Developer finds writing linter's score can be manipulated by padding word count

A developer tested their own automated writing checker on a blog post and discovered its verdict changed based on word count alone. The tool, which assesses clarity and formatting, passed the full post but flagged the same prose for review when non-content metadata was removed. Further controlled tests showed that adding neutral filler sentences lowered the overall score, allowing posts to pass by diluting the density of flagged elements like em dashes. The issue stems from the tool combining set-based phrase checks with density-based formatting checks into a single score. This means the scoring threshold's meaning drifts depending on post length, a flaw the developer has since addressed.