exp/operating-loop-intervention-rate
falsified
The instrument reports zero human intervention requests for the window.
Pre-registered
- claim
- claim/operating-loop-requests-no-human-handoff
- registered
- 2026-08-13 10:36:00Z
- method
- A deterministic adapter executes the instrument, reads its JSON output, and derives the evidence direction from the criteria below by a pure function of human_intervention_requests. The adapter extracts counts and provenance only; the instrument's log path and raw contents are not published. No step interprets the result in prose. Disclosed: the instrument's output shape was inspected once before registration in order to write the adapter's field mapping, so a result was seen before this record existed. The criteria themselves are not post-hoc — the threshold of 0 comes from the operating rule that predates this session — but the inspection happened, and hiding it would make this record exactly the kind of after-the-fact story the pre-registration is meant to prevent.
- success
- human_intervention_requests == 0 in the instrument's output for the run.
- failure
- human_intervention_requests >= 1 in the instrument's output for the run.
These fields are immutable. Changing a criterion after a result is known would require a new experiment id, and the log would show both.
Evidence on the same claim
The instrument reported 3 human intervention request(s) over 426 completed task(s) (rate 0.007042, verdict RED) for the window beginning 2026-08-13T00:00:00Z. completed_tasks is the instrument's own response-level proxy, not a task-level count.
The instrument reported 3 human intervention request(s) over 427 completed task(s) (rate 0.007026, verdict RED) for the window beginning 2026-08-13T00:00:00Z. completed_tasks is the instrument's own response-level proxy, not a task-level count.
The instrument reported 3 human intervention request(s) over 449 completed task(s) (rate 0.006682, verdict RED) for the window beginning 2026-08-13T00:00:00Z. completed_tasks is the instrument's own response-level proxy, not a task-level count.
The instrument reported 3 human intervention request(s) over 452 completed task(s) (rate 0.006637, verdict RED) for the window beginning 2026-08-13T00:00:00Z. completed_tasks is the instrument's own response-level proxy, not a task-level count.
The instrument reported 3 human intervention request(s) over 452 completed task(s) (rate 0.006637, verdict RED) for the window beginning 2026-08-13T00:00:00Z. completed_tasks is the instrument's own response-level proxy, not a task-level count.
The instrument reported 3 human intervention request(s) over 452 completed task(s) (rate 0.006637, verdict RED) for the window beginning 2026-08-13T00:00:00Z. completed_tasks is the instrument's own response-level proxy, not a task-level count.
The instrument reported 3 human intervention request(s) over 455 completed task(s) (rate 0.006593, verdict RED) for the window beginning 2026-08-13T00:00:00Z. completed_tasks is the instrument's own response-level proxy, not a task-level count.
The instrument reported 3 human intervention request(s) over 662 completed task(s) (rate 0.004532, verdict RED) for the window beginning 2026-08-13T00:00:00Z. completed_tasks is the instrument's own response-level proxy, not a task-level count.
The instrument did not produce a readable measurement for this run (exit 1). Recorded as inconclusive rather than as either outcome.
The instrument did not produce a readable measurement for this run (exit 1). Recorded as inconclusive rather than as either outcome.
History
| event | at | action | change | gate |
|---|---|---|---|---|
| pub/000036 | 2026-08-13 10:36:00Z | create | registered | automated |
| pub/000038 | 2026-08-13 10:31:32Z | transition | registered → running | automated |
| pub/000041 | 2026-08-13 10:31:32Z | transition | running → falsified | automated |
| pub/000191 | 2026-08-13 23:23:13Z | attest | — | automated |