One Lab is built toward the constitutional purpose of an autonomous research organism. The current production system is not an autonomous end-to-end researcher: Housing has governed forward records; Quant has capture evidence but no issuance authority; Brain is materialized; AION-D is shadow advisory; AION-A, AOK and Learning remain inactive.
An eligible forecast requires an exact authority, resolution basis and source. A locked vintage cannot be edited or deleted, and the database assigns its time. When the named outcome becomes eligible, the grader compares records that already exist; before then it reports pending.
Forecasting is the most mature public mechanism here, but the current scorecard has not established accuracy, calibration or comparative performance. The public ledger keeps failures and holds visible, and it does not turn an empty grade denominator into a score.
The daily Observer evaluates ten declared drift tests. Code and schema boundaries also carry CI and mutation checks. A machine-decidable failure is retained in the decision ledger; questions a machine cannot settle remain marked for human review.
The current system has append-only evidence, schedules, capture paths, forecast locks, daily grading attempts and bounded advisory components. Those mechanisms do not establish research performance before eligible outcomes arrive.
Official benchmark Attempt #1 remains permanently consumed with no system output and no evaluation. Attempt #2 has not occurred. Public-only remediation cannot substitute for fresh independent acceptance.
Use the live scorecard for current forecast and grade counts. Use the operating report for current drift, heartbeat and human-review status.