Round r62b
r62b
Source-reported narrativeRound README and results text are source material. Numeric values that are not tied to a claim ledger record remain source-reported.
Round source README · rounds/r62b/README.md
Round date: HISTORICAL (before 2026-10-09) Publication badge: INVALID; KEPT FOR AUDIT Why not VALID: The historical inventory retains INVALID because of COMPARATOR, NO_PREREG, UNMATCHED_CONDITIONS. See the round evidence and limitations below. Recomputation status: OBSERVED SOURCE ONLY
Recomputation update: The mined raw records now provide per-cell outcome, manifest token-counter, and wall-time recomputation for the observed cohort. This does not reconstruct missing planned cells, a full intention-to-treat denominator, or model execution.
Status
INVALID
- NO_PREREG — The README records no pre-registration.
- UNMATCHED_CONDITIONS — Baseline formatting checks failed for two tasks, so the comparison conditions did not match.
- COMPARATOR — The direct controls come from earlier rounds rather than a contemporaneous matched cohort.
Required reproduction metadata
- Kogen commit: `945548ba49b95079e3d17bba29d5d391dfd3d3af`, the round source's pinned build.
- Harness commit: not recorded in the cited round source as a Git commit; any adapter or runtime fingerprints are not Git commits.
- Model and effort: a complete per-cell mapping is not recorded in the cited round source; values present in the source summary remain as stated above.
- Historical execution command: not recorded in the cited round source.
- Raw records: the cited sources do not provide a complete public round-specific request and grade bundle; public delivery and outcome exports are linked above where applicable.
Terminology: public round glossary.
STATUS: INVALID
Data completeness: 55.7% (72 deliveries; PASS WITH DECLARED GAPS). Measurement contract; declared gaps.
Pre-registered: no
Lifecycle: Historical capture. Snapshot 2026-10-05; earlier reports may use earlier cutoffs.
Question: Can Kogen 945548ba match direct agents on held-out Elixir/Phoenix?
Venue: Fixed HELDOUT workers: US elx-02/syn-01/syn-20/syn-31; EU elx-04/syn-06/syn-15/syn-24.
Design: 8 tasks × 3 arms × 3 reps = 72; historical r53/r53b direct controls.
Decision rule: none predeclared beyond Round 58 ITT and fixed design.
Arms: kogen-direct-escalate, kogen-direct-shell, kogen-staged.
Planned tasks in the surviving round note:
- elx-02-ingest-supervision
- elx-04-queue-backpressure
- syn-01-live-ticket-filters
- syn-06-migration-ticket-numbers
- syn-15-validation-ticket-form
- syn-20-email-invite-flow
- syn-24-csv-import
- syn-31-inbound-email-webhook
Verdict: Recomputed 6 October 2026 from the public cell export: kogen-direct-escalate 19/24, kogen-direct-shell 16/24 and kogen-staged 17/24 passes; the measurement record reports 72/72 officially graded. The earlier 62/72 snapshot is stale. Base-wide formatting on elx-02/elx-04 confounds Kogen comparisons; no inferential result is reported.
Public delivery and outcome reconciliation
This captured-delivery summary uses the public run-record export, grouped by audit round, task ID, public host ID, and the exact exported arm label. Captured counts rows; graded counts rows marked graded=true; pass, fail, and other are the run-record outcome fields. Official outcome states are in cells.jsonl, whose classifications can differ; see the register source crosswalk. Planned n is shown only when source-reported; unknown allocations are not inferred.
| Task ID | Public host ID | Arm label in export | Planned n | Captured | Graded | Pass | Fail | Other graded outcome |
|---|---|---|---|---|---|---|---|---|
| elx-02-ingest-supervision | kogen-bench-us | kogen-direct-escalate | 3 | 3 | 3 | 2 | 1 | 0 |
| elx-02-ingest-supervision | kogen-bench-us | kogen-direct-shell | 3 | 3 | 3 | 3 | 0 | 0 |
| elx-02-ingest-supervision | kogen-bench-us | kogen-staged | 3 | 3 | 3 | 3 | 0 | 0 |
| elx-04-queue-backpressure | kogen-bench-eu | kogen-direct-escalate | 3 | 3 | 3 | 3 | 0 | 0 |
| elx-04-queue-backpressure | kogen-bench-eu | kogen-direct-shell | 3 | 3 | 3 | 3 | 0 | 0 |
| elx-04-queue-backpressure | kogen-bench-eu | kogen-staged | 3 | 3 | 3 | 3 | 0 | 0 |
| syn-01-live-ticket-filters | kogen-bench-us | kogen-direct-escalate | 3 | 3 | 3 | 3 | 0 | 0 |
| syn-01-live-ticket-filters | kogen-bench-us | kogen-direct-shell | 3 | 3 | 3 | 2 | 1 | 0 |
| syn-01-live-ticket-filters | kogen-bench-us | kogen-staged | 3 | 3 | 3 | 3 | 0 | 0 |
| syn-06-migration-ticket-numbers | kogen-bench-eu | kogen-direct-escalate | 3 | 3 | 3 | 2 | 1 | 0 |
| syn-06-migration-ticket-numbers | kogen-bench-eu | kogen-direct-shell | 3 | 3 | 3 | 1 | 2 | 0 |
| syn-06-migration-ticket-numbers | kogen-bench-eu | kogen-staged | 3 | 3 | 3 | 3 | 0 | 0 |
| syn-15-validation-ticket-form | kogen-bench-eu | kogen-direct-escalate | 3 | 3 | 3 | 2 | 1 | 0 |
| syn-15-validation-ticket-form | kogen-bench-eu | kogen-direct-shell | 3 | 3 | 3 | 3 | 0 | 0 |
| syn-15-validation-ticket-form | kogen-bench-eu | kogen-staged | 3 | 3 | 3 | 3 | 0 | 0 |
| syn-20-email-invite-flow | kogen-bench-us | kogen-direct-escalate | 3 | 3 | 3 | 3 | 0 | 0 |
| syn-20-email-invite-flow | kogen-bench-us | kogen-direct-shell | 3 | 3 | 3 | 1 | 2 | 0 |
| syn-20-email-invite-flow | kogen-bench-us | kogen-staged | 3 | 3 | 3 | 0 | 3 | 0 |
| syn-24-csv-import | kogen-bench-eu | kogen-direct-escalate | 3 | 3 | 3 | 3 | 0 | 0 |
| syn-24-csv-import | kogen-bench-eu | kogen-direct-shell | 3 | 3 | 3 | 2 | 1 | 0 |
| syn-24-csv-import | kogen-bench-eu | kogen-staged | 3 | 3 | 3 | 1 | 2 | 0 |
| syn-31-inbound-email-webhook | kogen-bench-us | kogen-direct-escalate | 3 | 3 | 3 | 1 | 2 | 0 |
| syn-31-inbound-email-webhook | kogen-bench-us | kogen-direct-shell | 3 | 3 | 3 | 1 | 2 | 0 |
| syn-31-inbound-email-webhook | kogen-bench-us | kogen-staged | 3 | 3 | 3 | 1 | 2 | 0 |
Recomputed from raw records
`recomputed.json`, mined cell records, and the token audit back observed official outcomes, per-arm pass counts/rates, manifest token counters, and wall medians. The source contains 72 rows (fail 20, pass 52); overall pass rate is 72.2% (52/72) among pass/fail grades. Per-arm detail and coverage counts are in the JSON.
Published-export outcome cross-check: 72/72 shared cell IDs match; 0/72 public-export rows have no mined source record. Exact IDs and results for unpaired rows are in recomputed.json.
No per-cell outcome mismatches were found in the shared IDs.
Token reconciliation under METHOD.md: 72/72 published metadata totals equal the manifest input + cached input + output sum; 0 differ (no mismatches). Reasoning remains a separate reported component and is not added to output.
The recompute keeps both numbers visible. 0 unexplained single-response numeric errors were identified; multi-response rows retain the published full-cell total and the manifest counter separately because per-response usage records are excluded.
Captured run records
Standard-schema records assigned to this round in the indexed snapshot.
Reproduction inventory
Published analysis replay and full experiment replay are reported separately.
- Replay type
unavailable- Script
- No per-round analysis script is published.
- Inputs
- No identified public input bundle.
- Full execution replay
- Unavailable
- Scope
- No public per-round analysis script with a complete, identified input bundle is published. See the round README, MEASURED.md, and MISSING.md for recorded scope and gaps; historical execution inputs/toolchains/grading are not bundled.
View the complete per-round reproduction inventory · Repository reproduction guide