"The mathematical models have failed." Here we are again, a luminary in his field, but from whose CV he does not seem to have any specific competence in the "mathematical models" decided for all of us. Journalists, always in search of sensation, did not mind to check who issued this sentence. The Italian mathematical union (https://umi.dm.unibo.it/2020/06/10/comunicato-dellunione-matematica-italiana) intervened, using "mathematical models" in the same sense of the newspaper article, in a generic way, so many colleagues commented making the distinction between the different types of models ... And so on.
This whole story (which is not over) points the finger at the real problem: the relationship between science and society. This relationship has, as often happens in relationships, a communication problem. Those who do science by profession speak a given language, those who do not make professional science do not understand this language, feel diminished, or take offense barking out the wrong tree. Furthermore, both tend to put their mouths a little too incisively on the skills of others. Then maybe the mother-in-law gets into the picture, represented here for example by dr. Zangrillo or by dr. Silvestri, and the disaster is done. Divorces.
If there was a failure, it did not concern the forecast models used. Among those proposed, many caught exactly what was happening and had excellent predictive performances (we indulge in the praise of ours, but also those of many others (https://sites.google.com/community.unipa.it/covid- 19, http://www.sis-statistica.it/ita/20/index.php?p=9996, https://www.facebook.com/Robbayes-C19-111971610458527/)), what failed was the communication.
Too often those who make models have little perception of the context in which they find themselves talking or of the medium they use. On the contrary, the kings of newspapers or social media are rarely scientists or, if they are, they often make instrumental use of the scientific result (see the mother-in-law above). We also need care and attention in communicating with the decision-maker. For example, let's take the much-reviled report provided to the government with 40 and more scenarios reported. Aside from the considerations on the possible failure of the model, which is evident in our opinion: providing those who know nothing about predictive modeling with all those scenarios, without grouping them, for example noting that in reality many provided statistically equivalent forecasts, it is like not giving any information. On the contrary, it generates confusion and offers the side to the instrumental use of the result.
In conclusion: in science, it is normal to make mistakes, it is normal to change opinion, and it is also normal that some models fail. Despite this, it is essential that scientists learn to communicate with the decision-maker and with society, always emphasizing, as they know how to do well, the limits of their approaches. And it is essential that policy decisions will always be based on the best validated scientific evidence available.
