Repetition occurs at two levels
Surface repetition includes the same short paragraphs, contrarian hooks, em dashes, and constructions that reject one phrase in order to install another. Deeper repetition occurs when many pieces make the same claim with no new evidence or mechanism. The second form is more damaging because a stylistic rewrite can hide it. The voice can be sanded until it sounds original and still say nothing.
An audit should therefore separate rhetorical similarity from contribution. Two posts may look different while offering the same unsupported prediction. Two posts may share a format while one contributes a dataset or a counterexample. Format is not the problem. Empty certainty is.
Score contribution, not novelty theater
For each piece, identify the central claim and ask whether it contributes at least one of four things: original evidence, a causal mechanism, a boundary condition, or an actionable decision rule. A piece with none of them is restating the feed, however confident the packaging.
This is an editorial screen, not a universal index of writing quality. Domain expertise, clarity, and synthesis still matter, and familiar ideas can be valuable for a new audience. The score only exposes when confident packaging is doing the work that evidence should do.
If the author is deleted and nothing remains, the line was branding
Claims of personal candor, authority, or courage often occupy the space where evidence should appear. Phrases that ask the reader to trust the writer's honesty or bravery are asking the reader to evaluate the writer instead of the argument.
A useful edit removes the self-description and checks whether the sentence still contains information. If nothing remains, the line was branding. Replace it with the observation, the method, or the limitation. The author can remain in the piece as a witness to a measurement. The author should not be the proof.
A five-minute editorial check
Underline every claim, number, and recommendation. Connect each claim to evidence, each number to a denominator and a consequence, and each recommendation to a boundary condition. Then mark any paragraph whose main subject is the author rather than the topic.
The remaining gaps identify the rewrite. Add evidence where it exists, narrow the claim where it does not, and delete sentences whose only function is to sound certain. If that process guts the piece, the piece was already hollow. Better to find that in edit than in public.