Receipts-based writing about AI writing. No detector snake oil, no humanizer spam. Every claim sourced, every experiment shown.
The mark proves processing, not authorship. A grammar-checked essay carries the same mark as a ghost-written one. The full honest breakdown.
The word is entering the dictionary in real time. Definition, origin, why it is not detector evasion, and the five-pass method.
The four situations every webpage falls into, with the exact reference and in-text citation for each. Covers no author, no date, and the site-name rule most generators break.
Detectors measure how predictable your writing is, not who wrote it. Researchers took a 5% flag rate to 57% by simplifying vocabulary alone. What the numbers say, and what evidence actually works.
Internal stages, the Completion of Procedures letter, then the free OIA route with its twelve-month deadline. Why to argue process rather than science.
The National Student Ombudsman has taken complaints since February 2025 and has no time limit. Plus the Australian universities that switched Turnitin detection off.
Harvard is a family of styles, not one style. The Cite Them Right pattern, the 2025 change to place of publication, and the accessed date that APA does not want.
When an organisation is both author and publisher, MLA removes the author entirely. Almost every free generator breaks this. Ours did too.
Our own experiment: AI copy scored 222 tells per 1,000 words. Lincoln scored 3.7. Darwin scored zero. The full dated lexicon, with the fix for each habit.
The false-positive numbers, the court cases, and what you can actually do. Coming this week.
See your own rhythm skyline and learn why flat rhythm reads robotic. Coming this week.