top of page

When wrong is always ready

Let's for a minute imagine that we are engineers and we are asked to build a wind farm in the North Sea. Let's also imagine that we are not experienced in this task, in fact, we have never built a wind farm in an ocean before. We need a strategy. One approach would be to build a couple of windmills, fasten them up to the grid and see how we get on. We know that there will be some problems but we will learn from these and this will help us as we roll out the next windmills. Sure enough, we hit problems. Our windmills collapse shortly after being built. However, we are ready to react and we change the cement, the depth of the foundations, and the size of the wind vanes. Hmm, this doesn’t work either so we try something else. You get the idea. Now let's imagine that we don’t work like this. We are engineers and so we spend time designing the windmills and, because we are uncertain about the constraints, we over engineer the whole thing. To be safe, we over engineer by quite a lot, possibly by 15 to 20% because we think it would be dangerous to fine-tune it with a lower level of overkill. Eureka! Just like all the stuff the Victorians built, our windmills work perfectly and they keep on going for year after year.

What I have been talking about here are two different frameworks. One is "constrained optimisation" and the other is a "robustness framework". The constrained optimisation framework is “let’s just build and see how we get on”. You strive for the best outcome while taking account of constraints – in the case of the windmills those constraints could be the materials we use to build them, the weather, the speed with which we have to complete the project and so on. We balance those (possibly changing) constraints against the optimal policy of getting our wind farm up and running as quickly as possible. The robustness framework is the over engineering approach that anticipates there will be severe risks so we build it like a brick sh*t house.

A constrained optimisation framework works when the risks are small. For example, imagine you decide to market your product to a new target audience and do so by "suck it and see". You could experiment and learn as you go and though mistakes will be made, it will be a learning process that eventually enables you to succeed. A constrained optimisation framework is fine when there is not much at risk - but it is dangerous if you are betting the farm.

There has been a huge amount at risk with Covid. And yet, most governments of the Western world have adopted a constrained optimisation framework, reacting as constraints change and failing to implement an optimal policy either for health or economics. A "robustness framework" would have involved a serious lockdown, one that lasted longer than most people thought necessary. It would have been politically difficult to apply but because it was over engineered it would have worked.

Sometimes in business, as in politics, luck is with us and sometimes it is not. So far luck has been against the politicians, certainly those in the libertarian West with their constrained optimisation policies. They must thank their lucky stars that a significant change is on the horizon. Vaccines have arrived. This hasn’t stopped governments pursuing constrained optimisation strategies. In the UK the second doses of the jabs have been delayed in order to inoculate more people more quickly. I'm not knocking this decision, it seems right to me, but I can only hope that luck stays with us. After all, there is no evidence of the efficacy of a single dose and if we find this doesn’t work we will have insufficient immunity in the herd. That's the thing about the constrained optimisation framework, you have to be strong on hope, you have to be optimistic, and you need nerves of steel because, guess what, wrong is always ready.


bottom of page