Define the Question
Begin with a neutral, specific, investigable question. Do not smuggle the conclusion into the framing.
Investigation Protocol v1.0
Neutrality applies to the method.
PARALLAX uses the same operating process for conventional and unconventional explanations. It does not require the final evidence to appear balanced.
Operating chain
From question to revision.
QuestionScopeInterpretationsSource planEvidence collectionVerificationStress-testLedgerState of evidenceChallengeRevision
Protocol steps
The core method.
Define Scope
Record the time period, event, actors, evidence categories, geographic limits, and excluded side questions so the case does not expand endlessly.
Record Initial Assumptions
Privately record starting belief, confidence, reasons, and what evidence would change the view. This is bias control, not evidence.
Establish Competing Interpretations
Name the strongest serious explanations, what each predicts, what would support or weaken it, and what could falsify it if possible.
Build the Evidence Plan
Identify the evidence categories most likely to distinguish the interpretations before collecting everything indiscriminately.
Verify Source Provenance
Ask who created each source, when, why, whether it is original or copied, whether the chain of custody is clear, and what it actually establishes.
Stress-Test Each Interpretation
Ask what supports it, what contradicts it, what would prove it wrong, and what alternative explanation accounts for the same evidence.
Separate Fact from Interpretation
Label statements as fact, supported inference, disputed, speculation, unknown, or unverified claim.
Maintain the Ledgers
Keep an evidence ledger and an open-questions ledger so contradictory evidence and unresolved issues cannot be quietly selected away.
Publish the State of Evidence
Answer what is established, disputed, strongest, weakest, unknown, and what evidence could change the assessment. No forced verdict is required.
Revise or Reopen
Reopen cases when new evidence, authenticated sources, discredited sources, expert analysis, or corrected errors materially change the record.
Source hierarchy
Evidence is weighted, not flattened.
Unverified does not mean false. It means unverified. Authority is context; evidence is the basis.
Primary / Direct Evidence
Original documents, raw datasets, official records, unedited footage, photographs, physical evidence, firsthand testimony, laboratory results, and transcripts.
Technical / Expert Analysis
Peer-reviewed work, engineering reports, forensic analysis, specialist reports, expert submissions, and technical critiques.
High-quality Secondary Reporting
Investigative journalism, reputable historical synthesis, well-sourced books, and serious long-form reporting.
Advocacy / Interpretive Sources
Campaign groups, activist publications, advocacy websites, partisan analysis, and official public-relations material.
Unverified Material
Anonymous claims, unattributed screenshots, unsourced social posts, rumor, and reposted material with unclear provenance.
Review structure
Every major evidence category follows the same pattern.
- Core question
- Established observations
- Evidence quality
- Conventional interpretation
- Unconventional interpretation
- Strengths and weaknesses
- PARALLAX assessment
- Open questions
- Sources
Publication requirements
Before a case goes public.
- The primary question is neutral.
- Scope is defined, including what the case is not trying to answer.
- Strongest serious interpretations are included.
- Major claims are sourced and primary sources are linked where possible.
- Unverified claims are labeled.
- Contradictory evidence remains visible.
- Source limitations are stated.
- Open questions remain visible.
- No private person is recklessly accused.
- Wording does not exceed the evidence.
Challenge and reopening
Reopening is the system working.
A case should be reopened when significant new evidence appears, an important source is authenticated or discredited, expert analysis materially changes, an error affects the overall conclusion, or a previously unresolved question becomes answerable.
Before publication PARALLAX asks: If someone who completely disagreed with us examined the report, could they understand how we reached our assessment?
They do not have to agree. They should be able to follow the evidence, reasoning, source trail, competing interpretations, and uncertainty.
Do not trust us because we say so. Check the sources. Truth should withstand a change in perspective.