The short note published with @HeikoEnderling is not a conclusion of a discussion, but a starting point. The feedback is helping us to shape our debate. 1/10

There is something complex, which we do not understand well; we arrogantly assume it to be simpler than it is; to then make predictions or gain understanding of it. A fascinating interplay of getting it wrong to get it right. 2/10

The main distinction to discuss models, it seems, is whether its purpose is prediction (now-casting, or forecasting), or whether the goal is to understand something, in order to explain it better: modelling as thought experiments. 3/10

Once you no longer change the structure and parameter values of your model, you start using it. What is then right or wrong about that model, also depends on the story of getting it right and wrong in the process of its creation. 4/10

The discussion about models, their quality, purpose, and value should therefore focus on the process of modeling. The problem is that we mostly focus on documenting the outcome, not how we got there. 5/10

We often expect reproducibility and numerical accuracy of a model, when actually we would benefit more from recapitulating the modeling, the process of constructing a model. 6/10

Epidemiological modeling illustrates the spectrum of purposes for modeling. A few equations suffice to explain the urgency of lockdown measures to avoid exponential growth. 7/10

Add a few more equations, and the model can help understand the consequences of alternative testing, lockdown and vaccination scenarios. Adding more equations, or choosing a statistical model, you can try to predict the number of required ICU beds. 8/10

For predictions, our goal for modeling is to get it right, while for understanding you benefit most from getting it wrong. A model, as a set of equations, is hiding this story. 9/10

For predictive modeling, we claim to know the variables that make it "right", while in modeling for understanding, not knowing the variables and how they interact, getting it wrong, is key to eventually explain a phenomenon. 10/10