Every draft is scored. Every
flagged section rewrites itself.

A section crosses the threshold when its phrasing, structure, and overlap with a single source exceed the configurable similarity ceiling. The default is set conservatively, well below the level where any reader would notice resemblance. Teams in higher risk categories, legal, healthcare, finance, can tighten it. Teams where consensus phrasing is unavoidable can loosen it. The team controls it.

Because fifty is where the comparison is genuinely comprehensive rather than superficially defensive. Comparing against the top three or ten catches the obvious overlap but misses the long tail: the competitor on page four whose article phrases a claim almost identically. Going further compares against pages with minimal relevance and produces false positives. Fifty is the calibrated balance.

When a section is flagged, Ranko generates a new version of just that section, with meaning, brand voice, and research intact but the phrasing shifted far enough from the matched source to clear the threshold. The new section is rescored against the top fifty plus the internal library. If it passes, the rewrite is accepted. If it still fails, it happens again with context on what did not work. The loop usually terminates in one or two iterations.

The loop terminates with a clear escalation to the team after a configurable maximum number of attempts. This is rare, but when it happens the team gets a flag showing the section that could not clear the threshold, the sources it kept matching, and the closest passing version Ranko produced. The team can decide whether the phrasing is unavoidable, whether to cut the section, or hand it to a human.

Yes, every rewrite is recorded in the audit trail with the before version, the after version, the similarity score at each step, and the sources the original section matched. The team can click any section of any article and see the full history: the first draft, what was flagged, how the rewrite shifted the phrasing, the new score. Compliance teams have a complete trail to point at. The check is never a black box.

Passive checkers tell you there is a problem and leave the rewrite to you. Ranko fixes it before the team ever sees it. Traditional tools produce a score, highlight the matched passages, and stop, then the team has to manually rewrite, rerun, and hope it passes. That is where problems accumulate. Ranko's check is active: section rewrites are part of the check. The article does not ship until originality is verified.