Skip to content

Robustness vs Optimality

Why It Matters

If you're uncertain about future conditions, robustness may be worth more than optimality. The question is how much expected performance to sacrifice for stability.

The Key Move

Ask: how confident am I in my model of future conditions? How bad is it if I'm wrong? If uncertainty is high or downside is severe, favor robustness. If you have good predictions and limited downside, optimize.

Classic Application

Engineering tolerances. Optimizing for exact load means failure at slight overload. Robust design handles a range of loads with safety margin.

Surprising Application

Career skills. Optimizing for your current job maximizes near-term performance. Building robust skills (general problem-solving, communication) performs worse in any specific job but survives job changes.

Failure Modes

  • Optimizing when you should be satisficing
  • Paying for robustness you don't need (over-engineering)
  • Confusing fragility with efficiency

Go Deeper

Taleb's "Antifragile." Robust optimization literature.