Design decisions so they work even if some things go wrong

Ask: does this plan require everything to go right? If so, redesign it.

Why it works

Complex plans with multiple sequential dependencies are highly sensitive to single-point failures — when any step goes wrong, the whole plan fails. Redesigning so that the acceptable outcome is achievable even if some components fail is the structural form of margin of safety. This is distinct from planning for failure; it is designing for robustness so failure of parts does not equal failure of the whole.

How to do it

1. Map the dependencies in your plan: which steps must succeed for the outcome to work?
2. Identify single points of failure — any step where one bad outcome ruins everything.
3. Redesign to create fallbacks or redundancies at single-point failures.
4. Accept a slightly lower expected upside in exchange for a substantially higher floor.

Evidence

Engineering reliability theory formalizes this: series systems fail when any component fails, while parallel/redundant systems tolerate component failure. The principle that reducing single points of failure increases system reliability is mathematically well established. (mechanistic)

Redundancy adds cost and complexity; the optimal level of redundancy depends on the probability and cost of failure, which are often hard to estimate in non-engineering contexts.

Common mistake

Adding a plan B that is only reachable if plan A partially succeeds — true robustness means the fallback works even when the primary path fails completely.