opuscript.com/authorship · In development
The same record, for written work.
A bound book can carry a provenance record. So can the text inside it. Who wrote it, over what period, from what sources, with what tools doing what — signed, dated, and structured against a published standard.
Why written work needs the same treatment
The provenance standard exists because generative tools made “handmade” costless to assert. The identical thing happened to “written by”, faster and at greater scale, and detection tools for text are demonstrably unreliable in both directions — they falsely accuse human writers and clear machine text routinely.
The answer is the same: not detection, but a record produced during writing, by the writer, of things only that person could know. Drafts and their dates. Sources consulted and rejected. The argument that was abandoned and why. Which tools were used, for what, and what they decided.
Who it is for
- Independent authors and publishers who want a defensible claim about how a manuscript was produced.
- Commissioning editors and clients buying written work who want the disclosure in writing rather than implied.
- Researchers and expert witnesses whose written output may be challenged.
- Educators who would rather require a process record than run a detector that does not work.
This will not certify that a text was written without AI, and it will not detect that it was. It applies the same disclosure rule as the craft standard: tools that execute need only be named; systems that made shaping decisions must be disclosed along with what they decided. A record showing disclosed, bounded assistance conforms. An undisclosed one is void.
Join the list
The standard is being drafted and a small number of pilot records will be produced with working authors before publication. No launch date is promised. If you would like to be told when it opens, or offered a pilot place, leave an address.