The Adversarial Method
We don't investigate to confirm. We investigate to find where independent evidence points to the same place — and then we put that evidence through every challenge we can construct.
Five Stages, No Shortcuts
Our research pipeline runs on a fully orchestrated network of specialized AI agents — an expanding collaborative intelligence designed to investigate the way no single mind can. Each agent holds a specific role: source analyst, cross-cultural comparator, linguistic historian, skeptical challenger, archaeological interpreter, and dozens more.
They work in concert but without shared context — isolated by design so that when independent agents converge on the same finding, that convergence means something. They challenge each other. They pressure-test conclusions. They read between the lines of what sources say and don't say.
The result is a synthesis that emerges from genuine adversarial collaboration, not a single perspective dressed up as research.
No agent publishes autonomously. Every report is reviewed by a human editor before it appears on this site.
What the Number Means
The score measures structural agreement across geographically isolated traditions — not the probability that a claim is true. A flood narrative that appears in 40 unconnected cultures with the same specific details scores higher than one found in 3 related cultures with vague similarities.
What We Commit To
We publish the evidence and the score. We do not tell you what to conclude.
Every claim is attributed. Anonymous assertions are not admitted.
Skeptical rebuttals are given equal prominence to affirmative cases.
Tradition-specific claims are labelled as such and excluded from convergence scoring.
We distinguish between what the evidence shows and what researchers have interpreted it to mean.
The pipeline is re-run when significant new evidence emerges.
Every piece of research that appears on UnraveledTruth has been reviewed by a human editor. Agents investigate. Humans decide what gets published.
What We Don't Do
Confirm existing beliefs
We run every topic through a skeptic. If the evidence doesn't survive, the score reflects that.
Publish anonymous claims
Every source is attributed. Unsigned assertions don't enter the pipeline.
Conflate correlation with causation
Convergence means independent agreement. It does not mean one caused the other.
Suppress inconvenient findings
If the skeptic's case is stronger, that appears in the report. We don't bury it.