Description
Studies why everyday work usually succeeds despite variability, contrasting work-as-imagined by designers with work-as-done by users to find adaptive strategies.
Safety-I defines safety as the absence of failures and studies incidents. Safety-II defines it as the ability to succeed under varying conditions and studies normal work. The core analytic move is comparing work-as-imagined (procedures, specifications, designers' flows) with work-as-done (observed practice), and also work-as-prescribed, work-as-disclosed and work-as-normal. Gaps are not treated as violations but as evidence of adaptations users make to cope with the system's shortcomings. Researchers then identify which adaptations are protective and should be supported, and which create latent risk. For UX this reframes workarounds as data about design debt and shows where rigid automation would remove the very flexibility that keeps work going.
Attributes
- Discipline
- Human Factors & Error
- Loss type
- Unmigrated
- Stage
- Discovery, Post-launch
- Data
- Qualitative
- Effort
- Weeks
Steps
- Document work-as-imagined: specs, flows, help content, training material and stakeholder descriptions of the task.
- Observe work-as-done in context across normal, busy and unusual days, focusing on successful outcomes.
- Catalogue adaptations and workarounds, with their triggers (time pressure, missing data, edge cases) and effects.
- Interview practitioners about trade-offs they make, such as efficiency versus thoroughness, using Hollnagel's ETTO framing.
- Classify adaptations as protective, neutral or risky, and map each to the design gap that prompts it.
- Design to support protective adaptations and make risky ones unnecessary; share findings with the people observed.
Sources
- Hollnagel, Safety-I and Safety-II: The Past and Future of Safety Management (2014)
- Hollnagel, The ETTO Principle: Efficiency-Thoroughness Trade-Off (2009)
- Hollnagel, Woods & Leveson, Resilience Engineering (2006)
Also known as
- Safety-II
- Work-as-imagined vs work-as-done
- Learning from what goes right
Categories
- Accessibility
- Human Factors & Error