Protocol · public summary

A research trail that can be followed.

This is the launch summary. It will be versioned when the full protocol source is added and whenever a methodological decision changes.

Phase 1

Recruitment gate

Collect expressions of interest from a trusted initial cohort. Review eligibility, willingness to attest candidly, and evidence availability before inviting full submissions.

Gate: do not proceed to full data collection without viable cohort buy-in.
Phase 2

Consent and intake

Record the active consent version and separate choices for attribution, excerpts, public-data inclusion, and future contact. Freeze the submitted article text at intake.

Gate: consent and article-specific attestation must be complete.
Phase 3

Detector run

Verify the frozen text, record the Pangram interface or version and scan date, retain the overall result and any segment-level flags, and attach a screenshot when practical.

Rule: rescans create new run records; they do not overwrite history.
Phase 4

Review

Compare detector output with the attested workflow. Record qualitative failure modes and a distortion score. Blind review may be used where practical.

Rule: exclusions require a recorded reason.
Phase 5

Freeze and analyze

Produce a coded analysis export, a separate contact export, an evidence manifest, and a publication-safe export. Preserve the analysis version used for reported findings.

Rule: identity fields do not enter routine analysis exports.
Phase 6

Publish and correct

Report aggregate results, representative cases within permission, limitations, detector version information, and any null or contrary result. Date later corrections.

Rule: methodological updates remain visible.

Protocol status

Pre-registration draft

Public summary version
0.1 · 22 July 2026
Recruitment
Interest phase
Results
Not yet available