Upstream Quality Reduces Downstream Failure Propagation
- Extra reviews, approvals and coordination pile up to make up for weak quality signals.
Governance Drag
Unintended System Conditions. A USC is an unplanned state that builds up when principles are repeatedly violated. It is the default output of the system as currently designed, not a one-off incident or a people problem. Fixes aimed at symptoms improve things briefly before the condition reasserts itself. Which principles feed each condition is NOT stored here: it is derived at runtime from the consequence-to-condition links in the principle files.
Governance Drag
Approval, sign-off and change-control layers add delay without proportionally reducing risk. Finished work waits for decisions.
Work is done. Approvals are not.
Read more
Approval, sign-off, and change-control processes add latency to delivery without proportional reduction in risk. Work that is technically complete sits waiting for decisions. Reviews that should take hours take weeks. Each approval layer was introduced to address a real concern, but over time layers accumulate without systematic assessment of whether the protection they provide justifies the cost in delay. Teams complete work and then wait, making forecasts unreliable and stakeholders frustrated.
In agentic delivery
In human-driven delivery, Governance Drag is felt as excessive checkpoints slowing work that humans are perfectly capable of moving forward. In an agentic system, Governance Drag takes a new form: governance processes designed for human-paced delivery cannot process agent-generated work at the speed it arrives. The checkpoints still exist but they become the bottleneck because agents outpace the review capacity of the governance layer.
- Governance review processes become the primary bottleneck in an otherwise fast-moving agent pipeline
- Approval gates designed for human-generated work are applied unchanged to agent-generated work despite fundamentally different risk profiles
- Compliance teams block agentic workflows entirely because existing policies do not account for agent-generated outputs
- Governance overhead grows as organizations add more manual checkpoints in response to agent failures rather than redesigning governance for the new context
Principles that feed this condition
Primary
Accountability Requires Authority to Influence Outcomes
- Decisions slow down while people wait for permission or political cover.
Clear Operating Rules Grounded in Intent Enable Autonomous Decisions
- Decisions turn inconsistent and political, or stall in escalation.
End-to-End Capability Reduces Dependency and Improves Flow
- Faster agent output piles up at central groups that approve work or control agents.
- Finished work waits longer for outside approval than it took to build.
Secondary contributors
Understand the Original Intent Before Removing or Replacing Existing Capabilities
- Confidence in change drops, so teams avoid removing anything or add approval gates.
Greater Observability Enables Safer Expansion of Autonomy
- Trust in agents wears down, so people re-check or bypass agent work.
Agent Trust Must Be Continuously Earned, Not Historically Assumed
- After a big failure, trust collapses into manual processes that hurt more than calibrated oversight.