Robustness > Optimization

The world is obsessed with optimization. There’s nothing wrong with improvement, efficiency, and effectiveness. But many times, optimization goes to far. As with anything, there’s a point where the costs of optimization exceed the benefits.

For example, corporations optimize their capital structure with substantial debt, but end up in bankruptcy when a recession hits. Investors optimize their portfolios through leverage and mean-variance analysis, but forget the inherent flaws embedded in the data and assumptions. Even individuals who optimize their health end up wasting time and money chasing useless biomarkers and other hacks.

At some point, optimization breaks down. Optimization often relies on the assumption of near perfect knowledge of the future. That’s why those optimized companies go bankrupt. They planned on one future but got another. Optimization is fragile. It’s great if the world unfolds just as you expected. It’s awful in every other scenario.

Since we don’t know the future, robustness, not optimization, should be the goal.

As Nassim Taleb states, “Layers of redundancy are the central risk management property of natural systems…redundancy is ambiguous because it seems like a waste if nothing unusual happens. Except that something unusual happens—usually.”

Accept the unpredictability of the future. It’s not about giving up. It’s about recognizing the limits of our understanding. Work with the world how it is, not how you want it to be.

Don’t worry about maximizing. Worry about surviving. Make sure you can live to see another day.

There’s a cost to robustness. Robustness means you won’t be as successful as others, at least in the short run. Someone out there will always be taking excessive risks, and you must be comfortable lagging behind them. But you will win in the long run, as you survive the tough times others won’t. There’s also a financial cost. For example, airplane manufacturers must pay for the double and triple redundancies they build into their aircraft. That work and material isn’t free. So why do they do it when they could get by on the cheap? Because no one would buy, or fly, those planes. Flying demands robustness, not optimization.

It’s the same reason why having a rainy-day fund is robust, but not optimal. Sure, you could live a little larger spending that money. Or investing it in higher risk assets. But by doing so you’re now at the mercy of an unknown future.

Investors understand this concept as margin of safety.

Warren Buffett explains:

Obviously, if you understood a business perfectly — the future of a business — you would need very little in the way of a margin of safety. So the more volatile the business is — or possibility is — but assuming you still want invest in it, the larger the margin of safety... Well, if you’re driving a truck across a bridge that holds — it says it holds 10,000 pounds — and you’ve got a 9,800 pound vehicle, you know, if the bridge is about six inches above the crevice that it covers, you may feel OK. But if it’s, you know, over the Grand Canyon, you may feel you want a little larger margin of safety, in terms of only driving a 4,000 pound truck, or something, across.

The problem isn’t understanding robustness. It’s the implementation. The incentives in the modern world push us to think only about maximizing today and assume the future will be just like the present. Robustness advocates are branded as afraid or alarmist. Optimization appears innovative and robustness appears boring. But I’d argue that ensuring that your portfolio or your company survives the unexpected is the most innovative thing you can do.