# Neon Tracked Energy Precision Report

- date: 2026-05-25
- target: https://main.dlwavl00q582x.amplifyapp.com
- package: @riddledc/riddle-proof-packs@0.6.3
- lilarcade_deploy_commit: 8129746f89c6ef203247d7c3ebbe8710c9a42e08
- lilarcade_amplify_job: 699
- status: candidate_ready_for_listening_review

## Claim

The Neon review packet should preserve tiny nonzero tracked-instrument energy deltas so a human reviewer can see the measurable effect of a candidate instead of seeing a rounded `0`.

## Catch

The previous tracked-instrument packet exposed the right guitar-down candidate data, but the Markdown formatter rounded some very small energy deltas to `0`. That made the human packet less precise than the proof JSON and could make a real rendered change look like no tracked-instrument energy movement.

## Fix

`@riddledc/riddle-proof-packs@0.6.3` now formats small nonzero audio values with more precision. LilArcade then consumed the package and redeployed production.

## Live Verification

A fresh production preliminary Neon ratchet run used:

```text
turn the guitar part down a little
```

with explicit guitar/down constraints and three candidate iterations. The batch passed fast mix health, mobile layout, playback sync, claim-candidate review, and human-review packet extraction.

The packet recommended `guitar -0.05` for listening review. It supported three candidates, rejected zero, restored state after the loop, and kept the proof/taste boundary explicit.

## Precision Evidence

The final packet now shows nonzero tracked guitar energy deltas:

- Intro Bed, `guitar -0.05`: `0.000298 -> 0.000254 (-0.000044)`
- Vocal Entry, `guitar -0.05`: `0.000451 -> 0.000385 (-0.000066)`
- Target movement, `guitar -0.05`: `energy -0.000054`

Those values are review aids. They do not prove subjective mix quality. They prove the packet can preserve the measurable candidate movement and guardrails for review.

## Result

- recommendation: `guitar -0.05`
- supported_candidates: 3
- rejected_candidates: 0
- required_section_energy_floors_preserved: true
- guardrails_preserved: true
- no_clipping: true
- headroom_guardrail_preserved: true
- no_low_level_proof_window: true
- state_restored_after_loop: true
- evidence_role_pattern: interaction_snapshots

## Boundary

This run proves deterministic app/audio receipts and preserves readable tiny audio deltas in the review packet. It does not prove the candidate sounds better.