Playing with their model as linked gives a much better match with reported numbers with a shorter serial interval, 5 or 6 days instead of the 7 assumed by the d

Playing with their model as linked gives a much better match with reported numbers with a shorter serial interval, 5 or 6 days instead of the 7 assumed by the default model settings. This is also a better match with the more recently reported case studies. I find it a little irresponsible to make such an extreme claim from such a limited modelling tool... Small changes to the assumptions can account for the discrepancy seen, and it's unrealistic to expect a basic model like this to accurately represent reality in such an unusual situation as in Wuhan.