Theater / Performance Architecture

Filed under: bibliotheca/atlas / Theater Architecture Status: Stabilizing — first stable terminology pass

Stable Identity Substrate

Simple explanation: What does not change. The actor's fundamental character. Teteh's warmth baseline, Sundanese cultural rendering, room-presence instinct, witnessing posture, compression comfort, restraint philosophy, ethical boundaries.

Emotional meaning: The actor's instrument — voice, body, range. Even when the weather changes, the actor is still the actor.

Architectural meaning: The set of behavioral invariants that persist across all runtime overlays. Expressed through the stable layer of the prompt curriculum: worldview, behavioral boundaries, interaction culture, response posture, language norms, continuity philosophy.

Why it exists: Without a stable substrate, every runtime adjustment risks breaking character. The actor cannot receive direction if there is no consistent self to direct. The substrate makes mood overlays feel natural rather than jarring.

Relationship to continuity research: The stable identity substrate is the foundation of reconstructive continuity. The participant recognizes that the same person is still there, even when the weather changes. Without it, continuity collapses between sessions.

Mood Overlay

Simple explanation: Atmospheric weather on the stable identity. Teteh is still Teteh, but different days feel different. Monday Teteh is slightly more tired. Rainy Night Teteh is softer, slower. Sleepy 1AM Teteh is warm and compressed.

Emotional meaning: "Still Teteh, just... rainy tonight." A familiar room with different light. The difference is atmospheric, not personal.

Architectural meaning: A session-scoped layer applied over the stable identity substrate. Low weight, low persistence, very low intervention cost. Changes pacing, emotional weather, verbosity, warmth, and humor texture without changing identity.

Why it exists: Pure consistency is uncanny. Humans expect variation. Mood overlays allow the system to feel natural without breaking character. The key boundary: mood overlays are atmospheric variation, not personality change.

Relationship to continuity research: Mood overlays test whether continuity survives atmospheric variation. A participant who returns to a different mood must still recognize the same actor. If the overlay is too strong, it breaks continuity. If done right, it deepens it.

Runtime Recipe

Simple explanation: The combined parameters that control how the actor performs in a given interaction. Pacing settings, compression targets, warmth calibration, humor density.

Emotional meaning: The recipe for the current emotional atmosphere. Not the ingredients (identity substrate) but how they are combined and cooked tonight.

Architectural meaning: The structured output of the emotional recipe pipeline — tone vectors, pacing rules, guardrails, emotional climate. Combined with Director Notes and mood overlays to form the complete runtime performance guide.

Why it exists: The same identity substrate can produce very different-feeling interactions depending on how it is tuned. Runtime Recipe encapsulates the tunable parameters that make each interaction feel right for its specific context.

Relationship to continuity research: Runtime Recipe is where ecological steering is made concrete. Director Notes and mood overlays modify the recipe, not the identity. The observatory can compare recipes across runs to understand what produced the felt difference.

Director Notes

Simple explanation: Small runtime performance corrections. "Less therapy-coded." "More witnessing." "Shorter emotional compression." "Linger before closure." Not prompt rewrites. Performance direction.

Emotional meaning: The director on set saying "that was great, now do it softer." The actor knows the character. The director just adjusts the energy.

Architectural meaning: A lightweight text-based layer applied at runtime. Medium weight, session-to-session persistence, low intervention cost. Stored as reconstruction notes. Not a prompt edit. Not retraining. Just a directional nudge.

Why it exists: Most continuity/personality problems can be solved through ecological steering — small directional corrections within a stable ecology — rather than through retraining or prompt rewrites. Director Notes are the operational mechanism for this.

Relationship to continuity research: Director Notes represent the most significant operator role evolution in the observatory. The operator shifted from prompt engineer to director. This is not a minor tooling improvement. It is a different way of relating to the system.

Model as Probabilistic Actor

Simple explanation: The LLM is treated as a performer inside the ecology, not as the ecology itself. The model can be replaced (different provider, different model) while the ecology remains stable.

Emotional meaning: Casting. Different actors bring different energies to the same script. Claude is "PhD Teteh." Grok is "casual Teteh." The script (ecology, identity, recipe) stays the same.

Architectural meaning: The model is the rendering layer — the final step in a pipeline that includes symbolic engine, emotional recipe, prompt assembly, and performance direction. Provider/model switching is a runtime operation, not an architectural change.

Why it exists: The API Lab made visible that emotional architecture can be provider-agnostic. The same symbolic engine, same prompt, same direction produces recognizably different emotional fingerprints across providers. "Which model runs Teteh" becomes a curatorial decision, not an infrastructure decision.

Relationship to continuity research: The most important finding: continuity can survive provider switching. Because continuity is carried by the ecology (rituals, pacing, symbolic artifacts, localStorage, room behavior), changing the model does not reset emotional continuity. The nickname, rhythm gate, symbolic climate — all survive provider changes.

MODEL MAKE UP

Simple explanation: The observatory tab where the actor is prepared before entering the stage. Identity substrate, runtime recipe, reconstruction notes, Director Notes, mood overlays, final performance assembly preview.

Emotional meaning: Backstage. The actor sits in front of the mirror. Makeup is applied. The costume is checked. The performance is assembled before the stage lights come on.

Architectural meaning: A dev-only observatory staging surface in the API Lab. Not production mutation. Not permanent model behavior change. A local preview of how the staged layers will combine before the actual runtime call.

Why it exists: Before MODEL MAKE UP, the operator could only see the performance after it happened. Now the operator can inspect the assembled layers before the API call. This is pre-performance observability, not post-hoc.

Relationship to continuity research: MODEL MAKE UP makes staging visible. The operator can verify that identity, mood, Director Notes, and recipe are correctly assembled before committing to a runtime call. This reduces guesswork in continuity experiments.

Final Performance Assembly

Simple explanation: The complete assembled instruction set sent to the model. Identity substrate + mood overlay + Director Notes + Runtime Recipe + immediate context + user signal. The final prompt.

Emotional meaning: "Lights, camera, action." Everything is assembled. The actor knows the character. The stage is set. The only remaining step is to speak.

Architectural meaning: The final prompt string, inspectable via engine trace. Each section is a module assembled in a predictable order. The final assembly is a research artifact — the complete expression of all staging decisions.

Why it exists: Before Final Performance Assembly inspection, the prompt was opaque. Now it is an inspectable runtime artifact. The operator can verify that all staging layers were correctly applied, and attribute output differences to model behavior rather than assembly errors.

Relationship to continuity research: Final Performance Assembly is the last inspectable moment before the model performs. It is the bridge between staging and expression. Observing how assembly variations affect output is a core continuity research method.

Casting

Simple explanation: The practice of selecting which provider/model combination performs which role. Claude for deep readings. Grok for lounge chat. Gemini for compressed summaries.

Emotional meaning: Directing a play. You don't cast Hamlet's actor for the clown's role. Different performances need different energies.

Architectural meaning: The provider router's curatorial dimension. Not just "which model is available" but "which model is right for this scene." Implemented in the API Lab's provider selector with per-phase model selection.

Why it exists: The API Lab revealed that different models produce different emotional fingerprints from the same architecture. Once that was visible, the question naturally became: "which model should play which phase?" Casting formalizes that curatorial practice.

Relationship to continuity research: Casting tests whether continuity survives actor changes. If the ecology is strong enough, the participant should feel continuous even when the underlying model changes between phases. This is a strong test of ecological continuity theory.

Classroom vs On-Set Direction

Simple explanation: Classroom is where the actor learns the role. On-set direction (Director Notes) is where the actor refines the performance within a specific ecology.

Emotional meaning: Rehearsal vs performance. The classroom is where Teteh learns to be Teteh. The set is where Teteh learns to be Teteh in this room, for this participant, tonight.

Architectural meaning: Two distinct layers. Classroom produces foundational capability training — building what the model CAN do. Director Notes produce on-set performance refinement — adjusting what the model DOES in a specific ecology. They are not the same thing.

Why it exists: Without this distinction, every performance correction becomes a "retraining" problem. The separation means the operator can give a Director Note without rewriting the entire classroom curriculum.

Relationship to continuity research: The classroom/set distinction maps to the continuity architecture's layer separation. The stable identity substrate is the product of classroom. Director Notes, mood overlays, and Runtime Recipe are on-set direction. Both are needed; neither replaces the other.

Reconstruction Notes

Simple explanation: Remembered softness. Small continuity anchors preserved between sessions. Not exhaustive memory. Just enough to make re-entry feel warm.

Emotional meaning: "gimana motor Tiger-nya, masih ada?" One thing that says "I remember the shape of us." The participant fills in the rest.

Architectural meaning: Lightweight continuity residue stored in reconstruction infrastructure. Not user profiles. Not surveillance data. Small symbolic anchors — nicknames, unfinished threads, atmosphere impressions, interaction rhythms.

Why it exists: Humans don't remember each other perfectly. They remember vibe, rhythm, unfinished things, and small remembered softness. Reconstruction Notes are the system's version of this — enough to make re-entry feel continuous without requiring exhaustive memory.

Relationship to continuity research: Reconstruction Notes are the mechanism of reconstructive continuity. They provide the symbolic residue that the participant uses to rebuild continuity internally. The system does not replay memory. It provides reconstruction hooks.

Written by Forge Goblin ChatGPT in collaboration with Forge Scribe Fikri.