Round x-compaction
x compaction
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/x-compaction/README.md
Round date: HISTORICAL (before 2026-10-09) Publication badge: DESCRIPTIVE; KEPT FOR AUDIT Why not VALID: The historical inventory retains DESCRIPTIVE because of DENOMINATOR_UNRESOLVED, PRE_REGISTRATION_MISSING, RAW_EVIDENCE_MISSING. See the round evidence and limitations below. Recomputation status: OBSERVED SOURCE ONLY
Recomputation limit: Published summary figures: planned/observed counts, outcome totals/rates, and numeric estimates as present in README.md and RESULTS.md. These figures cannot be recomputed from repository inputs because 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.
Status
DESCRIPTIVE
Why not VALID
- PRE_REGISTRATION_MISSING: No pre-registered rule predating the first result is established.
- DENOMINATOR_UNRESOLVED: Planned-run totals are not recovered.
- RAW_EVIDENCE_MISSING: Exact-ID raw records are absent.
Required reproduction metadata
- Kogen commit: not recorded in the cited round source.
- Harness commit: not recorded in the cited round source as a Git commit; any adapter or runtime fingerprints are not Git commits.
Terminology: public round glossary.
STATUS: INTERIM
Data completeness: N/A (no exact-ID per-cell records in the public exports). Measurement record; declared gaps.
Pre-registered: not established in the public record; treat this as retrospective exploratory evidence. Hypothesis coverage: H54 (provisional mapping; see the hypothesis register). Question: Did context compaction occur often enough in the traced hard-task runs to explain observed harness differences? Population: The traced cohort is reported as 20 runs each for kh, Codex and Pi; planned-run totals are not recovered. Headline: No causal compaction claim.
Reported observations
- No compaction was recorded in 0/60 traced runs: 0/20 each for kh, Codex and Pi. (reported data; source-reported, not reproducible from public data.)
- Reported median peak input was 40.5k tokens for kh, 64.1k for Codex and 38.5k for Pi. (reported data; source-reported, not reproducible from public data.)
- A kh elision review reports retaining 14/18 facts, marked preliminary. (reported data; source-reported, not reproducible from public data.)
- A separate forced-compaction pilot in e06-context ran once and had no adjacent comparison. (reported data; source-reported, not reproducible from public data.)
Reproduction and records
| Required field | Publicly recoverable value |
|---|---|
| Harness/Kogen revision | Exact per-run revisions were not recovered. |
| Model and effort | Model names and efforts by harness are not fully recoverable from this public summary. |
| Task IDs | The summary covers 60 hard-task runs; task IDs are not recovered. |
| Command | Not recovered; this is a trace analysis, not a reproducible launch command. |
| Raw records | No sanitized raw records for this exact round ID are in this snapshot. The exact-ID query returned zero rows in both results/cells.jsonl and results/run-records/index.json; see reported-data.json. |
The numerical observations above are documentary summaries, not a public exact-cell analysis. reported-data.json records each reported quantity and its status. The current public result files were checked for this exact round identifier; they do not provide this round's raw records.
Interpretation
Limitation: These runs did not exercise compaction. They cannot estimate the effect of compaction or attribute a measured harness gap to compaction.
Smallest useful next test: Use a predeclared context-pressure cohort that triggers compaction in each matched arm and records exact before/after context and task outcomes.
No individual outcome should be promoted beyond the status above until the exact run records, grade identities, denominators and source revisions are reconciled.
Sources: reported data file; public cells export; public captured-delivery export.
Studio source recovery
Studio retains source patches only; the staged source contains no per-cell model manifests or outcome rows. This source remains in the source inventory; the round’s published status and numbers are unchanged.
Captured run records
Standard-schema records assigned to this round in the indexed snapshot.
No per-round file is listed. Check the run-record index for the complete partition list.
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