Multiple Necessary Conditions¶

Why It Matters¶

Monocausal explanation is psychologically irresistible but systematically misleading. Finding one cause feels like completion—we stop searching, implement one fix, and remain vulnerable. This tool forces you to keep asking "what else had to be true for this to happen?" It prevents premature closure and reveals the full opportunity space for intervention.

The Key Move¶

When analyzing any failure, create a necessity test: "If we removed just this factor but kept everything else, would the failure still occur?" If yes, you haven't found a sufficient cause—keep searching for other necessary conditions. List all factors that pass this test. Only when you have multiple factors, each individually necessary, have you captured the causal structure. Force yourself to identify at least 3-5 necessary conditions before considering your analysis complete.

Classic Application¶

The Tenerife airport disaster (1977)—583 deaths from two 747s colliding on a foggy runway. Investigation revealed simultaneous necessary conditions: fog reducing visibility, radio communication interference, KLM captain's rush due to duty-time limits, ambiguous phraseology in pilot-controller exchange, airport congestion from a bomb threat diversion. Remove any single factor and the collision doesn't happen. This recognition transformed aviation safety from "pilot error" attribution to systematic barrier design.

Surprising Application¶

Medical diagnostic error. A patient misdiagnosis typically requires: time pressure limiting examination, cognitive anchoring on an initial hypothesis, missing information in the medical history, system design making test results hard to access, and fatigue affecting clinical judgment. Treating it as "doctor error" misses that fixing any one systemic factor prevents the error. Emergency departments now explicitly check for necessary condition alignment before attributing error to individuals.

Failure Modes¶

Explodes into infinite regress if you don't bound the analysis timeframe and scope. Everything is necessary if you go back far enough (Big Bang caused this accident). Paralyzes action if interpreted as "we must fix everything"—the goal is to identify multiple intervention points, not to require fixing all of them. Can become an excuse mechanism ("so many things went wrong, no one is responsible") if you don't distinguish necessary conditions from accountability.

Go Deeper¶

Reason, Managing the Risks of Organizational Accidents, Chapter 9; Dekker, The Field Guide to Understanding 'Human Error', Chapter 4; Perrow, Normal Accidents, pp. 62-100.

Anchoring - Behavioral Economics: Reasoning About Systematic Deviations