Nate Silver? Even after he flubbed it in 2016 and spent months arguing with other statisticians about which model they used and which one was "less wrong"? The problem with statisticians is that they are not scientists, and are not accustomed to challenging their data. Any social or physical scientist would have gone and checked the "pulse" of the data by talking to people (to realise that white people lie about race in public - on dating sites as well as in polls - they know what the "right" answers are), by looking for contradictory polls (plenty of those that showed Hurk not winning) and exploring the causes of difference, and built dynamics into their silly models to account for late behavior of both Hurk and voters.