Italian White Paper

Methodology

Methodology

How we depoliticize and verify the news.

The fact-fidelity scale (1–5)

We grade every original report. Our neutral article is the reference. The grade measures how closely a text sticks to verifiable facts, not which side it takes.

  1. Neutral and accurate

    Correct facts, non-evaluative language. No correction needed.

  2. Broadly accurate

    Correct facts with slight framing or marginal omissions.

  3. Mixed

    Contains verifiable facts alongside evaluative or selective framing.

  4. Distorted

    Predominantly evaluative framing or generalizations unsupported by the data.

  5. Heavily distorted

    Facts are subordinated to partisan judgment or misrepresented.

Source triangulation

We use no paid fact-checking APIs. Verification comes from comparing several outlets of differing orientation clustered on the same event. Agreement across sources is the signal of truth; disagreement lowers confidence and is reported as an open debate, not hidden.

AI transparency

Neutral texts are generated by language models and then reviewed by a human editor before publication. No article goes live without human approval. We always link to the original coverage: we publish only transformative summaries, never copies.

Outlet orientation

We label each outlet's editorial orientation as metadata for balance and transparency, not as a value judgment.