Editorial standards

Mantle shows how the news is being told across the political spectrum. These are the standards it holds itself to, and what you can expect when something goes wrong.

Independence

Mantle is created by Anthony Fiume and run with a team of editors. It is owned by Amo Per La Vita LLC, its creator's own company, and was built with its creator's own money. There are no outside investors and no billionaire owner.

Mantle takes no investment or funding from billionaires, political parties, campaigns or the outlets it rates. Nobody outside the team decides which stories Mantle covers, how it ranks them, how it rates an outlet or what its analysis says.

Mantle is paid for by its readers' memberships. When automated ads are added for readers who aren't members, political, gambling, weapons, alcohol and dating ads will be blocked, no ad will appear on a story that carries a warning, and members will never see ads.

How a story is made

Mantle reads the outlets on its outlet list, groups their articles about the same event into one story, and lays the coverage side by side: who is reporting it, what each side includes and how each side frames it.

The headline, summary and analysis are written by AI (Claude, made by Anthropic) under published rules. Mantle uses AI for one reason: to keep its own team's bias out. The same rules are applied to every story and every outlet, every time, and every story says that its headline and summary were written this way.

Mantle quotes only headlines and short excerpts from the outlets, and every one links to the original article. A story's image is chosen from the outlets' own images by published rules and credited to the outlet it came from.

People in charge

Mantle's editors read its stories and step in when something needs it:

• when the AI gets a fact wrong, or groups articles about different events into one story;
• when a story is driven by blatant hateful rhetoric, which is opinion, not journalism.

An editor who takes a story down must record why. The story leaves the site, the app, the feeds and the sitemap within minutes, and Mantle stops updating it.

Hateful rhetoric and rage bait

Mantle flags hateful rhetoric wherever it appears, quoted or not, with a "Read with care" warning, and it flags claims that the public record contradicts. The language guide explains the loaded and dehumanizing words it watches for.

Mantle does not profit from either: no ad appears on a story that carries one of these warnings.

Fairness

Every outlet and every part of the spectrum is measured with the same rules. Outlet ratings, the list of loaded words and the rules the AI follows are all published, and a change to an outlet's rating is recorded on that outlet's page. Mantle never tells you which side is right, and it never calls a single article biased: a rating describes an outlet's coverage as a whole.

Corrections

If something on Mantle is wrong, tell us and choose "A correction to a story". Include a link to the story and the source that shows what is wrong. A person reads every message.

When a story gets the facts wrong, an editor takes it down and records why. When the cause is one of Mantle's rules, the rule is fixed, and every change to the method is dated in the methodology's version history. A wrong outlet rating is corrected on the outlet's page, with the change recorded.

Privacy

Reading Mantle needs no account, and what you read is kept on your own device. The privacy policy says exactly what Mantle knows about its readers.