Our charter
A trust layer
for the open web.
RealFakeLies is an independent public-benefit corporation. We exist to make the credibility of any factual claim legible, traceable, and resistant to capture by any single party.
What guides us
Five principles. No verdict in isolation.
Evidence, not opinion
Every credibility score is a function of the evidence behind it. Every score links back to the sources that produced it.
Sources, weighted
Reputation is earned over time and across topics. No source carries equal weight on every subject.
Triangulation by default
A single source is a lead, not a verdict. We require corroboration across independent sources, weighted by credibility.
Independent and durable
No platform owner, advertiser, or government has special access. Our model and our scoring are auditable.
Calibrated, not binary
A claim is rarely simply true or false. We publish a calibrated state with explainable weighting that updates as evidence shifts.
Methodology
How a claim becomes a score.
-
01
Extract
Articles, posts, and broadcasts are parsed into atomic factual assertions, separable from opinion, framing, and tone. Satire and hypotheticals are flagged at this stage.
-
02
Match
Each claim is matched against the corpus of indexed sources by semantic similarity and entity resolution. Supporting, contradicting, and contextual statements are isolated.
-
03
Weight
Each source contributes in proportion to its calibrated credibility on the topic. Reputation is earned by track record, not by volume or recency alone.
-
04
Score
The weighted evidence produces a credibility state on the seven-point verdict scale. The state is recomputed whenever new evidence arrives.
-
05
Explain
Every score is traceable. The supporting and contradicting sources, the weights, and the underlying quotes are visible on every claim page.
Governance
Independence is structural.
RealFakeLies is incorporated as a public-benefit corporation. Our charter binds us to a stated public mission: legible, traceable credibility for the open web. No commercial relationship can override that mission.
Our scoring models are documented and our credibility weightings are published. Researchers can audit our methodology, replicate our scoring on a corpus of their choice, and dispute our verdicts through a public process.
We do not accept paid placement, advertiser-driven verdict adjustments, or platform partnerships that would compromise source neutrality.
Get in touch
Questions, corrections, or partnerships.
Newsrooms, researchers, and platforms can request API access or propose methodology changes. Anyone can contest a verdict.