Safety I: Humans are a Liability
In the Safety I worldview, the system is assumed to be well-designed and correct. When something goes wrong, a human must have deviated from the correct procedure. The logical response is to constrain human behavior more tightly.
- View: The system is perfect; humans introduce errors.
- Goal: "As few things as possible go wrong."
- Method: Constrain the human. Add more rules, checklists, barriers, and approval gates.
- Analysis: Root Cause Analysis (RCA) of accidents — find the human who deviated and fix the deviation.
Safety II: Humans are a Resource
In the Safety II worldview, complex systems are inherently underspecified. No process document can anticipate every situation. Humans are not the source of failure — they are the adaptive layer that keeps the system running despite its imperfections.
- View: The system is messy and underspecified; humans are the glue holding it together.
- Goal: "As many things as possible go right."
- Method: Empower the human. Give them better tools, visibility, and adaptive capacity.
- Analysis: Studying "Normal Work" — how work is actually done every day, not just during accidents.
Key Insight: In Safety II, success and failure come from the same source: human adaptation. The same shortcut that caused an outage on Tuesday is the same shortcut that saved the deadline on Monday. You cannot eliminate the shortcut without understanding why it exists.