Trust & Review Packet
A compact packet for outside review.
This page gives researchers, journalists, campus groups, civil-rights organizations, and technical reviewers a narrow way to inspect the public-source archive without treating review as endorsement. Use the Impact page for the proof summary and Updates for the chronological maintenance log.
Records selected for adversarial review are published in the challenge arena; queue order is for review workflow only.
Current Proof Package
- Dataset scale
- 4,000 public-source records across 947 schools, 25 source records, and 30 research briefs.
- Snapshot hash
sha256:78f91ad886c8cf3024e25989b79d251b519077afb4b87f8d44e95ea5413cbee2- Review artifacts
- quality page, methodology, public codebook, coverage limits, evidence robustness dashboard, evidence provenance page, methodology stress test, replication packet, credibility boundaries, downloads, impact summary, product updates, live source audit, release notes, and changelog.
- Responsible-use guardrail
- Research Guide explains why record counts are documentation signals, not rankings, safety scores, or prevalence estimates.
What A Reviewer Can Audit In 30 Minutes
- Read the inclusion, exclusion, source hierarchy, confidence, verification, privacy, and no-ranking rules in the methodology.
- Sample five event records and confirm that source URLs, school metadata, dates, categories, affected communities, and confidence labels match the public source basis.
- Check whether a research brief uses cautious language and avoids converting public documentation into incident prevalence.
- Inspect the source audit and changelog to see whether the dataset is reproducible and correction-friendly.
Review Tasks
Methodology review
Test whether the rules are clear enough for another reviewer to accept, reject, or correct a record without private judgment calls.
Source audit review
Check whether source types, publishers, dates, URLs, and source-to-record links are accurate for a small sample.
Classification review
Compare event category and affected-community labels against the public source text and report any overbroad or stale labels.
Research-guide review
Use the archive for one narrow question and identify where the guide should prevent stronger claims.
What Trust Signals Prove
- The public archive has deterministic files, generated pages, schemas, hashes, source audits, release notes, and launch checks.
- The project has a visible workflow for source submissions, corrections, duplicate reports, school metadata fixes, and reviewer feedback.
- Outside reviewers can inspect the standards before choosing whether to comment publicly.
What They Do Not Prove
- They do not prove that the archive is complete.
- They do not prove that one school is safer, more dangerous, better, worse, or more hostile than another school.
- They do not prove legal liability, institutional intent, or incident prevalence.
- They do not imply partner endorsement.
Reviewer Entry Points
- Impact proof summary
- Public codebook
- Coverage limits
- Evidence robustness dashboard
- Evidence provenance page
- Methodology stress-test memo
- Replication packet
- Credibility boundaries
- Press / Research Brief
- Public product updates log
- Reviewer Brief for outside critique
- Journalist Use Guide
- Markdown reviewer packet
- Reviewer checklist issue template
- One-page outreach email draft
- Partner acknowledgment policy
- Public acknowledgments page
Acknowledgment Rule
Campus Evidence Lab should list reviewers, organizations, or collaborators only after documented review or collaboration is complete. Prospective outreach, informal praise, private encouragement, or unanswered emails are not public impact.