Intelligent Pawns for Toy Worlds

  • output

Mr. Sum had let Mr. Node do his thing with COVID-19 data to find his feet with the pandemic. Because he understands that he is part of a complex adaptive system, Node has started to work with an instrument that is prepared for this. On November 28, his NetLogo program presents the following chart:

Fig. 1 Tableau for NLD as of November 24 with COVID initially considered 13% lethal

The black line shows the absolute number of new COVID-19 deaths per week for the Netherlands. The red line shows the number of registered infections, not in absolute numbers, but in a linearly adjusted form. In fig. 1, the red line shows 13% of the absolute numbers of registered infections. A vertical is drawn once every five weeks. The peaks of the first wave fall around week 15.

In a new pandemic like COVID-19, the first question is about how bad it is, or how deadly an infection is expected to be. COVID deaths follow infections. Some of those infections are deadly. A naive way of seeing how bad it is is to fit the curves together and see at what percentage of the infections they coincide. When this is attempted, it is clear that overlapping will succeed in the first 25 weeks of the pandemic. The pandemic then looks exceptionally dangerous: the red line fits on the black, when it has been reduced to 13% of itself.

But from week 25 this will no longer be the case. The question why not is an important one. Maybe it helps to try and see if another percentage makes the red line coincide with the black in the second half of the figure. That has been tried in Fig. 2. Actually it works fine, shaping the red line in such a way that the black one echoes it, with a lag interval of about 17 days. Now the percentage of registered infections leading to a death has been reduced to less than 1%, namely 0.875%.

Fig. 2 Tableau for NLD as of November 24 with COVID ultimately as 0.875% lethal

That’s quite a difference, from 13% to less than 1% lethality. Explanations for this have already been discussed. Most importantly, in order to obtain reliable observational lethality rates, infection numbers must be complete, accurate and reliable. And they are not at all, and certainly not at the start of a pandemic by an unknown virus.

What both figures show with certainty is that it is possible to influence the curve. Laws, norms, knowledge, morality, support and (social and economic) costs play a role. For the Netherlands, the curves show for the time being that the first wave has been collectively eliminated, and that although the second has been reversed, it is unclear whether this reversal will continue. For now it seems that the decline has come to a halt. It is then worth looking at whether these types of curves show different patterns in the rest of the world.

Those patterns are then the result of the behavior of citizens in jurisdictions. Mr. Node and mr. Sum can develop hypotheticals about such processes by looking around them. And they can play them out by populating their toy world with intelligent pawns end watch these play.

Intelligent pawns are very often referred to as “intelligent agents.” Node and Sum don’t like that for its human-responsibility eroding suggestion.

Computer says no?