CoMSES Net Discourse Forums

Beware of interpreting unvalidated models

Whilst playing with models is always good, please be careful about drawing any conclusions from these, unless and until it is empirically validated. I have seen a rash of corona virus models which are little more than playthings or computational analogies, but where the authors then suggest real-life conclusions from them! We have a responsibility to be cautious in what models can tell us.

1 Like

Countries are basing national policy on models that are not widely vetted. The US has taken the Imperial College study as the basis for new policy recommendations. It may be a very good model. But the report does not list anywhere someone can download and see the code. If someone knows that the code is indeed posted, it would be good to make everyone aware of it.

Chen Shen, Nassim Taleb, et al posted a critique of this model that I’ve also added to our curated resources wiki:

https://www.academia.edu/42242357/Review_of_Ferguson_et_al_Impact_of_non-pharmaceutical_interventions…_