The Case of the Single COVID
A stress test of COVID decision making
We are collectively faced with a wicked problem — how to respond to the COVID-19 pandemic. The main decision point right now is whether or not to put a community into lockdown. This decision is being made mostly by using models that predict what will happen and how likely it is.
Right now this is emergency decision making, but soon we will want to introspect — are we making the best possible decisions? To answer that, we have to examine our process of decision making itself.
To do this, we can use limit thinking — thinking about a hypothetical situation that is at one extreme of the spectrum of possibilities. Although unrealistic, these limit cases are extremely valuable because they stress test our decision making — i.e. they force us to confront the assumptions behind our decisions and the real implications of what we are deciding.
So the question is — would you put a community (country) into lockdown if there was exactly one COVID case within it (N=1)?
(Of course , we are assuming there that we are making this decision with the objective of minimizing harm to the entire community.)
There are only three possible answers to this question— you either lock down, you don’t lock down, or you do a partial lockdown. For each of these possible decisions, instead of predicting outcomes we sketch out the best case and worst case scenarios. We also project forward in time in the limit case that COVID is always present in the community (N > 0) forever.
Under each of these limit scenarios, we can then exactly specify the magnitude of harm to the community. Note that we are not predicting — magnitude of harm is not a calculation. We do not have to predict the exact amount of harm in each case. We are simply looking at whether the harm is zero, maximum, finite or infinite each case.
Full lockdown
In this case, the harm through COVID is reduced to zero. However, we maximize the harm to the community through lockdown (loss of freedom, decreases in wellbeing, increase in disease, deaths by crime, suicide etc.). This non-COVID harm is guaranteed under both best- and worst-case scenarios.
Integrating forward in time, this non-COVID harm is unbounded i.e. infinite, and harm from COVID is always zero. However, note that total harm from COVID is bounded even in infinite time because the virus cannot spread once it has infected 100% of the population. So in this decision path, we are inflicting guaranteed, infinite/unbounded harm overall.
Partial lockdown
In this case, in the worst case scenario the entire population is infected from the initial one case and harm from COVID is maximum. Also, the partial lockdown inflicts partial harm on the community (less than maximum). In the best case, the community escapes harm from COVID but still has partial harm from the lockdown.
Integrating forward in time, the community suffers non-COVID harm that is unbounded i.e. infinite. This is guaranteed. The harm from COVID is either zero or finite, so finite overall. So again in this decision path, we are inflicting guaranteed, infinite/unbounded harm overall.
No lockdown
In this case, in the worst case scenario again the entire population is infected and harm from COVID is maximum. But there is no non-COVID harm. In the best case, the community escapes harm from COVID as well and does not suffer at all.
Integrating forward in time, there is no harm from non-COVID lockdown measures. The harm from COVID is either zero or finite, so finite overall. In this decision path, we are inflicting a finite/bounded harm overall and it is not guaranteed i.e. it is possible there is no harm at all.
The Key Question
Why do all of this? Well, soon we’ll approach our version of this scenario once the virus is sufficiently suppressed— what do when there is a very small but non-zero risk of COVID-91 infection growing in the community?
In fact, all of the emergency decisions being taken right now are trying to get us to that point. That is the real wicked problem that lies ahead.
This stress test gives us a system for our decision making at that point. It doesn’t tell us which decision we should make, but it does suggest to us the one critical question that we need to ask ourselves. Given that we have two scenarios in which harm increases unboundedly in proportion to time, the question is— how long are we willing to stick with our decision?*
The longer we are willing to stick with it, the worse is the case for a full lockdown, as the total harm for that scenario is unbounded and will keep growing. The same applies for partial lockdown. Otherwise we will be essentially be inflicting growing (unbounded) harm on the community to save finite harm.
If we are wanting a long-term solution without a time limit, we will have to move away from lockdown. It is the only solution that will cap the total amount of harm to the community to a finite amount. If we decided to do that, the immediate corollary is to prioritize the develop of treatments above all — to bring that finite harm from COVID down to as low a level as possible.
*You might say that we are not able to answer that question because we don’t know when COVID will be vaccinated against. But the question is about intent under uncertainty, not relying on prediction.
