Contact.

Every score is meant to be challengeable. If we mis-graded evidence, mis-extracted a claim, or applied the rubric wrong, tell us. We publish the corpus so the fix is visible in the diff to cards.json or a specific post annotation.

01Direct email

• [email protected] — score disputes, missing citations, algorithm critique
• [email protected] — everything else

02Feedback form

Fill in what you can. Submit opens your default email client with the message pre-filled — nothing is stored in your browser or sent to a server.

03How to challenge a score well

For disputes, what helps us triage fastest:

1. The account handle and the specific claim family name (visible in the expanded claim row).
2. The exact component you disagree with (A Evidence / B Scoping / C Uncertainty / D Counterevidence / E Commercial / F Safety).
3. At least one stronger or more-recent source than the one we cited — guideline, Cochrane review, or landmark trial.
4. If you're the account owner: any post we missed context on (a follow-up reply, a thread we only read the first tweet of).

Feedback that fits this template gets a response + either a corpus update or a clear explanation of why we scored it the way we did.

04What we don't respond to

• Demands that a specific account be scored higher without citing evidence we missed.
• Requests to remove an account's report because the owner disagrees with it.
• General grievances that the methodology is wrong without pointing at a specific component.

The corpus is published; the scoring is deterministic Python; both can be inspected and diffed. That's how the system earns its name — not by pretending to be definitive, but by being correctable in the open.