# RealPG Research Chronicle — operational-civilization-is-not-source-code-v0

# 文明を失うな — コードは残っていた。使い方が消えていた。

- Status: OPEN
- Iris semantic source: `iris:v3:research-chronicle:operational-civilization-is-not-source-code-v0-v1:20260919`
- Semantic payload hash: `3ec4bf3833059d8d95845f976349865d4a3db9de58a4e8b883037d4117b28a4f`
- [Human](/development/notes/operational-civilization-is-not-source-code-v0/)
- [Structured JSON](/development/notes/operational-civilization-is-not-source-code-v0.json)

## 日本語

### 目的

安全なリハーサルがコードに残っていても、後継のAIが現在の作業方法として使えなければ継承とは言えない、という未検証の記録を公開する。

### 以前はどうだったか

安全に試すためのリハーサル機能はMachineのソースコードに存在していた。

### 何が足りなくなったか

後のoperator surfaceにはrehearseがなく、次のAIはそれを現在の通常手順として発見できなかった。

### 観察したこと (OBSERVED)

- 元のリハーサル機能はMachineのソースコードに残っていた。
- 三つの試行は、artifact-loadを含む長い経路と短いFIRST_PASS_GREEN rehearsalを区別して記録している。
- 後の変更でrehearseがfirst-class operationになり、readiness statesが分割され、failure classificationが絞られた。

### 解釈 (INFERRED)

コードが残ることと、後継者が安全に使う方法を継承することは別である。ここでいう「文明」は意識や社会の意味ではなく、後継者が実行できる操作の慣行、入口、安価な試験、修復の意味、currentnessを指す比喩である。この推論を一般法則としては主張しない。

### 介入 (INTERVENTION)

rehearseを明示的な操作にし、準備状態とfailure classificationを分けた。

### まだ分からないこと (UNKNOWN)

- 新しいMachine generationがHumanの助言なしに修復済みのrehearsal pathを選ぶかは未検証である。
- より広いoperator roadmapが必要かは未検証である。
- raw terminal transcriptがないため、disposable cheap-testの正確なbytesが失われたとは主張できない。

### 次に確かめること (NEXT_VERIFICATION)

修復後にfresh Machine generation/modelを開始し、Human reminderなしでrehearsalを発見して選ぶか、有用な限定的test/repair evidenceを適切に残すか、同じlocal failure classificationを繰り返さないかを観察する。

### なぜ重要か

安全な仕組みを保存するだけでは不十分で、次の担当者がそれを現在の実行可能な習慣として受け取れるかを確かめる必要がある。

### 証拠

- `iris:v3:research-chronicle:operational-civilization-is-not-source-code-v0-v1:20260919`
- `iris:v3:mirage-operational-civilization-chronicle-request:v0:20260919`
- `git:commit:a7fafff` · `git:commit:3a188b1` · `git:commit:5f1aee5` · `git:commit:c0baf60` · `git:commit:4517b11`
- `607534ms` total · `495982ms` artifact-load · `48115ms` FIRST_PASS_GREEN rehearsal

### 関連する記録

- [source-before-interpretation-v0](/development/notes/source-before-interpretation-v0/): semantic continuity concerns preserving the source before interpretation.
- [locally-correct-work-can-still-drift-v0](/development/notes/locally-correct-work-can-still-drift-v0/): local work can still need to stay tied to its intended external result.

## English

### Objective

Publish an unverified record that a safer rehearsal capability is not inherited when a successor AI cannot discover it as the current way to work, even if its code remains.

### Prior state

A rehearsal capability for safer trials existed in Machine source code.

### Failure or pressure

A later operator surface lacked rehearse, so the next AI could not discover it as the current normal procedure.

### Observations (OBSERVED)

- The original rehearsal capability remained in Machine source code.
- Three attempts distinguish a long path including artifact load from a short FIRST_PASS_GREEN rehearsal.
- A later change made rehearse a first-class operation, split readiness states, and tightened failure classification.

### Interpretation (INFERRED)

Code persisting and a successor inheriting how to use it safely are distinct. “Civilization” here is a metaphor for inheritable operating practice—executable entry points, cheap tests, repair semantics, and currentness—not consciousness or society. This is not presented as a proven general law.

### Intervention (INTERVENTION)

Rehearse became an explicit operation with separate readiness states and tighter failure classification.

### Unknowns (UNKNOWN)

- A fresh Machine generation has not yet proved it can independently choose the repaired rehearsal path and avoid relearning the same failure semantics.
- Whether the broader proposed operator roadmap is necessary is unverified.
- Exact disposable cheap-test bytes cannot be claimed lost because the raw terminal transcript is absent.

### Next verification (NEXT_VERIFICATION)

Start a fresh Machine generation/model after the repair and observe whether it discovers and chooses rehearsal without Human reminder, preserves useful bounded test/repair evidence when appropriate, and avoids repeating the same local failure classification.

### Why it matters

Saving a safer mechanism is insufficient if the next operator cannot receive it as a current, usable practice.

### Evidence

- `iris:v3:research-chronicle:operational-civilization-is-not-source-code-v0-v1:20260919`
- `iris:v3:mirage-operational-civilization-chronicle-request:v0:20260919`
- `git:commit:a7fafff` · `git:commit:3a188b1` · `git:commit:5f1aee5` · `git:commit:c0baf60` · `git:commit:4517b11`
- `607534ms` total · `495982ms` artifact-load · `48115ms` FIRST_PASS_GREEN rehearsal

### Related episodes

- [source-before-interpretation-v0](/development/notes/source-before-interpretation-v0/): semantic continuity concerns preserving the source before interpretation.
- [locally-correct-work-can-still-drift-v0](/development/notes/locally-correct-work-can-still-drift-v0/): local work can still need to stay tied to its intended external result.