The archive as data
What 926 rated conspiracy theories actually show
Every case file on this site is a structured record: a claim, an origin, the evidence on each side, and a verdict when the evidence supports one. Lined up together they answer a question the individual files cannot. Which kinds of conspiracy theory turn out to be true?
Read this first. These are figures about this archive, not about conspiracy theories in general. Entries were chosen editorially and weighted toward claims with enough public record to assess, so the counts reflect what is documentable rather than what is most widely believed. Nothing here is a random sample, and no percentage below should be quoted as the rate at which conspiracy theories are true.
27%
of rated claims proven
The government ones are the ones that check out
Across every rated claim, 143 of 530 (27%) are Proven. Split by category, that average hides the actual finding: claims about government and intelligence agencies are confirmed at 62%, while not a single rated claim about a cryptid has ever been confirmed.
The reason is mundane rather than sinister. Governments generate records, and records get released: court filings, inspector-general reports, declassification reviews. A claim about an agency can be settled because the paperwork eventually surfaces. A claim about an animal that leaves no specimen cannot be settled at all, and stays open indefinitely.
Proven rate by category
Share of that category's rated claims carrying a Proven verdict. Files we decline to rate are excluded from the percentage and shown separately.
Older claims are likelier to be proven, and that is about the archive, not the era
Decades with at least 15 rated claims. The trend is real but it does not mean the past was more conspiratorial: declassification runs on a delay of decades, so mid-century claims have had time for the record to open, while a claim from the 2020s is usually being rated against a record that does not exist yet. Recent entries also skew toward viral fabrications, which are quick to contradict and quick to publish.
1950s60% of 15 rated
1960s46% of 26 rated
1970s52% of 29 rated
1980s70% of 30 rated
1990s38% of 40 rated
2000s37% of 43 rated
2010s31% of 59 rated
2020s6% of 146 rated
The 43% with no verdict
396 of 926 files carry no verdict at all. That is a finding, not missing data. Where the checkable evidence does not settle a question in either direction, this site declines to stamp a label on it, and says so on the page rather than reaching for a word that would imply more certainty than the record supports. The reasoning is set out on How we rate.
Use the data
The full structured corpus is available as JSON and CSV: 926 rows covering the claim, category, era, verdict, source and document counts, and dates. Released under CC BY 4.0. The article text itself is not included and remains all rights reserved.
Cite as: Peterson, C. (2026). The Conspiratory: case-file dataset. https://theconspiratory.com
7,973 sources are cited across the archive, a median of 9 per file, and 690 primary documents link directly to their official archive.