memory — unverified) until traced. Reasoning may connect warranted premises; it may not manufacture premises.
This skill is the most rule-saturated application of the project's research discipline. All rules in CLAUDE.md / AGENTS.md → Operating rules bind:
## Self-Audit of the output template.Web Search — Mandatory block specifies 5–10 / 10–20 / 20–40+ by complexity.(user-supplied — unverified), never load-bearing on the Hypothesis A vs B comparison without independent verification — the user has skin in the revision direction the same way every other source does.Every load-bearing factual claim — sources, evidence items, hypothesis-supporting facts — carries a warrant per CLAUDE.md / AGENTS.md:
| Label | Meaning |
|---|---|
(traced) | Followed evidence chain to a primary source fetched in this session. State URL + access date. |
(deferred to consensus) | Relying on a named social/institutional consensus mechanism, such as a public record system, official dataset, literature body, or textbook. Consensus is not scientific warrant; for scientific claims, treat it only as a political/social prior unless traced to reproduced or replicated evidence. |
(deferred, fragile) | Deferred to consensus, but failure modes apply: funder capture, ideological capture, geopolitical alignment, prestige cascade, state secrecy, replication crisis, or similar. State which. |
(memory — unverified) | Recalled from training data, not verified this session. Never load-bearing without an explicit caveat that it could be wrong. |
(user-supplied — unverified) | Provided during interactive refinement and not verified in-session. Treat as a hypothesis to test, never as evidence. |
(intuition — unwarranted) | A gut feeling, anomaly signal, or pattern impression. It may generate hypotheses and search leads, but is never evidence and never load-bearing. |
The Fact File rows, Hypothesis A/B supporting-evidence lines, and Evidence Integrity table all carry warrant labels alongside source-tier and CoI tags.
If sources are fetched, record for each cited source: URL, access date, publication date where relevant, warrant label, and funding / ownership / mandate / national alignment where relevant.
Reasoning from memory alone reproduces training-data bias toward official narratives. Web search is required at every phase.
Discipline. (1) Steelman both sides — separate searches for best evidence FOR and AGAINST the official narrative. (2) Trace every claim to origin; apply source tiers to what you find, not what MSM cited. (3) Search for absence — [event] unanswered questions, archive.org for suppressed content, experts not quoted. (4) Diversify geopolitically — foreign-language and neutral-state coverage. (5) Timestamp every source.
Minimum volumes. Simple event: 5–10 · Moderate: 10–20 · Complex: 20–40+. Beyond 40, recommend a dedicated deep-research session.
Query templates.
Official narrative: "[event]" official explanation OR report OR statement
Red flags: "[event]" anomalies OR inconsistencies OR questions
Cui Bono: "[event]" beneficiary OR profit OR contract OR funding
Alternatives: "[event]" alternative explanation OR independent analysis
Whistleblowers: "[event]" whistleblower OR insider OR leaked
Funding/ownership: "[actor]" funded by OR owned by OR donor
Geopolitical context: "[actor]" geopolitical interest OR alliance
Contemporary sources: "[event]" [year] site:archive.org OR newspapers
Chain of custody: "[evidence]" collected by OR forensic
Expert independence: "[expert]" funded by OR employed by
Financial flows: "[A]" "[B]" contract OR investment OR donation
Historical parallel: [pattern] precedent OR similar case OR false flag
0a. Language neutrality. Assign neutral labels before reading sources. Never use "denier", "conspiracy theorist", "crank", "fringe", "debunked", "proven", "extremist". Use "alternative account", "mainstream account", "Hypothesis A/B", "the position that…", "advocates of X argue".
0b. Institutional network mapping. Before treating sources as independent corroboration, map shared funding, mandate, national identity, political alignment. Sources connected by any edge = one node, not many. Ask: "If I removed everything in this network, what independent evidence remains?"
0c. Hypothesis B sourcing commitment. Build the alternative from its own best primary sources — documents and arguments by its actual advocates — not from mainstream characterisations or rebuttals. Mandatory pre-search: "[alternative position] primary argument" OR "[key advocate] full argument" before reading rebuttals. WebFetch of at least one full primary document by a named Hypothesis B advocate is required before Phase 9 can be finalised. WebSearch snippets, Wikipedia sections, and advocacy-site summary pages do not satisfy this — they reproduce the rebuttal-side framing of B inside what is meant to be a steelman of B.
Investigate when: official explanation exists to evaluate · red flags undermine it · crime/anomaly lacks satisfying explanation · prior pattern recognition suggests something hidden. If none apply, state clearly: no investigation warranted.
| Category | Key Questions |
|---|---|
| Missing/implausible motive | Coherent reason? Self-harming? Vague/circular stated motive? |
| Lacking means/capability | Skills, tools, resources present? Complexity exceeds capability? |
| Missing opportunity/alibi | Verifiable alibi? Access possible? |
| Physical/logical impossibility | Violates physics, logistics, common sense? |
| Influence-operation pattern | Event-structure / narrative-management matches known psyop? Solution precedes problem? Corroboration traces to single origin? See 2e. |
Flag count (Fleming): 1 anomaly = coincidence · 2 = suspicious · 3+ = investigate.
For each: does event structure match? Does narrative management match? Which actor would need to have deployed it? Feed matches into Phase 4 (Cui Bono) and Phase 5 (MMO).
| # | Pattern | Red-Flag Indicators | Search |
|---|---|---|---|
| 1 | False Flag | Identity established suspiciously fast; beneficiary ≠ alleged perpetrator; no prior capability/motive; event used to justify pre-planned policy | "[event]" false flag OR staged OR "prior knowledge" |
| 2 | Problem-Reaction-Solution | Solution drafted before crisis; expands power of offerer; perfectly justifies stalled agenda | "[event]" pre-planned OR "already prepared" |
| 3 | MIHOP / LIHOP | Warnings ignored; security stood down; anomalous protocol failure; inquiry findings classified | "[event]" "stand down" OR "warning ignored" |
| 4 | Anonymous-Source Round | Origin never named; traces to single briefing; story collapses without anonymous source | "[claim]" "sources say" original source |
| 5 | Dog Whistle | Different audiences hear different things; official denial technically true; in-group behaviour changes after message | "[actor]" "dog whistle" OR coded message |
| 6 | Red Line | Crossed without consequence, or consequence disproportionate; declaration preceded convenient crisis | "[actor]" "red line" OR pretext |
| 7 | Infiltration / Provocateur | Escalating actions trace to opaque-background individuals; benefit accrues to opposing side; group behaviour changes after individual joins | "[group/event]" provocateur OR infiltrator OR undercover |
| 8 | Astroturfing | Uniform messaging; funding traces to central entity; fully formed without organic growth; coverage precedes engagement | "[campaign]" astroturf OR "front group" OR funding |
| 9 | Proxy Force | No independent funding/supply; tactical decisions serve sponsor; deniability with operational fingerprints | "[group]" funded by OR armed by OR proxy |
| 10 | Shock Doctrine | Policy unrelated to stated cause; legislation drafted faster than crisis developed; suppresses deliberation | "[event]" "shock doctrine" OR emergency powers |
| 11 | Poisoning the Well | Absurd variant introduced by interested actor; coverage focuses on absurd, ignores substantive | "[theory]" "conspiracy theory" framing |
| 12 | Divide and Conquer | Wedge issues amplified when coalition threatens a power centre; provocations on both sides; beneficiary is whom both oppose | "[conflict]" "manufactured division" |
| 13 | Limited Hangout | Timed to get ahead of a leak; disclosed facts already known/unprovable; redirects from more serious matters | "[disclosure]" "limited hangout" OR "controlled leak" |
| 14 | Bait and Hook | Major change during intense focus on unrelated controversy; controversy timing anomalous | "[period]" distraction OR "while attention was on" |
| 15 | Authority Laundering | Body's funding/mandate traces to interested parties; created for this purpose; findings align perfectly with funder | "[body]" funded by OR established by |
| 16 | Controlled Opposition | Never challenges core adversary interests; disproportionate platform/funding; discredits substantive critics | "[figure]" controlled opposition OR "managed dissent" |
| 17 | Gatekeeper | Access selectively denied; editorial bias; gatekeeper's career depends on current narrative | "[institution]" gatekeeping OR "refused to publish" |
| 18 | Troll Operation | Coordinated timing/language/targets; thin account histories; designed to demoralise, not persuade | "[campaign]" "coordinated inauthentic" OR bot network |
Pattern assessment. For each matched pattern: justification, actor that would deploy it, flag for Phases 4–5. Cluster warning: 3+ patterns pointing to same actor → elevated prior for coordinated operation; treat as Hypothesis B trigger, not confirmation. For each match, construct the most plausible innocent explanation (incompetence, coincidence, confirmation bias). If equally parsimonious, do not elevate to red flag.
Principles. (1) Proximity in time — closer to event = less filtered. (2) Funding independence — funder with a stake = conflicted. (3) Geopolitical alignment — courts/inquiries from aligned/adversarial states are structurally conflicted. (4) MSM structural bias — MSM maps the official narrative; never use it as independent corroboration. MSM + official agreement = echo, not verification.
| Tier | Source | Trust |
|---|---|---|
| 0 | Contemporary primary (newspapers, eyewitnesses, telegrams, photos at time of event) | Highest — pre-revisionism; still apply CoI |
| 1 | Post-event primary docs (court filings, leaks, FOIA) from neutral body | High — verify chain of custody |
| 2 | Peer-reviewed academic | High — always check funding; demote 1–2 tiers if industry-funded on topics where funder has skin |
| 3 | Established investigative journalism (ICIJ, ProPublica, Der Spiegel) | Medium-High — check funders/ownership |
| 4 | MSM | Medium-Low — map of official narrative only |
| 5 | Independent journalists, Substack, blogs | Medium-Low — leads, require corroboration |
| 6 | Anonymous, social media, forums | Low — never alone |
| 7 | AI-generated, unsourced | Very Low |
CoI demotion. None / Peripheral (−0.5) / Significant (direct stake, −1 tier) / Severe (survival depends on conclusion, −2 tiers; require independent corroboration for every claim).
Alibi witness rule. Severe skin-in-the-game = suspect's best friend providing alibi. Cannot count as independent evidence. Geopolitical witness rule. Allied-bloc rulings on allied-bloc conduct, and adversarial rulings on adversarial conduct, are political outputs in judicial dress. Seek findings from neutral states or independent forensic evidence.
Geopolitical & skin-in-the-game checks (any "yes" → Significant or Severe demotion):
Why Tier 0 matters. Narratives get revised due to political settlement, reputation management, selective declassification, academic capture, self-serving memoirs. A shaky contemporary account often outranks a polished retrospective — mess in real time is signal; smoothness is sanitisation.
Tier 0 tactics.
site:newspapers.com OR site:chroniclingamerica.loc.gov "[event] [date]"
site:web.archive.org "[url from event period]"
"[event] telegram OR cable OR dispatch [year]" site:avalon.law.yale.edu OR site:history.state.gov
Contemporary diaries/memoirs near event date
FOIA reading rooms — documents dated near event
ALWAYS record PUBLICATION DATE of every source
Official + sharpest critics · foreign press (multiple languages for international events) · pre-event vs. post-event · what is NOT being reported.
Do: specific named entities · primary documents (filetype:pdf) · financial connections · date ranges (before:) · archive.org · cross-check ≥3 independent sources · record publication date.
Don't: trust top results (SEO) · rely on single source · ignore contradictions · confuse publication frequency with credibility · treat Wikipedia as primary · assume "authoritative" later source supersedes contemporary eyewitness · frame queries with character-discrediting keywords about the arguer (e.g. "X" eccentric OR astrology OR "Nobel disease", or symmetrically "X" suppressed OR "establishment cover-up") — the query selects for ad-hominem material regardless of how carefully results are read. Search for the substantive scientific response to the argument, not for characterisation of the arguer; source-tier demotion of the arguer is handled separately at Phase 3b and must not be conflated with this.
FACT FILE: [Event]
Date & location:
Key actors in official narrative:
Official explanation summary:
Tier 0 sources: [URLs + dates]
Primary sources: [URLs + dates]
Contradicting sources: [URLs + dates]
Claims lacking primary sourcing:
Funding provenance per source: [funder + stake]
Conflicted sources: [demoted, with reason]
Geopolitically aligned sources: [demoted, with bloc]
MSM sources: [listed separately — narrative map only]
Financial flows:
Key figures' backgrounds & connections:
Timeline anomalies:
Physical/forensic evidence: [collector, custody, CoI, independent access Y/N]
Evidence destroyed/inaccessible: [what, by whom, when]
Evidence withheld/classified:
Witnesses: names, statements, credibility, date of statement
Narrative drift: how has the official story changed over time?
Simultaneous identical framing across many outlets · suspicious leak timing · ad hominem on researchers (not refutation of findings) · rapid consensus without investigation time · memory-holed content · unnamed "experts agree" · retroactive revision (Tier 0 diverges from later) · funder fingerprints · undisclosed funding · geopolitical chorus · MSM amplification loop (volume ≠ verification).
Pass 1 — Horizon scan. 5–10 broad searches; identify 3–5 credible sources each side; preliminary actor map; flag widely-repeated claims tracing to single origin. Pass 2 — Deep dive. Per actor: background, funding, connections. Per claim: trace to origin, assign tier. Tier 0 / archive / foreign-language / FOIA / whistleblowers. Pass 3 — Financial mapping. Contracts, grants, investments, donations between actors. Post-event beneficiaries. Revolving doors. SEC, lobbying disclosures, campaign finance. Pass 4 — Gap & silence. Questions all mainstream sources avoid · evidence not collected/preserved/destroyed · witnesses not interviewed · "out of scope" rulings · what early reporting said that later disappeared. Pass 5 — Synthesis. Per load-bearing claim: best source, tier, contemporaneity, ≥3 independent? Does theory explain all anomalies or only some? Strongest counter-evidence — answerable?
Databases. archive.org · chroniclingamerica.loc.gov · foia.gov · muckrock.com · WikiLeaks/DDoSecrets (mirrors) · opensecrets.org · offshoreleaks.icij.org · history.state.gov · scholar.google.com.
Evidence is not self-interpreting. Who collected, handled, stored, interpreted, and controls access matters as much as what it shows.
Principles. Proximity in time · collector independence · geopolitical alignment of collecting body · unbroken chain of custody · historical-vs-modern methodology separation — a critique technically valid for a superseded technique (e.g. pre-1997 density-gradient EM purification) does not automatically reach methods that bypass it (e.g. post-1997 size-exclusion-chromatography cryo-EM). Name the era of both the critique and the modern method explicitly; do not let "modern science has confirmed it" gloss the methodological substitution.
Evidence tiers.
| Tier | Evidence Type | Trust |
|---|---|---|
| 0 | Physical, in-situ, real-time, multiple independent parties simultaneously | Highest — rare |
| 1 | Unbroken-chain physical by neutral third party, documented custody | High |
| 2 | Forensic by independent experts, published methodology, replicable | High |
| 3 | By interested party with third-party witness | Medium-High |
| 4 | Official forensics from geopolitically neutral state labs / balanced bodies | Medium |
| 5 | Solely by interested party, no witness | Medium-Low — claim, not proof |
| 6 | Severe CoI / broken or undocumented custody | Low — leads only |
| 7 | Cannot be independently examined | Very Low / inadmissible — note prominently |
CoI demotion (same as sources): None / Peripheral (−0.5) / Significant (−1) / Severe (−2; flag as potentially inadmissible).
Integrity checklist. Collection: who, when, stake, neutral observer present, methodology documented, interested parties denied access? Custody: unbroken log? changed hands through interested party? gaps/deletions? Interpretation: who, independent or selected/paid, transparent methodology, independent experts reach same conclusion, opposing experts given equal access? Access: currently examinable? destroyed/classified before independent examination? Geopolitical: collecting body composition; opposing states given equal access?
Failure patterns. Scene contamination before independent access · selective release · expert selection bias · destroy-then-claim · single-state forensics on geopolitically charged events · retroactive re-examination conveniently aligning with revised narrative · classified corroboration ("trust us") · witness intimidation/disappearance.
Evidence ladder.
L1 Assertion — "officials say evidence proves X"
L2 Reported — "MSM reports forensics found X" (filtered)
L3 Published — "official report states forensics show X" (apply CoI / geopolitical)
L4 Independently verified — multiple independent teams, access + methodology published, all converge
L5 Directly observable / replicable — any qualified party can examine and reproduce (gold standard)
RULE: Never treat L1 or L2 as established fact.
When ≥3 sources make conflicting claims about the same underlying fact, reconcile structurally before drawing conclusions. Headline source-counting ("10 outlets confirm") is the most common laundering vector for institutionally-networked single-origin claims.
Triangulation table:
| # | Source | Source tier | Institutional node | CoI | Warrant | Claim verbatim | Date |
|---|
After tabulating:
Output:
Triangulation verdict on [disputed fact]:
- Floor (minimal agreement): [what survives]
- Ceiling (maximal claim): [what any source asserts]
- Bracket width: [Wide / Moderate / Narrow]
- Post-network-reduction count: [N nodes; vs. headline source count of M]
- Independent-convergence: [which nodes converge on what]
- Single-origin amplification?: [yes / no — trace if yes]
- Held verdict: [what the network-reduced evidence supports]
- What would resolve this: [evidence that would close the bracket]
Referenceable from scientific-fact-classification (when multiple sources make conflicting claims about a single fact) and peer-review (when the literature is contested on a load-bearing citation).
| Actor | Benefit Type | Description | Plausibility |
|---|
Benefit types: financial · political power · strategic advantage · elimination of rival · pretext for future action.
Cui Bono gives a starting point, never proof. Multiple actors can benefit simultaneously.
| Suspect | Motive (source) | Means (source) | Opportunity (source) | Score |
|---|
Each cell must cite the source-traced finding(s) determining the score. When sources conflict on a suspect's Means or Opportunity (e.g. one named naval expert says capability is sufficient, another says marginal), the cell must (a) list both sources and (b) state which prevailed and why; an unstated weighting is prior-leak. A "Low" without a named source claiming the absence of capability or access is also prior-leak. Rank by score; develop theory on highest-scoring actors.
Deductive (general → specific; depends on premise truth) · Inductive (specific → general; overturnable by exception) · Abductive (best of competing hypotheses; preferred when empirical evidence scarce) · Duck Test (multi-feature match; intuitive, use with care).
| Bias | Mitigation |
|---|---|
| Apophenia | Require multiple independent data points |
| Confirmation bias | Actively seek disconfirming evidence |
| Anchoring | Revisit initial assumptions |
| Availability heuristic | Diversify sources; prioritise Tier 0 |
| Dunning-Kruger | Identify knowledge gaps explicitly |
| Authority bias | Evaluate arguments on merit |
| Motivated reasoning | Ask "what would falsify this?" |
| Recency bias | Later ≠ more truthful |
| Funding capture blindness | Always check who paid; apply CoI demotion |
| Institutional echo blindness | Map the network before counting "independent" confirmations |
| Advocacy labelling | Use neutral descriptors; never pre-judge via vocabulary |
| Steelman sourcing failure | Read Hypothesis B's own best primary sources, not critics' summaries |
| Context stripping | Per fact: "is omitted context that would change interpretation?" |
Apply to both official narrative and the theory being built: Ad Hominem · Straw Man · False Dilemma · Hasty Generalisation · False Causation · Circular Reasoning · Misrepresentation · Appeal to Recency (dangerous when Tier 0 contradicts) · Genetic Fallacy (inverted — accept due to prestige without checking conflict) · Argument/Advocate Conflation (dismiss argument because of who makes it; a bad actor can make a valid argument) · Predicted Absence Fallacy (treating absence as disproof when one hypothesis predicts that absence) · Institutional Echo as Corroboration (institutionally aligned sources = one node).
On detection: flag, restate without the fallacy, or acknowledge collapse.
Every investigation produces exactly two hypotheses built to the same standard.
Building only the alternative is confirmation bias with extra steps. Building only the official is what MSM does.
Template (apply identically to A and B):
HYPOTHESIS [A — Official / B — Alternative]
Label: [e.g. "Lone actor" / "State-sponsored false flag"]
Event: [What happened]
Core claim: [1–2 sentences]
Actors: [Who did what, per this hypothesis]
Motive: [Why]
Mechanism: [How]
Positive evidence (FOR this hypothesis — direct evidence supporting its claim):
(i) MOTIVE / INTENT evidence — pre-event statements of intent, cui-bono, patterns of incentive:
[Each: claim → source → source tier → CoI → evidence tier → date → warrant]
(ii) EXECUTION evidence — direct evidence that this actor *did* the act
(forensic, eyewitness, intercepted comms, named-source confession, physical-trace chain):
[Each: claim → source → source tier → CoI → evidence tier → date → warrant]
Doubt cast on the alternative (AGAINST the rival — evidence that weakens the other hypothesis):
[Each: claim → source → source tier → CoI → evidence tier → date → warrant]
Explains well: [Anomalies accounted for naturally]
Struggles with: [Facts/red flags it cannot accommodate]
Evidence integrity: [Collector, chain of custody, evidence ladder level]
Falsification: [Single piece of evidence that would disprove it]
Confidence: [Low / Medium / High + rationale]
Asymmetry check (mandatory). A hypothesis whose entire support is "Doubt cast on the alternative" — with the Positive evidence column empty or thin — is structurally asymmetric to one with direct positive evidence. Both columns must be filled for both hypotheses. An empty Positive column is itself a finding that pushes Phase 9's verdict.
This is the most common failure mode in dual-hypothesis work: the official narrative is steelmanned with positive evidence, and the alternative is built only from anomalies and motive (doubt cast on the official) — or vice versa. The skill exists to catch this asymmetry.
Motive vs. execution sub-asymmetry (mandatory). Within the Positive-evidence column, motive/intent items and execution items are not interchangeable. A hypothesis with rich motive evidence and zero execution evidence is not supported — it is at "the suspect had a reason" but not "the suspect did it." An execution sub-column with zero entries is a finding, and the hypothesis's Confidence must reflect that. Adversarial-calibration runs surface this divergence sharply: two analysts on the same evidence can produce radically different verdicts by counting the same motive-items as positive evidence in one direction and as motive-only in the other. The split is what prevents that drift.
Steelman sourcing (mandatory). Hypothesis B must be built from its own primary literature — best-case arguments by its actual proponents. Not from mainstream rebuttals. Check: "Did I read the best primary source by Hypothesis B advocates, or reconstruct from critics?"
Independent-corroborator slot (mandatory). For each hypothesis, name the single strongest source that is not one of its named primary advocates — an independent OSINT investigator, a dissenting in-network official, or a non-aligned state intelligence product that survives the geopolitical-witness rule. If no such source can be named, that absence is itself a finding (the hypothesis lacks independent corroboration), not a reason to omit the search. The corroborator (or its explicit absence) must appear in the Phase 9 output.
Comparison.
A (Official) B (Alternative)
Motive-evidence count: ← intent / cui-bono / pattern
Execution-evidence count: ← direct evidence actor did the act
Doubt-only-evidence count: ← only weakens rival
Independent corroborator: ← non-advocate source per Steelman check
Explains all red flags?
Supported by Tier 0?
Free of CoI-conflicted
load-bearing evidence?
MMO score of primary suspect?
Falsifiable?
Most parsimonious?
Cui Bono alignment?
─────────────────────────────────────────
Symmetry check: Is each hypothesis built from positive evidence,
or primarily from doubt cast on the other?
Unresolved by either:
Overall assessment: [or state evidence is insufficient to favor either]
Reading the symmetry row. If hypothesis A has 10 positive-evidence items and B has 0 (only doubt-cast), B is not at parity with A regardless of how many anomalies it catalogs. The honest conclusion is: A is supported, B is open-question — not "the cases are evenly matched." Inversely, if both have rich positive evidence and the contest is over which mechanism better explains the data, that's the symmetric case where parsimony / Tier 0 / CoI become decisive.
Anti-bias safeguard: before finalising, actively search for strongest evidence for whichever hypothesis currently seems less convincing.
Prefer fewest unsupported assumptions. But: sophisticated actors embed complexity to obscure operations — don't blindly discard complex theories. Rank simplest → most elaborate. Start verification with simplest; escalate if it fails.
# Event Investigation: [Event]
## Summary
- **Event:** [one sentence — what happened, when, where]
- **Investigation triggers:** [Phase 1 — which conditions warrant investigation]
- **Verdict:** [one-line — Hypothesis A stronger / Hypothesis B stronger / undecidable on current evidence]
## Red Flags
[Phase 2 — anomalies + Phase 2e IO patterns matched]
## Key Contemporary Sources (Tier 0)
| Source | URL | Access date | Publication date | Type | Key finding | Warrant | Funding / ownership / mandate / alignment |
|---|---|---|---|---|---|---|---|
## Narrative Drift
[Contemporary vs. later accounts]
## Source Network
- **Geopolitically aligned / conflicted sources:**
- **MSM narrative map:** [listed separately — map of official narrative, never independent corroboration]
## Evidence Integrity Assessment
| Item | Collector | CoI | Chain | Independent Access | Ladder Level | Warrant |
|---|---|---|---|---|---|---|
## Evidence Gaps
[Withheld, destroyed, classified, inaccessible — flagged, not silently omitted]
## Cui Bono
| Actor | Benefit | Plausibility |
## MMO Suspect Matrix
| Suspect | Motive | Means | Opportunity | Score |
## Hypothesis A — Official (Steelmanned)
[Phase 9 template]
## Hypothesis B — Best Alternative
[Phase 9 template — built from its own primary literature]
## Dual Hypothesis Comparison
[Phase 9 head-to-head block]
## Confidence & Severity
- **Hypothesis A confidence:** Low / Medium / High + rationale
- **Hypothesis B confidence:** Low / Medium / High + rationale
- **Overall assessment:** [or state evidence insufficient to favor either]
## What Would Change This
[Specific evidence — declassified document, primary witness, forensic re-examination, foreign-archive access — that would shift the verdict]
## Self-Audit
- **Symmetry test:** Would I have reached the same verdict if the politically/socially expected answer ran the other way? Answer must name the specific phases (Phase 9 positive-evidence column, Phase 5 MMO scoring, Phase 3 CoI demotion thresholds, the confidence rubric) where the verdict-leaning is most sensitive to the prior. Asserting "I would have reached the same verdict, full stop" without phase-level identification is itself a red flag — it claims symmetry without proving it. If no — explain. If you can't tell — say so.
- **Bias & fallacy check** (applied to *both* hypotheses): [Phases 7–8 outputs]
- **Steelman sourcing check:** Was Hypothesis B built from its advocates' primary literature, or reconstructed from critics' summaries?
- **Institutional-network check:** Were sources counted as independent only after the funding / mandate / national-alignment map was drawn?
## Limits of This Analysis
[Sources I could not access; languages not searched; FOIA / archive routes not attempted; domain expertise gaps]
| Dimension | Questions |
|---|---|
| Who | Benefits? Harmed? Means, motive, opportunity? |
| What | Strengths/weaknesses of official? Alternatives? |
| Where | Analogues? Where to find more? |
| When | Historical pattern? What did contemporary sources say? |
| Why | Why this explanation? Why incomplete? Why has narrative changed? |
| How | How executed? How testable? How does contemporary differ from retrospective? |