What the Pack Is
The team evidence pack assembles the artifacts produced by team-level practice across all three Sub-Arc 3B units into a single, structurally organized collection. Its purpose is proof: evidence that the practice is real — that moral-hazard awareness is not theater, that cooperation is genuinely chartered, that enforcement operates as designed, and that the structure is being maintained against drift.
The pack is governed by the principle that runs through all of Arc 3: processes public by default; persons private by default. The team publishes what it looks for (processes), how it looks (structure), and what it finds (outcomes) — while protecting individual data from exposure.
Private Artifacts (Accountability Layer)
These are the team's own records — maintained for accountability to self, not for external exhibition.
From (2.3B.3) — moral-hazard awareness: Goodhart map assessments (completed quarterly), counter-gaming pattern audits, metric rotation logs, hazard scan worksheets, capture resistance test results, and cultural signal audit findings.
From (2.3B.2) — encoded objective law and selective transparency: Decision audit logs (every enforce-in-code decision with timestamp, decision-maker, reasoning), transparency matrix reviews (what changed in classifications and why), encoded objective law records (checklists used, violations with reasoning), and exception logs (privacy exceptions with justification, scope, and sunset status).
From (2.3B.1) — engineered and bounded interdependence: Team charter amendments (changes to roles, decision rights, interfaces, consent validity), entry and exit records (onboarding outcomes, clean-exit protocol adherence, portability issues), interface map updates, and payoff table reviews.
The private artifacts layer is the team's mirror. If artifacts exist and are being updated, the practice is real. If artifacts are stale or absent, the practice has evaporated — and the evidence pack is the mechanism that makes that drift visible.
Public Attestations (Verification Layer)
These are outcomes that external parties can verify — proof that the team's structure is holding.
Turnover and retention. Departing members report fair treatment. Exit interviews, anonymous surveys, or attestations from subsequent employers can confirm whether people believe they were treated fairly — clear entry and exit terms honored, portable work preserved, decisions transparent, standing to contest.
Quality stability. Shipped quality matches claimed metrics. Customers, downstream teams, or external auditors can confirm whether quality is stable or degrading. A widening gap between internal metrics and external perception signals stage-2 or stage-3 capture.
Metric honesty. External auditors look for the patterns described in stages 1 and 2 of capture: gaming discussed openly as strategy, scope creep in definitions, casual references to "padding estimates" or "narrowing defect categories." If these patterns are absent, metrics are probably honest.
Decision auditability. Decisions affecting team members are traceable to published criteria with visible reasoning. An external audit of a sample of decisions (hiring, promotion, compensation changes) can verify whether the team's process matches its claims.
Sub-Arc 3B Toolkit Sequence
Table 2.3B.3-4 — Complete toolkit inventory: structure → enforcement → maintenance
| Unit | Function | Toolkits | Cadence | Artifact Classification |
|---|---|---|---|---|
| (2.3B.1) | Structure | Team charter, entry/exit/portability terms, interface map, payoff table, participation test, restraint as cooperation boundary | Charter review: quarterly (new team) or annually | Private: charter, terms, map. Public: published charter |
| (2.3B.2) | Enforcement | Enforce-in-code decision tree, transparency matrix, dispute path, Under AI Conditions | Transparency audit: quarterly. Decision log: continuous | Private: decision logs, exception logs. Public: transparency matrix, audit summary |
| (2.3B.3) | Maintenance | Goodhart map, culture as team signal, capture resistance test, counter-gaming patterns | Goodhart map: quarterly. Cultural signal audit: quarterly | Private: hazard scans, audit worksheets. Public: metric honesty attestations |
Assembly and Maintenance
The pack is assembled incrementally — every discipline that produces an artifact feeds the pack. No additional documentation burden beyond what the practice itself generates.
Quarterly review cadence. Synchronized with the Goodhart map cycle. At each quarterly review: are private artifacts current? Have external parties confirmed the structure is holding? Are any artifacts stale — not updated in ninety days despite the discipline being active? Stale artifacts signal practice drift.
Privacy preservation. Names are removed from decision records. Dispute details are redacted to structural outcomes. Performance data is aggregated. The pack demonstrates structural integrity without exposing the persons whose work the structure governs.
Failure mode: pack theater. Maintaining artifacts without maintaining the practices they represent. A team can have perfect Goodhart maps and capture resistance tests if no one is actually running them. The test: Does the evidence pack reflect a living practice, or has it become a filing system for unused templates?
Change Log
- 19 Mar 26 — Initial publication.