Manuscript watermarking

Every manuscript delivered to a beta reader is watermarked with two invisible layers identifying both the author and the reader. This deters unauthorized sharing without altering the reading experience.

How watermarking works

Watermarks are embedded at the character level using Unicode variation selectors and zero-width characters. They are invisible in normal reading but detectable by both AI analysis and visual inspection tools.

What gets watermarked

The watermark is applied to each scene's rendered HTML at delivery time — not stored in the database. Every reader gets a uniquely watermarked copy of the manuscript.

Dual-layer encoding

Layer 1 encodes the author's account identifier.
Layer 2 encodes the reader's account identifier and campaign ID.

If a watermarked copy surfaces somewhere it shouldn't, you can decode both layers to determine which campaign and reader it came from.

Limitations

Watermarks survive copy-paste but do not survive re-typing from scratch or OCR-then-retype workflows. They are a deterrent and attribution tool, not a technical prevention mechanism.