Project Overton
Home
Analyze an Article CLOSED History PRO
Rubric
Reports SOON
How It Works Analysis Types How Scores Are Made Bibliography Glossary
About Project Overton FAQ & Challenges
Data Privacy
Membership & Pricing
Log in Get access
How a Score Is Made
Three independent readings of the article. A question gets a single score only when they agree closely.

How a Score Is Made

Every score on the site is a judgment the tool can state and defend: which arrangement the article treats as proper, and which rung of the rubric that arrangement fits. The judgment is made three times, independently. Where the three readings agree closely, the rung that two or all three of them share is the score. Where they are further apart, the page shows the range they span and explains what each reading found.

Step What happens
Observations The article is read through fourteen analytical lenses. Each lens records what the article does: what it assumes, whom it lets speak, what it presents as normal, and what it leaves out. Observations are evidence. They carry no score of their own.
One reading per axis For each of the seven questions, the tool makes one judgment. It names the arrangement the article treats as proper, such as who should own a business or who may overrule a decision, and matches that arrangement to the rung of the rubric whose definition it fits. It states why the article sits on that rung and not the one beside it. Where the article treats no arrangement on an axis's question as proper, that axis is marked No Signal.
Three readings, one result The article is read three times, independently. On each question, when two readings place the article on the same rung and the third places it on that rung, one step from it, or at No Signal, that rung is the score. When the readings are further apart and at least two of them place the article on a rung, the question is shown as a range, from the lowest of those rungs to the highest, and the page explains what each reading found. When two of the three readings find nothing on a question, it is marked No Signal.
A sentence for every rung The sentence beneath each axis on the card says why the article sits where it does, in the article's own particulars. Before an analysis is published, each sentence is checked against the rung it explains, and every quotation is checked against the article.

Until October 2026, scores were produced by several scoring models, each of which converted the observations into a number by a different method. Those models disagreed with one another, and the same article could score differently from one run to the next. They were retired in favor of the method above. An analysis scored before that date says so on its own page.