Layered Defense & Safety Erosion

Primes about how multi-layer safety systems are built, through bulkheads, defense in depth, and the Swiss cheese model, and how they quietly erode through normalization of deviance, exposure creep, benign-sampling drift, and inherited or residual risk.

9 primes in this family — primes that sit near one another in abstraction space (k-means over structural-signature embeddings). Each is shown with its short description.

• Benign-Sampling Safety Drift — A system near a hazard boundary reads a benign sample of outcomes — uneventful only because the hazard is rare, not because the margin is safe — as proof the safety margin is unnecessary, so each lucky round licenses the next erosion until a low-probability event finds the now-absent buffer.
• Bulkhead Pattern — Partition a shared critical resource into sibling compartments so one compartment's failure stays local instead of draining the whole.
• Defense In Depth — Stacking multiple independent protective layers between threat and asset so that only a correlated breach across all layers produces total loss.
• Expectation Violation — An active prediction or schema specifies what should occur; an incompatible realization crosses a mismatch threshold and triggers surprise, reorientation, or repair, making the violated expectation diagnostically visible.
• Exposure Creep — Between rare severe shocks, valuable assets accumulate inside a known hazard's impact zone because a quiet history reads as safety, so the next shock's loss is driven by the grown stake rather than by any change in the hazard.
• Inherited-Substrate Risk — A system built on a borrowed or inherited substrate carries forward latent conditions of the substrate's origin through an inheritance channel that the new system's audit boundary does not cross, so the risk concentrates exactly where attention does not.
• Loss And Damage — The residual harm that passes through after every layer of defense, treated as its own accounting unit and assigned a bearer and a policy.
• Normalization of Deviance — An operating standard drifts outward because small departures, each repeatedly observed without immediate catastrophe, are reclassified as normal.
• Swiss Cheese Model (Layered Defense with Aligning Holes) — Catastrophe occurs only when a hazard finds a trajectory through a hole in every serial defensive layer at once, so the key variable is the correlation of holes across layers.