RELEASE_OWNERSHIP_MATRIX.md

Release Ownership Matrix

1. Purpose

This matrix defines initial release ownership expectations for Permea Core and Permea-aligned projects.

The goal is to make release approval paths explicit while keeping exact person-by-person roster decisions deferred until governance assigns them.

2. Release Ownership Principle

Each release type should have named review roles before public release.

Higher-risk releases require more than routine maintainer review. Dataset, artifact, claim, paper, wet-lab, and derivative project releases require role-specific review and may require founder/manual approval.

Approval means the scoped release passed the recorded checks. It does not create legal, clinical, validation, authorship, or dataset redistribution conclusions.

3. Release Type vs Approver Matrix

Release Type	Maintainer	Claim Reviewer	Dataset / Provenance Reviewer	Reproducibility Reviewer	Paper / Source-to-Claim Reviewer	Release Approver	Founder / Manual Approver
docs-only	required	required if claims change	required if data/release wording changes	optional	required if paper/public wording changes	optional	required if public-sensitive
code-only	required	required if claims change	required if data handling changes	required if reproducibility is claimed	optional	optional	required if high-impact
benchmark/schema	required	required	optional unless data scope changes	required	optional	optional	required if public benchmark standard changes are high-impact
result artifact	required	required	required if row-level or source-derived artifacts are involved	required	required if paper-supporting	required	required if public-sensitive or high-impact
dataset	required	required	required	required if reproducibility is claimed	required if paper-supporting	required	required
paper-support	required	required	required if data or artifacts are discussed	required if rerun claims are made	required	required	required for public preprint/package decisions
preprint/publication	required	required	required	required where reproducibility is claimed	required	required	required
derivative project	required	required	required if data/artifacts are involved	required if benchmark or result claims are involved	required if paper/preprint-linked	required	required for official branding or high-impact status
wet-lab collaboration	required	required	required	optional unless computational rerun claims are made	required if paper/preprint-linked	required	required

If a release spans multiple types, use the strictest applicable row.

4. Role Definitions

Maintainer

Verifies repo scope, file scope, review completeness, merge mechanics, and recordkeeping.

Claim Reviewer

Checks docs/scientific-governance/CLAIM_REGISTRY.md and docs/EVIDENCE-LADDER.md.

Dataset / Provenance Reviewer

Checks source, license, attribution, redistribution status, row-level artifact posture, and partner or institutional constraints.

Reproducibility Reviewer

Checks run surfaces, environment capture, benchmark contract alignment, run manifest status, result artifacts, and reproducibility wording.

Paper / Source-to-Claim Reviewer

Checks manuscript, supplement, README, website, release notes, citation support, source-to-claim mapping, and public-readiness wording.

Release Approver

Confirms required gates have passed or blockers are recorded before release.

Founder / Manual Approver

Handles high-impact release, dataset, public preprint, authorship, partner, official branding, wet-lab collaboration, and unresolved governance decisions.

5. Required Approvers by Release Type

Docs-only releases may be approved by a maintainer when they do not change claims, data posture, release state, paper status, or public-facing scientific interpretation.

Code-only releases require maintainer review and reproducibility review when the code supports benchmark or rerun claims.

Benchmark/schema releases require maintainer, claim, and reproducibility review.

Result artifact releases require maintainer, claim, reproducibility, and dataset/provenance review when artifacts are source-derived or row-level.

Dataset releases require maintainer, claim, dataset/provenance, public-safe artifact, release, and founder/manual approval.

Paper-support and preprint/publication releases require paper/source-to-claim, claim, dataset/provenance, release, and founder/manual approval.

Derivative project releases require derivative review, release review, and founder/manual approval when official branding, public release, paper linkage, dataset release, or partner-sensitive status is involved.

Wet-lab collaboration releases require claim, dataset/provenance, partner-sensitive review, release approval, and founder/manual approval before public claims.

6. Escalation Rules

Escalate when:

• reviewers disagree on release readiness
• source/license/redistribution status is unresolved
• row-level artifacts may be exposed
• claims imply stronger evidence than documented
• paper/preprint readiness is unclear
• official Permea branding is requested
• wet-lab collaboration, partner, or institutional data are involved
• authorship, reviewer credit, or contributor credit is disputed
• sensitive, private, or confidential information may be exposed

Escalation should route to maintainers first, then founder/manual approval for high-impact unresolved items.

7. Recordkeeping Requirements

Release ownership records should include:

• release type
• release state
• repo and paths
• included files
• excluded files
• required reviewers
• completed reviews
• blocking findings
• unresolved caveats
• manual approval status
• decision
• commit, tag, pull request, issue, or report reference

Records must be public-safe and must not expose secrets, private data, restricted row-level artifacts, or confidential partner information.

8. Deferred Exact Roster Note

This matrix defines roles, not a final people roster.

Exact maintainer, reviewer, release approver, founder/manual approver, legal/manual, partner, and institutional approval assignments remain deferred until governance records them.

Until the roster is explicit, high-impact or ambiguous releases should remain Hold or Internal Review.

Claim-Boundary Reminder

Permea may use "AlphaFold for Delivery" as ambition or positioning only.

Permea must not claim AlphaFold-level performance, adoption, or standardization; completed wet-lab validation unless documented; clinical efficacy; universal delivery prediction; production-grade drug delivery platform status; or dataset redistribution permission without source/license approval.

Benchmark claims must remain scoped to dataset, split, metric, model, artifact, and evidence level.

Derivative projects must not inherit claim authority automatically. Permea-aligned does not mean Permea-validated. Wet-lab collaboration interest does not equal wet-lab validation.