Meanwhile, today is May 12 – I woke up with a cold this morning (but not yet “panicking together,” [incidentally an ideal titel], as Welmoed would say) – I am a long way further with the program, which is developing towards an instrument with which theories can be reenacted and adapted to what reality has been happening (calibration). The thing is already far too complicated to fully show in a single blog post. In short, this is going to be a whole series.
Fig 1 shows an interim result of an application: developing and modeling a theory about how the COVID-19 pandemic develops globally and what the influence of us humans is / can be. If we can and want to have influence, the results of this model must be convincing and reliable, especially if it involves political decision-making. A good reason to avoid higher mathematics.
What the graph in Fig 1 shows is the development of three things over time (one week periods). The red line (MOD-I) shows the results of the model for a first infection on December 29, 2019. The black line (JUR-I) shows the registered COVID-19 infections for a jurisdiction (in this case the world). The gray line (JUR-D) indicates the number of COVID-19 deaths in that jurisdiction. The numbers are cumulative. A horizontal line indicates that there are no more infections or deaths.
What can also be seen is that the model and reality move together for a while and then separate: the model in its current form can only be used in the run-up to the pandemic. The figures show that the moment when the ways of the model and reality separate comes around day 98, which is the fifteenth week of the pandemic. Why that is so must be seen. But for using the platform (I’m going to call that Epiframer) there is one thing clear. The algorithm 0 model is useful only in the early stages of pandemic.
Still quite long, 98 days. About three months.