Self-Preference
Asked to judge, a model scores text from its own family higher — it recognises its own habits as quality.
ExampleThe same essay wins when the judge wrote it and loses when a rival did.
What to doDon’t let a model judge its own work. Use a different model, a person, or a check with a known right answer.
Paste this into the chatI will show you two texts. Don’t guess who wrote them. Score each one against these criteria only, quoting the passage that justifies every score.