Moderation Policy

1. Evidence integrity — The authoritative corpus must not be contaminated by unreviewed submissions.
2. Human dignity — People experiencing housing instability are not data points; their experiences deserve respectful treatment.
3. Epistemic honesty — Advocacy and lived experience are valuable but distinct from peer-reviewed research; our labels must reflect those distinctions.
4. Non-retaliation — Rejected submissions are not publicly identified or shamed.

All community input submissions enter a PENDING_REVIEW state before any editorial action is taken. Moderators assess:

1. Spam check — Automated score ≥ 70 triggers silent rejection; borderline scores (40–69) are reviewed by a human.
2. PII assessment — Any submission flagged for possible PII is reviewed before any processing.
3. Doxxing assessment — Any submission containing apparent location information about private individuals is reviewed for safety.
4. Relevance — Is this within Bay Evidence's scope?
5. Duplicate detection — Does this duplicate an existing record?
6. Source quality — If a source is suggested, does it meet minimum standards for entering the discovery pipeline?

Regardless of the submission's other qualities, Bay Evidence will not publish:

• Personal identifying information (name, address, phone, email) of private individuals
• Precise locations of shelters, safe houses, or individuals experiencing homelessness
• Unsubstantiated allegations against named individuals
• Content designed to harass, target, or expose vulnerable people
• Information that could reasonably be used to locate or harm someone
• Political advocacy content without an evidence basis (advocacy research is labeled as such and its source's institutional position is visible)

Submissions that present safety concerns are escalated immediately to the senior editorial staff. Examples:

• A submission appears to contain a threat against a named individual
• A submission appears to include a precise location for a domestic violence shelter
• A submission appears to be an attempt to identify a specific person's current location

Escalated submissions are not processed further until a safety review is complete.

Submitters whose submissions were rejected may:

1. Resubmit with additional context (for relevance rejections)
2. Open a GitHub issue to request reconsideration (for source-quality rejections)

Bay Evidence will not reinstate submissions rejected for safety, spam, doxxing, or harassment violations.

Moderators commit to:

• Reviewing submissions based on their merit relative to the policy, not based on whether they agree with the submitter's perspective
• Not publicly identifying or shaming rejected submitters
• Maintaining confidentiality of contact information
• Escalating safety concerns promptly
• Documenting decisions in the moderation record for audit purposes
• Recusing themselves from reviewing submissions about topics where they have a personal conflict

Bay Evidence will maintain public transparency about the moderation system by:

• Publishing this policy
• Publishing COMMUNITY_INPUT_POLICY.md
• Maintaining the moderation workflow in EVIDENCE_PROMOTION_WORKFLOW.md
• Making submission states visible to the submitter where feasible
• Reporting aggregate moderation statistics annually (e.g., total submissions, approval rates, rejection categories) without disclosing individual submissions

See also the editorial standards and how to contribute.

2026-07-10 — Initial policy created.