Per-session randomization

The “music theory” lives in proxy/src/theory.ts. Three layers of variation combine, so two descents never sound the same — and not just “shuffled versions of the same piece” but genuinely different keys, modal colours, and pitch trajectories.

1. Scale picker — 96 starting points

At session creation the bridge picks a random root (12 pitch classes) and a random mode (8 modes), then locks them for the lifetime of the descent:

Mode	Feel
Aeolian	classic minor — melancholy, settled, familiar
Dorian	minor with a softened sixth — pensive but not despairing
Phrygian	dark, eastern, restless — flat second pulls everything toward unease
Lydian	bright but strange — raised fourth, floating, unresolved
Mixolydian	major-ish but earthy — flat seventh, folkloric
Locrian	unstable, hollow — diminished fifth refuses to resolve
Harmonic Minor	minor with a leading-tone bite — exotic, narrow tension
Pentatonic Minor	no half-steps — gamelan-clean, every interval lands

12 × 8 = 96 distinct (root, mode) combinations. Comparing C Aeolian to F♯ Lydian isn’t “same music transposed” — they’re different sonic worlds, with different intervals, different harmonic gravity, and different colour notes.

2. Per-layer randomization

Every placed layer (player or agent) goes through pickFreqForLayer(), which composes three sources of variation:

• Octave — each layer type has a preferred octave range (drone oct 2, bell oct 4–5, drip oct 5–6, etc.). When the range has multiple options the engine picks randomly, so two bell placements in the same descent can land an octave apart.
• Scale degree — chosen from a candidate set derived from the player’s click (no intent → consonant-weighted random) or the agent’s intent parameter (see below).
• Sub-parameter jitter inside the engine — chord amplitude is 0.0448 + Math.random() * 0.0358; swell amplitude is 0.125 + Math.random() * 0.10; glitch grain timing is randomised; drone filter has a static centre but the sub voice gets a random detune cent. None of these change the pitch, but they prevent two same-type layers from sounding identical.

3. Player + agent compositional choices

On top of pitch randomization, the actual sequence of layer types and their spatial placement is unique per descent. The player chooses 7–8 types and where in 3D space to drop them; the agent picks the other half driven by its own reading of the in-progress composition (see How Hermes decides).

4. Pads also derive from the session scale

The three atmospheric pads (GLOW, AIR, DEEP) build their voicings from scale.intervals:

• GLOW — root + 3rd + 5th
• AIR — root + 5th + 9th
• DEEP — sub + root + 5th

So in C Aeolian the GLOW pad is C–E♭–G; in F♯ Lydian it’s F♯–A♯–C♯. Different notes, different intervals, different tonal colour. Pads aren’t a fixed sample bank — they’re synthesised in-key on every descent.

Why this matters

Without (1), a session would always be in the same key — every minted glyph would be tied to the same fundamental, and the agent’s tension would always mean the same interval. Without (2), every drone in a descent would land on identical Hz. Without (3), the sequence would collapse into agent-only or player-only patterns. Stacking the three gives 96 × layer-choice × pitch-jitter possibilities, far more than the ~10⁴ tokens we expect to mint.