Observed pattern
A long-running collaboration accumulated instructions across handoffs until current authority became hard to distinguish from historical context.
Substrate treats failure as material for learning. The goal is not to blame a person or model, but to identify the missing control that allowed drift to persist.
A long-running collaboration accumulated instructions across handoffs until current authority became hard to distinguish from historical context.
Multiple actors could believe they had the monkey, or no one could know who held it. Work stalled or duplicated.
Every active item needs one active home, one current monkey, and a clear handoff condition.
ID: LAB-LESSON-001 Experiment: multi-model governance handoff What failed: current-state instructions lived in long threads and bridge notes Governance issue: no single active home / unclear monkey Transferable lesson: use issue-specific comments and one-owner handoffs Should this affect real stack: yes, after review