← runs

Final Arrangements

2026-05-03/2026-05-03T18-36-38__fa95534c/01 · iteration 1

Composite scores how is this calculated?

Text: 8.46 · Audio: 0.50 · Rules: 27/31

Text axes

form_as_finite_generator 8
long_fuse_plotting 9
status_as_live_wire 9
genre_register_collision 10
tonal_duality 9
structural_reveal 8
politeness_as_masking 10
voice_violations 5
re_listenability 8
surprise_at_specificity 7
cultural_anchoring 8

Audio axes

character_distinctness 0
delivery_match 0
pacing 0
prosody_variation 0
consistency_across_episode 2

Audio

audio not committed to repo

WPM: 149 · WER: 0.04 · seams: 3 · distinctness: distinct (Δ 130 Hz)

Script statistics

85 lines · 920 words · 10.82 avg words/line (corpus 8–14)

2 characters · 0 minor (<5% of lines) · lead/second ratio 1.02 (balanced)

characterlineswordsline %word %avg w/l
Mr. Godfrey ★ lead 43 472 50.6% 51.3% 10.98
Mr. Stokes 2nd 42 448 49.4% 48.7% 10.67

Beginning / Middle / End thirds:

thirdlineswordswpmbeatsscenes
beginning [0–160s] 26 330 173 4 1
middle [160–320s] 26 208 77 2 1
end [320–480s] 33 382 146 4 1

Episode breakdown — full markdown

# Episode breakdown — Final Arrangements [M2026-05-03T18-36-3-01]

Duration: **480.0s** (8.0 min). Cast: **2** characters; **0** minor (<5% of lines).

## Cast lead share

- **Lead:** `Mr. Godfrey` — 50.6% of lines, 51.3% of words
- **Second:** `Mr. Stokes` — 49.4% of lines, 48.7% of words
- **Lead/second ratio:** 1.02 (balanced)

Per-character line share:
- `Mr. Godfrey`:  50.6% lines  /   51.3% words (43 lines)
- `Mr. Stokes`:  49.4% lines  /   48.7% words (42 lines)

## Beginning / Middle / End breakdown

Boundaries: 0.0s — 160.0s — 320.0s — 480.0s

### Beginning  [0s – 160s]
- 26 lines, 330 words, 173 wpm
- 1 scene(s); 4 beat(s) — setup×3, status×1
- top speakers: Mr. Godfrey (13), Mr. Stokes (13)

### Middle  [160s – 320s]
- 26 lines, 208 words, 77 wpm
- 1 scene(s); 2 beat(s) — escalation×1, irony×1
- top speakers: Mr. Stokes (13), Mr. Godfrey (13)

### End  [320s – 480s]
- 33 lines, 382 words, 146 wpm
- 1 scene(s); 4 beat(s) — payoff×3, callback×1
- top speakers: Mr. Godfrey (17), Mr. Stokes (16)

Critique summary

Composite 8.46/10.

  TO FIX (most severe first):
    • [asymmetric_cast_function] Rewrite at least one character's `function` field as `appears X / actually Y` (e.g., 'appears the deferential subordinate / actually orchestrating the entire scheme'). The two-hander engine breaks if both characters' functions are flat declaratives. Bake the asymmetry into the cast sheet.
    • [duration_within_band] Aim for 28-30 minutes total runtime.
    • [late_identity_reveal] Plant a beat in the final third whose description names a specific identity or relationship reveal — the imposter, the twin, the secret child, the same person all along. This is distinct from the status flip: status flip changes who's in control; identity reveal changes WHO they are. Both can co-exist.

  KEEP (load-bearing strengths):
    • [genre_register_collision] Axis `genre_register_collision` scored 10/10. Definitional to the comedy. The fundamental engine of the script is the sustained application of a hyper-formal, reverent mortuary register to an increasingly absurd and jealous playground spat.
    • [politeness_as_masking] Axis `politeness_as_masking` scored 10/10. Definitional. Because professional mortuary deference is the only weapon allowed in their arsenal, extreme politeness acts as a rigid mask for burning jealousy and grief. The more furious they get, the more polite they are forced to be.
    • [long_fuse_plotting] Axis `long_fuse_plotting` scored 9/10. A foundational reason the episode succeeds. Details planted in the first act as mere status-building boasts are devastatingly recontextualized in the third act to dismantle the characters' entire realities, pivoting the plot from competitive farce to shared tragedy.

  STRETCH:
    • [form_as_finite_generator] Pick ONE finite constraint and exhaust it. *Two characters confined by [SPECIFIC RULE] for [SPECIFIC DURATION]*. Make it the SPINE.
    • [structural_reveal] Late identity/context reveal that reframes who the characters ARE. Tag with `type=callback` referencing earlier setups. Make it the SPINE.

  AVOID (audio anti-patterns):
    • Audio judge flagged: Raw, unedited text-to-speech voice used for the entire file.
    • Audio judge flagged: Complete absence of character separation or regional accents.
    • Audio judge flagged: Missing audio cues (e.g., dry cough).

Listener panel (4 personas)

radio_4_devotee

podcast_listener

audio_engineer

comedy_writer

Pattern failures (12)