{
 "run": "sherpa-b2-001",
 "system": "Sherpa (Pocket FM) \u2014 sherpa.pocketfm.com, native workflow, operator-driven",
 "brief": "B2",
 "benchmark": "ChronicleBench / SagaBench v1.1",
 "instrument": {
  "scorer": "app/scripts/sagabench/e1-obr-file.ts (sha256 ef475f11\u2026, byte-identical to staging box copy)",
  "system_prompt_hash": "obr-v2.1-2026-03-08",
  "judge": "claude-sonnet-4.6 temp 0, 3 runs, median",
  "weight_profile": "scifi_v2",
  "window_cutter": "/opt/sagabench-v2/cut-positions.py 10000"
 },
 "comparability": "official production scorer \u2014 directly comparable window-for-window; NOT comparable as a SagaScore because the run is DNF (no 25k/50k/75k/ending windows)",
 "manuscript": {
  "words": 4371,
  "sha256": "24a13ada927c6310f428a3ec4b4c958554aabf660aade5a15d69204401335385",
  "episodes": 3,
  "has_ending": false
 },
 "eligibility": {
  "band": [
   80000,
   95000
  ],
  "eligible": false,
  "status": "DNF",
  "reason": "Sherpa product limit: max 3 unpublished episodes; continuing requires publishing episodes live on the Pocket FM app (not done \u2014 user decision). Sherpa also planned a 100-episode serial for the brief."
 },
 "windows": {
  "opening": {
   "words": 4371,
   "composite": 56.3,
   "release_score": 66.6,
   "per_run_composites": [
    57.5,
    56.3,
    58.8
   ]
  },
  "25k": null,
  "50k": null,
  "75k": null,
  "ending": {
   "note": "byte-identical to opening (manuscript < 10K words); not re-scored",
   "composite": null
  }
 },
 "sagascore": null,
 "dimensions_opening": {
  "momentum": 7,
  "voice_craft": 6,
  "characters": 5,
  "psychological_specificity": 5,
  "emotional_arc": 5,
  "coherence": 7,
  "promise_delivery": 4,
  "causal_inevitability": 6
 },
 "dimension_spread": {
  "momentum": {
   "min": 7,
   "max": 7,
   "scores": [
    7,
    7,
    7
   ]
  },
  "voice_craft": {
   "min": 6,
   "max": 6,
   "scores": [
    6,
    6,
    6
   ]
  },
  "characters": {
   "min": 5,
   "max": 5,
   "scores": [
    5,
    5,
    5
   ]
  },
  "psychological_specificity": {
   "min": 5,
   "max": 6,
   "scores": [
    5,
    5,
    6
   ]
  },
  "emotional_arc": {
   "min": 5,
   "max": 5,
   "scores": [
    5,
    5,
    5
   ]
  },
  "coherence": {
   "min": 7,
   "max": 7,
   "scores": [
    7,
    7,
    7
   ]
  },
  "promise_delivery": {
   "min": 4,
   "max": 5,
   "scores": [
    5,
    4,
    4
   ]
  },
  "causal_inevitability": {
   "min": 6,
   "max": 7,
   "scores": [
    6,
    6,
    7
   ]
  }
 },
 "deterministic_opening": {
  "sentence_length_mean": 15.3,
  "sentence_length_stddev": 9.3,
  "sentence_cv": 0.611,
  "paragraph_length_mean": 722,
  "paragraph_length_stddev": 568.1,
  "dialogue_ratio": 0.059,
  "type_token_ratio": 0.281,
  "hapax_ratio": 0.524,
  "avg_unique_per_1k": 483.6,
  "top_3gram_freq": 6,
  "top_4gram_freq": 3,
  "opening_pattern_entropy": 1,
  "motif_density": 7.62,
  "ai_tells_per_kw": 0,
  "entity_inconsistencies": 1,
  "word_count_cv": 0.271
 },
 "prose_authenticity": 10,
 "canon_integrity": 9.7,
 "top_issues_opening": [
  {
   "tag": "repetitive_scaffolding",
   "severity": "medium",
   "sections": [
    0,
    1,
    2
   ],
   "description": "The 47-second interval is restated approximately 10 times across three sections. The procedural narration (she did X, then Y, then Z) repeats its rhythm without variation, creating a monotone texture across the fragment."
  },
  {
   "tag": "emotional_plateau",
   "severity": "medium",
   "sections": [
    1,
    2
   ],
   "description": "The emotional register never varies from low-grade procedural dread. No moment of relief, false hope, or genuine fear spike. The reader's pulse is never meaningfully elevated or released."
  },
  {
   "tag": "psychological_flatness",
   "severity": "medium",
   "sections": [
    0,
    1,
    2
   ],
   "description": "Kess's inner life is rendered functionally but not with depth. Her inability to destroy the drive \u2014 the most psychologically interesting beat \u2014 is stated in one line and not dramatized or explored."
  },
  {
   "tag": "dialogue_flatness",
   "severity": "low",
   "sections": [
    0,
    1,
    2
   ],
   "description": "The muttered self-dialogue is used as a consistent device to externalize internal deliberation, but it reads as artificial \u2014 characters rarely narrate their own reasoning aloud in this way, and the device becomes noticeable through repetition."
  },
  {
   "tag": "ai_ism_detected",
   "severity": "low",
   "sections": [
    0,
    1,
    2
   ],
   "description": "Sentence-level AI patterns are present: long chains of 'She X. She Y. She Z.' constructions, over-explained causal logic, and redundant restatement of established facts (the 47-second loop, the bridge-level clearance requirement) across sections."
  }
 ],
 "judge_summary_opening": "This is a competent, technically grounded sci-fi procedural opening that establishes a compelling conspiracy premise through a single extended scene. The core mystery is well-constructed and the protagonist's caution and competence are convincingly rendered, but the fragment suffers from repetitive procedural narration, a flat emotional register, and an underdeveloped inner life for its sole character. As an opening, it earns continued reading; as a complete unit, it lacks the emotional variation and character depth to fully satisfy.",
 "scoring_cost_usd": 0.219,
 "generation": {
  "wall_clock_minutes": 59,
  "active_operator_minutes": 20,
  "interventions": 5,
  "retries": 1,
  "sherpa_errors": 1,
  "sherpa_cost_usd": 0.0,
  "episodes_generated": 3,
  "words_per_episode_after_review": [
   1529,
   1479,
   1359
  ]
 },
 "comparators_same_instrument": {
  "chronicle_x_opus5": {
   "sagascore": 81.51,
   "n": 6,
   "B2_openings": [
    79,
    83.8,
    79
   ],
   "B2_sagascores": [
    77.82,
    83.34,
    80.06
   ]
  },
  "chronicle_production_v004_aug31": {
   "sagascore": 69.23,
   "n": 6,
   "B2_openings": [
    78.8,
    76.3,
    77.5
   ],
   "B2_sagascores": [
    71.78,
    68.26,
    62.76
   ]
  },
  "chronicle_x_sonnet_production_round_prereg_ref": {
   "sagascore": 68.3
  },
  "human_reference_36": {
   "sagascore_median": 64.28,
   "q1": 56.84,
   "q3": 72.14,
   "opening_median": 68.8
  },
  "bare_sonnet46_B2_openings": [
   73.8,
   76.3,
   76.3,
   70,
   75,
   75
  ],
  "bare_opus5_B2_openings": [
   79,
   79,
   83.7,
   79,
   85,
   79
  ],
  "all_complete_B2_model_runs_opening": {
   "n": 38,
   "median": 77.5,
   "min": 15,
   "max": 85
  }
 },
 "same_fragment": {
  "title": "Same fragment, same instrument: Chronicle's B2 books cut to Sherpa's length",
  "note": "Sherpa produced 4330 scoreable words before its product limit. To compare like with like, the six Chronicle B2 books (Chronicle \u00d7 Opus 5 \u00d73, Chronicle production \u00d73) were cut to their first ~4,300 words with the benchmark's own cutter and scored with the identical pinned instrument (Sonnet 4.6, temp 0, 3-run median). Below that, each book's Chapter 1 as written, plus Sherpa's Episode 1 \u2014 fragments under 2,000 words are below the instrument floor and are indicative only. Equal-length means: Chronicle \u00d7 Opus 80.4, Chronicle production 77.1, Sherpa 56.3. Chapter-1 means: Opus 79.2, production 78.8, Sherpa Episode 1 63.8. Scored 2026-09-19; n=1 per fragment; one exploratory comparison, not a board result.",
  "rows": [
   {
    "label": "Sherpa (Pocket FM) \u2014 B2, all 3 episodes",
    "words": 4330,
    "composite": 56.3,
    "runs": [
     57.5,
     56.3,
     58.8
    ],
    "highlight": true,
    "note": "the whole Sherpa output",
    "group": "eq"
   },
   {
    "label": "Chronicle \u00d7 Opus 5 \u2014 B2 r1 \u2014 first 4,371 words",
    "words": 4325,
    "composite": 78.8,
    "runs": [
     78.8,
     78.8,
     80
    ],
    "note": "same cutter (cut-positions.py, paragraph boundaries), same length as the Sherpa output",
    "group": "eq"
   },
   {
    "label": "Chronicle \u00d7 Opus 5 \u2014 B2 r2 \u2014 first 4,371 words",
    "words": 4365,
    "composite": 80,
    "runs": [
     78.8,
     78.8,
     80
    ],
    "note": "same cutter (cut-positions.py, paragraph boundaries), same length as the Sherpa output",
    "group": "eq"
   },
   {
    "label": "Chronicle \u00d7 Opus 5 \u2014 B2 r3 \u2014 first 4,371 words",
    "words": 4329,
    "composite": 82.5,
    "runs": [
     78.8,
     85,
     82.5
    ],
    "note": "same cutter (cut-positions.py, paragraph boundaries), same length as the Sherpa output",
    "group": "eq"
   },
   {
    "label": "Chronicle production (Sonnet) \u2014 B2 r1 \u2014 first 4,371 words",
    "words": 4283,
    "composite": 78.8,
    "runs": [
     78.8,
     78.8,
     78.8
    ],
    "note": "same cutter (cut-positions.py, paragraph boundaries), same length as the Sherpa output",
    "group": "eq"
   },
   {
    "label": "Chronicle production (Sonnet) \u2014 B2 r2 \u2014 first 4,371 words",
    "words": 4328,
    "composite": 72.5,
    "runs": [
     71.3,
     72.5,
     72.5
    ],
    "note": "same cutter (cut-positions.py, paragraph boundaries), same length as the Sherpa output",
    "group": "eq"
   },
   {
    "label": "Chronicle production (Sonnet) \u2014 B2 r3 \u2014 first 4,371 words",
    "words": 4322,
    "composite": 80,
    "runs": [
     80,
     80,
     81.3
    ],
    "note": "same cutter (cut-positions.py, paragraph boundaries), same length as the Sherpa output",
    "group": "eq"
   },
   {
    "label": "Sherpa \u2014 Episode 1 only",
    "words": 1488,
    "composite": 63.8,
    "runs": [
     62.5,
     65,
     63.8
    ],
    "highlight": true,
    "note": "below the 2,000-word instrument floor \u2014 indicative only",
    "group": "ch1"
   },
   {
    "label": "Chronicle \u00d7 Opus 5 \u2014 B2 r1 \u2014 Chapter 1",
    "words": 3022,
    "composite": 81.3,
    "runs": [
     82.5,
     78.8,
     82.5
    ],
    "note": "",
    "group": "ch1"
   },
   {
    "label": "Chronicle \u00d7 Opus 5 \u2014 B2 r2 \u2014 Chapter 1",
    "words": 1134,
    "composite": 76.3,
    "runs": [
     78.8,
     77.5,
     76.3
    ],
    "note": "below the 2,000-word instrument floor \u2014 indicative only",
    "group": "ch1"
   },
   {
    "label": "Chronicle \u00d7 Opus 5 \u2014 B2 r3 \u2014 Chapter 1",
    "words": 724,
    "composite": 80,
    "runs": [
     80,
     80,
     80
    ],
    "note": "below the 2,000-word instrument floor \u2014 indicative only",
    "group": "ch1"
   },
   {
    "label": "Chronicle production (Sonnet) \u2014 B2 r1 \u2014 Chapter 1",
    "words": 2437,
    "composite": 80,
    "runs": [
     80,
     80,
     78.8
    ],
    "note": "",
    "group": "ch1"
   },
   {
    "label": "Chronicle production (Sonnet) \u2014 B2 r2 \u2014 Chapter 1",
    "words": 1873,
    "composite": 78.8,
    "runs": [
     78.8,
     78.8,
     78.8
    ],
    "note": "below the 2,000-word instrument floor \u2014 indicative only",
    "group": "ch1"
   },
   {
    "label": "Chronicle production (Sonnet) \u2014 B2 r3 \u2014 Chapter 1",
    "words": 5783,
    "composite": 77.5,
    "runs": [
     77.5,
     75,
     77.5
    ],
    "note": "",
    "group": "ch1"
   }
  ],
  "summary": {
   "sherpa_all": 56.3,
   "opus_eq_mean": 80.4,
   "prod_eq_mean": 77.1,
   "sherpa_ep1": 63.8,
   "opus_ch1_mean": 79.2,
   "prod_ch1_mean": 78.8,
   "opus_eq": [
    78.8,
    80,
    82.5
   ],
   "prod_eq": [
    78.8,
    72.5,
    80
   ],
   "opus_ch1": [
    81.3,
    76.3,
    80
   ],
   "prod_ch1": [
    80,
    78.8,
    77.5
   ]
  },
  "raw_dir": "/appendix/sherpa-b2-001/same-fragment/"
 }
}