Our Methodology
Every theory on Conspirafy goes through the same five-stage process. This page describes that process and walks through a worked example.
The five stages
- Scope and steelman. We identify the strongest plausible formulation of the claim. If the original claim is multi-part, we split it into testable sub-claims. We record the claim as believers would describe it.
- Source gathering. We collect primary documents (government reports, court records, FOIA releases), peer-reviewed papers, investigative journalism from established outlets, and — when relevant — on-the-record testimony from named sources. We do not use anonymous-source claims as dispositive evidence. Minimum 10 sources per theory.
- Evidence weighting. Each evidence point gets a strength rating (weak / moderate / strong) based on source credibility, sample size, methodology, and reproducibility. Supporting and debunking evidence must both be presented even when the verdict is clear; we do not suppress a weak supporting point, we label it weak.
- Verdict.One of: confirmed, partially_true, unsubstantiated, debunked, ongoing_investigation. Accompanied by a confidence percentage and an explicit “what would change our verdict” clause. If we cannot assign a confidence ≥60%, the verdict is ongoing_investigation.
- Review and update. Each theory has a next-review date. If new primary evidence emerges, we update and log the change. Corrections are tracked at /corrections.
Worked example: MKUltra
Take MKUltra. In the 1960s-70s, claims about a CIA mind-control program would have been a classic conspiracy theory. Today it's a confirmed historical fact. How did the verdict shift?
- Steelman stage: The original claim was specific — the CIA administered LSD and other drugs to non-consenting subjects, including US civilians, in programs researching coercive interrogation. We noted which sub-claims were testable and which were vague.
- Source gathering:The 1975 Church Committee report, the 1977 Senate Select Committee hearings, declassified CIA files (Project ARTICHOKE / MKULTRA subproject records), and Admiral Stansfield Turner's 1977 testimony are all primary-source documents.
- Evidence weighting: Primary-source declassified documents are strong evidence. On-the-record Senate testimony by a sitting CIA director is strong evidence. Competing claims that MKUltra never existed have zero strong evidence.
- Verdict:Confirmed with 98% confidence. “What would change our verdict”: a credible finding that the entire Church Committee record was fabricated, which has never been asserted by any serious source.
- Review:MKUltra's verdict is stable. New details continue to emerge (e.g. Canadian MKULTRA sub-project disclosures in 2018), which we append to the update log.
How verdicts are decided
Every verdict is drawn from one of five tiers. Here is exactly what each tier means, what evidence threshold triggers it, and a real example from our index.
Confirmed
Threshold: Primary-source documentation (declassified files, court records, official acknowledgment) plus independent corroboration. Confidence ≥ 85%.
Example: MKUltra — confirmed by the 1977 Senate Select Committee hearings and declassified CIA files.
Partially True
Threshold: A verifiable core fact exists, but the broader claim over-reaches. Proponent evidence supports part of the claim; debunking evidence negates the rest. Confidence 60–84%.
Example: Operation Mockingbird — real CIA media-influence programs existed, but claims of total mainstream-media control go beyond the evidence.
Ongoing Investigation
Threshold: Active official inquiry, significant new primary evidence has emerged in the last 18 months, or we cannot assign a confidence ≥ 60%. Verdict is provisional and will be updated.
When it resolves: We reassign to confirmed, partially_true, or debunked once primary evidence settles. The update is logged at /corrections.
Unsubstantiated
Threshold: Insufficient evidence to confirm or deny. The claim is plausible but unproven; available evidence is ambiguous; or only low-credibility sources support it.
Not the same as false — it means the claim has not cleared our evidence bar. We may revisit if new sources emerge.
Debunked
Threshold: The core claim is conclusively disproven by scientific evidence, documented records, or logical analysis. Confidence ≥ 85% that the claim is false.
Example: Flat Earth — contradicted by independent satellite imagery, circumnavigation records, physics, and first-hand astronaut testimony.
Confidence scores
Each verdict carries a confidence percentage (e.g. Confirmed — 98%). This is our editorial estimate of how likely the verdict would survive an independent review given the available evidence. It is not a statistical probability calculated by a model; it is a calibrated human judgment.
Factors that raise confidence: more primary sources, stronger corroboration, official acknowledgment, reproducible scientific evidence. Factors that lower it: conflicting primary sources, reliance on anonymous testimony, rapid new developments, or a single point of failure in the evidence chain.
If confidence falls below 60%, the verdict defaults to Ongoing Investigation regardless of the direction of the evidence, reflecting genuine epistemic humility. Learn more at /about/confidence.
What would change our verdict
Every theory page includes a What would change our verdict field. This is an explicit, editor-written statement of the evidence that would cause us to revise. For confirmed theories it typically names the type of document that would need to be fabricated for the consensus to collapse; for debunked theories it names the primary evidence that would need to emerge to reopen the question.
This mechanism exists to prevent the site from becoming a static archive of concluded opinions. Verdicts are conditionally held, not permanent. If you believe new evidence meets the stated threshold, use the Suggest a correction button on that theory page and link the source.
Evidence and source weighting
We assign every evidence point a strength rating. Here is how the tiers map:
| Strength | What qualifies |
|---|---|
| Strong | Peer-reviewed studies, declassified government documents, court records, named on-the-record testimony, reproducible scientific consensus. |
| Moderate | Credentialed investigative journalism (with named sources), academic books with citations, official government reports not yet declassified but publicly confirmed. |
| Weak | Anonymous sources, uncorroborated first-hand accounts, circumstantial pattern-matching, or claims from outlets with documented accuracy problems. We include weak evidence with a label rather than suppress it. |
No single strong piece of debunking evidence automatically overrides multiple moderate supporting pieces — we weigh the full picture. Verdicts are the product of the aggregate, not a mechanical scoring rule.
When we decline to publish
Some claims are on our exclusion list and we do not give them their own theory page. These include: theories targeting specific private individuals with accusations of crimes they have not been credibly linked to, theories that repackage antisemitic or other racist tropes, and theories whose central claim is legally defamatory in jurisdictions where we operate.
Red flags and fallacies
Our verdicts also note which common logical fallacies and conspiracy red flags appear in the public discourse around each theory. These are tools for readers to evaluate future claims, not just the ones we cover.