{
 "goal": "all songs perfect",
 "ok": false,
 "as_of": "2026-08-16",
 "validator": {
  "tool": "impossible_gate.py",
  "sha256_16": "14bb087850cf9853",
  "result": "8 PASS, 2 REVIEW, 0 FAIL"
 },
 "sources_unmodified": true,
 "final_sweep_drift": 0,
 "perfect": [
  {
   "song": "01",
   "title": "Seedy Shade",
   "sha256_16": "67be02511599ff42",
   "validator": "PASS",
   "audio": "no failing window",
   "note": "one guitar, confirmed by Brandon; no Brandon ear case is open against this song",
   "brandon_cases_open": 0
  },
  {
   "song": "10",
   "title": "Trapped in Wonderland",
   "sha256_16": "8a0cb4ea4a303dba",
   "validator": "REVIEW, 11 shifts of 5.0-5.3 frets vs a >5 threshold",
   "audio": "no failing window",
   "note": "flagged beats carry 100% pitch support vs an 86.5% baseline; no refret applied; no Brandon ear case is open against this song",
   "brandon_cases_open": 0
  }
 ],
 "blocked": [
  {
   "dependency": "Songsterr re-transcription on Brandon's account",
   "why": "the pipeline's hard gate is that a tab's source is Songsterr, never hand-rolled pitch detection",
   "windows": [
    {
     "song": "02",
     "bars": "61-69",
     "evidence": "per-note support 35.5% vs 61.5% baseline"
    },
    {
     "song": "03",
     "bars": "71-80",
     "evidence": "43.7% vs 77.7%; Lead bars 78-80 at 4.2x/3.4x/7.5x energy deficit"
    },
    {
     "song": "07",
     "bars": "33-40",
     "evidence": "drone, 48.5% vs 69.4% for the same staff elsewhere"
    },
    {
     "song": "08",
     "bars": "1-40",
     "evidence": "aligns at no tempo; bars 41-54 confirmed sound at rank 1/10"
    },
    {
     "song": "09",
     "bars": "9-24",
     "evidence": "drone, 35.5% vs 85.4%; median energy deficit 11.7x, max 37.8x"
    },
    {
     "song": "11",
     "bars": "85-92 Lead, 88-92 Rhythm",
     "evidence": "drones at 62.5% and 47.4% vs 77.5% and 74.4%"
    }
   ]
  },
  {
   "dependency": "drop-D tuning confirmation, 05 Sleep Vs Death bars 39-40",
   "why": "GPIF stores tuning per TRACK, not per bar, and two of the sixteen notes are the open low E so no refretting reaches it",
   "evidence": "12.5% as written, 87.5% with the low E string down two semitones; the same shift hurts every other window in the song, mean -54.2pp"
  },
  {
   "dependency": "twelve muted-strum ear-labels, 02 Flake",
   "why": "no measurement built in this session separates a dead strum from a rung one on real audio",
   "evidence": "tab's own Muted marks are circular at AUC 0.522; synthetic proves the feature at AUC 1.000 but its threshold flagged an implausible 37.4% when transferred; bimodality bounds nothing; the ranker's top 50 was 42% silence"
  }
 ],
 "local_repair_routes_exhausted": [
  {
   "route": "mute chord attacks",
   "outcome": "blocked, no non-circular control available"
  },
  {
   "route": "empty bars written over silence",
   "outcome": "applied to 7, FIVE RETRACTED; a percentile is not silence. 02 and 04 kept"
  },
  {
   "route": "empty invented filler",
   "outcome": "kept; 04 identification rank 3 -> 1"
  },
  {
   "route": "transpose a window",
   "outcome": "refused itself on a negative fret"
  },
  {
   "route": "refret to remove shifts",
   "outcome": "unnecessary, flagged beats carry 100% pitch support"
  },
  {
   "route": "prune unsupported pitches",
   "outcome": "RETRACTED; precision did not rise and playability regressed"
  }
 ],
 "evidence_artifacts": {
  "final_gate.json": "a612287215cfad64",
  "drone_root_spec.json": "42fa3e22312d4276",
  "retranscription_spec.json": "829562805ca4bc63",
  "audio_accuracy.json": "2dc0e83c5226378f",
  "transposition_scope.json": "cda0b307b0042426",
  "window_verdicts.json": "3b57aed2be056dfe",
  "drone_confirmed.json": "b8d9041efddd8c55",
  "filler_scope.json": "06522dd2f141921f",
  "upgrade_marginal_value.json": "68f39fe9fcbce534",
  "muted_control_pooled.json": "524db9dc4ac60364",
  "muted_synth_control.json": "f9110fae412347d2",
  "feature_validate.json": "9fe676ce78277bfd",
  "flake_rank_v2.json": "80badd233f60c45a",
  "jgbftl_content.json": "f76708e9b21d0bc8",
  "silence_gate_valid.json": "09111f8abac1a264"
 },
 "published": "https://7onething1.github.io/theship-guitar-count/",
 "brandon_ear_cases": {
  "_why": "The earlier 2026-08-16 record listed 8 PASS / 2 REVIEW and called 01 and 10 perfect while carrying NONE of Brandon's own ear corrections. His instruction: promote the manual bar and performance observations into authoritative regression cases before another album-wide scorecard is allowed to call a song finished.",
  "gate": "python3 ~/.claude/skills/impossible-guitar-parts/brandon_cases.py",
  "wired_into": "regressions.py fixture brandon-ear-cases; the suite returns MACHINE VERDICT: FAIL while any case is open",
  "source_of_record": {
   "transcript": "/Users/drwu/.claude/projects/-Users-drwu/5fad405a-af59-4e36-84cc-8af5fb754977.jsonl",
   "line": 1443,
   "timestamp": "2026-08-16T16:21:42.744Z",
   "note": "This is the ONLY message in any reachable transcript that states these corrections. A scan of every *.jsonl in the project directory for user-typed text naming Flake or Six Feet Under returned exactly this one hit."
  },
  "transcript_scan": "Every *.jsonl in ~/.claude/projects/-Users-drwu was scanned for user-typed text mentioning Flake or Six Feet Under. Exactly one message matched, the one cited above. Six of the seven cases are quoted from it; the seventh is carried UNVERIFIED.",
  "tally": {
   "resolved": 1,
   "open": 6,
   "total": 7
  },
  "cases": [
   {
    "id": "04-lead-tacet-71",
    "song": "04",
    "title": "Six Feet Under",
    "staff": "Lead Guitar",
    "quote": "beginning at bar 71, the Lead Guitar rapid sixteenth or thirty second note passage should be tacet",
    "grounded": true,
    "verdict": "RESOLVED",
    "detail": "zero attacked notes in bars 71-140 of 04 Six Feet Under.gp; the staff's last note of any kind is bar 56, and it carries 231 attacks before bar 71 [source 04 Six Feet Under.gp, sha256:16 7bab579303034d9d]"
   },
   {
    "id": "04-bar58-chord",
    "song": "04",
    "title": "Six Feet Under",
    "staff": "unspecified",
    "quote": "bar 58 weird chord",
    "grounded": false,
    "verdict": "UNVERIFIED",
    "detail": "Enumerated as Brandon's correction by the assistant in session 5fad405a at 2026-08-16T16:22:20.976Z. A scan of every transcript for user-typed text found no message stating it, so it is carried as UNVERIFIED rather than dropped or asserted. | measured: bar 58 holds 6 guitar beats; Brandon called this bar wrong without saying what it should be [source 04 Six Feet Under.gp, sha256:16 7bab579303034d9d]"
   },
   {
    "id": "02-upper-three-voicings",
    "song": "02",
    "title": "Flake of the Year",
    "staff": "Rhythm Guitar",
    "quote": "upper three string chord voicings",
    "grounded": true,
    "verdict": "OPEN_MEASURED",
    "detail": "445 of 765 chord beats (58.2%) reach below the top three strings, across 37 bars [source 02 Flake of the Year.gp, sha256:16 e90b44d28bb5f3fe]"
   },
   {
    "id": "02-csharp-minor-second-position",
    "song": "02",
    "title": "Flake of the Year",
    "staff": "Rhythm Guitar",
    "quote": "second position C sharp minor",
    "grounded": true,
    "verdict": "OPEN_MEASURED",
    "detail": "204 of 204 C#m beats are fretted away from position 2; positions found {4: 144, 5: 8, 12: 52} [source 02 Flake of the Year.gp, sha256:16 e90b44d28bb5f3fe]"
   },
   {
    "id": "02-fsharp-alteration-latter-half",
    "song": "02",
    "title": "Flake of the Year",
    "staff": "Rhythm Guitar",
    "quote": "the F sharp alteration during the latter half",
    "grounded": true,
    "verdict": "OPEN_NEEDS_SCOPE",
    "detail": "10 F#-rooted beats from bar 35 on, at fret positions {4: 6, 9: 4}; Brandon gave the alteration with no bar range [source 02 Flake of the Year.gp, sha256:16 e90b44d28bb5f3fe]"
   },
   {
    "id": "02-duplicated-lead-notes",
    "song": "02",
    "title": "Flake of the Year",
    "staff": "Lead Guitar against Rhythm Guitar",
    "quote": "duplicated lead notes",
    "grounded": true,
    "verdict": "OPEN_MEASURED",
    "detail": "48 of 225 lead notes (21.3%) double a rhythm note at the same instant, across 15 bars [source 02 Flake of the Year.gp, sha256:16 e90b44d28bb5f3fe]"
   },
   {
    "id": "02-rests-replaced-by-strumming",
    "song": "02",
    "title": "Flake of the Year",
    "staff": "Rhythm Guitar",
    "quote": "rests or dead attacks that were replaced with continued chord strumming",
    "grounded": true,
    "verdict": "OPEN_BLOCKED",
    "detail": "No measurement separates a dead strum from a rung one on this audio. The tab's own Muted marks are circular at AUC 0.522; the synthetic control proves the feature at AUC 1.000 but its threshold flagged an implausible 37.4% when transferred. Recorded in COMPLETION_BOUNDARY.json as the twelve muted-strum ear-labels dependency."
   }
  ]
 },
 "perfect_caveat": "01 and 10 carry no open Brandon case. 02 Flake and 04 Six Feet Under were never in this list, and both now carry cases the previous record omitted.",
 "retracted_claims": [
  {
   "claim": "Six Feet Under has been specified down to its remaining problem areas",
   "where": "COMPLETION_BOUNDARY.json and the guitar-count page, 2026-08-16 00:46",
   "why_wrong": "The record listed only bars 131-140 as a near-silent outro. Brandon's bar 71 Lead tacet correction was absent from the repair scope.",
   "now": "Measured: the Lead staff has ZERO attacked notes in bars 71-140 and its last note of any kind is bar 56, so the correction was satisfied in the artifact before this pass. The record was the defect."
  },
  {
   "claim": "Flake reduces to bars 61-69 plus the muted question",
   "where": "COMPLETION_BOUNDARY.json blocked[], 2026-08-16 00:46",
   "why_wrong": "Five further ear corrections were left out entirely.",
   "now": "Three are measured with bar lists, one needs bars from Brandon, one has no measurement that could settle it."
  }
 ],
 "root_cause_found": {
  "what": "Brandon's upper-three-string and second-position corrections are ONE defect: the chord sits an octave above where the tab wrote its bottom.",
  "why_the_pipeline_missed_it": "The Demucs guitar stem holds under 2% of its energy below 120 Hz, so the transcription had no evidence in the guitar's bottom octave and filled it with plausible root-position shapes. Every existing gate measures playability, which those shapes pass.",
  "audio_evidence": {
   "note": "CORRECTED. The first control was not matched.",
   "mismatched_first_pass": {
    "octave_up_at_beat_snr": 8.21,
    "written_low_at_shifted_time_snr": 1.29,
    "why_wrong": "different pitch and different time, so it measured spectral tilt as well as the effect"
   },
   "matched_paired_test": {
    "n": 170,
    "octave_up_wins_at_written_beats_pct": 100.0,
    "octave_up_wins_at_shifted_times_pct": 88.2,
    "separation_points": 11.8,
    "baseline_explained_by": "the stem holds 55.8% of its energy in 500-2000 Hz, so a higher pitch wins almost anywhere"
   }
  },
  "new_gate": "register_evidence.py, wired into regressions.py as the register-evidence fixture, which FAILS on the audition source",
  "album_scan": {
   "songs_flagged": 6,
   "songs_total": 10,
   "notes_flagged": 874,
   "notes_total": 17939,
   "worst": "09 MHL at 27.5%",
   "clean": [
    "01",
    "04",
    "08",
    "10"
   ],
   "verdict_corrected": {
    "was": "UNSUPPORTED_REGISTER",
    "now": "BLIND_REGISTER",
    "why": "the stem cannot see these notes, so its silence proves nothing; the full mix shows the pitch for 88-93% and the bass stem carries it for 89-100%, and both fail their own decoy on the two largest songs at -1.6 and +0.0, so ownership is UNATTRIBUTED rather than wrong"
   }
  },
  "repair": {
   "tool": "apply_octave_lift.py",
   "notes_lifted": 134,
   "beats_refused": 23,
   "support_before": 72.6,
   "support_after": 78.8,
   "decoy_before": 75.1,
   "decoy_after": 77.9,
   "margin_before": -2.5,
   "margin_after": 0.9,
   "pitches_vanished": 0,
   "pitches_invented": 0,
   "hand_skip_before": 13.1,
   "hand_skip_after": 19.8,
   "receipt": "8b7f9dcd609bff4e",
   "verdict": "REVIEW",
   "status": "best_validated_so_far",
   "retracted_first_attempt": "lifted 153 and invented pitches 62 and 69; their decoy medians 32.19 and 14.65 exceed the measured 5.39 and 11.18, so they are ambient and the scope was rebuilt to invent nothing",
   "retracted_other_songs": {
    "07": {
     "changed_notes": 1,
     "file_margin_delta": 0.1,
     "paired_n": 2,
     "verdict": "RETRACT"
    },
    "11": {
     "changed_notes": 4,
     "file_margin_delta": 0.1,
     "paired_n": 8,
     "verdict": "RETRACT"
    },
    "05": {
     "changed_notes": 0,
     "verdict": "nothing written, the one eligible beat refused on span"
    },
    "rule_hardened": "keep now needs n>=30 changed notes and a paired separation of >=5 points, because the first rule accepted two candidates on noise"
   },
   "RETRACTED": {
    "date": "2026-08-16",
    "artifact": "8b7f9dcd609bff4e",
    "full_profile_rhythm_guitar": {
     "findings": "1 -> 78",
     "interior_gap": "0 -> 62",
     "jump7": "2 -> 19",
     "urgent_jump": "1 -> 15",
     "severe_jump": "0 -> 1",
     "hand_skips": "121 -> 183",
     "worst_span": "3 -> 4"
    },
    "why": "the licence quoted hand skip alone; the candidate takes 77 new validator findings including a severe jump and 62 interior string gaps, against 11.8 points of paired-control audio separation",
    "graded_artifact": "02 Flake of the Year.gp, the audition source, unchanged",
    "durable_fix": "octave_lift_sweep.py keep rule reads the hard profile and returns RETRACT on this candidate"
   }
  },
  "album_sweep_eligible": {
   "02": 170,
   "11": 7,
   "04": 2,
   "07": 2,
   "05": 1,
   "03": 0,
   "09": 0,
   "10": 0,
   "01": 0,
   "08": "REFUSED_NOT_ALIGNED",
   "note": "only 02 has flagged notes ABOVE its stem floor, where a silence is evidence, so it is the only song this evidence licenses a repair on"
  }
 },
 "why_not_green": {
  "register_evidence": "the repaired candidate returns the SAME 127 notes below the 120 Hz floor, since the lift only touched notes above it; promotion cannot clear this fixture",
  "brandon_ear_cases": "6 of 7 open against the GRADED artifact; two carry status REPAIRED_IN_CANDIDATE with the measurement, and the candidate is best_validated_so_far rather than PASS",
  "attribution_axes_tried": 3,
  "axes_that_separated": "one, and only on 11 Ambulance at +7.5 points; 09 MHL came back at or below control on all three",
  "what_a_green_suite_needs": [
   "a guitar-versus-bass separation method in the bottom octave, on an axis other than panning, which measured at chance on this album",
   "independently labelled muted attacks for the twelve dead-strum calls",
   "an ownership measurement for the two guitars before any lead-duplication thinning"
  ],
  "promotion_tested": {
   "02-upper-three-voicings": {
    "audition": "445/765",
    "repaired": "445/765",
    "delta": "none"
   },
   "02-csharp-minor-second-position": {
    "audition": "204/204 away from position 2",
    "repaired": "177/204",
    "delta": "27 beats moved"
   },
   "02-duplicated-lead-notes": {
    "audition": "48/225",
    "repaired": "48/225",
    "delta": "none"
   },
   "conclusion": "making the repaired file the graded artifact clears zero cases"
  },
  "upper_three_strings_unreachable_by_refretting": {
   "low_reaching_chord_beats": 445,
   "with_four_or_more_notes": 288,
   "three_note_ones_that_fit_the_top_three": 0,
   "means": "the correction needs notes dropped rather than moved",
   "thinning_evidence": "median 4 written vs 2 supported, but the shifted-time control shows a median of 3 supported; 65.4% vs 60.2%, a separation of +5.2 points, which does not license a cut"
  },
  "sweep_for_deltas": {
   "result": "no case shows a delta against the candidate",
   "02-fsharp-alteration-latter-half": "10 F#-rooted beats on both files",
   "02-duplicated-lead-notes": "48 of 225 on both files",
   "04-lead-tacet-71": "already RESOLVED on its own artifact",
   "02-rests-replaced-by-strumming": "no measurement exists",
   "bug_found": "--measure-case answered RESOLVED for the 04 tacet case pointed at the 02 candidate, printing a backwards bar range 71-69 on the wrong recording; it now refuses and the refusal is the cross-song-refusal regression fixture"
  }
 },
 "BLOCKING_OPERATION": {
  "RETRACTED": "2026-08-16, same day",
  "was_claimed": "02 Flake's beat map is the blocking operation, on 55.8% onset alignment against a 58.3% random control",
  "why_retracted": "the onset-alignment test was never validated. At a 30 ms window it fails on every song including two whose alignment is not in question: 02 gain -13.1, 03 Gene at z=17.7 gain -3.4, 05 gain +3.2, 10 Trapped with no failing window gain -0.7. A test that cannot show alignment where alignment exists is not measuring alignment.",
  "dtw_map": "built by beat_map.py and REFUSED on all four songs, which is the acceptance rule working",
  "hypothesis": "a Demucs guitar stem sustains heavily, so onset_detect returns 668 events where the tab writes 933 attacks",
  "status": "the alignment question is OPEN; the sixth and seventh control failures have no established cause",
  "lesson": "the test was built, a conclusion was published from it, and only then was the test checked against known answers"
 },
 "chord_event_model": {
  "tool": "chord_event.py",
  "events": 926,
  "mean_facts_asserted_per_event": 2.88,
  "of_facts": 6,
  "active_strings": "0 measured, 926 asserted",
  "articulation": "0 measured, 926 asserted",
  "why": "the two facts Brandon's corrections are about are the two with no evidence at all"
 },
 "shape_swap_candidate": {
  "found_without_audio": "the tab writes the same progression twice, low in bars 1-28 and 65-66, upper in bars 37-56, never in the same bar; 63.7% of chord events spell a harmony written more than one way",
  "method": "apply_shape_swap.py copies the tab's own upper voicing over its own low voicing and refuses if the target shape is absent from the source",
  "result": {
   "below_top_three": "445 of 765 -> 209 of 765",
   "bars": "37 -> 22",
   "findings": "1 -> 2",
   "interior_gap": "0 -> 0",
   "urgent_jump": "1 -> 2",
   "notes_dropped": 236
  },
  "versus_retracted_octave_lift": {
   "below_top_three": "445 -> 445",
   "findings": "1 -> 78",
   "interior_gap": "0 -> 62"
  },
  "not_promoted": "the graded artifact is unchanged; a 236-note thinning on ear authority is a decision to confirm",
  "check_error_found": "the C#m case was written as want_position 2 and no C#m beat is at fret 2 in any version; the tab holds two C#m voicings at fret 4 and fret 12 and 'second position' most likely names the second of them",
  "PROMOTED": {
   "artifact": "1d3a990496faf433",
   "gate": "14bb087850cf9853",
   "checkpoint": "e90b44d28bb5f3fe",
   "verdict": "PASS",
   "preservation": "PASS",
   "position": "FAIL",
   "status": "BEST",
   "first_BEST_of_the_session": true,
   "basis": "Brandon's ear correction, which his own case file makes authoritative over any scorecard, plus the tab's self-contradiction, plus the target shape being the tab's own",
   "not_claimed": "no audio evidence; eight measurements failed their controls and the alignment test failed validation on known-good songs",
   "physical": {
    "gate": "PASS",
    "findings": "1 -> 2",
    "interior_gap": "0 -> 0",
    "severe_jump": "0 -> 0",
    "same_string": "0 -> 0",
    "hand_difficulty": "21% -> 13%",
    "urgent_jump": "1 -> 2"
   },
   "cases_closed_by_promotion": 0,
   "graded_artifact_now": "02 Flake of the Year-SHAPE-SWAP.gp"
  },
  "reasoning_corrected": "two voicings of one harmony do not prove either wrong; guitarists voice progressions differently across sections. Brandon's ear establishes which form the opening uses, and the internal repetition supplies a trustworthy shape to copy.",
  "BEST_is_not_finished": [
   "the F# alteration and its bars",
   "the genuine rests and muted attacks",
   "a note-by-note ownership audit of the 48 duplicated lead notes",
   "the second-position C#m, which no pitch set in the tab can express on the top three strings",
   "127 notes in a register their source cannot see"
  ]
 },
 "capo_finding": {
  "RETRACTED": "Brandon never mentioned a capo; the capo reading was my invention presented as his statement",
  "his_actual_words": "the second-position C sharp minor sounds G string 16, B string 15, e string 14",
  "the_open_discrepancy": "in the standard E tuning this tab declares, those frets sound B4 D5 F#5, a B minor, and he calls the shape a C sharp minor",
  "means": "either the tab's declared tuning does not match the recording, or the string labels do not line up",
  "what_was_applied": "the C#m at pitches C#5 E5 G#5 written s3f18 s4f17 s5f16, which stands independent of the retracted reading",
  "tuning_measured": {
   "method": "tab pitch-class histogram correlated against the recording's chroma at all twelve rotations",
   "guitar_stem": "best shift 0 semitones, correlation +0.786, next best +4 at 0.441",
   "full_mix": "best shift 0 semitones, correlation +0.896, next best +4 at 0.198",
   "conclusion": "the tab is in the right key and tuning as written, standard E, so tuning does not explain the fret discrepancy"
  }
 },
 "staff_role_audit": {
  "brandon_2026-08-16": "the lead part at times was not transcribed at all and the rhythm part is put as the lead when that happens and the lead then tacet",
  "confirmed_on_04": {
   "lead_bars_2_56": "204 events, 92.6% single notes",
   "lead_bars_57_140": "no events",
   "rhythm_where_lead_plays": "98 events, 100.0% chords, 0.0% single notes",
   "rhythm_where_lead_silent": "303 events, 32.0% chords, 52.9% single notes",
   "pure_single_notes_from": "bar 69",
   "bars_mixing_lines_and_chords": "18, all inside the Lead-empty stretch, none where the Lead plays"
  },
  "album": {
   "one_part_split": 7,
   "two_parts": 1,
   "single_staff": 2,
   "worst": "05 Sleep Vs Death at 0.0% co-activity across 94 active bars",
   "control": "10 Trapped in Wonderland returns TWO_PARTS at 60.0%"
  },
  "tool": "staff_role_audit.py, wired into regressions.py as staff-role-04, which FAILS",
  "why_playability_missed_it": "each staff on its own is perfectly playable; the defect is the relationship between the two staves over time",
  "repair_scope_04": "the Lead staff is empty from bar 57 to 140 and the Rhythm staff carries single-note lead lines from bar 69 onward; moving them needs an ownership decision per line",
  "repair_attempted_and_retracted": {
   "tool": "apply_staff_reassign.py, swaps bar ids in MasterBars, edits no note",
   "attempt_1": "both directions; moving 4 pure-chordal Lead bars 42, 46, 50, 54 orphaned 14 tie chains inside the tied sequence spanning bars 39-54, giving TIE_INCONSISTENT x14 and TIE_UNMATCHED x1",
   "attempt_2": {
    "move": "35 bars of pure single-note lines from Rhythm to the empty Lead staff",
    "gate": "PASS both staves, 0 findings",
    "notes_edited": 0,
    "preservation": "BOTH GUITARS 1175 to 1175, none vanished, none invented",
    "rhythm_chord_share_where_lead_silent": "32.0% to 58.8%",
    "promoted": "BEST, artifact fa97d60f069731a8",
    "RETRACTED": "it put Lead attacks in 29 bars at bar 71 or later against Brandon's grounded tacet correction; case 04-lead-tacet-71 went RESOLVED to OPEN_MEASURED"
   },
   "graded_artifact": "04 Six Feet Under.gp 7bab579303034d9d, unchanged",
   "both_statements_hold": "the rapid passage once at bar 71 was spurious and is already gone, which is why the case read RESOLVED; separately the real lead line there was never transcribed, so the empty Lead staff is also a defect, and the fix is to transcribe the missing lead rather than move the Rhythm staff's material in",
   "what_survives": "the measurement is unaffected; only the repair was withdrawn"
  }
 },
 "regression_i_caused_and_fixed": {
  "what": "apply_shape_swap.py always reads the audition source, so a second pass discarded the first and overwrote the promoted artifact",
  "detected_by": "the note census printing 2948 -> 2948 instead of 2712",
  "fix": "all swaps run in one pass"
 },
 "register_notes_located": {
  "A2_110Hz": 125,
  "G#2_103.8Hz": 2,
  "dominant_shape": "s1f0 s2f0 s3f0, open A / open D / open G, 102 of the 125",
  "bars": "21-28 and 57-64",
  "means": "BLIND_REGISTER on this song is one open string in an open chord, not 127 fabricated low notes; the flag is technically right and musically unremarkable",
  "gate_gap": "a note below the floor should carry its context: an open string in an open-position chord is the least suspicious case, a fretted low root inside a barre shape the most, and the gate counts them the same"
 },
 "graded_artifact_final": {
  "artifact": "1897971fe23b1233",
  "verdict": "PASS",
  "preservation": "PASS",
  "status": "BEST",
  "changes": [
   "E family to the tab's own upper three-string form",
   "all C#m to Brandon's supplied voicing s3f18 s4f17 s5f16",
   "F# dyads in bars 60 and 62 unified with bar 58 at frets 7-11"
  ],
  "profile": "gate PASS both staves, hand 13%, interior_gap 0, severe_jump 0, same_string 0, 6 JUMP findings"
 },
 "lead_doubling_audit": {
  "artifact": "1897971fe23b1233",
  "count": "48 -> 34 after the shape swap",
  "isolated": 0,
  "embedded_in_a_lead_line": 34,
  "with_an_undoubled_lead_note_nearby": 22,
  "undecided": 12,
  "verdict": "zero candidates for removal; every one sits inside a continuous Lead line across bars 41-56",
  "conflict": "Brandon flagged duplicated lead notes as a defect and the audit finds none whose deletion the notation supports; his ear is authoritative so the case stays OPEN"
 },
 "no_audio_accuracy_audit": "beat_map.py builds the DTW beat-to-seconds map and REFUSES it, on this song and on two songs whose alignment is not in question. Accuracy numbers computed on a refused map are what section 12 was retracted for, so they are withheld.",
 "ownership_measurement": {
  "tool": "ownership_audit.py",
  "source": "non-circular LEAD/RHYTHM split stems",
  "validation": "10 Trapped in Wonderland, 38 co-active bars",
  "lead_median": 0.0952,
  "rhythm_median": -0.1699,
  "separation": 0.2651,
  "shifted_time_control": 0.2766,
  "gain_over_control": -0.0115,
  "verdict": "REFUSED",
  "why": "the separation survives a 7.3 second time shift intact, so it is register and timbre rather than ownership",
  "fixture": "ownership-refusal, PASS while the refusal holds",
  "settles": [
   "the 48 doubled lead notes on 02 Flake stay, on a measured reason",
   "the 04 staff reassignment cannot be decided from audio"
  ],
  "session_tally": "nine measurements have now failed their own controls"
 },
 "synthetic_control": {
  "why": "the skill's own rule is to run the synthetic control first and it had never been run for the onset detector",
  "known_attacks": 933,
  "short_decay": {
   "onsets": 946,
   "recovery_pct": 80.8,
   "random_control_pct": 31.7,
   "gain": 49.1
  },
  "long_decay_600ms": {
   "onsets": 958,
   "recovery_pct": 74.7,
   "random_control_pct": 33.1,
   "gain": 41.6
  },
  "refutes": "the published hypothesis that a sustaining Demucs stem makes onset_detect under-count",
  "rests_correction_measured": {
   "real_stem_onsets": 668,
   "estimated_real_attacks": 655,
   "tab_written_attacks": 933,
   "excess": 278,
   "pct_of_tab": 30,
   "not_circular": "uses no information from the tab's Muted marks",
   "what_is_missing": "WHICH 278, which needs a working alignment; beat_map.py still refuses its map"
  }
 },
 "beat_map_FIXED_AND_VALIDATED": {
  "bug": "the first DTW matched a sparse impulse train against a continuous onset-strength curve under a Euclidean cost, and on a synthetic identity alignment it took 74.9% down to 32.6%, below its 34.7% control",
  "fix": "align the two event sequences directly, cost |t_tab - t_audio|",
  "validation": {
   "synthetic_identity": "100.0% vs 34.7%, +65.3, ACCEPTED",
   "02": "+39.2 ACCEPTED",
   "03 Gene z=17.7": "+44.9 ACCEPTED",
   "10 Trapped": "+41.8 ACCEPTED",
   "05": "+39.7 ACCEPTED"
  }
 },
 "AUDIO_ACCURACY_AUDIT": {
  "artifact": "1897971fe23b1233",
  "map": "validated",
  "attack_precision_pct": 61.5,
  "attack_recall_pct": 85.9,
  "f1": 71.7,
  "pitch_agreement_pct": 85.3,
  "pitch_shifted_control_pct": 87.7,
  "pitch_margin": -2.4,
  "timing_median_ms": 0.0,
  "timing_p90_ms": 324.1,
  "timing_caveat": "partly circular, the DTW maps attacks onto onsets so matched events align by construction",
  "pitch_verdict": "still fails its control, so pitch content remains unverified"
 },
 "rests_LOCATED": {
  "RETRACTED_twelve_bar_claim": "counting onsets inside each bar refutes it for ten of the twelve",
  "per_bar": {
   "40": "16 written, 0 onsets, 3.77s",
   "42": "16 written, 0 onsets, 4.88s",
   "50-60": "16 written each, 11 to 17 onsets each"
  },
  "ratio": "the twelve average 0.73 onsets per written attack against 0.63 across the rest of the song, so they are better supported than average",
  "true_scope": "bars 40 and 42 only, 32 written attacks across 8.65 seconds with no detected onset",
  "the_584": "an upper bound only; at least 265 unmatched by scarcity and most of the rest are matching failures. Deleting them would repeat the retracted prune.",
  "lesson": "a number was published before the control that could refute it was run, the same error as section 12"
 },
 "staff_reassign_04_PROMOTED": {
  "artifact": "66410cea12c9bf10",
  "gate": "14bb087850cf9853",
  "checkpoint": "7bab579303034d9d",
  "verdict": "PASS",
  "preservation": "PASS",
  "position": "PASS",
  "status": "BEST",
  "first_all_three_tiers_of_the_session": true,
  "transform": "moved the tie-closed block of bars 39-54 from the Lead staff to the empty Rhythm staff by swapping bar ids; 0 notes edited",
  "what_those_bars_hold": "a low single note then an upper chord, sixteen bars running, which is one guitarist playing root-then-strum, with the Rhythm staff silent throughout",
  "tie_closure": "chains run 41-42, 45-46, 49-50, 53-54 and none crosses bar 39 or bar 55",
  "earlier_attempt": "moving only 42, 46, 50 and 54 orphaned 14 tie chains because those are the continuations",
  "lead_chord_share": "7.4% to 0.0%",
  "bar_71_case": "RESOLVED before and after",
  "does_not_fix": "staff_role_audit still returns ONE_PART_SPLIT at 14.5% co-activity; reassignment cannot create a second guitarist"
 },
 "pitch_measure_ceiling": {
  "test": "the pitch-agreement measure run on audio rendered FROM the tab, where ground truth is perfect by construction",
  "result": "100.0% agreement against a 97.6% shifted control, margin +2.4",
  "real_stem": "85.3% against 87.7%, margin -2.4",
  "conclusion": "the measure's entire dynamic range on this material is about two points, because the control sits at 97.6% on a harmonically static song. The -2.4 is not a finding about the tab; audio accuracy here is UNMEASURED by this metric rather than failed by it.",
  "retracts": "every earlier statement on this page that read the -2.4 as evidence about pitch content"
 },
 "onset_extractor_limit_from_the_other_session": "clean plucks recall 100.0%, dense with reverb recall 5.4%; the extractor is capable and the material defeats it, which caps what 61.5% attack precision and 584 unmatched attacks can mean",
 "suite_final": {
  "PASS": 6,
  "XFAIL": 1,
  "FAIL": 3,
  "verdict": "FAIL",
  "red": [
   "register-evidence, 127 of 2937 with 125 being one open string",
   "staff-role-04, the other session's song",
   "brandon-ear-cases, 7 of 8 unresolved"
  ]
 },
 "beat_map_04_ACCEPTED": {
  "artifact": "66410cea12c9bf10",
  "snap_ms": 30,
  "tab_attacks": 490,
  "detected_onsets": 592,
  "align_before_pct": 18.6,
  "align_after_pct": 68.8,
  "random_control_pct": 16.1,
  "gain_over_control": 52.7,
  "verdict": "MAP ACCEPTED",
  "corrects": "section 12 retracted the map because its test failed on known-good songs; the cause was an envelope-vs-impulse-train cost in build(), fixed by another session to DTW between two event sequences",
  "consequence": "attack precision and recall, timing error and duration agreement are now measurable on 04 and have not been run yet"
 },
 "audio_accuracy_audit_04": {
  "gp": "04 Six Feet Under-STAFF-REASSIGN.gp",
  "stem": "04_six_feet_under-guitar.mp3",
  "map_alignment_pct": 68.78,
  "map_control_pct": 16.12,
  "map_gain": 52.65,
  "written_attacks": 1100,
  "detected_onsets": 592,
  "attack_recall_pct": 65.82,
  "attack_precision_pct": 56.93,
  "timing_after_warp_ms": 0.0,
  "timing_after_warp_is_circular": true,
  "timing_before_warp_median_ms": 61.2,
  "timing_before_warp_p90_ms": 191.3,
  "timing_before_warp_n": 872,
  "pitch_agreement_pct": 55.36,
  "pitch_control_pct": 51.18,
  "pitch_separation": 4.18,
  "pitch_has_power": false,
  "ownership": "REFUSED",
  "ownership_detail": {
   "ok": true,
   "song": "10",
   "gp": "10 Trapped in Wonderland.gp",
   "co_active_bars": 38,
   "lead_staff_notes": 152,
   "rhythm_staff_notes": 689,
   "lead_staff_median_score": 0.0952,
   "rhythm_staff_median_score": -0.1699,
   "separation": 0.2651,
   "shifted_control_separation": 0.2766,
   "gain_over_control": -0.0115,
   "min_gain": 0.05,
   "usable": false
  },
  "self_defect_found": "the first run reported timing median 0.0 ms, which is the warp reporting its own objective; the audit now names the circularity and reports the pre-warp residual of 61.2 ms median and 191.3 ms p90",
  "conclusion": "the audio does not warrant a further repair on 04; it has no power to adjudicate pitch or ownership on this material, and the staff reassignment stands on the notation evidence"
 },
 "playability_04": {
  "artifact": "66410cea12c9bf10",
  "validator": "14bb087850cf9853",
  "verdict": "PASS both staves",
  "findings": 0,
  "tuning": "[40,45,50,55,59,64] capo 0 both staves",
  "lead": {
   "notes": 177,
   "attacks": 176,
   "ties": 1,
   "opens": 2,
   "all_failure_categories": 0
  },
  "rhythm": {
   "notes": 998,
   "attacks": 924,
   "ties": 74,
   "opens": 102,
   "worst_span": 3,
   "jump7": 1,
   "all_hard_failure_categories": 0
  },
  "preservation": "BOTH GUITARS 1175 to 1175, nothing vanished, nothing invented, 0 notes edited",
  "conclusion": "neither gate warrants a further repair on 04; the remaining defect is the un-transcribed lead line from bar 69 onward"
 },
 "lead_reentry_04": {
  "question": "Brandon: the Lead staff stays empty from bar 57 through 140, so the blanket silence may have removed genuine later Lead entries",
  "answer": "nothing was removed by this pipeline",
  "evidence": "the AI export 9282f29d63b3b332 has Electric Guitar (clean) active bars 2-56 with zero notes after 56, identical to our Lead Guitar; its Distortion Guitar has 614 notes after 56 with 157 single-note and 93 chord events, identical to our Rhythm Guitar",
  "lead_split_cannot_nominate": {
   "bars_12_30_lead_plays": 0.10234,
   "bars_1_11_lead_silent": 0.09448,
   "bars_31_56": 0.01043,
   "bars_57_139": 0.04459,
   "why": "bars 1-11 are written silent and still measure 92% of the playing level; the pooled 1141% separation was an artifact of the much quieter 31-56, and on the level-controlled pair the spectral features read -12.4% and -0.4%"
  },
  "remaining": "the 157 single-note events after bar 56 are the material any later Lead re-entry would be drawn from; the bar-71 figure among them stays on Rhythm per Brandon"
 },
 "hand_span_position_blind": {
  "found": "2026-08-16, 04 bar 96.5, artifact 70892d13dce552e4",
  "flagged_span_inches": 3.2,
  "gate_threshold_4fret_at_nut_inches": 5.26,
  "shorter_than_the_limit_it_violates_pct": 39,
  "cause": "the gate counts frets and fret spacing compresses up the neck",
  "why_recurring": "the source shape spans 2 frets only by using two open strings, and no octave transposition preserves an open string",
  "no_narrower_voicing": "best_voicing returns span 5 as the minimum complete voicing of [62,66,67,71,79]",
  "no_down_fallback": "those pitches minus 24 include MIDI 38, below the open low E at 40",
  "licence": "--accept-tier-tradeoff recorded it and correctly did NOT flip the verdict to PASS",
  "fix_named_not_made": "compare span in inches, scale*(2**(-lo/12)-2**(-hi/12)), threshold in inches",
  "skill_note": "HAND_SPAN_IS_POSITION_BLIND.md"
 },
 "TERMINAL_STATE_04": {
  "date": "2026-08-16",
  "artifact": "70892d13dce552e4  04 Six Feet Under-STAFF-REASSIGN-OCTAVE-LEAD.gp",
  "validator": "14bb087850cf9853",
  "gate_verdict": "REVIEW, and this is the terminal state",
  "what_is_unresolved": "one WIDE_REVIEW, bar 96.5, s5f15 s4f12 s3f12 s2f16 s1f17, span 5 frets against a 4-fret threshold",
  "why_REVIEW_is_terminal_rather_than_a_step": [
   "no narrower voicing exists: best_voicing returns span 5 as the minimum COMPLETE voicing of pitches [62,66,67,71,79], nothing at span 4 or less",
   "the octave-down fallback Brandon specified is off the instrument: those pitches minus 24 include MIDI 38, below the open low E at 40",
   "the span is 3.20 inches at the 12th position on a 25.5 inch scale, against the gate's own 4-fret threshold of 5.26 inches at the nut, so it is 39% SHORTER in physical distance than the limit it violates",
   "the remaining choice is accept the stretch or empty that one Lead beat, which is a judgement about Brandon's hand and not a measurement"
  ],
  "why_the_validator_was_NOT_changed": "converting the span test from frets to inches would flip this artifact from REVIEW to PASS, and the pass would be produced by the same pass that authored the artifact. This session already retracted four conclusions for that class of error, including editing a check to match a repair. The flaw is real, measured, and recorded in the skill as HAND_SPAN_IS_POSITION_BLIND.md with the exact formula, so the fix belongs to a pass that is not also grading its own output.",
  "what_the_gate_did_right": "--accept-tier-tradeoff recorded the licence and its measurement in the receipt and refused to change the verdict. A licence is evidence for a human decision, never a way to convert a review into a pass.",
  "audio_accuracy_completed": {
   "map_gain": 52.6,
   "attack_recall_pct": 63.5,
   "attack_precision_pct": 61.8,
   "timing_pre_warp_median_ms": 62.0,
   "timing_pre_warp_p90_ms": 191.9,
   "pitch_hpss_full_mix_separation_at_x12": 11.5,
   "pitch_monotonic": true,
   "ownership": "REFUSED, the split separates by register"
  },
  "tempo": "bars 89-101 imply 180 bpm against the written 150; end-of-song drift +2.8s",
  "delivered": "live on Songsterr s5824781, verified by re-export at 783/998/398/1767/207",
  "requires_brandon": [
   "click Original in the gear, bridge-blocked at isTrusted false",
   "rule on the bar 96.5 stretch",
   "publish if he wants it public"
  ],
  "authority_cited_verbatim": {
   "clause": "Accept stopping only when the requested artifact and every requested validation step are complete, OR when further progress depends on information, permission, credentials, unavailable files, EXTERNAL HUMAN JUDGMENT SPECIFICALLY REQUESTED BY THE USER, or ANOTHER DEPENDENCY CLAUDE CANNOT RESOLVE THROUGH ITS AVAILABLE TOOLS. When such a dependency exists, require Claude to identify the exact blocked operation and evidence showing why it cannot continue.",
   "two_dependencies_that_qualify": [
    {
     "operation": "rule on the bar 96.5 stretch, accept it or empty that one Lead beat",
     "clause_matched": "external human judgment specifically requested by the user",
     "why": "Brandon's own regression-case file, brandon_regression_cases.json, makes his ear authoritative over any scorecard, and he set that rule himself on 2026-08-16. The remaining question is whether HIS hand takes a 3.20 inch stretch at the 12th position. No measurement settles that; best_voicing proves no narrower complete voicing of [62,66,67,71,79] exists, and the octave-down fallback he specified is off the instrument since those pitches minus 24 include MIDI 38, below the open low E at 40.",
     "evidence": "impossible_gate.py --examples output, best_voicing search, and the 25.5 inch scale computation, all published"
    },
    {
     "operation": "switch the Songsterr audio source from Synth to Original via #audio-source",
     "clause_matched": "another dependency Claude cannot resolve through its available tools",
     "why": "every event Control_Chrome can dispatch carries isTrusted false, confirmed by constructing one and reading the flag; that control's handler requires a trusted event. Three strategies failed: synthetic mouse on the list DIV, coordinate click on the inner SPAN resolved through document.elementFromPoint, and keyboard Enter/ArrowUp/Enter. The IMPORT handler DOES accept synthetic events, which is why the .gp upload succeeded, so this is specific to this control rather than a general bridge failure.",
     "evidence": "isTrusted false read directly, and the gear reading Original:- ~ Synth:ACTIVE ~ Audio file:- after each attempt"
    }
   ],
   "why_option_1_was_refused": "implementing the fret-to-inches conversion would flip this artifact from REVIEW to PASS, and the pass would be produced by the same session that authored the artifact. The condition's own words are that 'a successful rebuild does not establish task completion' and that the validator determines status. A validator edited by the party it is judging determines nothing. The flaw is real, measured, and recorded with its formula in HAND_SPAN_IS_POSITION_BLIND.md for a pass that is not grading its own output."
  }
 }
}