Current Status
CogniPrint currently has a working empirical evidence package supporting a follow-up manuscript. It should not be presented as a publication-ready empirical paper. The current readiness remains descriptive_only until a valid external non-owner review response is recorded.
What the Five Campaigns Mean
The five campaigns are controlled perturbation runs: each run compares a baseline text with one or more edited or variant texts in order to observe profile shifts, metric deltas, and stability tendencies.
They are not five independent publications. They are a cumulative empirical package used to prepare a follow-up manuscript.
- Campaign set: repeated controlled perturbation workflows.
- Comparison rows: individual baseline-versus-variant comparisons generated by the workstation.
- Campaign 004: the latest controlled series, contributing 1 series and 11 comparison rows.
Recommended Review Order
- Current state summary
- Public evidence snapshot
- Evidence visibility dashboard
- Evidence dossier
- Validation upgrade plan
- Public benchmark v1.1
- Public empirical growth layer
- Independent holdout layer
- Wave-005 descriptive validation
- Validation v1.2 correction layer
- Conventional stylometry baseline
- External review dispatch
- Evidence package index
- Public review bundle
- Colleague review checklist
Frozen Reviewer Packet
The current reviewer bundle is generated from the latest repository state with make reviewer-bundle. Initial archive-attachment attempts were blocked by Gmail policy; corrected no-attachment link-only reviewer requests were sent to two candidates on 2026-05-11, but no valid external response has been recorded yet.
Structured Feedback Path
Review feedback should preferably be submitted through the structured GitHub issue form so that framing, benchmark, and validation comments can be triaged consistently.
The external-review gate is satisfied only when a real non-owner response is saved under docs/external-review/responses/ and passes the repository checker.
Feedback Requested
- Is the empirical framing clear and appropriately cautious?
- Does the benchmark-versus-campaign bridge help interpret the current validation layer?
- Does the independent holdout layer improve confidence in source separation without overstating the result?
- Does validation v1.2 improve auditability, or does it add unnecessary complexity?
- Are the multi-draw random baseline and threshold sensitivity notes understandable and appropriately bounded?
- Do the cross-metric threshold summaries make the validation layer easier to review?
- Are the Methods, Results, and Limitations drafts readable enough for manuscript development?
- Is the evidence table useful for review?
- Which additional corpora or campaign types would make the follow-up manuscript stronger?
Review Loop Focus
The next review loop should treat the validation layer as manuscript support rather than as a separate results claim. The main question is whether the current benchmark-linked summaries, holdout layer, correction layer, and baseline comparison make the evidence easier to interpret for a scientific reader.
- Do the random-reference summaries clarify the contrast between within-campaign perturbations and cross-baseline mismatch?
- Do the threshold summaries help, or do they overload the narrative?
- Does the benchmark-versus-campaign bridge help discuss which perturbation families appear more or less corpus-bound?
- Does the benchmark-shift note suggest moderate stabilization, or does it still show strong benchmark sensitivity?
- Does the current external-review packet give enough information for a concise methodological review?
Interpretation Guardrail
The correct phrase is: working empirical evidence package supporting a follow-up manuscript.
The current package should not be described as a source-finality system, a forensic instrument, or a settled empirical paper.
Public site review note
The public site is limited to the research framing, evidence summaries, and review route. It does not change the scientific posture of the project.