Claims nobody has tested yet. Blinded tests anyone can run, published either way.
Some claims have no trial behind them at all — not a bad one, none. We don’t grade those; a grade needs evidence. Instead we list the claim, who makes it, and an experiment designed to test it: pre-registered, blinded where it can be, and published whatever it shows. The experiments run on HiveJournal, where anyone can copy one and submit a result.