On the aftermath of the US presidential election, observers were shocked by the results they apparently did not saw coming. Nate Silver, the widely known American statistician, immediately wrote an op-ed for his website fivethirtyeight.com where he explains what may have gone wrong with their predictions – based on the accuracy of polling averages in presidential elections dating back to 1972 – normally renowned for being very accurate.
Besides political elections forecasts, the article is interesting in general for Data Scientists because it stresses the importance of being able to comunicate correctly was stats can and cannot do. Nate points out that, at some point, you have to trust the data, even if common sense may try to bring you on another playground.