It exports as TWFF (Tracked Writing File Format) open container format. We are aligning the format with W3C Data Privacy Vocabulary (DPV) terminology where appropriate.
The motivation: GenAI text detectors are terrible.
What researchers actually need is process-level data: what changed during writing, where GenAI was used, and how the author engaged with it.
We are preparing multi-site case studies for higher education and would appreciate feedback on the methodology and format.
Demo: https://colophon.firl.nl/demo
Chrome extension: https://chromewebstore.google.com/detail/colophon/manpdfemof...
TWFF specification: https://github.com/Functional-Intelligence-Research-Lab/twff
I'm particularly interested in feedback from people working in HCI, learning sciences, writing research, educational technology, or anyone who has built research infrastructure on top of Google Docs.