Last I've checked, this was an open forum, right?
I believe it's moderated - I'm not sure how strict the stance on off-topic posts is here. But as for me - I ask people politely using the keyword 'please' to return to the topic of the thread - which is math. That's all - doesn't really
seem to warrant invoking something like freedom of expression, does it now? .
How would the model appear, if you have hidden infections? Like for example, the pandemic spreading wildly in a small population, that does not get tested, either because this population is mostly ignored (east-european low-wage workers in Europe) or does not want to be tested (alternative medicine believers).
By definition the model has perfect knowledge of what happens, so we know simulated truth and there are no hidden infections. You'd have to add a testing model to post-process the data to see how it appears under different assumptions how perfect your tests are. When looking at reality, I'd make the assumption that testing catches (with fluctuations) a subsample of what is actually happening, and so allows to statistically infer the actual numbers. So I'd consider e.g. averaged growth rates more reliable than absolute numbers.
What about having a graph of population nodes, which have their own SIR parameters, and edges describing how such a node interacts with other nodes, like a reproduction number for one infected from node X infecting a number of persons in node Y?
I've actually planned to include the concept of 'domains' into the grid to see how that - qualitatively - changes the dynamics. Quantitatively... see below.
Could we have a pandemic of different speeds, not just different strategies between countries explaining the development?
If you're specifically interested in Coronavirus- yes, I believe we see different speeds in different countries - largely driven by the different social connectivity. People who live in a favela don't go on skiing holidays - their social interactions are restricted to their immediate neighbourhood, and that makes a difference in spread - which you can get in the model by setting up different social mobility.
But as in the previous case - the caveat is that (to my surprise) we do not really know much of any of these things in reality. I had expected that someone has assembled a model of different social groups in e.g. Germany - how often they go out, how many people they meet on average, how easy people from different social groups mix - that'd be highly relevant data for an accurate model for how a disease spreads in Germany. But such data does not seem to be there (or if so, it is not mentioned). Augmenting a simple grid model with the capacity to compute such effects in essence introduces a large number of new parameters - but that does nothing for realism if you end up guessing the parameter values.
So there's that.