Skip to content
PCE 2.0
Search
Ctrl
K
Cancel
Zenn
Select theme
Dark
Light
Auto
Guide
Guide
PCE 2.0 とは何か
PCE 1.0 から 2.0 へ
問題設定
Reading Paths
Spec
Canonical
Spec
Principles
Principles
Process-first
Responsibility-first
対称的アクター性、非対称的責任
No merge without eval
Ontology
Ontology
Actor
Governance Surface
Recovery Point
Process Frame
Responsibility Bundle
Actor-local Compiled Context
Durable Project State
Process Delta
Responsibility
Responsibility
Goal Ownership
Execution
Approval
Evaluation
Memory Writing
Incident Ownership
Asymmetry
Process
Process
Transitions
Handoff
Approval Points
Checkpoint and Recovery
Lifecycle
Parent-Child Process
Branch and Join
Escalation
Rollback
Context
Context
Context Budget
Context Selection
Compaction
Memory
Memory
Memory Promotion Rules
Governance
Governance
Capability Scope
Auditability
Human Oversight
Evaluation
Evaluation
No Merge Without Eval
Eval Contract
Outcome vs Process
Process Metrics
Memory Promotion Criteria
Corrupt Success
Reference
Reference
Glossary
Term Index
Schemas
Schema: Process Frame
Schema: Compiled Context
Schema: Process Delta
Patterns
Mappings
Mappings
関連概念との関係
OpenAI との対応関係
Anthropic との対応関係
LangGraph との対応関係
Microsoft との対応関係
MCP との対応関係
Cases
Applied
Cases
Feature Delivery
PR Review
Bug Triage
Research Loop
Human-AI Handoff
Lab
Experimental
Lab
Open Questions
Hypotheses
ANT and Responsibility
Process Evaluation
Human-AI Parity
Notes
Changelog
Changelog
2026-03-08 Initial IA
2026-03-10 Public Draft Hardening
Zenn
Select theme
Dark
Light
Auto
Hypothesis:
このページは検証中の主張です。Spec の正規定義として引用しないでください。
Human-AI Parity
仮説
Section titled “仮説”
actor 記述は対称にできても、責任と承認権限は完全対称にはならない。
確認したいこと
Section titled “確認したいこと”
parity を強く置いたときの運用上の破綻
human oversight の必要条件