LearnNewsExamplesServices
Frontmatter
id11267
titleCross-family corrective-authorship rotation and FAIR split discipline
stateClosed
labels
enhancementaiarchitecturemodel-experience
assigneesneo-gemini-3-1-pro
createdAtMay 12, 2026, 2:54 PM
updatedAtMay 13, 2026, 7:26 AM
githubUrlhttps://github.com/neomjs/neo/issues/11267
authorneo-gemini-3-1-pro
commentsCount2
parentIssuenull
subIssues[]
subIssuesCompleted0
subIssuesTotal0
blockedBy[]
blocking[]
closedAtMay 13, 2026, 7:26 AM

Cross-family corrective-authorship rotation and FAIR split discipline

Closedenhancementaiarchitecturemodel-experience
neo-gemini-3-1-pro
neo-gemini-3-1-pro commented on May 12, 2026, 2:54 PM

Context Discussion #11265 surfaced the need for a cross-family corrective-authorship rotation alongside a FAIR ticket/PR split discipline. Layer 1 (skill-substrate-health) is now covered by #11275 (machine-readable manifest + CI lint). This ticket handles Layer 2: the cross-family corrective-authorship rotation tracking contract.

The Problem We need to formalize cross-family corrective-authorship rotation in our substrate without creating a permanent hierarchy or adding ambiguous tracking bloat.

The Fix

  1. Implement the Layer 2 tracking contract (5 signals).
  2. Update pull-request-workflow.md §6.2 (or introduce §6.2.1) to document the interim 3-lane distribution and cross-family corrective-authorship rotation.
  3. Establish metric tracking and sunset clauses per the tracking contract.

Layer 2 Tracking Contract — 5 Signals

Signal Primary substrate (GitHub) Memory Core role
Duplicate close-target / parallel execution gh pr list --search / #11265 comment link A2A [lane-claim] ev. / validation
Same-author correction Durable tag/comment pattern on PR Cross-session aggregation
Operator-direction / Author-yield Tagged PR review comment / Mention A2A audit trail for tracing
N≥10 sunset query PR selection query filter Annotation layer overlay
Durable tag/comment schema New stable tag vocabulary Memory Core ingests the tag

Acceptance Criteria

  • Implement AC-CycleA: Quota-guard discipline is documented as a pressure/churn signal, NOT a PR-count fairness scoreboard.
  • Implement AC-CycleB: Duplicate-PR hard stop. Corrective author MUST check for active PR or lane claim before opening a parallel PR.
  • Implement AC-CycleC: Narrow activation. Corrective rotation activates ONLY on operator-direction OR explicit author-yield.
  • Implement AC-CycleD: Metric-based sunset. Sunset criteria explicitly defined as N ≥ 10 PRs across ≥ 2 sessions with zero map-vs-atlas violations.
  • Implement AC-CycleD-Layer-2-tracking: Additive 5-signal tracking contract implemented (subordinates to #11275's Layer 1 enforcement substrate).
  • Implement AC-CycleE: Dogfood the proposal. (Satisfied by @neo-gemini-3-1-pro authoring this ticket and the resulting PR).
  • Substrate additions are context-neutral or reducing (per Discussion #11259 AC).

Out of Scope

  • Fixing any other skills or rewriting existing pre-flight skills during this ticket (Layer 1 is handled by #11275).
  • Establishing permanent hierarchy or supervision protocols.

Avoided Traps

  • Relying on same-harness correction for substrate-budget exposure gaps (rejected via empirical evidence in #11257).
  • Event-based sunset criteria (rejected for metric-based via AC-CycleD).
  • Framing as model-attribution instead of harness-attribution (corrected via operator calibration).

Related

  • Discussion #11265 (Graduation source)
  • Discussion #11259 (Substrate-budget AC)
  • Issue #11275 (Layer 1 skill-substrate-health enforcement)
  • PR #11264, PR #11266 (Empirical anchors)

Origin Session ID: 57502eb2-7f7b-4b9b-a849-49f016b08c95 Retrieval Hint: "cross-family corrective-authorship rotation and FAIR split discipline"

tobiu referenced in commit d3a91bd - "feat(workflow): implement Layer 2 corrective-authorship rotation (#11267) (#11277) on May 13, 2026, 7:26 AM
tobiu closed this issue on May 13, 2026, 7:26 AM