Root Cause Analysis and the Five Whys
Goal: find the underlying (root) cause of an event, not the proximate one — then redesign so it can't recur or has minimal impact.
The Five Whys (Sakichi Toyoda / Toyota Production System)
When you find a reason, don't stop — ask why that was the case, and again, until you reach the true underlying cause. "Five" is symbolic; it emphasizes not stopping early.
Example (F-22 crash):
- Q1 Why did the plane crash? → uncontrolled dive.
- Q2 Why no recovery? → pilot failed to initiate timely recovery.
- Q3 Why? → he may have been unconscious / oxygen-deprived.
- Q4 Why? → unknown — must investigate. (See "Cases: Error & Accident Investigations".)
Two failure modes of root cause analysis
- Stops as soon as a human error is found. When a machine breaks we keep asking why the part failed — we should do the same for human error. (The electric-utility electrocutions: committees blamed the workers and stopped; never asked why; "no changes… still injuring people.")
- Assumes a single cause. Most accidents have multiple causes — the Swiss cheese model (James Reason): an accident happens only when holes in all slices line up. So "the" cause is usually a myth; investigations rightly take a long time. See Designing for Error for the Swiss-cheese remedies.
Same technique for design (ch.6)
Finding the right problem = root cause analysis applied to design (the drill→hole→bookshelf chain). Drives the first diamond and radical innovation.
Caveats
"Why" is ambiguous; investigators still tend to stop at the limit of their understanding and to over-seek a single cause. Powerful but not foolproof.
Confidence
1.0 — explicit procedure.
Sources: ["DOET ch.5 'Root Cause Analysis' / 'The Five Whys' p.163-167; ch.2 root cause"]