Jyotisha/docs/research/vedastro_live_replay_conflict_2026_07_17.md

VedAstro Live Replay Conflict 2026-07-17

Status

blocked: a fresh official VedAstro full snapshot for the public Steve Jobs same-chart case does not reproduce the previously recorded D1 longitude packet. This report records an external-oracle conflict. It does not authorize prediction claims, production tuning, or a three-engine closure claim.

Reproducible Public Request

• Case: steve_jobs_public_aa / steve_jobs_public_1955_lahiri
• Birth time: 19:15 24/02/1955 -08:00
• Location: 37.7749, -122.4194
• Ayanamsa / node mode: lahiri / mean
• Fresh local raw artifact: scratch/local/vedastro_adapter/official_full_snapshot-1ff276e8fdee-ab172556e932.json
• Fresh official raw response SHA-256: ab172556e9327f55ce5a14d2a2da91c7736690c5181ba59627d0acbf0d1b0ef8
• Runner output: scratch/local/three_engine_parity_vedastro_live/three_engine_parity_replay.json
• Independent repeat raw response SHA-256: 64c784ddfce22b83f2ba4a666983c1ca3a01f54c6f4d49dd1070167dbf850d54
• Repeat runner output: scratch/local/three_engine_parity_vedastro_retry/three_engine_parity_replay.json

Fresh Result

metric	value
normalized D1 rows	15
matches	9
mismatches	5
blocked rows	0

Mismatched fields: Sun.longitude, Moon.longitude, Mars.longitude, Mercury.longitude, Venus.longitude.

The independent repeat used the same visible request body and produced the same five normalized mismatch values. The raw response hash differs, but the normalized external discrepancy is stable for this endpoint replay.

Conflict Arbitration

The prior commercial manifest referenced a research-local scratch artifact with the same visible birth-time request but a different official response hash. Two fresh official responses converge on the same five normalized discrepancies. The mismatch must remain visible until the VedAstro endpoint/version/method contract explains the drift. Do not replace the old packet with a pass result or tune local calculations to match either response.

Required Next Evidence

1. Capture official API version and method identifiers alongside the raw packet.
2. Replay the same request after a bounded delay and compare response hashes.
3. Publish a redacted, versioned public packet only after the response is stable.
4. Require a fresh normalized 15-row comparison before changing three_engine_parity_replay_manifest.json from its current blocked boundary.

Contract Probe Resolution Attempt

Research commit c91eea2 executed three complete versioned contract probes. Each probe compared the negative-offset birth time with its UTC equivalent and a positive-offset control, then cross-checked AllPlanetLongitude, AllPlanetData.PlanetRasiD1Sign, and PlanetNirayanaLongitude.

• Negative offset and UTC-equivalent inputs agreed.
• Positive-offset behavior was not reliably distinct.
• Cross-method results changed between runs; one run temporarily agreed while the other two exposed different planet-level discrepancies.
• The third probe also showed instability within its three baseline repeats.
• The service returned Server: Kestrel but no explicit API version header.

Therefore the requested contract evidence was attempted and did not resolve the conflict. VedAstro.D1.longitude remains field-level blocked. Separately validated VedAstro BAV, SAV, and Shadbala raw evidence is not invalidated by this longitude-method boundary.