Zero-Risk Bias, explained.
Zero-risk bias favors eliminating one risk completely over an alternative that prevents more comparable harm overall while leaving some risk in place.
Why it happens
A clean zero is emotionally and operationally attractive. But when costs and severity are equal, total expected harm may be the relevant comparison. Eliminating five expected incidents prevents fewer than reducing a different source by fifteen.
Zero-risk bias describes favoring the complete elimination of one risk over a larger overall reduction that leaves some risk remaining.
Read the result
Compare remaining totals after each choice: eliminating A leaves 50 expected incidents, while reducing B leaves 40. These are expected counts per period, not probabilities that should sum to one.
A worked example
Two reliability fixes
Source A contributes five expected incidents and source B contributes fifty, all with the same assumed cost per incident.
One equal-cost fix removes A; another prevents fifteen incidents from B.
The second prevents more incidents overall even though neither source reaches zero. Different severities could change the preferred allocation.
OPTIONAL DEEPER DETAILGo deeper: inside the model
Inside this model
Two fictional sources produce expected counts of 5 and 50 comparable incidents per period. Option A eliminates the first; option B reduces the second by 15. Equal cost and equal incident severity are explicit assumptions. These are expected counts, not probabilities of mutually exclusive events.
Where this idea is useful
A practical use
A reliability team can compare expected prevented incidents across equal-cost fixes instead of treating a component's zero-error target as the only objective.
A common misconception
“The option with zero risk in one component is safest overall.”
One component's zero can coexist with a larger remaining total elsewhere. Define the overall objective and compare comparable consequences.
What this explanation leaves out
- Unequal severity, duties, dependencies and uncertainty can justify a different choice. The game compares expected incident totals only and is not a real safety assessment.
Could complete elimination still be the right choice?
Yes. Particular duties, severe consequences, dependencies or uncertainty may favor it. The simplified game deliberately holds those factors aside.
Are you minimizing total harm or pursuing a zero in the most visible category?
Further reading
Explore the original research or the teaching reference behind this experiment.