Most authors will never be asked to demonstrate that they wrote their book, and the record that would answer it is produced automatically by ordinary work. The only requirement is not deleting it, and knowing which parts are worth keeping. This is a practical note rather than legal advice.
What a trail looks like
| Evidence | Strength |
|---|---|
| Dated drafts across months | Strong; shows development |
| Version control history | Strong; timestamped and granular |
| Notes and outlines | Strong; shows thinking |
| Research material | Supporting |
| Correspondence about the book | Supporting, and dated by the service |
| Beta reader exchanges | Supporting, and third-party dated |
| A single final file | Weak on its own |
| Anything produced after the question | Weak |
The pattern is continuity. A book that exists in twenty dated versions with notes and correspondence around it tells a story that a single file cannot.
The habit
- Keep dated drafts rather than overwriting one file.
- Use version control if you can, which does this automatically and thoroughly.
- Keep your notes, including the abandoned ones.
- Keep correspondence with readers and editors.
- Do not tidy up. The mess is the evidence.
- Back it up, since the trail is only as durable as its storage.
Point five deserves stating plainly. Authors delete old drafts to save space or to feel organised, and the old drafts are precisely the record that would answer this question.
A small supporting artefact
The Novel Kit exports a dated term list from any draft: a compact record of the invented vocabulary a manuscript contained at a point in time. It's a minor item in a trail, and it is small, dated and specific, the properties evidence wants.
The wider documentation practice is in how to document your writing process.
A practical next step
Stop overwriting your working file. Save dated versions instead, and the trail builds itself.
Sources checked
- How to document your writing process
- Manuscript backup and version control
- What publishers now ask about AI use
Frequently asked questions
Who asks this question?
Publishers with a warranty clause, competition organisers, and occasionally a dispute about authorship or originality.
What is good evidence?
A trail: dated drafts, version history, notes, research, correspondence. Volume and continuity matter more than any single item.
What is weak evidence?
A single finished file, and anything you produced after the question arose. Contemporaneous records are what count.