Safety Gradient Mapping¶

Why It Matters¶

We think of safety as binary—either you're safe or you've had an accident. But systems exist in a continuous space between these extremes. Organizations operate closer to or farther from the boundaries of safe operation, and this distance changes through gradual drift. Mapping the safety gradient makes visible that you're becoming less safe well before an accident proves it. It enables intervention during drift, not just after catastrophe.

The Key Move¶

Define what constitutes the boundary of safe operation (these are often regulatory limits, design specifications, or operational procedures). Track operational metrics over time: how close are you to these boundaries? Are you trending toward them or away? Create a visual timeline showing distance from boundaries across multiple dimensions (time pressure, resource constraints, maintenance backlog, training adequacy). Look for gradual trends that wouldn't trigger alarms in any single measurement but show drift in aggregate. Set alerts not just for boundary violation but for velocity toward boundaries.

Classic Application¶

Aviation safety metrics. Airlines track "leading indicators" of safety gradient movement: unstabilized approaches (coming in too fast/high), go-around rates (pilots aborting landings), maintenance deferrals (flying with non-critical equipment broken), crew fatigue reports. None of these alone constitute an accident, but trending in the wrong direction shows drift toward the accident boundary. Safety management systems now explicitly map these gradients rather than just counting accidents.

Surprising Application¶

Personal burnout prevention. Burnout isn't binary—there's a gradient from energized to tired to exhausted to burned out. Track leading indicators over time: sleep quality, exercise frequency, social connection, work hour trends, ratio of reactive to proactive time, frequency of emotional volatility. None of these alone means burnout, but trending wrong shows you're drifting toward the boundary. Intervention is easier at "tired" than at "burned out."

Failure Modes¶

Requires defining boundaries, which are often unclear or contested. What is the boundary of "safe" operation? Different stakeholders define it differently. Leading indicators may not actually predict trailing indicators (accidents)—correlation without causation. Can create excessive conservatism—operating far from boundaries is safer but may not be economically feasible. Measuring drift requires longitudinal data collection, which is costly and often not done until after an accident motivates it.

Go Deeper¶

Dekker, Drift into Failure, Chapters 6-7; Rasmussen, "Risk Management in a Dynamic Society: A Modelling Problem," Safety Science 27, 1997; Amalberti, "The Paradoxes of Almost Totally Safe Transportation Systems," Safety Science 37, 2001.