# Receive a candidate comparison · individual record

Use an invented alias. Keep a separate record per person and retain the first
account before discussion. Published expectations are practice, not an unseen test.

- My alias, role and prior reading: [not recorded]
- Program copy, exact command and original output file: [not recorded]
- Prediction before execution or feedback: [not recorded]
- Baseline/candidate versions, planned roster and rubric: [not recorded]
- Overall and routine/complex correct/executed counts: [not recorded]
- Matched IDs, improvements and regressions: [not recorded]
- Missing IDs; duplicate/refused records; actual unrun work: [not recorded]
- Observed subset result versus possible completed-roster range: [not recorded]
- Possible match-count bounds and direction claim before execution: [not recorded]
- Full-plan conclusion supported by every completion; remaining real-work assumptions: [not recorded]
- One changed missing ID, prediction and actual result: [not recorded]
- My independent explanation, correction, help and observation-only work: [not recorded]
- Peer role swap and their independently retained account: [not recorded]
- Evaluation owner — reconcile IDs, labels and all pending trials: [not recorded]
- Domain owner — review critical errors and the receiving workload: [not recorded]
- Service owner — candidate/rubric versions, reset state and rollback evidence: [not recorded]
- Receiving team — capacity, stop conditions and accepted scope: [not recorded]
- Pod lead — next discriminating test, responsible person and review point: [not recorded]

An absent output remains unknown. A successful command does not authorize release,
establish independent field performance or confer a credential or client placement.